Meisel1966a PDF
Meisel1966a PDF
Meisel1966a PDF
- L
)
...:.I
PRINCIPLES . OF
ELECTROMECHANICAL·
ENERGY CONVERSION
CONTENTS
l
preface ix
Lumped-element Relationships 23
State Functions 24
Functions 32
'probkms 44
Lagrange's Equation 48
state Functions 50
xiii
2-11 Summary 91
Problems 92
System 106
System 113
Functions 133
Forms 139
Problemil 142
Operator 156
Locations 202
------_..._
Contents
Problems 210
Transducer 238
Problems 258
Parameters 283
Parameters 294
Problems 306
Problems 330
xvi Contents
ProblemB 376
Problems 425
Distributions 436
ProblemIJ 463
Content. It
r.:
Chapter 11. The Synchronous Machine 467
~
11-5 Sinor Diagram of the Synchronous Machine 485
Problems 530
Parameters 569
Problems 612
Bibliography 623
Answers to Problems
627
Index 633
which states that the current through the capacitance element is propor
tional to the time rate of change of the voltage across the element. For a
given element this proportionality constant ha.s units of farads if the cur
rent i is in amperes, the potential rise v is in volts, and the time t is in
seconds. These are all consistent mks unit's. Equation (1-1) can be
Coordinate., element., and Itate func:tioM 3
c
Fig. 1-1 The IUII'\ped-eapa.citanee symbol.
The current function i(t) is defined as the time rate of flow of positive
charge, or
Taking q( - co) equal to zero, since our element was not even constructed
at that time, Eq. (1-5) gives the integral in Eq. (1-3) as being some
quantity of charge at time t. Physically, this charge q(t} is found to be
accumulated on the capacitor plates, with positive charge on the top plate
in Fig. 1-1 and a negative charge of equal magnitude on the bottom plate
for positive i(t). The current is defined as the rate of flow of positive
4 Principle, of electromechanical-energy canverr;on
which says that the voltage across the capacitor is linearly related, or
simply proportional, to the charge on the capacitor plates. From a physi
cal standpoint the capacitance element has the ability to accumulate
charge, and in the process a voltage or potential difference appears across
its terminals. The charge appears as a fundamental coordinate in this
mechanism, and it is chosen as the first of our basic coordinat~. The
first time derivative of the charge, dq(t)/dt, or more simply, q(t) , is the
current through the capacitor, namely, i(t). We have taken i11e charge
as the basic coordinate, with its first time -derivative, current, being one of
the two common circuital variables. In succeeding discussions involving
the inductance parameter the second fundamental electrical coordinate,
called flux linkage, is introduced. The first time derivative of the flux
linkages turns out to be voltage. Thus the usual network variables, cur
rent and voltage, are to be obtained as the first time derivatives of our
basic coordinates charge and flux linkage.
Equation (1-6) gives a general experimental scheme for describing
this capacitance element. For a physical device havi'ng the ability to
accumulate charge, an expression relating the voltage across the termi
nals of the device to the charge it accumulates is required. If the charge
is plotted as a function of the voltage, then the capacitanceC(v) is the
slope of this curve at any point, as shown in Fig. 1-2. If the plot were a
straight line, then the capacitance parameter would be constant and we
would then have a linear capacitor. However, if the plot is not a straight
line, as shown in Fig. 1-2, then the slope changes as a function of the
charge, or the voltage, and the element is nonlinear. "Linea.r" or IInon
linear" simply means the charge vs. voltage plot is straight or curved,
respectively. We shall call a curve such as Fig. 1-2 the defining-charac
Coordinate., element., and "ote function. S
teristic curve for the lumped-capacitance element, since this curve com
pletely defines the element.
All of the characteristic curves will be restricted to single-valued func
tions. Only one value of charge q in Fig:i-2 corresponds to a 'specific
terminal voltage v. Certain dielectric materials exhibit a double-valued
characteristic as sfiown in Fig. 1-3. The value of q for any specified value
of v depends on the past history of the voltage v. -This behavior whereby
1!.he value of q 'for increasing v does not correspond to the value qf q for
decreasing v is termed hysteresis. A large number of hysteresis loops, in
the form of Fig. 1-3, are actually needed to describe the device completely.
As an approximation for such characteristics a single-valued curve is
drawn through the centers of the hysteresis loops. The energy loss ass0
ciated with cycling around the hysteresis loop is then taken into aCcount
by a lumped-resistance element.
I
Example 1-1
The characteristic curve for a parallel-plate capacitor with a Rochelle-ealt dielectric
has the form shown in Fig. 1-4. The lack of symmetry results from dielectric polari
zation giving rise to reSidual charge analogous to residual magnetism for ferromagnetic
materials. For voltage values between v. and I'" the characteristic curve can be
A spherical capacitor.
C(v) - dq(,')
do
However, the given expression has vasa function of q; thus the capacitance is found
more conveniently from
1 dv(q)
C(q) - 1 + Aq + Bq'
Example J·2
Let us now determine the capacitance of a physical configuration consisting of two
perfectly conducting concentric spheres. The radius of the inner sphere is R and
that of the outer sphere, R.. The space between the two spheres is half filled" with
polystyrene having a relative dielectric constant of 2.7. The remainder of the space is
filled with air, &8 shown in Fig. 1-5.
Solution: Let us begin by connecting two leads to the inner and outer spheres, &8
shown in Fig. 1·5. Terminal variables q and v are arbitrarily assigned to these two
leads, also &8 shown in the figure. The only restriction on the assignment of these
terminal variables is that energy flows into the device if the two variables have like
signs. These reference conditions correspond to the original conditions in Fig. 1-1.
Coordinat••, .I.ment., and .fat. function. 7
Now if we construct a. concentric spherica.l Gaussi80n surface between the two spheres,
the electric displacement vector isl
D_.-L ..
4orT.
where q is the eh80rge on the inner sphere, with an equal and opposite charge existing
on the outer sphere, r is the arbitrary radius of the Gaussian surface, and a, is the unit
ra.dial vector.
The respective electric fields in the polystyrene and air are given by
D q
Es,017. - - - -
2.7'0 4'1'1'1(2.7'0)
a.
and
The potential of the inner sphere with respect to the outer sphere is designated by v in
Fig. 1-5. By definition, this potential difference is given as
v - - !cE'dl
where for the contour C we shall choose a radial path from the outer sphere to the inner
sphere, thus making
dl - -dr a,
On solving for v by substituting the electric fields into the contour integral, we have
q ( ~ (R.+ R.)f2 --+
v--
4....
R.
dr
2.7r'
f R.
(R.+R.)f2
dr)
r'
which reduces to
q [ 1 + 3.4 1]
v - 4.... 2.7R. 2.7(R. + R.) - R.
The capacitance, ILl! seen at the two terminals, is given by the slope of the q versus v
characteristic curve. For fixed values of R. and R. this characteristic is a stra.ight
line, and therefore
c_ 4..••
farads
1/2.7R. + 3.4/2.7(R. + R.) - l/R.
By using fa - 8.854 X lO-u'farad/m, and by expressing R. and R. in meters, we
would have the capacitance C in farads.
eind = - ~ /s B . dS (1-9)
(1-11)
Fig. 1-7
law.
Illustration of Faraday's
-----2~~J i
surface integrals taken for each turn will be involved in Eq. (1-11). The
total of these surface integrals represents the total flux linkages of the
entire multiturn coil. Therefore, symbolically
Mt) - X( - 00) (1-12)
where the flux linkages X have units of webers. The unit weber-turns is
also used; but the two are really equivalent, since turns are dimensionless.
The flux linkages in the coil at time - 00 is taken to be zero~ thus
Eq. (1-8) becomes
upon substituting Eq. (1-11) and then Eq. (1-12) into Eq. (1-8). For the
linear-inductaQce parameter the flux linkages are proportional to the
applied current. An inductance has the ability to accumulate flux link
ages, and in the process a current must somewhere exist.! A description
of this element is given by plotting the flux linkages as a function of the
current setting up this linking field. The defining-characteristic curve
for the lumped-inductance element is shown in Fig. 1-8. The slope of
this curve at any point is L(i), the inductance. As before with the capaci
tance element, a straight-line plot of X versus i gives a constant value for
L and the element is linear, whereas a curved plot of X versus i makes L a
function of either Xor i and the element is nonlinear. "Linear" or "non
linear" simply states that the X versus i plot is straight or curved, respec
tively. The fundamental coordinate is that of flux linkage. The first
1 Notice that the statement does not say that this current must exist in the same
inductance element. In our work with electrical circuits, the mutual-inductance
parameter is formulated from flux linkages in one element due to current in another
element.
10 Principl•• af elecfromecmlnieal•.,..,.gy conver.ion
Example 1·3
Figure 1-9 shows a cross-sectional view of a cylindrical iron frame. On the round
center post of the frame a coil of N turns is wound. The diameter of this cylindrical
center post is d meters, and the height is I meters, as shown in Fig. 1-9. A88uming
the magnetic-intensity vector to be very nearly zero everywhere except in the center
post, determine the inductance of the winding. Figure 1-10 gives a B-B hysteresis
loop for the iron.
Solutwn: The inductance can easily be determined once we have the proper
characteristic curve for the device. This involves a. plot of the coil flux linkages as a
function of the coil current.
Ampere'" circuital law is very helpful in determining magnetic fields in this type
of system. This basic law, in integral form, is given by
¢cH • dl - Is J . dS
where H is the magnetic-intensity vector, in ampere-turns per meter, and J is vector
current density, in amperes per square meter. The law tells us that the line integral
of the magnetic-intensity vector around any closed contour is equal to the surface
integral of the vector-eurrent density taken over any surface whose boundaries are the
- d
r-- .....---,
I I
+
~ lax
,..-Contour
I - Ii I
I
I
I
I :{ ~
~
N turns
I
•I
L ____ - - J+
.0--
Third-order opproximotion
Fig. 1-10 Normal magnetization curve and hysteresis loop for the iron core. A
third-order approximation to the magnetization curve is also shown.
chosen contour. Once a direction has been chosen for walking around the contour,
thus defining the vector dl, the incremental-surface vector dS has its direction estab
lished by the right-hand rule as follows: If the fingers of the right hand are curled in
the direction of the contour, then the thumb of the right hand defines the direction
for dS. This is a very important physical law, and the reader should spend sufficient
time to review itS meaning.
By applying Amp~re's circuital law to the contour shown in Fig. 1-9, we have
The dashed curve in Fig. 1-10 terminates this series after the cubic term. On sub
stituting the third-order approximation for H in Eq. (1-14), we have
(l-U5)
12 Pr;rK;;p'e. of eredromechoni«l'-energy connrs;on
By fonning the surface integral of B over each of the N turns, the total flux
linkages can be expressed &8
N
where the symbolism S. is used to mean the surface over the kth turn. The vector
direction of each dS" must be chosen to correspond to the assumed current direction in
the kth turn and the right-hand rule. Thus, referring to Fig. 1-9, wrapping the
fingers of the right hand around the ooil in the direction of the current, the thumb
points up; hence
dS. - -dS"..
Now since B is constant in the center P08t and therefore is constant over the area of
each t1frn, the total flux linkages are given by
). __ NB....d"
4
where ...d"/4 is the area of the center post. Multiplying this area by B. gives the total
flux in the center post linking all N turn8, so by multiplying by N we have the total
flux linkages. Upon solving for B" in terms of ). and substituting into Eq. (1-15), we
have
(1-16)
where
and
By using Eq. (1-16), the )...q characteristic curve can be drawn. The incremental
inductance or slope of the characteristic curve is given by
henrys
In this example the inductance is expressed &8 a nonlinear function of the flux linkages
rather than the current. In theory, Eq. (1-16) oould be solved for). &8 a function of q
and the inductance could then be expressed &8 a function of q.
Example 1-4
U8ing the magnetic a.ssembly shown in Fig. 1-9, con8ider cutting an air gap of g meters
in the center post &8 shown in Fig. 1-11. Con8idering the permeability of the iron to be
much greater than that of air, and neglecting fringing fields at the air gap, now deter
mine the inductanee of the winding.
Solu.tion: Since the penneability of the iron is many orders of magnitude greater
than that of air, it is reasonable to assume that the magnetie-vector intensity in all
of the iron, includip.g the center post, is very nearly zero compared with that in the air
gap. Again using Amp~e's eireuital law with the contour shown in Fig. 1-11, we
have simply
H"g - -Nq
Ir,:~,::::"
..
I
t Contour
I'
r
-q
f
I
where H. is the component of the magnetic-intensity vector in the air gap in the direc
tion of the unit vector a.. Thus, the magnetic-intenaity vector in the air gap is ,iven
by
H - - !!ja.
9
and for the vector-magnetic-flux density we have
B - "oil - - "oNq a.
9
where ". - 4r X 10-' hIm is the permeability of air.
Since fringing fields in the air gap are being neglected, the magnetic-f1ux-ciensity
vector in the central iron cylinder must be equal to that in the air gap. By forming
the surface integral of B over each of the N turns, the total flux linkages can be deter
mined to be
N
A- 2: Is. B•. dS. - "oNl(~/4) q
I-I '.\
The plot of Aversus q is now a straight line with constant slope. The inductance,
which is the slope of this ).. vel'8us q plot, is therefore
L _ ~ _ (rNd)1 X 10-' henrys
dq 9
upon substitution of the value for ,... Notice that cutting a small air gap in the iron
path results in a linear inductance eliminating the nonlinear characteristics of the iron.
The magnitude of the inductance is greatly reduced by the air gap.
systems dissipate some energy per unit time in the form of heat. The
term "dissipate" is used because the heat energy is usually lost by the
Ielectrical system to its physical surroundings. A third element is required
to account for this energy dissipation. In a conducting medium, the
current~density vector J is related to the electric-field vector E by 'an
expression of the form
J = O'E (1-17)
In this equation, 0' is the conductivity of the medium, in mhos per meter.
:For a linear medium, 0' is constant; but, in general, q may be a function
of either J or E.
Figure 1-12 shows an incremental tube of the conducting medium of
cross-sectional area I:J.S and length I:J.l. The tube is so selected that J is
normal to the surface I:J.S and is relatively constant over the surface.
Also, E does not change over the length I:J.l. The two ends of the tube are
equipotential surfaces, and any current entering one end exits through
the other end. No current Bows through the side walls. The surface
vector I:J.S is collinear with J and E; so by forming the scalar product of
I:J.S with Eq. (1~17), we have
after multiplying and dividing by I:J.l. The quantity J I:J.S is the total cur
rent in the incremental tube, and E I:J.l is the potential difference from one
end of the tube to the other end.
The incremental tube can be modeled by the lumped-resistance el~
ment whose symbol is shown in Fig. 1-13a. By using q for current and >.
for potential difference, Eq. (1-18) can be written as
q = G(>')>' (1-19)
where the admittance G = q I:J.S/l:J.l has units of mhos and I/G = R is the
resistance with units of ohms. The defining-characteristic curve for the
resistance or admittance element, as shown in Fig. I-13b, is a plot of q as a
Coordinot••, ./ement., and .tate function. 15
1
i
Gmhos Slope = NIH =GO,)•
Rohms
(a) (b)
Fig. 1-13 (4) The lumped-resistance symbol. (b) The defining-charaeteristio
curve.
with the momentum and velocity simply proportional in this linear case.
The defining characteristic for the lumped-mass element in general is
a plot of its momentum as a function of its velocity as shown in Fig. 1-15.
The slope of the momentum vs. velocity plot at any point is the mass of
the element. If the plot is a straight line, then the mass IS constant and
we have a linear-mass element. If the plot is curved, the element is non
linear with the mass being a function of the momentum, or of the velocity.
As with all of the electrical elements, "linear" or "nonlinear" is deter
mined by the defining characteristic being straight or curved.
Momentum is chosen as a basic mechanical coordinate for a transla
tional system. Taking the first time derivative of the momentum gives
us the force applied to the mass. As before with the electrical coordinates,
K m/newton
(complioncel
the first time derivative yields a variable with which we have had more
previous association, in this case a mechanical force. Since the velocity
v has been used in the description of the mass element, the second basic
mechanical coordinate should be the integral with respect to the time of
the velocity, or simply position.
The lumped-linear-spring element, shown symbolically in Fig. 1-16,
can be described by an equation of the following form:
f(t} I'
= K1 _.. vet) dt (1-25)
Taking the position of the top measured with respect to the bottom at
time - 00 to be zero gives us
1
f(t) = K x(t) (1-27)
Slope =K(fl
Fig. 1·17 The defining characteristic for the
f
lumped-spring element.
A plot of position vs. applied force, as shown in Fig. 1-17, defines the
general lumped-spring element. The slope of the characteristic at any
poibt is the compliance of the spring. A straight-line plot gives a constant
compliance K and a linear spring. Similarly, a curved plot yields K as a
function of the position x or the force f and the spring is nonlinear. As in
all previous cases, "linear" or "nonlinear" is a description of the defining
characteristic.
The fundamental mechanical coordinates are momentum and posi
tion. The first time derivatives of these coordinates are force and velocity,
respectively. 1 The dissipative, or viscous, damping element shows the
use of these first time derivatives. Figure 1-1& pictorially shows this
element. The force applied to the element is shown as p, and the velocity
t represents the velocity of the top side with respect to the bottom side,
as indicated by the brace and arrow in Fig. 1-18a. The defining rela
tionship for a linear-damping element is given by
p = Dt (1-28)
where D is in units of newton-seconds per meter, using consistent mks
units. A plot of p versus X, shown in Fig. 1-18b, gives the general defining
relationship for the damping element. The slope of the plot at any point
is the damping coefficient D(x). Again a straight-line plot represents a
linear element and a curved plot represents a nonlinear element.
I Very often in beginning mechanics courses the scheme of selecting force and
position as basic variables is used. In view of the discussion in this chapter, this
certainly appears to be a rather hybrid selection. In beginning electrical-circuit
theory a more consistent choice of current and voltage is used.
Slope =f) Ii )
Fig. 1-18 (a) The sym
bol for the lumped-viacous
damper. (b) The defining
(al (bl characteristic curve.
Coordinates, elem8lll$, ond stote functions 19
Example 1-5
A common electromagnetic audio loudspeaker is shown in Fig. 1-19. The voice coil is
wound on a cylindrical member attached to the apex of a right~ircular paper cone.
The edges of the paper cone are rippled and attached to a rigid-eircular frame. A
constant radial magnetic field (not shown in Fig. 1-19) generates a force on the voice
coil proportional to the coil current. This mechanical force of electrical origin is given
by !(t) in Fig. 1-19. As the apex of the cone moves horizontally in response to the
force !(I), the front and back surfaces of the cone push and pull on the surrounding air
and generate sound waves. Determine a lumped-element model for horizontal
translational motion of the voice coil.
Solution: The mass M of the moving system consists of the voice coil, the cylin
drical coil form upon which it is wound, and some portion of the paper cone. Owing
to the rippled edges of the paper cone, a spring compliance will tend to restore the
moving system to some particular equilibrium point in the absence of a force I<t).
The air surrounding the paper cone contributes a damping force opposing the motion
of the system. In Fig. 1-19 the variable i; is used to define the velocity of the voice coil
with respect to the stationary frame of the speaker. The small arrow on the brace
indicates the velocity to be positive when the voice coil moves to the right.
Figure 1-20 shows a lumped-element model for the system assuming only transla
tional motion. The model system in Fig. 1-20 is so drawn that the motion is in the
vertical direction with i; defined positive when the mass M is moving downwa,d.
The force !(I), when positive, is trying to make i; negative; thus!(t) is shown upward.
The compliance of the paper cone and the damping effects of the air are given by the
lumped elements K and D.
The compliance coefficient K could easily be determined experimentally by plac
ing a static force on the voice coil and measuring the static displacement. By plotting
displacement VB. force, a eurve similar to Fig. 1-17 is obtained. The slope of the
curve at any point is the compliance. The effective mass M and damping coefficient
I D are not quite so easy to determine. The defining~haracteristic curve for the m8.88
\,
~
\\
20 Principles of e/ectromechanical·energy conversion
.J
,l
Fig. }·20 A lumped-element model of the loud
speaker.
element, given by Fig. 1-15, requires a plot of the momentum of the moving systems as
a fun~tion of its velocity. Both momentum and velocity are difficult quantities to
measure. Simply weighing the elemente in the moving system would give a reasonable
approximation to the effective mass. Similarly, the damping coefficient is defined as
shown in Fig. 1-180, where we must plot force VB. velocity. Such measurements are
not easily made. Certain indirect tests, such as step- and frequency-response
measurements, are generally used for determining these parameters. These methods
are discussed in detail in later chapters.
The force or gravity is not included in the model system shown in Fig. 1-20. We
are not hanging a mass M by a spring and damper from a rigid support in a gravita
tional field. Figure 1-20 is a lumped-element representation of the actual system
shown in Fig. 1-19. The lumped-element model experiences translational motion in
only one direction, and for convenience we have chosen this direction to be vertical.
The equilibrium equations for the model have a correspondence to the actual system
only over a limited operating range for which the model parameters are valid. Usually,
the most difficult part of any analysis is the determination of the model and the evalua
tion of the parameters.
:{ ----+---if ~/t----
Fig. 1-21 The symbol for a lumped-inertia
element. 'WffH/rHHH/ffHrRHWffff///J.
w(t) = 1/1
J _00 T(t) dt (1-30)
Ir curved, then the element is nonlinear and the inertia is a function of the
angular momentum or the angular velocity. Angular momentum is the
basic coordinate being used. Notice that the first time derivative of this
coordinate 1 gives the common torque variable. The angular velocity w
is used in the defining characteristic of Fig. 1-22. The second basic coor
dinate is angular position 8. The first time derivative of this coordinate,
namely, 9, is the angular velocity w.
T = i. 8 (1-33)
o
~
--1].. . -
-:)I-,-i
Fig. 1-24 (a) The
symbol for the lumped
rotational viscous-damp
ing element. (b) The
(al (b) defining characteristic.
CoordilKlte., elemenfl, and Itate function' 23
TABLE 1-1· Summary of the basic coordinates and their first time derivatives
Electrical
~
J C(i)=d~aA ] "il.d~ all ] alii. :i
Mass Spring Damper
Mechanical
(rotationol) ~ ~
. a9
~
. ai
J(Q)=~ K,(l) '" (if 0,(91 = d8
For each of the three physical systems, three lumped elements have
been introduced. Table 1-2 summarizes these nine elements, showing
their schematic symbol and their individual defining differential relation
ship. For example, the lumped-capacitance element is defined by plotting
the charge q on the plates of the capacitor as a function of the applied
terminal voltage >.. Thus the capacitance for any particular charge or
voltage is the slope of this plotted curve as given by dq/dX. A Iinear
capacitance element has a straight-line plot, whereas a nonlinear element
has a curved characteristic. However, in either case the capacitance is
still clearly defined as the slope of the characteristic curve. Notice in
Table 1-2 that all other elements are defined in a simila'l: fashion.
known. Here only one basic coordinate has to be specified, since the
characteristic curve provides the second coordinate. Consider, as a sec
ond case, two mutually coupled inductances. The flux linkages in either
coil are a function of both the current in the same coil and the current in
the magnetically coupled coil. In order to specify the instantaneous
operation of the first coil, both its total flux linkages and coil current would
have to be known. The characteristic curve for the element would give
the portion of the flux linkages due to the current in the same coil. A
second characteristic curve is required to provide the flux linkages in the
first coil due to current in the second coil. For a complete description of
the first coil, currents in both coils are required with the characteristic
curves in order to specify the total flux linkages and thereby describe the
operation.
The general procedure for determining the system coordinates as
functions of time generally involves solving a set of simultaneous differ
ential relationships known as equilibrium equations. These equations
dynamically balance specific terminal variables for the lumped elements
according to the interconnections of the elements in the system. For
electrical systems, the equilibrium equations are formulated by means of
Kirchhoff's current and potential laws. These two laws generate formal
ized procedures yielding loop equatians or nodal equations, depending on
how the coordinates are chosen. Equilibrium equations for mechanical
systems are generally formulated by means of Newton's second law or
D'Alembert's principle. involving a summation of forces or torques. For
noninteracting electrical or mechanical systems these procedures are
adequate to yield the system equilibrium equations.
The variables used in the formulation of the equilibrium equations
are often called slate variables. The term "state variable" is used because
the equations specify the state of each element and therefore the state of
the system. For example. consider a capacitance element having a char
acteristic curve as shown in Fig. 1-25. At a given time instant the charge
on the capacitor has a value qQ' The state of the capacitor at this time
instant is therefore defined by point P in Fig. 1-25. The quantities qa
and the corresponding ~.. are the values for the state variables q and ~
at the specified time instant. The point P on the characteristic curve is
termed a 8tate point. As time progresses, the charge on the capacitor will
vary, and equivalently the state point will move along the characteristic
curve. Thus the motion of the state point as a function of time also
describes the operation of the element.
For electromechanical systems, involving an interaction between the
electrical and mechanical parts of the system, the formulation of the
equilibrium equations directly using Kirchhoff's laws and D'Alembert's
principle is often very difficult. To aid in the formulation problem, an
energy technique will be used. This technique requires the formulation
of energy-8tate functions. These functions of the state variables at any
sp~cific time instant have values determined by the values of the state
variables at that time instant. Therefore, the values of the state func
tions at a specified time depend only on the state points (or equivalently
on the state variables) at that time instant and in no way on the past
history describing how the state points were reached. As the state func
tions are formulated in the following sections, these ideas will be discussed
further. The energy formulation gives the same equilibrium equations
as obtained by Kirchhoff's laws or Newton's law. The main justification
for this energy method rests in the routine procedures involved, par
ticularly in the case of electromechanical systems.
The energy analysis will first be carried out for con8ervative electrical
and mechanical systems. The term "conservative" means that the system
has only the ability to store energy; energy cannot in any way be con
sumed or dissipated. Thus, for electrical systems, only capacitance and
inductance elements are to be included; initially, no resistance elements
will be considered. Similarly, for mechanical systems, damping elements
will be omitted. However, bear in mind that suitable procedures will
finally be discussed to include these dissipative elements. Also, we shall
limit our considerations to electrical and mechanical systems which are
noninteracting. These are the types of systems considered in previous
courses in circuits and mechanics, respectively. This will afford an oppor
tunity to check and compare the energy formulation with formulations
that are familiar to us. In the next chapter the broader class of elec
tromechanical systems involving electrical and mechanical interactions
will be discussed in detail.
where q is the charge at time t and qo is the charge at time to. Notice how
the functional dependence of the voltage >" changes from t to q' as the
integral is manipulated to an equivalent form. In Eq. (1-37) the primes
are used to depote the variables involved in the integration. The
unprimed variables are the limits of the integral, and in the case of q they
define the state of the element at any time t.
Usually we shall be interested in the total energy stored in the electric
field of this capacitance element. If time to is so chosen that qo = 0, then
the energy stored at this time is zero, meaning
W.(t o) = W.(qo) = 0 (1-3S)
Then Eq. (1-37) reduces to
q'
Characteristic curve
be given by
I Since each of these functions specifies the state of the element and the element has
one unique state, the sta.te functions quite obviously cannot be independent.
general case the voltage across anyone of the capacitors could be a func
tion of the charge on all n capacitors. For the usual capacitance element
the voltage can almost always be expressed as a function of the charge
on that capacitor alone. However, let us keep open the possibility of the
mutual-capacitance parameter. The total energy stored in the system
of Fig. 1-28 can be given by
(1-42)
witH the final energy stored being a function of the charges on all n capaci
tors. The prime on the coordinates is used to indicate the general charge
or voltage on the capacitance elements, and the unprimed coordinates refer
specifically to the state-point coordinates. Therefore, the unprimed coor
dinates will specify the actual state of the elements and, ultimately, the
state of a system of interconnected elements. This added complication of
having primed and unprimed variables arises because we have to perform
integrations with respect to a certain coordinate from one value of the
coordinate to another value of the coordinate. We must distinguish
between the variable over which we are integrating and the limits on this
integration. Hence the primed and unprimed variables.
In order to evaluate Eq. (1-42), we must give some consideration to
how the final state of the system is reached. Suppose, for simplicity, that
the total final charge on each capacitor is deposited in sequence. Thus,
with q;, q~, . . . ,q~ all equal to zero, q~ is raised from its initial value of
zero to its final value, at time t, indicated as simply q" without the prime.
The first integral on the right side of Eq. (1-42) becomes
(q~ ~ I I , " ( ql . , , ,
Jo }.\(qltq:, . . • ,q,,) dql -+ Jo }.\(qltO, . . . ,0) dq\
ments care how they got there. 1 As an abbreviation, Eq. (1-42) is written
as
.
=
.-12: fo'l' X:(q~,q;, . . . ,q~) dq:
W.(ql,q" . . . ,q..) (1-43)
Example ,·6
Three nonlinear capadtors have the following defining characteristk-s:
ql - 3(kd~ q. - 4(k.)~i q. - 7{k.)~i
Determine the energy- and coenergy-state functions for this system of capacitors.
Also, show that the sum of these two state functions satisfies Eq. (l-45).
Solution: Let' us start with the coenergy-state function. If we increase the
three voltages successively from zero to their final-state values kIt k., k., then Eq.
(1-44) for the coenergy becomes
,
W.(k"k.,k.) -
r~' "
10 r A"I A.
ql(k"O,O) dA, + 10 q.(k"k.,O) dA. + 10 q.(""k.,k.) U".
f , ( I. I J{ I
W'(k I, k 2, ka) -
Al 3(k')~' dA' + hAt hAl
0 4(k')~i dA' + 0 7(k')~ dA'
•
h0 1 t 2: , J •
Solving the defining relationships for the voltages in terms of the charges and sub
stituting lead to
W(
,qt,q.,q. ) (fl.' (1j"q,'
-}o (q'(l1;i'.' ) *
') dq,• +}o (Il' (tfJ.)'
dq.' +}o l ' dq.•
thU8 verifying the first term in the summation of Eq. (1-45). In a similar fashion the
remaining two terms of this summation can be checked for these two state functions.
Just for practice, complete this verification.
Oqk
'aX/I;
= at i !o~' q;(X:,x~,
4: i - I
. . . ,X~, . . . ,>.~) dX~ (1-49)
Thus the change in the coenergy-state function with respect to the voltage
on the kth capacitor, when all other voltages are held fixed, gives the
charge on the kth capacitor. Taking the total first time derivative of both
sides of Eq. (1-50) would yield the current through this kth capacitor.
Nodal equations are formulated by summing currents at each node.
Therefore, selecting X's or X's as coordinates of interest will give rise to
nodal voltage-equilibrium equationlf when partial derivatives are formed
with respect to these coordinates. After the energy and coenergy func
tions for all of the remaining electrical and mechanical elements are
formulated, the procedures whereby equilibrium equations are obtained
from state functions will be systematically formalized.
I See W. Kaplan, "Advanced Caleulus," p. 220, Addison-Wesley Publishing Company,
(1...51)
where all >"'s in the last two terms are available as functions of the q's.
Expanding Eq. (1-51) we have
(I-52)
Notice how the chain rule is used in taking OW;;Oqk. Using the result
given by (I-50), the summation is seen to be equal to zero, and thus
Eq. (1-52) gives the same result as (1-48). Equation (1-51) clearly
shows that a partial derivative of an energy function is not equivalent to
the same partial derivative of the corresponding coenergy function. The
result given by Eq. (1-50) can also be obtained in a similar fashion from
Eq. (1-45) using the energy-state function instead of the coenergy
function.
(1-54)
Coordinat••, ./emenl., and .tal. funclion. 35
ancel.
~-----}::
If the energy stored at t = 0 is taken to be zero, then the total energy
stored from time 0 to time t is given by
(1-56)
where X" X2, • • • • X". without the primes, are the respective flux linkages
at the time t. By using the symbolic notation of the preceding section,
.
Eq. (1-56) can be condensed to
.
W...(X 1,X 2 • • • • ,X..) = 2 Io~' q~(X~,X;,
,.1 .
• • • ,x:) dX; (1-57)
Notice the complete duality between Eq. (1-57) for the magnetic-field
energy storage in a system of n inductances and Eq. (1-43) for the electric
field energy storage in a system of n capacitances. In the magnetic-field
case, the coordinates X and q are involved, as opposed to q and Xfor the
electric-field case. This is not surprising, since the original defining char
acteristics of Figs. 1-28 and 1-29 have the same symmetry.
The coenergy-state function for the system of n inductances is given
by
(1-58)
36 Principl•• of ./ectromechanico/-en.rgy conversion
Also, the energy and coenergy are related through an expression similar
to Eq. (1-45) as follows:
.
W...(X I ,X 2, ••• ,X,,) + W~(ql,q2' ... ,q,,) = L: X;q;
i-I
(1-59)
aXk
(1-60)
I
which reduces to
aw.. (XItX" •.• ,Xk, .•. ,X,,)
aX =
. ('\ '\
qk Al,A2, •••
'\ '\ )
,Ak, • • • ,A"
k
(1-61)
Thus, taking the change in the total magnetic energy stored with respect
to a change in the flux linkages of the kth element, when all other flux
linkages are held constant, gives the current in the kth inductor. Again
observe how the selection of X as the variable of interest, for a nodal
formulation, leads to a current term. Performing a similar operation
with Eq. (1-59) and the coenergy-state function gives the same result.
The partial derivative of the coenergy function with respect to a
particular current leads to
aW~(qlJq2, ... ,qk, . . . ,q,,)
Oqk
which, when the first total time derivative of both sides is taken, gives us
the voltage across the kth inductance. Loop equations involve currents
as variables; thus, taking a derivative with respect to a current gives us a
voltage term which we can then sum in the loop formulation of equilibrium
equations.
Thus far we have considered the applicable state functIons for the
capacitance and inductance elements. The admittance or resistance ele
ment has a different type of state function. In the capacitance and
inductance elements, energy is stored in electric and magnetic fields,
Coordinates, e'ements, and stote function. 37
where :t.(t) is the velocity of the ith mass and p.(t) is the force applied to
the ith mass element. In an increment of time dt the kinetic energy
received by the system is given by
where T is the symbol for kinetic energy, the Xi are the respective veloci
ties, and the Vi are the respective momenta. Taking the kinetic energy
at time tOto be zero, the total kinetic energy stored at some later time
t is given by
which becomes
+ . : . + 10 ( "• • /(' I
x.. Vl,V2, • • . (1-67)
Using our abbreviated notation, the total stored kinetic energy is given by
..
T(Vl,VS, •.• ,V..) = L fo"; x~(v:,v~, ... ,V:) dv~
i-I
(1-68)
38 PrillC;pl•• of .lectromechonicol-energy conversion
p'
Characteristic curve
p(t}
~ a
which is a general statement that the total rectangular area for a given
state point is the sum of the respective energy and coenergy functions.
Equilibrium equations for mechanical systems are most often formu
lated on the basis of Newton's second law or D'Alembert's principle for
both translational and rotational systems. Equations in this form are
summations of forces or torques, respectively. With these mechanical
systems, the dual formulation corresponding to loop equations for elec
trical systems exists, but it is not commonly used. We sum forces from
free-body diagrams, but very seldom do we sum velocity-type terms.
CoOl'Clinotee, e1_eb, and date fund/one 39
The reason for this preference lies in the fact that velocity terms would
involve the momentum as coordinates, and our interest is not usually
centered around the system's momenta. The force formulation, on the
other hand, involves position as coordinates, and most often the sys
tem's configuration is of prime interest.
With the position coordinates of major interest, all partial derivatives
of state functions should be taken with respect to z's and x's. Thus
aT' (XI,X" • • • ,:tk. . . . ,:tIl)
aXk
a ~
= ax ( ~, ,( ., .,
~ Jo Pi Xl,X"
.,
• • • ,X/o, • • •
.') "'...,
,x" ai i (1-71)
"-1
which gives us simply
aT'(x"x" ... ,x/o, ... ,x,,) (.. . . ) (1-72)
a:t,. = Pit XI,X" . . . ,X/o, . . . .x"
Taking the first total time derivative of both sides of Eq. (1-72) gives us
the required force on the kth inertia element as a function of the n veloci
ties'in the system. With all of the mass elements that we shall consider,
the momentum will be a function of only the velocity of that element.
Thus in Eq. (1-72) Pic will be only a function of XII. The formulation of
the kinetic-energy- and -coenergy-state functions is presented in a com
pletely general way in order to parallel the development of the correspond
ing functions for the electrical elements.
For the translational spring, energy is stored in the form of potential
energy. A system of n springs, in an increment of time dt, could change
its potential-energy storage by an amount
dV(t) = PI~t) d Xl(t) dt + p,(t) dz,(t) dt
dt dt
which is equivalent to
(1-75)
With the abbreviated summation notation used with the previous ele
ments we can write Eq. (1-75) as
..
V(Xl,X2, •.. ,x..) .... 1 /0'" P:(X~,X~,
_-I
(1-76)
I
Notice the complete symmetry of Eq. (1-76) for the potential-energy
storage in a system of n springs and the kinetic-energy storage in a system
of n inertias, given by (1-68), with the interchange of p's and x's between
the two state functions.
For completeness, we shall also formulate the potential-coenergy
functions as being
.
" {Pi '( , .,
1., )0 Xi Pl,P2' . . . ,P..') d Pi
I
(1-77)
i-I
aXk
= i
a~·,_1 /0'" pHxl'x~, ... ,x~, ... ,x~) dx; (1-79)
which is the force across the kth spring as a function of the position
coordinates.
Coordinal••, .I.menl" and .101. fllndion. 41
Table 1-2, which indicates the defining relationships for all the elements,
shows that the conservative elements are always defined as a relationship
between one basic coordinate and the first time derivative of the other
coordinate. Thus, for example, the rotational inertia element has a" = lk
and Pk = ~".
In evaluating the functional relationship of the form
n
remember that the system can arrive at its final-state points in any fashion
with relationship (1-82) still maintaining the same value. This means
that with all the other a'S equal to zero, a~ could be increased from zero to
its final value al. Then by holding a~ constant at this final value aI, a~
could be increased from zero to its final value a2. In turn, each of the
other a~ could be increased to its final value as the summing index i
proceeded from 1 to n. Any otlier scheme leading to the same final-state
points ai, a2, . . . , an would give the same value for (1-82). For
example, each of the a;could be increased in turn from zero to half its
a;
final value. Then, with each at half its final value, each could in turn
be increased to the final value. The result would be the same as with the
total increase previously described. Very often, the actual system will
have all its coordinates changing simultaneously from zero to their final
values.
t;
TABLE 1·3 Summary of the conservative element energy· and coenergy-state functions
Capacitance
W.(ql,qt, . . . ,q.) "" L" f'~' ,
i-I
0 ,(q\>q"",
1 " dq,
,q.) Electric-field energy
W.(~I'~"
1
. . . • ~.) "" L" ~~' , 0 q,(~u~t.
11 . . ,~~) d>.~ Electric-field coenergy
i-I
Inductance
W...(>'I,>'., . . . ,>'.) '" L
n
~X' ,
0 q'(>'I'>"" , d>.,f
.. ,>'.) Magnetic-field energy
i-I
W''"(q l,qt,
. * ••
q) =
J" Lf'
i-I
n
0 >.'(qf
i uq,,···
.f .f) dq'i
,q" Magnetic-field coenergy
.
Mass (translational)
T(pl,PI, ..• ,p.) "" Lt
i-I
0
Pi
Xi(PIIP",
" 0 0 1 dp,t
,P.) Kinetic energy (translational)
..
T
too
(XI,XI, • 0 • ,x.) -
0 L t;t, 0
flo' of
p,(tux . . . ,x.) dx,
"
0'
Kinetic coenergy (translational)
i-I
"'----_..........
TABLE 1·3 Summary of the conservative element energy· and coenergy·state functions (continued)
Spring (tran8Iational)
. Lt'"'
" , ,,
V(x\,Xt, ••• ,x.) - 0 p,(x\,x, • •.. ,x.) dx. " Potential energy (translational)
i-I
,. . . L" t
V (PI,P" ••. ,P.) -
P '.'.'
0 ' x,(PltP,•.•. . ' dp.
,P.) .' Potential coenergy (trane1ational)
.-1
Inertia (rotational) "
T(ll.l" ••• •1.) - Lf"'11
.-1 ('
8.(lI,l" . . • , ) dl. Kinetic energy (rotational)
t
44 Prirn:iple$ of electromechanical-energy conversion
PROBLEMS I
1.1 For values of applied voltage between zero and v.. a certain capacitor can be
dellCribed by the characteristic equation
L",/2 ----J------,
i Fig. Pl-3
1-4 Two windings are placed on the same iron core. The first winding has 800
turns, and the second winding has only 500 turns. With a current in the 800-turn
winding the following data are taken:
Current in 800 turns, amp o 0.1 0.3 0.4 0.5 0.6 0.7 0.8 0.0 1.0
Flux linking 800 turns, kilolines o 13 38 49 56 60 64 67 69 71
Flux linking 500 turns, kilolines o 13 38 47 53 57 59 61 63 64
Carefully draw the following curves on one sheet of graph paper.
a. Plot the characteristic curve for the 8oo-turn coil in mks units.
b. Plot the flux linkages in the 500-turn coil as a. function of the current in the
800-turn coil.
I Answers to odd-numbered problems are listed at the end of the book.
Coordinate" elements, and ,tate functions 45
;
l c. From part!! G and b determine the self-inductance of the 800-turn coil and the
[ mutual inductance between the two coils. Plot these inductances as a function of the
t
!"
current in the 800-turn coil.
1-5 The radio telescope shown in Fig. Pl-5 consists of a large antenna attached
by means of a tower-like structure to a large base. The telescope can be turned by
means of a motor connected through appropriate gearing to the base. Let the tor
sional compliance of the tower be K, rad/newton-m. The base and antenna inertias
are J Band J .. kg-m', respectively. The torque applied to the base by the motor is T ..
newton-m. Assume that the wjndage torque on the antenna is proportional to the
angular velocity with a coefficient of D, newton-m-ilec/rad.
G. Draw a lumped-element model for the radio telescope. Indicate all sources
and the angular velocity of the base and antenna on this model system.
b. In tcrms of WB and "'A write expressions Cor the kinetic-cnergy- and -eoenergy
state functions T(lA,l,,) and" T' (WA''''8) or T'( dA,b,,).
c. Write expressions Cor the potential-energy- and -eoenergY-iltate functions
V(8.. ,IJs) and v'(iA,IB).
I
t
II Fig. Pl-5
I
1-6 The charge VB. voltage characteristic Cor a particular electric-field device can
t be approximated by
v - aq' + bq
where and b are constant!!.
G
G. Determine the capacitance of the device as a Cunction oC the charge.
b. Find the clcctric-encrgy- and coencrgy-statc fun('tions. Evaluate these CURC
tions Cor q - 3 coul.
c. Check Eq. (1-48) from the results oC part b.
I 1-7 The slope of the characteristic curve for a certain capacitance element is
given by
I ~ - aq./l-.'
f•
I where G and q:
are constant!!. When the voltage v acrO!!ll the capacitor is zero, the
I
cbarge II is al80 zero.
Coordinate" element" and dote function' 47
1-11 The average length of each turn on the inductor in Prob. 1-10 is 7.0 in.
Il.Find the resistance of the winding if copper wire w.ith a diameter of 0.028 in. is
used. (The conductivity of copper is listed in the table of physical constants.
Appendix.)
b. Sketch the characteristic curve describing this lumped-resistance element.
c. For a certain temperature increase the conductivity decreases by 10 per cent.
Does this temperature change affect the curve plotted in part b? If it does. sketch
the new curve.
1-12 When a step-current function is applied to a certain lossy capacitor, the
voltage reaches 63 per cent of its final value in 2.5 sec. The capacitor can be modeled
by a linear lossless capacitor in parallel with a linear resistor. With a d-c voltage
source across the lossy capacitor the power dissipated is 85 watts.
Il. Determine the electric-field energy stored in the system, with the voltage
source connected.
b. Also find the electric-field coenergy under the conditions of part Il.
1-13 According to Einstein's special theory of relativity. the mass m of a particle,
instead of being a constant. is given by
m.
m - -:r:==='F'F7.
i VI - (vic)·
I
where m. is the rest mass (or the m&88 at zero velocity), c is the velocity of light, and v
!
i·
is the velocity of the mass m.
Il. Draw the characteristic curve for the relativistic mass element by using the
relativistic velocity vic as the abscissa. Remember that II cannot be greater than c.
Since the system can reach its final state in any manner, assume that the kth element
is the last to be assembled. Therefore all terms in the summation will not contain q.
except
46 Principle. of electromechanical-energy cOlIYersion
Coil current, rna 5.4 8.6 13.5 22.5 27.4 38.2 54.0 65.2
Core flux, kilolines 66 100 133 160 167 173 180 183
a. Plot the characteristic curve for this inductance element. Plot flux linkages,
in weber-turns, as a function of the coil current, in amperes.
b. For a coil current of 30 ma graphically determine the magnitude of the energy
and coenergY-fltatc function.
c. Graphically determine the inductance, in henrys, as a function of the coil
current. Plot the inductance VB. coil current on the same graph as used in part a.
d. If the inductor has a resistance of 100 ohms, determine the approximate
terminal voltage for a coil current of (30 + 2 sin 377t) milliamperes. Note: 1 weber
10· lines ~ 10' kilolines.
1-10 An inductor is constructed by winding 10,000 turns uniformly around a
toroidal iron core with a rectangular cross section. The toroid has an inside diameter
of 4.0 in. and an outside diameter of 6.0 in. The height of the toroid ia 2.0 in. The
magnetization curve for the iron core is as follows:
LAGRANGE'S EQUATION
Coord i-
Element nates or Proper partial derivative Type of analysis
interest
ax.
= P.(x•• .. . ,x.) Force summation
Spring
lPa
flows into the element when both terminal variables have the same signs.
Thus for both x" and Po positive, potential energy is being stored in the
spring. Also for both Xb and Pb positive, kinetic energy is being stored by
the mass element.
The spring element is described by a characteristic plot of the dis
placement x" as a function of the applied force po. For simplicity assume
that the spring is at its free length (that is, p" is zero) when x" is equal to
zero. The mass element is described by a characteristic plot of its
momentum Pb as a function of its velocity Xb. From the characteristic
plot the potential-energy-state function for the spring element is given by
(2-1)
(2-2)
The primes are used to indicate the variables over which the integration
takes place. The unprimed variables are the final state variables defining
the state point on the characteristic curves.
Taking the partial derivative of both sides of Eq. (2-1) with respect
to x" leads to • ,
(2-3)
which is the external force applied to the spring element. Similarly, the
external force applied to the mass element can be recovered from the
kinetic-coenergy-state function. By forming the total time derivative of
the partial derivative with respect to Xb of both sides of Eq. (2-2), we have
-d
dt
[fJT'(Xb)]
- - - - - = -d [ - 8
8Xh dt 8Xb
!c:b '(") d"] =
0
Pb xI> XI> . ('Xb )
Pb (2-4)
where x is the single velocity coordinate shown in Fig. 2-2. If xo , x., and
x are so arranged that the spring is at its free length when theya.re equal
to zero, then Eq. (2-.;) can be integrated to give
Xo=Xb=X (2-6)
52 Principle. 01 e/ectromed!onical-energy conversion
I(
Mi + ~K = J(t) (2-9)
Now let us see if we can extend these ideas to a system composed of many
lumped mass and spring elements. Before these elements are attached
together, individual force and velocity terminal variables are assigned.,
For the mass elements a kinetic-coenergy-state function can be written
in the general form
T '("Xo,Xb, • • •
~ Jo
) = L (s. p"'(""
X ..,Xb, • ••
) d"XI< (2-10)
Ie
If the constraint equations given by (2-12) are holonomic, then they can
be integrated to yield relationships between the original element coordi
nates and the new node coordinates. Substituting the integrated equa
tions of constraint into the state functions, given by (2-10) and (2-11),
yields
(2-13)
and
f2-14)
In the next section we shall examine cases where the state functions'have
a form more general than is proposed by Eqs. (2-13) and (2-14).
S4 Principle.: of electromechanical-energy conversion
which is the total external force applied to all spring elements connected
to the kth mechanical node. If Q" is a total of all forces externally applied
io the kth mechanical node of the system, then D'Alembert's principle of
dynamic equilibrium requires Q" to be balanced by the mass and spring
forces at the kth node as given by Eqs. (2-15) and (2-16). Thus we have
!!:.. [aT'(X1,X2, .' .. ,X,,)] + aV(XI,X2, .•. ,x,,) = Q"
dt aXI; aXk
fork = 1,2, .. _ ,n (2-17)
Example 2·J
Write the equilibrium equations for the linear m8.88-8pring system shown in Fig. 2·3.
Assume that springs K K t , and K. are stretched distances a, b, and c, respectively,
when x. and Xb are equal" to zero.
Solution: In Fig. 2-4 the system is broken apart to show each lumped element
separately. Individual terminal variables are defined for each element. For
example, compliance K I has the velocity of its bottom end with respect to its top end
given by x.. The small arrow on the x, brace means that the velocity is taken posi·
tive when the bottom moves away from the top. The force pIon the compliance K, is
defined as positive when it causes a positive velocity, or when it also is directed down·
ward. The terminal variables on the mass elements have their velocities defined with
respect to the stationary support. Both velocities x. and x. are positive when directed
downward. The terminal forces applied to these m&lJSe8, namely, p. And P., Are alflo
positive in the downward direction and are applied by pushing from the stationary
support.
Lagrange', equation 55
t
t'
i
f }:.
I
t.
;"
•
The relationships betwecn the momenta and the velocities for the linear-maBII elements
®iven by
• XI
PI - K.
where Zit X., x. are the amounts each,spring is stretched. When X., X., and x. are zero,
the forces exerted by the springs are respectively zero. The potential-energy function
therefore expands to give
56 Principles of electromec:nanic:al.ener9Y conversion
By connecting the elements together, as shown in Fig. 2-3, constraints are placed upon
the individual terminal variables. We can write the relationships between the
coordinates x. and i. and the original coordinate variables as follows:
Xl .., x.
x, - -:to + Xb
xa =- :4
By integrating the first three of these equations, we have
XI - +a
x.
x. - +
- x. x. + b
x. - x. + c
with a, b, and c as the respective stretched positions when x. and Xb are equal to zero.
The kinetic..j)oenergY-l!tate and potential~nergy-8tate functions are now given by
T'(:t.,X!.) - !M,x.· + !M.x.' (2-18)
V(X.,Xb) _ Lx,,-±~~~ + ( -x. + Xb + b)' + (Xb + e)1 (2-19)
2K, 2K, 2K.
upon substitution of the new coordinates. The external forces acting on the mechani·
cal nodes are seen by inspection to be
Q. - -j,(t)
with a negative sign indicating opposition to the assumed positive sense for the
coordinate and a positive sign indicating that the force is in the same direction as the
positive sense for the coordinate.
Equation (2-17) for. the first coordinate becomes
By expanding these partial derivatives with the use of Eqs. (2-18) and (2-19), we have
(2-20)
~ [ilTJ~.•,Xb)] + ilV(X.,Xb) _ Qb
Equations (2-20) and (2-21) are the two required equilibrium equations. A simple
free-body diagram of each of the masses shows these equations to be valid force summa
tions according to Newton's laws or D' Alembert's principle. The purpose of this
example is t~ illustrate the detailed procedures involved in the use of the energy
formulation suggested by Eq. (2-17). For a problem this simple, such an involved
method is certainly not warranted. However, (or more complex systems the same
general routine will yield equilibrium equations when not all of the forcet! are quite 80
obvious.
Lagrange'_ equation 51
Example 2·2
Figure 2-5 represents a motor driv.ing a rotating loed. The motor is represented by
the rotating inertia J ,. The torque developed by the motor is some explicit function of
time TI(t). The loed, represented by the inertia J., is coupled to the motor by means
of a drive shaft with compliance K,. In addition, the load contains a resisting torque,
labeled T.(t), which is some function of time. The combination of an automobile
engine, drive shaft, and rear wheels, in its most basie form, eould be represented by
the lumped system of Fig. 2-5. For this simple rotational system determine the
equilibrium equations by using the energy formulation.
Solution: By breaking the system apart, as shown in Fig. 2-6, terminal variables
are defined for eaeh of the elements. For the motor inertia J I , the angular velocity
'I.
with respect to a stationary reference is designated as II. and the torque applied to the
inertia is Similarly, the angular velocity of the left side of the shaft compliance
with respect to the right side is I, and the torque applied is I,. In the same manner,
terminal variables are defined for the loed inertia J •.
The kinetic-coenergy function for the system is formulated by referring to Table
1-3. For this example, we have
2
T'(II,.I.) - 2: !c/' t,(I~,I~) dl;
.-1
The angular momenta are related to the respective angular velOcities by the linear
relationships
Similarly, from Table 1-3, the potential energy is given for this system by
V(e,) - 10
( " I' I
,,(8,) dB,,
For the linear shaft-compliance element, the torque /, is related to the angular twist by
1
1. - -II.
K,
and so the potential function becomes
When these three lumped elements are connected together into the system shown
in Fig. 2-5, constraints are placed on the individual element variables as follows:
The assumption has been made that, when II• ... II., the shaft compliance is in its free,
or untwisted, position.
The external forces for a rotational system are in the form of torques. From
Fig. 2-5, the externally applied torque on the a mechanical node is given by
d 0. - II.
dt (J.fJ.) + -1[6 ... T.(t) (2-24)
which is one of the equilibrium equations. In a similar fashion, Eq. (2-17) for the
second coordinate becomes
which reduces to
(2-25)
Lagrvnge', equation S9
This is the second equilibrium equation obtained after substituting Eqs. (2-22) and
(2-23).
Obtaining the equilibrium equations by the energy method seems to be a lengthy
job. For a problem as simple as this one, the equilibrium equations could be written
by inspection. The major use of the energy formulation is for systems involving
certain interaction force terms. These force terms are often difficult to postulate by
inspection, and the energy formulation becomes a more useful tool. Consider this
example and the one preceding it to be illustrations of a new technique and not a
suggestion for complicating a problem you could have solved quite simply by using
your previous experience.
\.+ OV(XI,X:,.
i !i IOT'(XI,X2,
dt
•
OXt
,x,,)
ox"
• ,x,,)
=
Q
t
for k = 1, 2, . . . ,n (2-26)
XI = XI(~I,~2, • • • ,~m,t)
X2 X2(~I,~2, • • • ,~ ... ,t)
(2-27)
where ~ is the symbol used for these m new coordinates and the possibility
of an explicit dependence on the time t is included. I The total time deriva
tive of anyone of the old X coordinatcs is given, in general, by
(2-28)
Equation (2-28) shows that the old velocities Xt are functions of not only
the new velocities E; but also the new position coordinates ~;, where i is
equal to any integer, from 1 to m. 011 taking the partial derivative of t,he
old velocities Xt with respect to one of the new velocities Ei, we have
(2-29)
(2-30)
I The letter ~ is a general coordinate which, for our electromechanical systems, could
be charge, flux linkage, position, or momentum, both angular and translational.
62 Principles of electromechanical-energy conversion
aXk/iH;" gives us
iJXk ~ [aT'(XI,X2, .~ .. ,X,,)] + aXk [aV(XI'X2" . ,X,,)] = aFt Qk
a~, dt aXk a~, ax" a~,
for k 1, 2, . . . ,n (2-31)
(2-32)
I
Also, the total time derivative of the last term in Eq. (2-32) is equivalent
to
(2-33)
If the old velocities X/o are not a function of the new position coordinates
~i, then the last term in Eq. (2-32) will reduce to zero.
Observe that the first term on the right-hand side of Eq. (2-32) is
identical with the first term in the equilibrium equation (2-31). Substi
tuting Eq. (2-32) into Eq. (2-31) with the identity of Eq. (2-33) yields
!
for k = 1,2, .. ,n
(2-34)
II
For each value of k one equation similar to Eq. (2-34) can be written.
Adding all of these n equations together gives us as a result j
!
I
i
i
!
Lagronge', equation 63
By using Eqs. (2-27) and (2-28), the state functions T' and V can be
expressed in terms of the m new coordinates and the time t. After making
the substitution of coordinates, the following identities can be written:
,x,,) aXk
a~.
(2-36)
aT'al'~2' ... ,~... ,tl,t2' ,~ ... ,t)
at.
n
" aT' (XI,X2. . . . ,xn ) aXk (2-37)
L,
• -1
aXk at•
aV(~I,t2' ... ,tm,t) aV(XI,X2, .. ,Xn ) aXk
(2-38)
at. aXk at.
(2-39)
The djdt and summation operations in the first term of Eq. (2-35) are
mutually exclusive and can be interchanged. Therefore, after substitut
ing Eqs. (2-36) to (2-39) into Eq. (2-35), we have
Example 2·3
A classical problem illustrating the real power of this method is the double pendulum
shown in Fig. 2-7. A mass M, is hung by means of a massless rod of length r, from &
Lagrangci', equation 65.
rigid support. A second mass M. is hung by a similar massless rod of length r. from
the first mass M ,. Find suitable equilibrium equations for this system. Assume both
pivot points to be frictionless.
Solution: A simple xy cartesian-coordinate system is set up with its origin at
point 0 in Fig. 2-7. The position of each mass at any instant in time is given by
XI, 1/, for mass MI and X•• ,¥. for mass M.. Now the Lagrangian for this system is
given by iii
.e(x,,:t••/iI./i.,x.,X•• II ..!I') ... T'(x.,:i:,./i,,/i,) - V(x••x,.!I',1I.)
However, with this system, no spring elements are involved; thus, the potential func
tion V is equal to zero, and all we must find is the kinetic-coenergy function T'. t
Since all these masses are linear elements, the kinetic-coenergy function is readily
derived to be
T'(x.,:i:"y,.y.) ....!M.x,· + !M,y.' + !M,x.' + !M.Y.'
and with V ... 0 the Lagrangian is simply
.e - T' ... !M.(x,· + !i,') + !M.(x,' + Ii.') (2-45)
The external forces acting on this system arise from gravitational forces acting in
the 11 direction on both masses. On mass lIf. the external force of gravity is simply
Mig, and on mass M. the gravitational force is M,g. Therefore, the external forces
for the cartesian set of coordinates are 68 follows:
Q" - M.g Q., - 0 (2-46)
•Q., ... M,g Q., ... 0
Positive signs are used, since the forces are tending to increase the cooroinate.
t Many authors, in dealing with tllis classic problem. place the gravitational
effects into the potential function with the concept tha~ change in elevation causes
changes,in this form of potential energy. Ho\~ever. in our treatment we- shall restrict
V to containing only spring (orees and consider gravitational forces only 68 external
forces. For an alternative treatment sec 8. W. McCuskey, "Advanced Dynamics,"
p.54, Addison-Wesley Publishing Company, Inc., Reading, Mass., 1959.
66 Principle, of elec:fromec:honicol-energy conversion
The cartesian set of coordinates do not represent a generalized set suitable for
Lagrange's equation. The pendulum rods impose holonomic constraint equations
which can be expressed as
and
These are holonomic constraints because they are relationships between the coordinates.
The angles /I, and /I, shown in Fig. 2-7 represent a set of generalized coordinates.
Either of these coordinates can be varied independently of variations in the other
coordinate. The system has only 2 degrees of freedom, and the two coordinates 8,
and /I, specify the complete configuration.
The original cartesian coordinates can be transformed to the new /I coordinates
by the relationships
z, - r, sin /I, z, - r, sin /I, + r, sin /I, (2-47)
~, - r. cos /I. y, - r, cos /I. + rt cos /I,
These transformation equations automatically satisfy the two constraint equations.
Notice also how these equations correspond to the general transformations of Eqs_
(2-27).
The first total time derivatives of the old coordinates are derived to be
x. - r.iJ. cos /I, i, os r.iJ. cos /II + r,iJ, cos /I,
li. - -r.iJ. sin 8, iI. - -rliJ. sin 8. - r,iJ, sin 8,
On substituting these relationships into the Lagrangian given by (2-45), we have
£(8,,8,,/1,,/1.) .. !M.r.'8. 1 + j-M.[r,'8I' + r,'8,'
+ 2rlr,818. cos (/I, - 8,)1 (2-48)
using the trigonometric identity for the cosine of the difference of two angles. The old
velocities are functions of the new velocities 8. and 8, and also the new coordinates 81
and /I.. Thus the Lagrangian becomes a function of both the new COOrdinaU3 and the
velocities, even though the potential function V is equal to zero.
The generalized forces in the new generalized coordinate system can be evaluated
by using Eq. (2-39). For the /II coordinate we have
oy. Q oz. Q oy, Q oZ,
Q,," Q" 08. + 'I 0/1. + " 081 + ", 08.
By use of Eqs. (2-46) for the original forces and the transformation equation (2-47),
the first generalized force becomes
(2-49)
Since the new Lagrangian, given by Eq. (2-48), is a function or angular coordinates
rather than translational coordinates, the external forces must be in the form of torques
rather than lineal forces; therefore the form of Eqs. (2-49) and (2-50) should seem
Lagrange'll equation 61
reasonable. The minus signs in (2-49) and (2-50) indicate that these torques are
tending to decrease the coordinates BI and B,.
Lagrange's equation given by (2-42) can now be used with the Lagrangian func
tion (2-48) and the generalized external forces given by Eqs. (2-49) and (2-50). The
procedure is as fol1ows:
For i - 1:
Substituting these quantities into Lagrange's equation, using Eq. (2-49) for Q." gives
us finally
For i - 2:
In a similar fashion', by using the expressions for the Lagrangian and Q,., we obtain
M,r,I" + M,rlr./Il cos (B. - B,l M,rlr,6 1'sin (B. - B,)
- - M t(/TI sin B. (2-52)
as the second equilibrium equation.
These two equilibrium equations are very complicated nonlinear differential equa
tions in the two coordinates BI and B,. An attempt to formulate these equations
directly from Newton's second law or D' A1embert's principle is certainly a very diffi
cult job, at the least. By using Lagrange's equation, we have acquired these equations
in a very systematic fashion.
Exampl.2-4
The pivotsl point for the ideal pendulum shown in Fig. 2-8 is constrained to move at a
constant angular velocity of w rad/sec around a circle of radius a 8.8 shown. At time
, - 0 8.8I!Iume that the frictionless pivot point is located at the bottom of its circular
path. Find the equilibrium equation for this system.
68 Principle. of e/edromechallical·ellergy cOllversioll
,
x
We have two coordinates and one equation of constraint, so the system has only
1 degree of freedom. The 11 coordinate shown in Fig. 2-8 is a suitable generalized coor
dinate. The transformation equations from the original cartesian coordinates are
thus expressing the old coordinates in terms of the new coordinate. Notice that both
x and yare not only functions of (J but also explicit functions of the time t. For the
first time derivatives we have
Q. - Mg Qz - 0
Lagrunge's equatjon 69
These forces can be transformed to the generalized coordinate system by using Eq.
(2-39), which expands to the form
a'll ax
Q. - Q. a(l + Q. a(l
thus giving us
Q. - Mg( -r sin (I)
The minus sign indicates that the torque is opposing an increase in the generalized
coordinate angle (I.
Lagrange's equation for this single coordinate is simply
~ [a.e(6,(I,t>] _ a.e(6,(I,t) _ Q.
tit a6 a(l
;ud [a.e(6,(I,t)]
~ - Mr" + ~Mawr[
1 •
-w sm (wt - (I) + 6 sm
.
(wt - (I)l
Upon substituting these results into Lagrange's equation, after some simplification,
we fin&lly have
a,w'
, - Trsin (wt (I) - - ~ sin (I (2-54)
r
The second term on the left side of the equilibrium equation (2-54) results from the fact
that the kinetic-coenergy function, in terms of the new polar coordinates, is a function
of not only 6 but also e and t. Thus, we have the additional term -aT'(6,(I,t)/ae
introduced into the equilibrium equation. This represents a mechanical inter
action of the forces of constraint on the motion of the system.
tJ ") ) tJ
tJ ") ") tJ
~ tJ tJ tJ Fig. 2-9 Loop-current variablee as generalized
coordinates.
where til, q2, . . . , qlt are an independent set of loop currents in the sys
tem. Taking the partial derivative of the magnetic coenergy with respect
to the kth loop current gives us the flux linkages in the inductors in the
kth loop, as seen from Table 2-1. Taking the total time derivative of
these flux linkages would yield the voltage across these inductances. Thus
the first term on the left side of Eq. (2-55) represents the voltage across
the inductors in the kth loop of the network as a function of all 11. loop
currents in the n-Ioop network.
Also from Table 2-1 we see that, by taking a suitable partial deriva.
tive of the electrUt.f!eld energy, the voltages across the capacitors are
obtained. The genmlized external-force term Ql on the right-hand side
of Eq. (2-55) involves the algebraic sum of voltage sources in the kth loop,
taken positive if directed aiding the assumed loop-current direction and
negative if opposing that direction. These voltage sources are generally
not functions of the coordinates, but are usually explicit functions of time.
Remember that a constant or d-e voltage source is just a very special
function of time. Equation (2-55), in terms of our energy-state functions,
tells us simply that the sum of the voltages across the inductors and capaci
tors in the kth loop must equal the algebraic sum of the voltage sources in
the kth loop. This is simply a statement of Kirchhoff's potential law.
Notice, however, thAt the state functions in Eq. (2-55) are formulated as
functions of the 11. loop currents til, ti2, . . . , qlt and their integrals
ql, q2, . . . ,q,.. To be used in Eq. (2-55), these variables must represent
a generalized set of coordinates. In general, we might find it more con
venient to formulate these functions in terms of the branch currents in the
network and then relate the branch currents to the loop currents.
Example 2-5
By using an energy formulation, determine the equilibrium equations, on a loop basis,
for the linear electrical circuit shown in Fig. 2-10.
Solution: First let us define voltage and current terminal variables for each of the
lumped-linear inductance and capacitance elements in the system. These variables
-- are designated by the a, b, c, and d subscripts in Fig. 2-11. Notice in each CMe that the
reference directions for the terminal voltage across each element and the current
through the element are 80 chOl!en that, if both are of like sign, energy flow is into the
element.
72 Principles of eledromechonicol-_rgy conversion
For a loop analysis, we must formulate the electric-energy-state function and the
majpletic-eoenergy-state function. From Table 1-3, the electric-field storage is given
by
W.(q.,q4) - )0
( q• • "
A.(q.,O) dq.
,
-r )0
( q• • ' ,
Ad(q.,qd) dqd
,
since this form of en~rgy storage takes place only in the two capacitors. The capaci
tors are linear. Therefore we have
W.(q.,qd) - 2C. + 2C
t
as a function of the element coordinates.
Also from Table 1-3, the magnetic-eoenergy function for this network is given by
, .( q. I I I (q. I I ,
W..(q.,q.) .. jo h.(q.,o) dq. +)0 h.(q.,q.)dq.
with the magnetic-field IItorage taking place in the inductance elements. From FiV;.
2-11 we have
The minus sign precedms the mutual term in these two expressions because the mutual
field contribution to the flux linkages is in opposition to the self-induced flux linkages
in each of the two inductances. The current ~b in inductor L. is 8.88umed to flow into
the dotted side of this element. This current sets up flux linkages in a certain direction
in the inductance L,. Current q. is 8.88umed to flow into the nondotted side of induct
ance L" and therefore, according to the dot convention, it sets up flux linkages opposite
to those set up by the current ~b. Substituting these flux-linkage (unctions into the
integral expressions gives us .
,rq· ,rq· I
10 (L.~.-O)d~6+ 10 (-M~.+L~.)d~.
W.,(4b.~.) -
I I
with ~. in the second integral acting as a constant. Finally, the magnetic coenergy as
a function of these element coordinates becomes
W~(~.,~.) - tL'~b' - MM. + j-Lt4.'
In order to obtain the desired loop formulation, the loop-current coordinates must
be substituted for the element coordinates as follows:
q.(t) - ~,(t) q.(1) - ~.(t) - ~.(t)
These four equations represent holonomic constraints which can easily be integrated.
The only charge coordintiles appearing in the state functions are q. and q4. By
integrating the first and lail. constraint equations, we have
for all time after the network is connected together. The energy-state functions in
terms of the loop coordinates now become
q.' q.'
W.(q.,q,) - 2C, + 2C. (2.56)
The equation for Q. is made up of all voltage sources in loop 1 taken positive if aiding
~, and negative if opposing ~.. Similarly, for Q, we have all sources in loop 2 taken
positive if aiding ~. and negative if opposing ~•.
By using Eq. (2-55), we have
Fork-I:
~[aW~(~I'~')] + aW.(q"q.) _ Q.
dt a~. aq,
aq, - C.
using the expression for Q,. Taking the total time derivative and rearranging terms
gives us, finally,
1
(L , - 2M + L.}ij. + C I q. + (-L. + M)lb - fll(t) - lI,(t) (2-58)
By lusing Eqs. (2-56) and (2-57) for the state functions and the expression fev Qt, we
have •
which becomes
.:{ ~$
i
'----v---"
63 +
coordinates as follows:
Example 2·6
Determine the nodal equilibrium equations, by using an energy formulation, for the
net"'ork shown in Fig. 2-13.
8olutwn: First the terminal variables for each of the lumped elements are desig
nated as shown in Fig. 2·13. The only rule on their selection is that energy flow into
the element if both the respective;f.. and q are positive. The original formulation of our
state functions, leading to Table 1-3, is based on this arrangement of the terminal
variables for each of the lumped elements. Observe the way elements LI and C 1 have
their terminal variables defined in an opposite fashion.
For a nodal formulation, we must find the electrie-field-coenergy and magnetic
field-energy storage. By use of Table 1-3, and the fact that all elements are linear,
these funetions are given respectively by
W~(4;f..,,) - iCI~' + iCt:f..,,'
)..1 ~.
W -().•• ~) - 2LI + 2L.
for the system shown in Fig. 2-13.
The generalized nodal coordinates must be substituted for these individual terminal
coordinates in order to use the nodal energy equation (2-62). The eoordinate rela.
lOne is reminded of Humpty-Dumpty if the king's horses and men had been more
successful.
c,
.",. DQtum
Fig. 2-13 A linear network.
Logronge', eqIJOtion 77
These constraint equations are holonomic and can eMily be integrated. The magnetic
energy-state function requires only A. and >.e, so by integrating the first and third rela
tionships we have
A.(t) - A,(t) ,
By substituting th, relationships into the state ~unctions, we have
W~()",,)..,) .. -tCI).." + -tC.)..,' (2-63)
W ..( 'A',.
A) - A.'
2L,
+ (A. 2L,
-- A,)' (2-64)
with current sources into the respective nodes taken 118 positive and out of the nodes
118 negative.
By using Eq. (2-62), tie have
For 11: - 1:
d[OW:()",,)...>] (lW..(A.,A.) ... Q
dt (1)..1 + (lA, I
0)...
oW...(A.,A.) ... ~ A, -- A,
+
(lA. L . · L,
A, A, AI
.
Thus, Eq. (2-62) with 11: ... 1 gives us, finally,
..
C,A, + 1:; + L. -- L. ... i,(l) -- i.(t)
after we take the total time derivative and substitute for Q,. Thus we have the
correct nodal equation for node 1.
For 11: ... 2:
Equation (2-62), with 11: ... 2, gives the second nodal equation 118
where the /j's are the loop currents in the network. Substituting this
Lagrangian into Lagrange's equation (2-42) gives us Eq. (2-55). This
latter equation is the proper energy-state formulation yielding loop equilib
rium equations.
For a nodal formulation the proper Lagrangian is given by
", (.x,x,)..,
"
q't) -- T'('X,x, t) W~(q)
+ W;()..) loop
(2-67)
nodal
where the upper line indicates a loop formulation on the electrical part of
the system and the lower line is for a nodal formulation. The total sys
tem energy-state function indicated by script 'l) is given by
loop
(2-68)
nodal
The total system Lagrangian is defined by
.c (x ,x')..,A,
4,Q't) = (x ' x')..,q't) - (x 'A,q't)
3' 'l)
loop
nodal
(2-69)
again using the upper set of symbols for a loop formulation and the lower
set for a nodal formulation. The equilibrium equations for the system
are then obtained by using Lagrange's equation, which, in a general form,
is given by
for this element, shown in Fig. 2-14b, relates the restraining force p to the
velocity i of one side of the damper relative to the other side. An expres
si~m for p as a function of i can be written, and the force as a function of
velocity can then be included in the equilibrium equations. Such a
procedure is familiar to us from previous work in mechanics. For the
electrical resistance or admittance element the same idea can always be
used. For a loop analysis the relationship expressing the terminal poten
tial across the dissipative elements as a function of the loop currents is
needed. This potential term is then properly inserted into the appropriate
loop equations. For a nodal analysis the current through the dissipa
tive elements as a function of the node voltages is placed into the proper
nodal equations.
A second approach can also be used to introduce the dissipative terms
into the equilibrium equations. The defining-characteristic curves for
the dissipative elements show the first time derivative of one basic coordi
nate as a function of the first time derivative of the other basic coordinate.
Thus for the translational damping element shown in Fig. 2-14b the force
p is plotted as a function of the terminal velocity i. For the electrical
admittance or resistance element the current q is plotted as a function of
the terminal potential X, as shown in Fig. 2-15. The state point P defines
a particular state of the element where the current is q and the terminal
potential is }.. (both without the primes).
For the electrical element, the Rayleigh dissipation function is defined
as the shaded area above the characteristic curve in Fig. 2-15. Thus for
the state point P the electrical Rayleigh function is given by
(2-77)
(2-79)
(2-80)
In both Eqs. (2-79) and (2-80) the possibility of having a lllutual resistance
element is included. For example, Eq. (2-79) has the potential across the
ith element as a function of all the n terminal currents. Normally, the
voltage across a resistance element is only a function of the current in that
single element.
In a similar fashion Rayleigh and co-Rayleigh dissipation functions
can be formulated for mechanical systems in translational and rotational
motion. For a system of n lumped viscous damping elements the mechan
ical Rayleigh and co-Rayleigh functions are defined as follows:
. ) " (in.,(.t .,
\'
,1'" = "")0 Xi 1'hP%, (2-81)
..
i-I
F'"' (.Xl,X2,
. • • •
.)
,X,. = \' )(t.
"" "(" .,
0 Pi Xl,X%,
.,) .> ...'
• . • ,X" (.I.;l;i (2-82)
i-I
and
(2-84)
(2-85)
(2-86)
Example 2·7
A P-N junction diode is shown in Fig. 2-16a. The terminal characteristics for the
dioue arc shown in Fig. 2-16b. The terminal voltage is assumed to be limited such
that reverse breakdown in the Zener region does not occur. The characteristic shown
in Fig. 2-160 can be described by the equation
r 1 N
i.
P
+
Zener
breakdown
(al (bl
Fig. 2-16 (a) A P-N junction. (b) Its terminal ehara.eterietic.
86 Principle, of electromechanical-energy conversion
it ~ -----i'" ~
i~Pl {i ip",
m, mechanical dampers
(both tranSlational and
j { 10 , 10", j
rotational)
n, eleCfrical admi!!ances
t
Fig. 2-17 A system of linear meehani('ul and electrical dissipative elements.
which reduces to
Example 2·8
For the mechanical system shown in Fig. 2-18 write the equilibrium equations by
using Lagrange's equation. Include the force of gravity on each of the two m&88e8.
The coordinates x. and x, are 80 designated that, when x. - x, - 0, spring K. is
stretehed a distance a, spring K, is compressed a distance b, and spring K, is stretched
a distance c. Motion is constrained to be in only the vertical direction.
Solution: Because of the simplicity o{ this problem, let us not define terminal
variables for each element. In other words, let us just formulate the state functions
directly in terms of x" Xl, x., and t" which are clearly a set of generalized coordinates.
The coenergy (unction for the system is given by
3'(x,.x,t) - j-M,x,' + iM:i:.'
and is actually only a function of the x's.
For the total energy (unction we have
1 1 1
'll(x.t) - 2K, (x. + a)' + 2K. (x, - x, - Ill' + 2Ka (x, + c)'
which depends on only x for this case. Each of the terms in parentheses represents
the displacement of one end of the spring with respect to the other. Combining
the coenergy- and energy-state fl!nctions, the Lagrangian for the system is given by
Notice that each term in the Rayleigh function is simply one-half the damping
coefficient D times the square of the velocity of one side of the damper with respect to
the other side. This represents one-half the power dissipated by each damping
element.
The external forces are as follows:
Q. - M.g -- f(t)
where g is the acceleration of gravity.
Now, using Eq. (2-88) with k - I, we need
(J~ _ M,x,
(Jz,
ii£ 1 1
(Jz, - -- K, (z, + a) + K. (z, -- z, -- b)
a~ _ Dd, __ D,(:t. - x,)
(Jz,
which combine, according to (2-88), to give
M,i, + i.. (z, + a) + ~. (x, + b z,) + D,x, + D,(xl -- x,) - M,g (2-93)
x, - X, and z, - X,
where X, and X, are constant!!. With the forces of gravity as the only external fornell,.
.
(2-95)
it is certainly reasonable to assume that the system with damping will reach some ,
constant, non-time-varying configuration. Substituting Eqs. (2-95) into the equi~"
librium equations (2-93) and (2-94) gives us . ~
1
K. (Xl + a) + K,1 (Xl + b - X,) - MIg
1 l '
Logrange', equotion 89
+
+
vlttlr:
"Ql
Fig. 2-19 A lumped-lin
ear electrical network.
with all time derivatives of constants equal to zero. If the spring compliances, the
two m8.8EleS, and the constants a, b, and c are known, the static configuration of the
system can be determined.
Example 2-9
I·'or the lumped-linear electrical network shown in Fig. 2-19, determine the equilibrium
equations on a loop basis by using Lag.range's equation.
Solution: By using the three generalized loop current coordinates. shown in Fig.
2-19, the following total state functions are determined.
Total coenergy function:
The external forces. or voltage sources for a loop formulation, are given by
Q. - -Vt(t) + v.(l)
The total Lagrangian for this system is
90 Principle. of e/edromechanical·energy conversion
Using Lagrange's equation, given by (2-88), we need the following terms for
k - 1:
Combining these terms with the external source Ql according to Eq. (2-88) gives us
L,(lb - ii.) - M(ij, - ij.) + -(~ ql + -c~- (q. - q,) + R.(q. - q,) + RMI - q,)
J' a
- VI(t) - v,(t) (2-96)
aftet we take the total time derivative d/dt required by the equation. Equation (2-96)
is readily seen to be the correct loop equation for the first loop.
By using the Lagrangian, the Hayleigh dissipation function, and the external
sources, derive the remaining two equilibrium loop equations. At this point in the
development of these energy techniques, it is important that you be able to manipulate
these functions properly, particularly in regard to partial and total derivatives.
Therefore, take flufficient time to complete this example carefully.
Notice in this example that we have written the state functions :I', 'll, and !f
directly in terms of the loop variahle8 without defining a preliminary set at the terminals
of each element. With a small amount of experience in settihg up these functions for
simple linear systemll, a great deal of time can he saved hy rememhering the form of
these functions. Their original definitions, given by TaMe 1-3, are important when a
new, unfamiliar situation is presented.
Example 2-10
For the lumped-linear network shown in Fig. 2-20 determine the equilibrium equa
tions on a nodal basis by using the Lagrange formulation.
Solution: For the nodal variables shown in Fig. 2-20, the Lagrangian and Ray
eigh dissipation functions are given respectively by
1 1 1 1 1
,£, - 2 CIX,' + 2 C.(Xs - X,), - 2L. Xl' - 2L. (X, - X.), - 2L. XI'
!f -4Gl~'lS + j-Gt(X, - X,), + jGaX.' + jG.(X, - X,), + jG,(X, - Xa)2 + lG.x,'
The external forces, or current sources for a nodal formulation, are
Qt ... it(t) Q, == it(t)
The first nodal equation is obtained by letting k = 1 in Eq. (2-88). For k =- 1
we require the following terms:
t
I By substituting into Lagrange's equation, we have
I
I
l
including the external for('e Q.. This equation is readily seen to be the correct nodal
equation for the first node.
Just for practice, complete the example by obtaining the remaining two nodal
equations.
2·11 SUMMARY
i Let us see where ~e now stand. Our objective has been to develop a
systematic procedure for determining a consistent set of equilibrium equa
PROBLEMS
2-1 The rotational-mechanical system shown in Fig. P2-1 consists of two fly
wheels with inertia.s J\ and J, coupled by a shaft having a. compliance K\. A third
Fig. P2-1
lagrottge'$ equation 93
Fig. P2-3 y
2-4. Formulate the equilibrium equations for the system shown in Fig. P2-3.
Assume that the top of the spring is moved vertically up and down according to the
equation y, - a sin wt, where '" is a constant angular frequency, a is II. constant ampli
tude, and 1/1 is the distance from the top of the spring to the origin of the X1/ coordinate
system.
2.5 A spherical pendulum, shown in Fig. PZ-5, consists of a mass 11'/. suspended
from a frictionless pivot by a weightless rod of length I. The pendulum is not con- .
strained to swing in a plane. The only constraint is that the mass m rema.in on the
surface of a sphere of radius I.
a. In terms of the xyz coordinates, write the kinetie-coenergy-etate function for
the system.
94 Principles of e/ectromechc:mical.energy conversion
b. Explain why the xyz coordinates are not a suitable set of generalized coordi
nates. Show that the two spherical-coordinate angles (6,4» are a suitable set of gen
eralized coordinates.
c. Write the constraint equations between the two sets of coordinates. Are
these equations holonomic? Why?
d. Determine the Lagrangian for the system in terms of the generalized
coordinates.
,/ y
/
/
/
/
"
----y /
l
Jf
m Fig. P2-5
2-6 A flyball speed governor is illustrated in Fig. P2-6. The four rods of length
l are pivoted at the three point masses and at the origin. The rods have negligible
mlU!8. The system rotates about the x axis, causing the two identical masses M. to
move away from the axis and thus pull mlU!8 M 2 toward the origin against the spring
having compliance K. The position of Mt is a function of the angular velocity of the
system; thus, the device can be used as a speed governor. AU pivot points are fric
tionless, andM. slides along the x axis without friction. When the two maBSe8 M.
are on the x axis, the spring is at its free length.
a. Write a Lagrangian for the system in terms of the three cartesian coordinates
x, y, z. The z axis is out of the paper.
b. Can we use the Lagrangian of part a in Lagrange's equation? Explain your
answer.
Fig. P2-6
l Lagrunge', equation 95
Fig. P2-7
2-9 The four elements shown in Fig. P2-8 a.re joined to form the network shown
in Fig. P2-9.
G. Why a.'re the origina.l termina.l currents q., q~, q" and q4 not suitable genera.lized
coordina.tes? Ca.n loop currents q, and ti. serve as generalized coordina.tes? Why?
b. Write four constraint equa.tions expressing the origina.l currents in terms of the
loop currents til and tho Are these constraint equa.tions holonomic? Expla.in your
a.nswer.
C. Substitute the constra.int equa.tions into the La.gra.ngian found in Prob. 2-8b
a.nd find the loop equilibrium equa.tions by using La.gra.nge's equa.tion.
L,
.~C'
Fig. P2-9
YIlIQ 0
96 Principles of e/ectromeehollieal.enerllY cOllversioll
2-10 Repeat Prob. 2-9 with the four elements connected into the network shown
in Fig. P2-lO.
Fig. P2-10
2-b The four linear-lumped elements shown in Fig. P2-8 are now joined to form
the circuit shown in Fig. P2-11. Assume that the mutual-coupling coefficient Mis
equal to zero.
a. Write 'the electric-coenergy. and magnetic·energy-state functions in terms of
the individual voltage variables he, }.bJ he, and }.•.
b. Write a Lagrangian suitable for a nodal analysis in terms of these four voltage
variable!! of part a.
c. After the elements are connected into the circuit of Fig. P2·11, can the four
variables of part a serve as generalized coordinateS in Lagrange's equation? Why?
d. Write suitable constraint equations between the nodal variables and the
original four variables.
11. Transform the Lagrangian in part b in such a way that it becomes a function of
generalized coordinates, 'and use Lagrange's equation to find the two nodal equations
for the network;
Fig. P2.11
2.12 For the linear mechanical and electrical systems shown in Fig. P2-12 write
an appropriate Lagrangian and Rayleigh dissipation function and determine the
equilibrium equations by using Eq. (2-88). Do you notice any similarity between
these three systems? Hint: Look at the state functions.
2-13 A certain fan can be described by the following torque-speed data:
Speed, rpm o 200 400 600 800 1000 1200 1400 1600 1800
Torque, lb·ft o 0.4 1.0 1.7 2.7 3.7 5.0 6.7 8.5 11.2
a. Draw the characteristic curve for this diBBipative element. Use mks units.
b. For a speed of 1500 rpm find the magnitude of the Rayleigh and co-Rayleigh
dissipation functions.
c. Evaluate the rotational viscous damping coefficient D, at 1500 rpm. Compare
jD",,' with the state functions found in part b.
Lagrange'$ equation 97
Ie)
Fig. P2-12
!
Gravity
Fig. P2-14
98 Principles of electromechanical-energy conversion
FORMULATION OF EQUILIBRIUM
3
EQUATIONS FOR
ELECTROMECHANICAL
SYSTEMS
;
,..---....,........::
Notice that the device has a set of electrical terminals in addition to "
movable I1Hl(!hanical member. The term port is generally used to describe
the place in the system where either electrical or mechanical energy can be
supplied or extracted. The system in Fig. 3-1 has a single electrical port
and a single mechanical port. The voltage and current variables for the
electrical port are designated by ). and q, respectively. Similarly, the
velocity and force variables for the single mechanical port are given by
x and p, respectively. The velocity x of the movable member is meas
ured with respect to the stationary member, and x is taken equal to zero
when the air gap betwecn the stationary and movable members is equal
to zero.
Now we would like to examine the various state functions describing
the magnetic-field system shown in Fig. 3-1. For this purpose, let us
assume that the velocity :i; of the movable member is equal to zero. In
other words, the movable member is held at some fixed position x. As
far as the electrical portion of the system is concerned, we now have a
simple lumped-inductance element. As discussed in Sec. 1-2, this element
is conveniently characterized by a plot of the total flux linkages X as a
function of the coil current q. Figure 3-2 shows typical characteristic
plots for various fixed positions of the movable member. These positions
are designated by x"' Xb, and xc. Thus, with a current qb and with the
movable member held at X6 the state of the system is described by point B
in the X'-Ij' plane of Fig. 3-2.1 Holding the current constant at the
value qb and repositioning the movable member to position x" causes the
state of the system to move to point A with an increase in the flux linkages
to a value of X For any final current q and position x a single point on
G•
the plane of Fig. 3-2 is determined. In fact, specifying two out of the
three variables X, q, and x is sufficient to describe the particular state point
of the system. The plot of Fig. 3-2 gives the third variable once the other
two are specified. Notice that nothing has been said about how the state
1 The primes on the coordinates are used to indicate integration variables. Coordinate
symbols without the primes refer to system state points, or state values of the
coordinates.
\
point ls reached. The only thing that matters in determining the state
point of the system is the particular final values of ~, q, and x, and not how
these values were obtained.
With the mechanical member held at a fixed position x the magnetic
field energy storage can now be computed in the usual way as being
in accordance with Sec. 1-11. The only difference here lies in the fact
that the current q is a function of not only the flux linkages>. but also the
fixed position x of the movable member. Thus the magnetic-energy
storage W .. becomes a function of both ~ and x. The integral given by
Eq. (3-1) is readily interpreted as the shaded area above the particular
characteristic curve for the particular final position Xb, as shown in Fig.
3-3. The magnetic energy stored in the system, for a given final state of
the system, is a function of only the final state coordinates. These state
coordinates define a state point, a.nd the shaded area above the curve
going through this state point is the energy stored in the system.
Equations (3-1) to (3-3) are not very different from the corresponding
equations for the simple electrical-inductance element. The mechanical
part of the system offers a small additional complication in that the
position coordinate influences the state (unctions. However, in all these
integrations x is treated as a constant and is simply carried along, and it
appears in the final results.
Exomple 3·1
The device pictured in Fig. 3-4 is a common electromechanical actuator, or force gen
erator. Hupplying a current to the electrical winding will cause the plunger to tnove
upward, thereby pulling the mechanical load to which the plunger might be attached.
The device provides a means of controlling mechanical motion by an electrical signal.
Connecting the plunger to the valve stcm of a hydrl\ulic or pneuml\tic valve gives us a
remotely operated, electrically controlled valve. Connecting 0. pair of electrical
contacts, one to the stator and one to the plunger, makes this device the common
electrical relay. A very large number of rommon devices are founded upon the simple
mechanism shown in Fig. 3-4.
Let us, at this poil\t, find the magnetic--energy- and -eoenergy-state functions for
the electromechanical system shown in Fig. 3-4. Assume the permeability of the iron
r----.,
I
I
I Iron
t stator
I
I
I
Contour • Iron
plunqer
Fig. 3-4 An electromechanical sys
tem with magnetic-field energy storage.
10" Principlea of electromechanical· energy conversion
ill many orders of magnitude greater than that of air. Also, take the permeability of
the brMII to be equal to that of air. Neglect all fringing fields on the ba.sia that the
air-ga.p length x and thiekness d of the bra.ss are small eompared with the radius of
the plunger.
Soluli<m: The energy- and eoenergy-etate funetions ean be readily computed onee
the eharacteristie eurves for the deviee are known. We require the total flux linkages
X in the eoil a.s a function of the eurrent I} in the eoil and the Position x of the plunger.
The flux link:agee are determined in the following manner.
The magnet ie-intensity vector in the bra.ss and air ean be found by using Amp~re'8
eircuitallaw, whieh can be written in the form
¢c B • d1 - Is J . dS
This vector equation tells us that the line integral of the magnetic-intensity vector H
arounlany e10sed contour C must equal the total current croesing a surface 8 with
this eo tour a.s its periphery. When the contour shown in Fig. 3-4 is ehosen, Amp~re's
eircuital law becomes
H.x + H"d - -Nq
where H.a. and H"a. are the magnetic-intensity vectors in the air and bra.ss, respec
tively. The unit vector a. is positive in the positive x direction. Notice that we are
traversing this contour in the positive x direction, as shown by the arrows on the
contour in Fig. 3-4. By placing a surface over the contour, we !lee that the total eur
rent NI} croeses the surface in a direction out of the paper. By the usual right-hand
convention, currents into the paper are positive for the contour direction we are using. 1
Therefore, the total current NI} croesing the surfaee ha.s a negative sign in Amp~re's
circuital law. The magnetic-field intensity in the iron is taken to be zero compared
with that in the bra.ss or air. This approximation is based on the permeability of
iron being many orders of magnitude greater than that of bra.ss or air.
Neglecting fringing, the magnetic-f1ux-density vector in the bra.ss equals that in
the air gap between the plunger and the bra.ss. Therefore, the magnetic-f1ux-denllity
vector is given by
B - ,.,II.a. - ,.,II"..
where the permeability of air and bra.ss is taken to be equal to,.". By substituting
into Amp~re'8 cireuitallaw, we have for the magnetic-f1ux-density vector
B ___ ,.oNl} ..
d+x
The total flux ,., existing in the bra.ss and in the air gap is defined by
,., - IsB'dS
where the surface 8 croeses through either the bra.ss or the air gap. By evaluating the
surface integral, we have
,., _ __ ,.oNAI}
d+x
I Place the fingers of your right ha.nd along the contour pointing in the direction of
the arrows. Your thumb, which points into the paper, gives the correct convention
for positive currents.
EI.ctromechon;cal••yllfem equofiOM 105
where A is the face ares of the plunger or the brass, as shown in Fig. 3-4. Positive
flux is taken in the positive z direction; thus, the negative sign indicates that the flux
is in the negative z direction.
Since all of the flux. links each of the N turns, the total flux linkages are given by
(3-4)
where the negative sign arises from the fact that A, as shown in Fig. 3-4. would become
positive if the flux 'I' were increasing in the upward or negative x direction. In other
words, A equals -N d./dt. since. is positive for Rux in the positive x direction, or
downward.'
Figure 3-5 shows a plot of h as II. function of q for various fixed values of x. Notice
that h is a linear function of q; thus, all these characteristics are straight lines.
From Eq. (3-2) the ma.gnetic-coencrgy-state function is given by
k(h ) _ (d + z».
'I ,x PeNtA
, This discussion of signs is usually quite confusing. An easy check is simply to use
the right-hand rule as follows. Curl the fingers of your right hand around the coil in
the assumed direction for positive current. Your thumb, which now points upward,
gives the corresponding flux direction for positive current. This direction corresponds
to the positive sign on >.; therefore, h should be positive if 9 is positive, as given by
Eq. (3-4).
106 Principle. of electromechanical-energy con"e,.,ion
W...(A,:l;:) - 2p..N"A
If the current q(A.:l;:) i8 substituted into the expression for the coenergy-state function
given by Eq. (3-5), we see that
w... - W~
This result is to be expected, since the A versus q characteristics shown in Fig. 3-5 are
all straight lines. For any state point P the energy is equal to the area of triangle
OPQ and the coenergy is equal to the area of triangle OPR. Obviously, area OPQ
equals area OPR, since each equals half of the area of rectangle OQPR. Also, we can
show that
following the general form of Eq. (1-56), where the mechanical ports are
not present. Notice in the first integral of Eq. (3-6) that all At coordinates
except A~ are zero. The second integral holds A~ constant at its final
value AI with A~, A~, ..• , A: still equal to zero. The same sequence con
tinues for each successive integral. Again let us remember that even
though the system does not reach its final state (as represented by AI,
Xl, . • • ,X",XIIX2, • • • ,x"') in such a manner, the energy-state function
calculated this way is still correct, since it depends only on the final state,
and not on how the system got there. When we use the abbreviated
Electromechanical-system equation, 107
notation introduced in Chap. 1, Eq. (3-6) for the magnetic energy stored
becomes
+ )0( q, "2
1 (. .1 °
~ Ql,Q" . . . , ,X"Xt, .•. ,x.. ) d Q,
.1
Here we have fixed each of the m mechanical coordinates with all currents
equal to zero. Then, in sequence, we increase each current to its final
value and hold it there as we adjust the succeeding current.s. The coen
ergy expression can be abbreviated simply by
.
lV~(q,x) = I loti' A~(q',X) dq:
i-I
(3-9)
Example 3-2
A particular magnetic-field electromechanical device having two mechanical ports and
three electrical ports has the following characteristic relationships:
>-.(q"q"q.,x,x,) - .!
X,
ql + ~
XIX,
q, + 9xl'x,"h
We have fixed the mechanical coordinates at some particular values x. and x.. Then,
in succession, we increase the currents q;, q;,
and q;
from zero to their final values,
given by ql, ql, and q., respectively. The state of the device at any time instant is
given by specifying q.. q., q., x., and x.. Analytically we have taken the system to its
final state and computed the coenergy in the process. The resulting expression for the
coenergy is correct even though the system does not actually reach the final state in this
manner. On evaluating the coenergy, we have
As an additional exercise let us evaluate the coenergy by raising the currents from
q;
zero to their final values, taking first, then q;,
and finally q;.
The general coenergy
expression, from Eq. (3-9), now becomes
' . ('I"., I
W .. (q,x) - )0 X.(O,O,q.,xt,x.) d4. + )0(v, X.(O,q.,q.,x.,x.) dq.
I I ,
('ii
+ )0 I I I
X,(q"4'.,q.,x,,x.) dq,
When the defining flux-linkage relationships are substituted, the coenergy is given by
When these three integrals have been evaluated, the resulting coenergy is seen to be
identical with Eq. (3-12).
The original characteristic relationships, given by Eqs. (3-11), have a certain
symmetry. Notice that the contribution to X, from current q. is identical to the con
tribution to x. from q,. A similar statement can be made for all other mutual terms,
or terms involving flux linkages in one portion of the system due to currents in another
portion. This property of too defining relationships must be true in order for the state
functions to be independent of the path by which the final state is reached.
If the magnetic-energy-state function were required, then Eqs. (3-11) would have
to be solved for the currents in terms of the flux linkages in order to use Eq. (3-7) for
the energy-state function. The defining relationships of (3-11) make this look like a
very tough job. A simpler procedure would be to use Eq. (3-10), which gives the
magnetic-energy-state function for this device as
Substituting Eqs. (3-11) for }.I, }.o, and }.I and Eq. (3-12) for W~ would give the mag
netic-energy-etate function in terms of the currents ql, ql, and q, and mechanical
coordinates :1:1 and :1:,. This is not the usual functional representation of W.., since
normally the flux linkages rather than the currents are used. However, with the
defining relationships (3- \I) 80 difficult to manipulate, we are practically forced to
accept W .. as a function of the q's and x's rather than the }.'s and x's.
This discussion is included to point out a slight difficulty that we have not, as yet,
faced with our energy formulation. The difficulty arises when the energy and coenergy
are correctly formulated but are functions of variables differing from their defining
relationships. As we proceed with the energy formulation, keep this problem in mind
and we shall see what its consequences and the required correction factors are.
For state point B in Fig. 3-8 the coenergy integral is the area in the
ql> by ~ rectangle below the x' Xo curve. The energy and coenergy
defined by the integrals of Eqs. (3-13) and (3-14) must give the total area
I For an excellent discussion of flux plotting see R. P!onsey and R. E. Collin, "Principles
and Applications of Electromagnetic Fields," chap. 5, McGraw-Hill Book Company,
New York, 1961.
Electromechanicol-,y"em equation, 111
of the rectangle when t.hese state functions are added together. Thus we
can, as usual, write
W.(q,x) + W;(X,x) (3-15)
showing that the energy and (!oenergy are not independent state functions.
fxampl.3-3
An electrostatic voltmeter is shown pictorially in Fig. 3-9. The device consists of a
set of stator plates electrically connected together. A set of rotor plates, al80 con·
nected electrically together, are interspaced between the stator plates as indicated in
the figure. However, the stator and rotor are electrically insulated from each other.
The rotor plates are mounted on a shaft to which are attached a torsional spring and
a pointer. A later example in this chapter shows how the device can serve as a volt-
Pointer
LII Ir----""'!{---:"",-,...,,.:
A
....f-+'--- • + + +
1------';;.;;,-;;;..;;;
..
• + ....
........
.. .. .. +
.... + ..
.. .. + ..
-
q
.. . . .. .. .... +
Stator
plotes ..,..-Pivot
Rotor
plates
meter. For the time being, let us simply determine the electric-energy- and -coen
ergy-6tate functions for this electric-field electromechanical device. Assume that
there are four stator plates and four interspaced rotor plates. Neglect fringing fields
at the edges of the plates.
Solution: The energy- and coenergy-state functiQns can be readily determined once
the characteristic curves for the device are known. These curves are plots of the
charge on either the stator or rotor as a function of the voltage applied between stator
and rotor for various fixed values ofthe angular position of the rotor. In other words,
we need the function q().,II).
Figure 3-10 shows an edge view of the device. Suppose we place a positive
charge q on the rotor plates by means of the current q indicated at the terminals of the
voltmeter. Then an equal amount of negative charge will be found on the adjacent
IlUrfaces of the stator plates. The charge distribution is shown in Fig. 3-10 by the
respective plus and minus signs. GaU88's flux theorem can be used for determining
the electric field in the space between the stator and rotor plates. This theorem can
be given in integral form by
D. [
a +
(...R,· - ... R,') -z;- II] - 14q
where D. is the magnitude of the electric-displacement vector in the direction of the
unit vector a.. The bracketed term OD the left side of the equation is simply the
adjacent area between the stator and rotor plates in the space labeled A. Refer to
Fig. 3-9 for the definitions of R" R" and a. The charge q is divided into 14 equal
Electromechanica/.system equotions 113
parts on the rotor plates, as shown in Fig. 3-10.' The electric-field vector in the space
A is therefore given by
E-~- q ....
'0 7' 0(R,' - RI')(OI + e)
where <0 is the permittivity of air, equal to 8.854 X 10- 11 farad/m.
The voltage of the rotor with respect to the stator is designated in Fig. 3-10 as~.
By definition, the potential difference is related to the electric-field vector by the
following line integral:
~ - - Ie E·dl
where the contour C starts at the stator plate and ends at the rotor plate. In the
IIpace A we can choose a simple straight-line contour; thus
d1 - -/h ••
The minus sign is necessary because the selected contour goes from the stator to the
rotor, which in the space A is in the negative x direction. Evaluating~, we have
>- _ _ ~_~_!Lh____
where h is the separation between stator and rotor, as shown in Fig. 3-10. The char
aeteristic curvCl! of the electrostatic voltmetcr arc thcrefore obtainable from
W (>-)
•' ,e - 7••(R." - RI") (
2h a + 9)>-. (3-17)
We can easily show, by substituting Eq. (3-16) into Eq. (3-18), that here W. - W:.
The equality of the energy- and coenergy-state functions is expected, since q is a linear
function of >- for fixed values of Ii. Thus the characteristic curves would be straight
lines. We can also verify Eq. (3-15), which shows that the sum of and W.
must W:
equal q>-. This last step is left for you to perform as a brief exercise.
(3-19)
I
This equation for the electric-field energy stored is identical with Eq. (3-6)
if ). and q are replaced in (3-6) by q and X.
The first integral on the right-hand side of Eq. (3-19) has all q' coor
dinates except q;
equal to zero. Therefore, this first integral evaluates
q;
the energy stored as is raised from zero to its final value ql with all other
q' coordinates equal to zero. All the mechanical coordinates are held at
their final values. The second integral has q;
held at its final value ql
with all other q' coordinates still equal to zero as q; is' increased from zero
to its final state value q2, thus evaluating the contribution to the final
energy stored during the second process. The same scheme then con
tinues for q~, q~, and on up to q:. Throughout the entire process the
mechanical coordinates are held fixed at their final state values. Since
the energy stored is a state function, the result obtained by Eq. (3-19) is
identical with that obtained by the coordinates reaching their final state
values by means of any other path. Equation (3-19) can be written more
conveniently by using the abbreviated notation introduced in Chap. 1.
Thus, we can write
n
W.(q,x) = I
i-I
q
fo , X:(q',x) dq: (3-20)
W~(X,x) = i
i-I
fo'A' q:(X',x) dX: (3-21)
Elec,romechonica/ •• y.,.m equation. 115
Thus, if we can calculate either the energy or the coenergy, then we can
find the other state function. If the q' versus }", characteristics are all
straight lines for various constant values for the mechanical coordinates,
then the energy- and coenergy-state functions are equal. However, if the
device is not linear, then these two state functions are not in general
equivalent, or equal, and, as we shall see, care must be exercised to select
the proper function.
Electric
Fig. 3-11 A two-port \0881_ electric- Electromechonicol device ,,'
field electromeehanicalllYlltem.
116 - Principl.s of electromechanical· energy conversion
p. = Xq (3-23)
p ... = xp (3-24)
1 A pra.ctical inductive element stores magnetic energy but also has some dissipa.
tion. UsUAlly we account for this dissipative component by placing a resistance in
series or pa.rallel with a lossless inducta.nce. The II!.me procedure is used with a
practica.l capacitive element.
Electromechanical''Y.tem equation. 117
dW (q x) = aW.{q,x) dq
• , aq
+ a~-"(q,~
ax
dx (3-26)
{3-27}
Fig. 3-12. If the system is operating at state point A, then the energy
which reduces to
I. aW.(q,x) I
P= ax (3-29)
q'
order for the member to be held at zero velocity, the external force p
would have to be equal and opposite to the force of electrical origin.
Figure 3-14 illustrates this situation, withf. as this force of electrical origin.
The force of electrical origin can be determined by using Eq. (3-29) as
being
(3-30)
since p = f.. Notice in Fig. 3-14 that the force of electrical origin is
positive if it is in a direction opposing the assumed positive direction for x.
In other words, positive f. tries to make the velocity of the mechanical
member negative according to the sign convention being used.
The procedure we have used to determine the force of electrical origin
bears the imposing title of the principle of virtual work. Since the velocity
of the mechanical member is held at zero, meaning x = there can be no 0:
displacement or change in x. Therefore, no work can actually be done,
or no energy can be supplied to the system through the mechanical port.
However, in order to achieve the result of Eq. 3-29, we assume a virtual
or infinitesimal displacement in x as p struggles to hold x as near to zero
as possible. We are always sufficiently successful in this manipulation of
p that the mechanical kinetic energy is negligible, but not quite successful
enough to have dx equal to zero, causing a question to be raised concerning
its cancellation in going from Eq. (3-28) to Eq. (3-29).
at state point A, the energy, area OAC, plus the coenergy, area OAD, sum
to give the rectangular area ODAC.
The total differential of the energy-state function can be expanded
as follows:
dW.(q,x) = d[q>.. - W;(>",x)]
which reduces to
. aW;(>--,x)
p = - ax (3-36)
q'
f. = _ aw~(x,x)
(3-37)
ax
as the force of electrical origin from the coenergy-state function.
Notice the negative sign in Eq. (3-37). Figure 3-15 shows the coen
ergy increases as x is increased by an amount ax. On the other hand,
Fig. 3-13 shows that the energy-state function decreases (i.e., the change
in the energy is negative) as x goes to (x + ax). Therefore, a negative
sign must appear in Eq. (3-37) for f. in terms of the coenergy if it does not
appear in Eq. (3-30) for the force of electrical origin in terms of the energy
state function.
There is one further word of caution concel1ling these results.
Observe that the energy-state function in Eq. (3-30) is a function of q
and x. The aW.(q,x)/ax requires that q be held constant. Therefore W.
must be a function of only q and x. Expressing W. as a function of X
and x would not, in general, give the correct force of electrical origin.
The sanJ(lll'9;rgument holds for expressing the coenergy W~ in Eq. (3-37) as
a function of only Xand x, not q and x. In Sec. 3-10, we shall reformulate
the force of electrical origin by using the energy as a function of X and x
and the coenergy as a function of q and x. Sometimes these functions can
be calculated only in this form, and in these cases it is important to be able
to find f. from the reversed functional dependences. For the time being
let us work with the more straightforward forms given by Eqs. (3-30)
and (3-37).
Let us now generalize the results of Secs. 3-5 and 3-6 for an electric-field
system having m mechanical ports and n electrical ports. Applying the
first law of therodynamics to such a system gives us
n '"
L Xi dqi + L Pi dXi = dW.(q),q2,
i-I i-I
. . . ,q.. ,XI,XS, ••• ,x...) (3-38)
again assuming that all energy supplied to the electrical and mechanical
ports goes into electric-field storage. The first term on the left-hand Side
of Eq. (3-38) represents alfUteeileTgy pouring in through the n electrical
ports. The second term is the energy through the m mechanical ports.
The right-hand side is the change in the electric-field energy storage.
122 Prin<;ip/e. of eledromeehoniccr/-energy conversion
~
'-
[ .
Pi -
oW.(q,x)] d
ox- Xi
0 (3-42)
i-I '
The only way to satisfy this equation for any arbitrary set of virtual dis
placements dXi is to have the bracketed term equal to zero for each i.
Therefore, we have
We can also calculate the forces of electrical origin by using the coen
ergy-state function instead of the energy-state function. Equation (3-22)
gives the following relationship between the energy and coenergy:
n
W.(q,x) + W:O"x) = L q,'''i
,-1
(3-45)
I • ••
Be sure that you understand this expansion of the total differential of Eq.
(3-46) to give Eq. (3-47). This is not trivial, and it is a very important
step.
By using Eq. (3-21), we can show that
aW;(X,x}
~~--- = q. (3-48)
a>.;
Now we substitute Eq. (3-48) into the third term on the right side of Eq.
(3-47) and ·cancel the first and third terms. Substituting the remaining
expression for dW. into the first-law formulation of Eq. (3-38) gives us,
finally,
I [pi + ~w;~>.,x}]
i ... l ...
dx. = 0 (3-49)
after canceling like sup1mations on both sides of the equation and slightly
manipulating the result. As before, in order to satisfy Eq. (3-49) for all
dx. virtual displacements, the bracketed term must itself be equal to zero.
Therefore, since (f.). = Pi, the force of electrical origin on the ith mechani
cal member is given by
. (I') = _ aW;(X,x) ~ = 1,2, . . . ,m (3-50)
j J.... ax,
Notice again the negative sign when the coenergy-state function is used.
Remember that the force of electrical origin is positive if its sense oppo8e8
the assumed positive direction for Xi.
Equation (3-44) gives us the forces of electrical origin calculated from
the energy-state function for an electric-field device having n electrical
ports and m mechanical ports. Similarly, Eq. (3-50) gives us these same
forces calculated from the coenergy-state function. The results from
these two equations are identical except for one small point. Equation
124 Principle. of e/ectromechanical·energy conversion
(3-44) gives all forces of electrical origin as functions of all the q's and x's.
This would be suitable for a loop formulation on the electrical part of the
system where the q's or q's are the coordinates of interest. In Sec. 2-6 we
showed that the electric-field energy W. is always used when loop equa.
tions are desired. Similarly, Eq. (3-50) gives the forces of electrical origin
as functions of the X's and x's, which is proper for a nodal analysis on the
electrical portion of the system. In Sec. 2-7 we demonstrated that the
el~ric-field coenergy-state function is used when a nodal analysis is
desired.
The pieces of this puzzle should be starting to fall into place. Notice
that no new state functions have to be introduced when electromechanical
interactions are present in the system. The only thing new is that the
electrical state functions are dependent on both electrical and mechanical
coordinates. Then partial derivatives of these electrical state functions,
taken with respect to mechanical coordinates, yield the mechanical-inter
action forces. All of the derivations in the preceding section have been
for a multiport electric-field system. Since a complete duality exists
between the electric- and magnetic-field state functions, we can predict
the magnetic-field results directly. The forces of electrical origin on the
ith mechanical member of a magnetic-field electromechanical system are
given by
(3-51)
(3-52)
where the 4's and x's are involved. As a review of the principle of virtual
work, carefully develop the magnetic-field interaction forces, given by
Eqs. (3-51) and (3-52), for a general n-electrical-port m-mechanical-port
system starting from the first law of thermodynamics. Also give some
personal thought to the virtual displacement concept as you perform this
derivation.
• :l!1
When a loop formulation for the electrical part of the system is desired,
the upper line of symbols is used; and when a nodal formulation is
required, the lower line of symbols is used. The total system coenergy
state function is designated by 3'; the mechanical-coenergy function is T';
and the magnetic- and electric-coenergy-state functions are given by W:
and W:.respectively. Remember that the mechanical position coordinate
x stands for both the translational- and the rotational-mechanical position
coordinates.
Equation (3-54) is identical with the original formulation of the total
system coenergy function for separate electrical or mechanical systems,
as given by Eq. (2-67). except for one small point. In Eq. (3-54) we are
including the possibility that the electrical-coenergy functions and W:
126 p,.inciple. of electromechanical·energy conve,.,;on
where :I' and 'U are defined by Eqs. (3-54) and (3-5.1), Now suppose we
have an electromechanical system with electric-energy storage only in
magnetic fields. In addition, we have decided to formulate our electrical
equilibrium equations on a nodal basis. For this particular case the
Lagrangian would be given by
.c(x,x,X,t) = [T'(x,x,t)] - [V(x,t) + W..(X,x)] (3-57)
f/ec'romechonic:o/·,y.'em equalion. 177
Notice that none of the electric-field terllls are present and that the mag
netic energy, as a function of>.. and x, is used, since a nodal formulation is
desired. Substituting the Lagrangian of (3-57) into F.q. (3-Sa) gives us,
for the kth mechanical node,
~ [aT'<x,x,t)] _ ~T'(x,x,t) + ~V(x,t) + ~W...(>..,x)
dt aXil: aXil: ax" aXil:
+ as:(~,).) = Q/t (3-58)
aXil:
The first three terms in Eq. (3-58) are the inertial forces and the
spring forces acting on the kth mechanical node. The as:/ax" adds the
dissipation forces, and QII: includes all non conservative external forces on
the kth mechanical node. The remaining term is the aW .. (>",x)/ax•.
Equation (3-51) tells us that this term represents the force of electrical
origin on the kth mechanical node. In a very natural way, it arises in the
mechanical-equilibrium equation in the proper place with the proper sign.
For the same magnetic-field electromechanical system, suppose we
desired a loop formulation of the ('lcctrical-equilibriul1l equations. The
Lagrangian would therefore he given hy
.£(t,x,q,t) = [T'(x,x,t) + W~(q,x)J - (V(x,t)] (3-59)
here using the magnetic-coenergy-state function W~. t The equilibrium
equation for the kth mechanical node, from Lagrange's equation (3-53),
would now hecome
!i [aT'(X,x,t)] _ ~T'(x,x,t) _ aW~(q,x) + ~V(x,t)
dt aXil: aXil: a:tk ax.
+ as:(~,q) = Q. (3-60)
ax.
This equilibrium equation is identical with Eq. (3-58) except for the force
of electrical origin. However, Eq. (3-52) has shown us that -aw:/ax/t
is the correct force of electrical origin when the coenergy-, rather than the
energy-, state function is used.
A similar exercise for a system containing an electric-field electro
mechanical Jumped element would again show that the forces of electrical
origin are correctly genel'ated and inserted into the equilibrium equations
by the Lagrangian method. The beauty of this whole scheme is that we
really did not have to alter Lagrange's equation in any significant way to
include the interaction forces. The electrical state functions became
slightly more involved for electromechanical elements, but the Lagrangian
formulation did not really change. The fact that the equilibrium equa- .
t Carefully examine Eq. (3-59) and be sure you understand each and every deta.il of
the terms and their functional dependance.
128 Principle. of electromechanical-energy convel"'ion
Example 3-4
For the electrostatic voltmeter shown in Figs. 3-9 and 3-10 determine the system
equilibrium equations by \l8ing a nodal basis for the electrical portion of the system.
Include a viscous damping torque at the rotor pivot points. Also include the possibility
of some ohmic loss due to a nonzero conductivity in the medium between the stator and
rotor members. Use Lagrange's equation to aid in this formulation.
Solution: The electrostatic voltmeter is shown schematically in Fig. 3-16. A
source, whose voltage is to be determined by the device, is connected to the electrical
tenninals. The ohmic loss due to the nonzero conductivity of the dielectric is repre
sented by the lumped linear admittance G, with units of mhos. The total viscous
damping baused by the pivots has a net value of D newton-m-sec/rad. The total
inertia of the rotor and pointer about the pivoting axis is J kg-mt, and the compliance
of the torsional restraining spring is K rad/newton-m. The spring exerts zero torque
when B equals zero.
In order to \l8e Lagrange's equation, we m\l8t first formulate the system's La
grangian function. The Lagrangian is composed of the difference between the total
system coenergy-state function and the total system energy-state function, as indi
cated by Eq. (3-56). Equation (3-54) tells us how to formulate the total coenergy
function. Since we are using a nodal formulation on the electrical portion of the sys
tem, the total system coenergy function becomes
The mechanical kinetic coenergy for the linear inertial element is simply
T'(~) - !J~.
In Example 3-3, Eq. (3-17) gives the electric-field coenergy for the system 88
Refer to Fig. 3-9 for an explanation of the physical dimensions of the device.
as seen from Eq. (3-55). However, since no magnetic-field storage is present, the
system energy function reduces simply to
III
'!J(8) - V(II) - 2K
which contains only the one linear spring element. Thus the Lagrangian for this
system is giv~n by
, 1 III
The only remaining terms required for the Lagrangian formulation are the externally
applied forces. In the mechanical portion of the system, these forces take the form of
applied torques. Since there are no externally applied torques,
Q, - 0
where the symhol~. _"ed for the externally applied torques influencing the (I
coordinate.
The externally applied forces on the electrical part of the system take the form of
current sources when a nodal formulation is used. However, in this system we have
only an externally applied voltage source. If the voltage source had some series
impedance, we might tran.sform to an equivalent current source and shunt impedance.
With no series element present, we have an even simpler situation. The node voltage
" in Fig. 3-16 is not really an unknown generalized coordinate, since the external
voltage source v specifies this coordinate for every instant in time. Thus, in reality
we do not even have to write a nodal electrical-equilibrium equation, and Q"" the
external electrical force, is not required. If we had decided to use a loop formulation,
then the one mechanical equation plus two electrical equations would have been
required to describe the system. The nodal formulation is far simpler, requiring only
the one mechanical equation.
Lagrange's equation for the mechanical coordinate is simply
By using the Lagrangian and the Rayleigh function, the required terms are evaluated
as follows:
Similarly, we have
it£(11,/I,1.) _ tV.' _ !
0/1 K
and for the dissipation term
Now substituting these results into Lagrange's equation, we have as the desired
equilibrium equation
Example 3·5
Determine the equilibrium equations for the electromechanical actuator discussed in
Example 3-1 and shown in Fig. 3·4. A schematic representation of the device is shown
in Fig. 3-17. The resistance of the N-turn coil is shown as a lumped-linear resistor R•.
To the terminals of this winding a voltage source v(t), with internal resistance R .. is
connected as an electrical driving function. The mechanical plunger is spring-loaded,
80 that when the position coordinate x equals a distance b, the spring exerts no force on
the plunger. The equilibrium equations are to be determined by using a loop basis for
the electrical portion of the system.
Solution: Since a loop formulation for the electrical part of the system is to be used,
the total system coenergy function, according to Eq. (3-54), becomes
w:,(q,x) - d +x
Mg 9rovifofionol
force
Fig. 3-17 A schematic
diagram of the mag
netic-field electromechani
cal actuator.
upon gathering terms from the expressions for 3' &nd 'U.
For the Rayleigh dissipation function we have
ax - - Cd + X)I - -r
iJ;J(x,q) _ Dx
iJx
Notice again how naturally the force of electrical origin has entered into the equilibrium
equation.
The eleetrical-equilibrium equation is obtained in a similar fashion from
!! [~,c(:t'X,q)] _ iJ'c(i:,x,q) + iJil'(x,q) .. Q
dt iJq - di d + x - d +
x q - (d x)!
+
iU!(:j;,x,q) _ 0
ilq
t Carefully verify that this result is correct. Remember that dldt is a total. not a
partial, derivative.
Electromechanical-system equation, 133
(3-65)
W ..(X,x) (3-67)
(3-68)
Considering Example 3-2, this looks like a very difficult job, but the
magnetic-field energy can still be evaluated by knowing the coenergy
because the two state functions are related by
"
If{... + W~ = L
10-1
A"q" (3-69)
Notice, however, that by use of the relationship of Eq. (3-70) the magnetic
energy will turn out to be a function of the current and position coordinates
and does not have its usual A, x dependency predicted by (3-67).
Can we still find the force of electrical origin from W",(4,x)1 The
answer is yes, but we must proceed with some caution. Suppose we take
the a/ax, of both sides of Eq. (3-70) while holding all currents constant
during the partial differentiation. Thus we have
(3-71)
(3-72)
Example 3·6
To illustrate these ideas, consider a. m8ognetic-field electromech8onic8ol system with the
following cha.r8octeristic rel8otionshipll:
(3-74)
The device has two electric80l ports 80nd two mech80nical parts. Determine the force
of electrica.l origin on the first mechanical node from the coenergy-state function Bnd
then from the energy-st8ote function.
Solution: Expanding Eq. (3-65), the coenergy-state function can be found by
ev80luating
Upon substituting the defining flux-linkage relations given by (3-74), we obtain for the
magnetic coenergy
The force of electrica.l origin on the first mechanical node, according to Eq. (3-66), is
given by
(/.). - (3-75)
Notice that when we use this relationship between the two state functions, the energy
is a function of the current and position coordinates rather than the more usual flux
linkage and position coordinates.
136 Principles of electromechanical· energy conversion
Upon substitution of Eqs. (3-74) and the magnetic eoenergy, tbe energy-state
fundion becomes
We must now use F.£!. (;J·72) to evaluate tbe forre of electrical origin. Simply taking
the aW../ax, will not give the correct (I.), because of the unusual functional repre
sentation of W... On expanding Eq. (3-72) we have
Hubstituting th{,lIe quantities into the expression for (f,)" after som{' lIimplification of
terms, gives the for('e of electrical origin on the first mechanical node as
. 12x,'q.q. 7x,'q,'
(j,), "" -x,X,I[" - -~ - -'--4
(3-78)
Electromechanical-system eqwtions f37
using Eq. (3-76). Again we see that expressing the state function in
terms of the alternative electrical coordinate requires a second term in the
expression for the interaction force.
In the case of an electric-field electromechanical device similar argu
ments apply. Assume that we were able to obtain defining relationships
expressing the charges as functions of the voltage and position coordinates.
This means that a set of expressions in the form
ql = ql(X h X2, ,Xn,XIoX2, . . • ,x..)
q2 = Q2(}.I,X 2, • • • ,X",XI,X2, . . . IX..)
(3-79)
(3-80)
Notice that the electric eoenergy is a function of the voltage and position
coordinates. From this coenergy expression, the force of electrical origin
on the ith mechanical node is given by
(3-81)
W.(q,x} = l
i-I
10
9
' X:(q~,q~, ... ,q~,XIpX2' .•• ,X.. ) dq: (3-82)
When Eq. (3-83) is used, the energy-state function depends on the}. and x
coordinates and not on the q a.nd x coordinates as in its defining relation
138 Prim;iples of electromechanical-energy conver$ion
ship (3-82). Taking the partial derivative of both sides of Eq. (3-83)
with respect to x, gives us
(f.). (3-84)
L .-
), ali.(A,X}
az,
L.
.1:-1
iJWM'(ti,..,}
Loop analyaia (/.), - ~W ~(q,Xl - 4. oA.(li,.,) (f.); -
aZt lb • ~
.1:-1
. .1:-1
Nodal analy.i.
(J.l< _ ow.cA,.,) _
oz. 2:
.1:-1
.-
A aq.{A•..,)
OZi
(J.); -
aw.'{~,.,)
- --,;;
ElflCtromechanical-sysfem equations 139
When the a/ax,. of (3-86) is taken, the force of electrical origin is given as
Compare Eqs. (3-81) and (3-87), and notice the correction or additional
term that needs to be included if we have W~(q,x) rather than W:(X,x)
for the coenergy-state function.
Table 3-1 summarizes all these relationships for the force of electrical
origin and shows how they can be obtained from 'the various functional
dependencies of the state functions.
where the top line of symbols is used for a loop analysis on the electrical
part of the system and the bottom line is used for an electrical nodal
analysis. The total.system energy-state function is similarly given by
'U (x, q,
h,
t) = V(x,t) + W.(q,x)
W .. (h,x)
loop
nodal
(3-89)
where the same scheme is again used for the upper and lower sets of
symbols. The Lagrangian state function is defined as (3' - 'U) according
to Eq. (3-56). Then by using Lagrange's equation (3-53), the system
equilibrium equations can be found.
Notice that Eq. (3-88) defines the system coenergy function to be a
sum of only coenergy-state Junctions. The coenergy function 3' excludes
energy-state Junctions in its definition. Similarly, the system energy-state
function 'U contains only energy-8tate Junctions, excluding the coenergy
Junctions. Thus, if a particular system has magnetic-field-energy storage
and a loop formulation for the electrical part of the system is desired, then
the magnetic coenergy W~ goes into the system coenergy function 3'. On
1M) Principle. of electromechanical-energy conversion
the other hand, if a nodal analysis were desired, then the magnetic-energy
function W", should be inserted into the system energy function 'U.
Also observe that Eqs. (3-88) and (3-89) carefully detail the proper
functional dependence of each state function. The W~ that is substituted
into the J' function must be expressed in terms of the <j's and the x's.
Similarly, the magnetic energy W ... that goes into 'U is a function of the >.'8
and x's. Suppose that in a given problem we require a loop analysis for
the electrical portion of a system containing magnetic-field elements.
And further suppose that we have available only the magnetic-energy
fUllction in terms of the <j's and the x's. Since a loop analysis is desired,
having the q's in the state funetion seems to be a step in the right direction.
Equations (3-88) and (3-89), however, do not have a space for W",(<j,x).
Norm..Uy, for a loop formulation we would use the magnetic-coenergy
function W~(q,x), and indeed that is what we must substitute. By using
the relation
n
L qk>'k -
.\;-1
W",(q,x) (3-90)
oW~(q,x)
(3-92)
ox.
tions can be found, these are functions of the q's and x's, and the magnetic
coenergy or its equivalent in terms of the magnetic-energy-state function
must be used in Lagrange's equation, thus resulting in a loop formulation.
If the defining relationships for the electromechanical elements cannot
be inverted, then the form of these relationships dictates our choice of
a loop or nodal formulation.
3-12 SUMMARY
The energy- and (~oenergy-8tate functions introduced in Chap. 1 are gen
eralized to include electromechanical lumped elements. With the elec
trical coordinates all equal to zero the mechanical coordinates are set at
their state values. While the mechanical coordinates are held fixed at the
state values, the electrical coordinates are then taken from zero to their
final state values and the energy- and coenergy-state functions are cal
culated. Since the state functions depend only on the final state point,
and not on the manner or path by which the system reached the final
point, the energy and cocncrgy funetions calculated in t.he described man
ner are valid. The state functions for electromechanical elements are
similar to the simple lumped elements discussed in Chaps. 1 and 2 with a
dependence on the mechanical coordinates in addition to an electrical
coordinate dependence.
By using the principle of virtual work, an experiment is analytically
conducted to determine the mechanical forces of electrical origin. The
experiment requires that an external force be applied in such a fashion as
to be equal and opposite to the force of electrical origin. By applying
the first law of thermodynamics with the virtual work principle, the exter- .
nal force is obtained III terms of partial derivatives of the state functions
describing the electromechanical elements.
A study of Lagrange's equation shows that the mechanical forces of
electrical origin are correctly inserted into the mechanical-equilibrium
equations simply by using the Lagrange formulation in the systematic
manner described in Chap. 2. For electromechanical systems the elec
trical state functions used in the Lagrangian are dependent on both
mechanical and electrical coordinates. However, the same state func
tions are required for the Lagrangian, and the formulation introduced in
Chap. 2 for separate electrical or mechanical systems is applicable for the
electromechanical systems. For the electrical-:-equilibriumJlQ.uatioIls,
Lagrange's equation also gives the electrical voltages or currents of
mechanical origin.
The state functions for the lumped electromechanical elements, in
general, can have two functional forms. The mechanical forces of elec
142 Prim;;p/es of electromechanical-energy conversion
trical origin can be found from the electrical state functions expressed in
terms of either of two electrical coordinates. Table 3-1 summarizes these
forces for all of the possible functional dependencies of the energy- and
coenergy-state functions. Since the Lagrangian formulation requires
prescribed state functions with prescribed functional dependencies in
synthesizing the two functions 3' and 'lJ, the interaction forces given in
Table 3-1 are calculated in a routine fashion by using Lagrange's equation.
PROBLEMS
3,l In the derivation of the energy- and coenergy-l'ltate functions for lumped
elemen'ts with both electrical and mechanical ports, the mechanical coordinates are
analytically fixed at their state-point values. The state functions are then calculated
by raising the electrical coordinates to their final values. Discuss the energy storage
during each step of this process, and explain why we follow such a procedure.
3-2 A parallel-plate capacitor is constructed from two circular conducting plates
each having a radius of 5 cm. A lO-volt doC 8upply is connected across the two plates.
With a spacing between the plates of 0.5 em:
a. Calculate the magnitude of the energy- and coenergy-l'ltate functions.
b. Calculate the force, in newtons, tending to pull the plates together.
c. If the spacing is slowly reduced to 0.01 em, calculate the new force tending to
pull the plates together and the work done by an external agent in repositioning t.he
plates to this new spacing.
3-3 The magnetic-field transducer shown in Fig. P3-3 has an N-turn coil wound
on an iron core. The iron rotor has a cylindrical surface and is axially mounted in the
N turns
air gap with 181 S 01/2. A torsional spring of compliance K, restrains the motion' of
the rotor. When 8 - .,/8, the spring is in its free position. The rotor has inertia J,
and its pivot points otTer a viscous restraining torque with a coefficient D,. The gap
length g is much less than the rotor radius r, so fringing fields can be neglected.
Assume that the permeability of the iron is much greater than that of the air.
a. Determine the coil flux linkages as a function of the' rotor angle /} and the coil
current i.
b. Find the inductance of the coil. Is the inductance linear?
c. Determine the magnetic coenergy W~(i,8) and tbe magnetic energy W ..(>.,8).
3-4 a. For the magnetic-field transducer discussed in Prob. 3-3 find the torque
of electrical origin exerted on the rotor if the coil current is held constant at i-i.,
Use the appropriate coenergy-state functions.
b. Repeat part a, but use tbe appropriate energy-state function bolding the flux
linkages constant at >. - >'0.
c. Compare the answers obtained in parts a and b.
d. Graphically illustrate the significance of the partial derivatives in parts a and b.
3-5 For the magnetic-field transducer discussed in Prob. 3-3:
a. Write a suitable Lagrangian for a loop formulation on the electrical portion of
the system.
b. Find a suitable Rayleigh dissipation function.
c. Determine the system's equilibrium equations by using Lagrange's equation.
3-6 ti-he electric-field transducer shown in Fig. P3-6 consists of a cylindrical
copper t~be of radius a with a solid copper plunger of radius b held axially within the
tube bYI an inSUlating bushing. The plunger moves only in the .:I: direction and is
restrain4jd by a spring with compliance K and a linear damper D. When x - 0, the
plunger~ halfway intO the tube and the spring is at its free length. If a is only slightly
larger th~n b. we can neglect all fringing fields.
a. Dl\termine tbe charge on tbe plunger as a function of the position of the
plunger and., tbe voltage between the plunger and the tube.
b. Find"the capacitance of the transducer. Is the capacitor linear?
c. Determine the electric coenergy W.(q,x) and tbe electric energy W:(~,.:I:).
_x
K
t "I
T f(tl
2a 211 AI
L
G
~
Insulating 0
bushing
Fig. P3-6 -
iltl
1.44 Principles of electromechanical-energy conversion
3-7 Q. For t.he electric-field transducer discussed in Prob. 3-6 find the force of
electrical origin exerted on the plunger if the capacitor voltage is held constant. Use
the appropriate coenergy-state function.
b. Repeat part a, hut use the proper energy-state function holding the charge
fixed.
c. Compare the answers obtained in parts Q and b.
d. Graphically illustrate the significance of the partial derivatives in parts a and b.
o(the system.
- b. Find a suitable Rayleigh dissipation function.
c. Determine the system's equilihrium equations by using Lagrange's equation.
3-9 For the two parallel iron pole faces shown in Fig. P3-!l, determine the
attract\ve force hetween the poles in terms of the air-gap flux density n, in webers per
square meter, and the pole-face area A, in square meters. Assume that the iron has
infinite permeahility compared with air and neglect fringing around the air gap.
Hinl: Remember that the energy stored in a volume V is
W .. - Iv H~B·dV
where B - "oH in air and H - 0 in the iron.
Fig. P3-9
3-10 A eylindrical iron rod of radius Q is inserted into the cylindrical iron frame,
shown in Fig. P3-lO, through two cylindrical hrass hushings of thickness t. The flux
density established by the coil wound around the iron rod is B webers/m' as shown.
Bross
--B
bushing ..L:::.-._,.....j,.,.,.,,---''--------=,.,.,.,'''----'
00000000
0000000
0000000
Fig. P3-l0
Electromechanical-system equations 145
Assume the iron has infinite penneability compared with the brass. Take the
permeability of the brass to be equal to that of air, ",.. Neglect fringing on the basis
that t « a and assume the field in the brass to be radial.
a. Find the total magnetic-field energy- and coenergy-state functions for the
system in terms of the flux density B and the physical dimensions.
b. Find the force of electrical origin from the energy-state function.
3.1 I For the electromechanical system shown in Fig. P3-11 the capacitance
element C is described by the following relationship:
q - A)"(x. - x,)-'
where q is the charge on the plates and A is a constant. All other elements, both
mechanical and electrical, arc linear.
a. Find the Rayleigh dissipation function 5().,x.,x,) for the entire system.
b. Find the total ele"tric-cocnergy storage W;().,T"T,).
c. Find a suitable potential function 'U().,x"x,) for the entire system.
d. Formulate the system's Lagrangian.
e. Derive the equilibrium equations for the system. Expand all total derivatives_
Fig. P3-11
'Flexible
connection
Fig. P3-12
146 Principle6 of e/ecfromechonico/-energy (;On.,er6;on
mutual coupling M between the fixed and movable coils is a function of the rotational
coordinate 8.
a. Find the functional relationship for M such that the torque of electrical origin
is independent of 8.
b. Show how the device can be used as a true rms current instrument.
c. By using Lagrange's equation and the functional relationship of part a,
determine the equilibrium equations for the system.
S-IS The electric-field force transducer in the electromechanical system shown
in Fig. P3-13 can be described by the following characteristic equation relating the
charge to the terminal voltage and plunger position;
q - aA~~ + b},.{x - C)4
When x - c, the plunger is centered between the two plates and the spring is at its
free length.
a. Write a Buitable Lagrangian for the system.
b. Write the Rayleigh diBBipatiol) function and the external generalized forces.
c. Find the system equilibrium equations.
d. From the equations in part c write the force of electrical origin and the gen
erated current of mechanical origin. Indicate the reference sense for these two
interaction terms.
ir------lI------Jt------1
t
i(tl
Fig. P3-13
3-14 For the electromechanical system shown in Fig. P3-14 equilibrium equa
tions which formulate the circuit containing the electric-field device on a nodal buill
and the circuit containing the magnetic-field device' on a loop basis are required_
The characteristic equations for the coil and capacitor are lioii folloWJI:
Coil flux linkages - ). - Q2,(A + R1:21)
Capacitive charge - q - AI' D .;E:r:1
Electromechanica/·system equations 147
a. Write a suitable LagrllJlgian for the system. Take the spring force to be zero
when:l:, - :l:J.
o. Determine the Rayleigh dissipation function and the external BOUrces.
c. By u8ing Lagrange's equation, find the four equilibrium equations.
X,
A
Fig. P3-14
3-1S A certain electric· field transducer having two electrical ports and two
mechanical ports haa the following characteri8tic equation8:
a. Determine the energy 8tOred in the electric field W.(A"At,:I:,,:I:.). Notice the
functional dependence of W •.
o. Determine the force of electrical origin on the two mechanical nodes from the
energy-atate function found in part a.
c. Repeat part 0, but use a 8uitable coenergy-atate function.
d. Write a Lagrangian for the 8ystem using the energy-8tate function found in
part a.
e. Find the currents bf mechanical origin flowing oul of the two electrical nodel.
3-16 The characteristic equation for a magnetic-field electromechanical device
is given by
X - X.(l - e-·b )
where X. and a are constants.
a. Show that the magnetic-field stored energy i8
THE ANALYSIS OF
4
LINEAR SYSTEMS
\
.4-1 CHARActERISTIC PROPERTIES OF THE LINEAR SYSTEMl
The input function (3(t) is usually some known function of time, and thus
the complete right-hand side Of Eq. (4-J-) can be determined as some
'. Input
PI t)
Response
x It)
Fig. 4-1 , A generallumped-ltnear system.
150 Principles of electromechanical-energy conversion
with Q(t) representing the known time function, which is the right-hand
side of Eq. (4-1). Suppose that a specific forcing function Q,(t) drives the
system represented by Eq. (4-2). The response to this forcing function
could, in general, be represented by some function X,(t); thus, we can write
1 In Fig. 4-1 the linear system modified the input function (jet) according to the
right-hand side of Eq. (4-1), thus giving a general Q(t), In Fig. 4-2 we are assuming
that this modification of the input does not take place, since Q(t) is actually being
inserted.
t:i,{t) X, (t)
x2{t) (e)
(b)
By adding Eqs. (4-3) and (4-4), we can simplify the resulting sum into the
form
an (
t) d"(xl
dt"
+ X2) + a,,-l(t) d,,-l(XI + X2)
dtn 1
+. + al(t) d(Xl: X2) ao(t)(Xl + X2) = Ql(t) + Q2(t)
(4-5)
which is represented by Fig. 4-2c. We can see how (4-5) results by taking
the first term as an example.
(4-6)
If the coefficient an were some function of the response Xl for the first
excitation and X2 for the second excitation, then this combination of terms
as indicated by Eq. (4-6) would not be valid. The fact that an is a function
of t, the independent variable, does not invalidate the combination of
terms.
Equation (4-5) and Fig. 4-2 point out the first important property of
a linear system, namely, that of superposition. Input Ql(t) gives rise to
response Xl(t), and input Qz(t) gives rise to response xz(t). Now inserting
both Ql(t) and Q2(t) simultaneously gives us as a response the sum of the
original two responses Xl(t) + X2(t). In other words, the responses to the
two inputs superimpose themselves one upon the other. The linear sys
tem handles each individually in such a way that the second input in no
way affects what happens to the first input, and similarly the first input
does not alter the response to the second input. Actually, no physical
system can successfully meet the superposition criteria for all ranges of
its terminal variables.
Take, for example, the simple case of a lumped-linear resistance
described by the relationship
(4-7)
Consider the current ti to be the input and the voltage Xto be the response.
As a first input function let
a constant current
Then the response voltage is
Xl = RIo (4-8)
giving a response
X2 = RI sin wt (4-9)
Now inserting a current equal to
(it + (12 = 10 +I sin wt
should give a response
Xl +X 2 R(lo +I sin wt) (4-10)
according to Eq. (4-7). The total response given by Eq. (4-10) is seen to
be the sum of the two responses given by Eqs. (4-8) and (4-9).
Suppose we allowed the direct current 10 to become very very large.
A physical resistor might in fact overheat sufficiently to actually disinte
grate and make an added sinusoidal response impossible. Thus super
position does not hold, since we can always postulate a large enough
current to bum up any physical resistor. When an equation in the form
of (4-7) is written, we are always implying that there is a certain allowable
range for the variables over which the relationship is a reasonable analytic
approximation for the physical device. The same statement should be
made in connection with all of the elements we have discussed in the
preceding three chapters, namely, that relationships between coordinates
are valid only for a limited range of coordinate values. The statement is
true even for nonlinear relationships.
The linear system, which can always be described by an equation in
the form of (4-2), has a second important property illustrated by Fig. 4-3.
If an input kQ(t) is applied to the system, where k is any constant, then
the response must be kx(t). The response x(t) is assumed to be the
response to Q(t) acting alone. This property, whereby a linear system
with an input k times as large will simply yield a response amplified by the
same factor k, is known as homogeneity. We can show that Eq. (4-2)
obeys the homogeneity property by simply multiplying both sides by the
constant k, thus setting
d"x d,,-IX dX]
k [ an(t) + an-let) dt,L-l + ... + al(t) dt + ao(t)x
= kQ(t) (4-11)
Since k is a constant, we can rearrange Eq. (4-11) into the form
+ al(t) d~t)
+ao{tHkx) = kQ{t) (4-12)
clearly showing that the response to an input kQ(t) is equal to kx(t).
OUI xU)
Nonlinear system
NO.2
Fig, 4-4 A system possessing superposition but not homogeneity.
d
p (4-13)
dt
tive is given by
(4-15)
showing that the operator raised to the nth power indicates that the nth
derivative shall be taken.
A general function of the differential operator can be formed as
follows:
(4-16)
where the a", a ..-l, . . . , ao coefficients can, in general, be functions of the
independent variable time. By using this notation, the linear differential
equation (4-2) can be written as simply
L(p)x = Q(t) (4-17)
where x is the unknown coordinate and Q(t) is the forcing function.
The linear differential operator defined by Eq. (4-16) obeys certain
common algebraic rules. Given three distinct linear differential operators
L1(p), L 2(p), and L3(P), the commutative and associative laws of addition
are valid. This means that
Commutative:
(4-18)
Associative:
(4-19)
Also, in general, the distributive and associative laws of multiplication are
valid, as follows:
Distributive:
(4-20)
Associative:
(4-21)
However, the very important commutative law of multiplication is not
generally valid, except under certain important circumstances to be dis
cussed later. This law is given by
The analysis of linear systems 155
Commutative:
(4-22)
and is not true for the general linear differential operator.
An example will serve to illustrate what these two additive and three
multiplicative laws mean.
Example 4-1
For the three linear differential operators
L,(p) + [L.(p) + L.(p)] "" (5p + 2t) + (3 sin t p' + 2p) == 7p + 2t + 3 sin t p'
[L,(p) + L.(p)] +
La(P) = (6p + 2t + 3 sin t pO) + p = 7p + 2t + 3 sin t p'
+
== (3 sin t p' 6p 2t)x
+
dx
= 3sint d'x
at l
+6
at
+2tx
Also, we have
dx . dlx dx
The linear differential operator will obey the commutative law for multi
plication provided the an, a,,-l, . . . ,ao coefficients are not explicit func
tions of time but are simply constants. This would mean that Eq. (4-16)
defining the operator becomes
(4-23)
with constant coefficients. Having the operator in this form, the four
laws, Eqs. (4-18) to (4-21), are still valid, and the commutative law for
multiplication, Eq. (4-22), is also valid. Equation (4-17) uses the linear
differential operator in a general description of a linear system. By use of
the constant-coefficient operator, given by (4-23), the description reduces
to a linear differential equation with constant coefficients: the type of
differential equation we can readily solve. The techniques used to obtain
solutions essentially stem from the fact that the constant-coefficient linear
differential operator obeys all of the algebraic laws of ordinary coeffi
cients, including the commutative law of multiplication. The operators
The analysis of linear systems 157
(4-24)
Example 4-2
For the circuit shown in Fig. 4-5 write out the nodal equilibrium equations and obtain
a single equation involving only the first node variable.
158 Prim:iple. of eledromedJanica/.en8rflY conversion
T Datum
Fig. 4-5 A lumped-linear electrical network.
Solution: In terms of the nodal variables ),1 and ),1 (the integrals of the respective
node voltages) the equilibrium equations are
1 J (c • (iii l
[(C1 + C•) dl),l G d),1
dt s + di + L ),1 -
1
d ),.)
= •
-h(t) + • (
$. t)
1
I
\
which reduces to
linear differeiltiaJ. operator, in this case A(p), is defined only when ithaa something
upon which to opera.te.
or· the input to the block x(t) is operated upon by L(p), giving the output
Q(t). .
The 8'Ummation symbol, in Fig. 4-6b, adds (or subtracts) the two
variables Xl and X2. Plus and minus signs must be placed on each of the
~ L(p) ~ == L(p)x=Qlt)
( Rp + ~) q = v(t) (4-30)
where q is used as the integral of the loop current i. Figure 4-8a shows one
block-diagram representation of Eq. (4-30). An entirely equivalent repre
sentation is shown in Fig. 4-8b, where the forcing function v(t) is shown
as the input to the system block and q(t) is shown as the output or response.
Figure 4-8b seems like a more natural representation, since the input
to the block is known and it is the output which we desire. If we want
the current i, rather than its integral q, a tandem block can be joined to
Fig. 4-8b, as shown in Fig. 4-8c, with p operating on q to yield i.
~I
qU) vlt)
Rp + lfc •
(a)
I' (t) 1
·1 Rp+1Jc
(b)
(Rp)q
Still another way to represent Eq. (4-30) is shown in Fig. 4-9. The
output of the summing point is given by
1
C q =.v(t) - (Rp)q (4-31)
X2 = L1(P)Xl
X3 = L 2 (p)X2 = L 1L 2 (P)Xl
(4-32)
x, = La(P)X3 LIL2La(P)XI
All the combinations given by Eqs. (4-32) are valid, because all the
constant-coefficient linear differential operators obey the associative and
comrnutll,tivelaws of mUltiplication given by Eqs. (4-21fand (4~22).
Summing points may be separated, reversed in order, and combined
as illustrated in Fig. 4-11. The manipulation of the summing point
Fig. 4-10
-
Reduction of blocks in cascade.
162 Principlef of electromechanical-energy cooverfion
"'2
Fig. 4-11 Manipulations involving summing points.
- 1--_~"'2
(a)
(0)
Fig. 4-12 Interchanging the positions of blocks and splitting points.
I
I
The analysis of linear systems 163
J
-
(al.
-
Fig. 4-13 Interchanging the positions of blocks and summing points.
which are obviously equivalent equations. Figure 4-13b stands for the
equations
Xa = L(p )XI - L(p )X2 (4-35)
which again are identical equations. N otice th~t the transmissions
from Xl and X2 to Xa in each of the diagrams in Fig. 4-13a and b are kept
identical.
The only additional manipulation we shall need involves the feed
back configuration shown in Fig. 4-14a. For this common diagrarn, in
terms of the operators G(p) and H(p), the following equations cu¥ be
written:
Xa = G(P)(XI =1= X2)
Therefore, we have
Xa = G(p)Ixl =1= H(p)xaJ
which reduces to
G(p) ] (4-36)
Xa = [1 ± G(p)H(p) Xl
_xf'--_..../ G(p) X3 ..
• __1_1_G_(p_I_H_~_I~r--~
(ill
(al
Fill:. 4-14 Reduction of the feedback eonfil!:Uration.
164 Principles of electromechanical-energy conversion
(a) (b)
~ KI
~ (el
O+Jp
~
and 4-9 for the R-C circuit are equivalent by simply applying the feed
back reduction given by (4-36).
Example 4-3
-K,il +(R. +
L.p)i. +
K.8 = v.(t) (4-37)
-K1i• .+ (D +
Jp)8 = T(t)
All the forcing functions VI, V2, and T are assumed to be known functions of time.
The unknown dependent variables are ii, i., and 8. All coefficients are constants, and
the linear differential operator p equals dldt. Draw the block diagram for the sys
tem and reduce to a form which clearly shows the relationships between the response 8
and the three forcing functions V" Vii and T.
Solution: The overall block diagram of the system represented by Eqs. (4-37) is
constructed by combining diagrams for each of the equations. Figure 4-15a, b, and c
represents each of th~ three system equations. Notice in Fig. 4-15b how the second
equation has been manipulated to have i 1 as the input to the diagram and i. as the
output. Similarly, Fig. 4-15c has i. as the input and 8 as the output. Therefore,
quite naturally, the three parts of the figure can be combined to give the complete
block diagram shown in Fig. 4-100. The output of the diagram is the response of
interest, in this case 8. The input to the left-hand summer K.8 is obtained by the
feedback path as shown in the figure. Also notice how the three forcing functions
appear as explicit inputs to the diagram.
Figure 4-16a can be reduced to a more compact form. Combining the blocks in
cascade and moving the block Kd(R. + L.p) past the right-hand summer gives us
Fig. 4-16b. Upon closing the feedback loop according to Eq. (4-36) or Fig. 4-14, we
have the final result shown in Fig. 4-1&. From this figure the transfer relationships, or
transfer functions, between the response 8 and the three forcing functions can be
The analysis of linear systems 165
The diagram of Fig. 4-1& gives the same relationship in a neater and more compact
form. The mathematics involved in manipulating the diagram from Fig. 4-16alto b
and finally to c is simpler than evaluating the determinants necessary when using
Cramer's rule on the original equations (4-37).
Cal
(lI)
+
+
+
(e)
Fig. 4-16 Combination and reduction of the three separate diagrams.
A linear system can always be represented by an equation of the form
L(p)x(t) = Q(t) (4-39)
where the linear" differential operator L(p) is defined by Eq. (4-16), Q(t)
is a specified time-dependent forcing function, and x(t) represents a
\ general unknown system coordinate. In Sec. 4-3 we have seen that a
sy'ktem represented by a simultaneous set of differential equations
involving all of the coordinates can always be reduced to a single equa
tion in the form of (4-39) provided the L(p) operators contain only con
stant coefficients. Thus for a set of constant-coefficient linear differential
equations, a single constant-coefficient equation involving only one
coordinate can always be obtained.
Let us now examine the solutions of Eq. (4-39) for certain particular
Q(t) forcing functions. Suppose Q(t) were simply a constant rather
than a general time-varying function. Further, let us assume that the
constant forcing function, which we shall represent by Qo, is applied to
the system for a time of sufficient duration to allow the system to reach
its steady-state response. If the linear system represents a physical
system, then certain loss elements, such as resistors and dampers, must
be present. The steady-state response of such a linear system to a
constant d-c input must also be a constant. At the time instant at
which Q(t) = Qo is applied, the system will usually respond with some
time-varying motion. However, after a sufficient time duration, the
transient motion decays to zero, leaving only a steady-state d-c response.
Analytically we can express these concepts in the following way.
The general coordinate x(t) in Eq. (4-39) must be a constant. Therefore
we have
L(P)xo = Qo (4-40)
where Xo is the constant d-c response to the constant input Qo. Equation
(4-40) assumes the response x in (4-39) is equal to a constant Xo in the
steady state. Since the operator p = djdt operating on a constant gives
zero as a result, Eq. (4-40) can be simplified to give
L(O)xo = Qo
and thus
Qo
Xo = L(O) (4-41)
;1
tions involving only the various steady-state responses, which can readily
be solved by using determinants and Cramer's rule.
Example 4-4
Determine the differential equilibrium equations describing the linear translationa.!
mechanical system with 3 degrees of freedom shown in Fig. 4-17. Each of the
rolling masses is assumed to have a viscous friction force on its wheel bearings, as
shown in the figure. The position coordinates are assigned in such a manner that the
three springs lIJ'e at their free lengths when Xl, X" and X3 are zero. Assumingf(t) to be
equal to a constant force fo, find t~e steady-state position of the three masses.
Solution: Let us use Lagrange's equation to formulate the system equilibrium
equations. The Lagrangian for the mechanica.! system, in terms of the coordinates
shown in Fig. 4-17, is given by
Assuming the applied force f(t) to be a constant fo, the steady-state position
response can be found by setting all p's equal to zero. This leaves the following set of
algebraic equations:
The additional subscript zero is being used to indicate the d-c steady-state value for
each of the coordinates. Cramer's rule can now be used to solve these equations for
each of the final positions. ThuB for (XI)O we have
1
0 0
~(XI)O = .l..+.l.. 1
0 - Ka
Kz Ka
1 1
fo - Ka Ka
~(X,)o = l.k a
where ~ l/K,K zK a.
The final steady-state position of the first mass is therefore
Similarly, the final positions of the remaining two masses are given respectively by
and
The d-c stea.dy-sta.te "response is obtained by letting p = dldt go to zero in each of the
original equilibrium equations and then using Cramer's rule to obtain the individual
steady-state responses.
statement is true provided the system contains some loss element, such
as a resistor or viscous damper. All physical systems have loss elements.
Therefore, the response can, in general, be given by
x(t) = "\1'2 X sin (wt + 1')
= 1m ("\1'2 Xei("'t+-r» (4-47)
where X is the unknown nns magnitUde of the response and l' is an
unknown phase angle. Equation (4-47) gives the complete form of the
steady-state response, with only the magnitude and phase angle as
unknowns.
Before SUbstituting the assumed steady-state response of (4-47)
into Eq. (4-46) let us see how the differential operator affects such a
function. The first time derivative of the steady-state response is
px(t) = p [1m ("\1'2 Xei("'t+-r»]
= 1m [p("\I'2 Xei(",H-r»l (4-48)
upon interchanging the operations of taking the time derivative and
taking the imaginary part. Equation (4-48) further reduces to
px(t) = 1m (jw V2 Xei("'t+-r» (4-49)
when the derivative is actually taken. In a similar fashion we can show
that the nth derivative is given by
p"x(t) 1m [(jw)" "\1'2 Xei{",t+-r)l (4-50)
Now, if the assumed solution (4-47) is substituted into the system
Eq. (4-46), using (4-23) for L(p), we have
1m ([an(jw)n + a _l(jW)n-l + ... + al(jw) + aol "\1'2 Xe;(..t+-r) I
n
= 1m ("\1'2 Qei(",t+a» (4-51)
Notice that the term in square brackets on the left side of (4-51) is
actually just L(jw). Therefore, Eq. (4-51) can be condensed tol
1m [L(jw) "\1'2 Xei("'t+-r)] 1m ("\1'2 Qei(",Ha» (4-52)
Taking the total time derivative of both sides of Eq. (4-52) and
dividing by the constant angular frequency w gives us
(4-53)
Since the imaginary part of j times a complex number is equal to the
real part (Re) of that number, Eq. (4-53) becomes
(4-54)
lOne might be tempted to simply drop the 1m operations on both sides of (4-52).
However, in general, such a manipulation is not valid. For example, 1m (3 + j2) =
1m (5 + j2), but certainly (3 + j2) r6 (5 + j2).
The analysis of linear systems 171
On comparing Eqs. (4-52) and (4-54) we see that both the real parts
of two complex numbers and the imaginary parts of these numbers are
respectively equal. Therefore the numbers themselves are equal, and
we have
(4-55)
upon expanding the exponential. Since e iloll has a constant amplitude
of unity, it may be canceled, along with V2, from both sides of (4-55),
leaving ,
L(jw)(Xeh) = Qe ia (4-56)
The right-hand side of Eq. (4-56), representing the sinusoidal forcing
function, is actually a shorthand notation for the original function of
(4-43). The term Qe ia is a complex number whose magnitude is the
rms amplitude of the sinusoidal time function and whose phase angle
is the initial angle of the sine wave. The abbreviation of the original
sinusoidal time function is termed a sinor or sinor quantity, and the
boldface symbol Q is used to denote the sinor. Thus, sinor
Q = Qeia
stands for
Q(t) = y2 Q sin (wt + a)
Notice that we are saying that Q standsjor Q(t), not that Q equals Q(t).
Once w.e know sinor Q and the original frequency w, we can always obtain
the corresponding time function Q(t). In a similar fashion a sinor
quantity for the response can be written as
x = Xeh
standing for
Q (4-57)
X = L(jw)
The result is very similar to the d-c steady-state solution given by
Eq. (4-41), except that here p = d/dt becomes jw instead of zero and
sinor quantities instead of d-c quantities are used.
172 Principles of electromechanical-energy conversion
Example 4-5
A particular mechanical system is described by the following constant-coefficient
linear differential equation.
where Xl is the sinor response to the first forcing function on the right side of the
equation. Notice that the sinor representation always uses the rms magnitude of the
timc function, and not the peak magnitude. On solving for Xl, we have
lO /60° lO /60°
where we use the polar representation of the complex quantities with degrees, rather
than radians, as the angular measurement. The corresponding time response is
therefore
Xl(t) = V2 2 sin (2t + 6.8°)
The second portion of the steady-state response is obtained from
[3(j4)' + 2(j4) + 15jX. = 20e-irls
and therefore we have
20~ 20~
(15 - 48) + j8 34 /166.4°
= 0.59 / -196.4°
The complete steady-state time response is the sum of these two components, which is
x(l) = x,(t) +
x.(t)
x(t) = V2 [2 sin (2t + 6.8°) + 0.59 sin (4t 196.4°)1
Remember to reinsert the V2 when the sinor is changed back into the corresponding
time function.
A(jw) (4-60)
...... . . . . . 9 . ~ ...
Lnl(jW) Ln2(jW) Lnn(jw)
Expanding Eq. (4-59) and dividing by A(jw) gives us
(4-61J
as the kth response in sinor form. The terms Ai.!; represent the i-k
cofactor of the determinant. The corresponding time response Xk(t)
can easily be recovered from the sinor form of Eq. (4-61).
Example 4·6
A motor driving an inertia load can be modeled by the lumped rotating system shown
in Fig. 4-18. The load has inertia J. kg-m' and a viscous friction, or windage,
component D, newton-m-sec/rad. The shaft coupling between the motor and the
load has a compliance of K rad/newton-m. The motor has a shaft inertia of J, kg-m 2
and windage coefficient of D, newton-m-sec/rad. The developed torque of the motor,
174 Principles of electromechanical-energy conversion
given by T(t) newton-m, has a sinusoidal component of the form (V2 T sin wt).
Determine the steady-atate angular-velocity response of the load to this component of
the driving torque.
Solution: In operator notation, the differential equilibrium equations are found to
be
(JIP + Dl + ~p) 01- ~p 0, = T(t)
- ~P 01 + (J2P + D. + ~p) 0, = 0
The division by P has not actually been defined. The terms involving (I/Kp>,O should
be interpreted as being (I/Kp)(pfJ), which reduces to simply (I/K)fJ. Since the
angular velocities of the motor and load are of primary interest, the equilibrium
equations have the th and fJ. coordinates multiplied by p, yielding 01 and 0., respec
tively. The four coefficients are then divided by P to maintain equality.
Using sinor notation, the steady-state angular velocity of the load is given by
J 1 (jw) + Dl + K(~w) T
1
K(jw)
o
where
1 1
J 1 (jw) + Dl + K(jw)
+ D. + K(~w)
1
- K(jw) J.(j",)
e =O.h
which has a corresponding time response
W2 = 0, = V2 O. sin (wt + or)
Suppose that the two windage coefficients Dl and D. were small enough to be
negligible. The steady-atate angular velocity of the load, in sinor form, would then
reduce to
The analysis of linear systems 175
Wr
then the denominator becomes zero and the velocity of the load oscillates between plus
and minus infinity. This infinite resonant response occurs only if the windage
coefficients are neglected; in practice, they never can be entirely. However, large
torsional oscillations can certainly cause great damage to the physical components in
the system.
(4-62)
'T'hp kt.h rp.<lnonI'lP in sinor form to a set of 0; sinor forcina: functions can be
A'k
X k = ~ Q. (4-63)
Equation (4-63) is the kth sinor response to the ith sinor forcing function
acting alone. The solution of Eq. (4-63) gives us the sinusoidal steady
state response to the sinusoidal forcing function Q.(t).
The frequency-response characteristics are plots of the magnitude
and relative phase angle of the response Xk(t) as the frequency of Qi(t)
is varied while the magnitude and phase angle of Q,(t) are held constant.
We desire, therefore, a plot of the magnitude and phase angle of the
transfer relationship or transfer function A.k/ A as a function of frequency.
The term "transfer function" is quite descriptive in the sense that it
gives the transferring recipe whereby the system "cooks up" the response
for a given forcing function.
Let us now consider the most general form that the complex transfer
function A,k/A could have. Both the cofactor and the determinant are
made up of the products of polynomials involving jw to various powers.
Therefore, we should always be able to write Aik/A in the following form:
A,k bm(jw)m + b..._1(jw)m-l + ... + b1(jw) + bo (4-64)
A a,,(jw)n + a.._l(jw),,-l + ... + al(jw) + ao
where m and n are integers. Factoring bo from the numerator and ao
from the denominator changes (4-64) into
Aik =K (bm/bo)(jw)m + (bm_tlbo)(jw)m-l + ... + (b1/bo)(jw) + 1
A (a,./ao)(jw)n + (an_t/ao)(jw)n-l + ... + (at/ao)(jw) + 1
(4-65)
W1 2 WI
I~:I (4-67)
I!!:
L....S1i. = tan- 1 nw (4-68)
'Here an wand an w, are being used, where w stands for the frequency, of the
forcing function and w, is a coefficient or parameter of the system. \..
178 Principles of electromechanical-energy conversion
0.001 -60
0.01 -40
0.1 -20
1 0
10 20
100 40
1000 60
The analysis of linear sysfems 179
40
,/'
/'
V
20
/'"
Actr' curve~
--?"
/ - High-frequency
asymptote
+ o Slope = 2.0 db/decode
a
....., LO~-freq~ency
~
asymptote I I
--20 I - I I I I I
40
I
I I I I
i
20
.0
'0
Low-frequency
/ asymptote Lgh-freqUency
asymptote _
~l~
0
Slope; -20
~
Actua~7 ~
~~ db/decade
~
curve
-20
............
i'--
-40 I !
' 0__----i----t----1"---..:::--____1,,;;:----t---t
~...
X
-201__~-4--_r----1__-+---·____1,
-40~--~--~----~-~-~---~--~
lIT, 1/T2 1fT3
41, rod/ sec
Fig, 4-21 Total asymptotic and corrected magnitude plots for the transfer function
X./Q. K(jTIW + 1)/0T"" + 1)(jr3w + 1) with T3 > Tz > T"
The third term in Eq. (4-70) can be plotted in exactly the same
manner as the second term, except that the slope of the high-frequency
asymptote is negative 20 db/decade in frequency. The correction
quantities are still 3' db at the corner frequency w = 1/72, and 1 db at
half and twice this frequency. The only difference is that these correc
tions must be subtracted rather than added to the asymptotes. Figure
4-20 shows a plot of the third term in (4-70). Notice again that the
corner frequency and the asymptotes are all we need to obtain the plot.
The high-frequency slope and corrections are standard for the distinct
factor.
According to Eq. (4-70), the complete plot of 20 logio of the magni
tude of the transfer function is simply the sum of the individual loga
rithmic plots. In Fig. 4-21 we begin by drawing the asymptotes of each
of the terms in Eq. (4-70), with 20 logIo K as simply a constant horizontal
line for all frequencies. All the various asymptotic curves can be added
together, giving the total asymptotic curve shown in Fig. 4-21. Just
by using the major corrections of 3 db at each of the three corner'fre
quencies, the approximate curve can be corrected to give a more exact
plot, as shown also in Fig. 4-21.
The phase angle of the transfer function given by Eq. (4-68) involves
the adding or subtracting of a standard angular characteristic. Figure
4-22 shows a normalized plot of tan- 1 a as a function of the normalized
angle ct, again using semilog graph paper. For values of ex much less
than unity tan- 1 a is approximately zero, and for a much greater than
unity tan- 1 a approaches 90°. From Eq.(4-68) the phase angle of the
transfer function can easily be plotted by using this normalized charac
teristic. Dividing the normalized angle by the individual time constants
182 Principles of electromechanical-energy conversion
TI, T2, and Ts allows the phase contributions of each term to be read
directly from Fig. 4-22. Adding and subtracting these contributions,
according to whether the factor is a numerator or denominator term in
the transfer function, yields the total phase response as a function of
frequency.
Figure 4-23 shows this resulting phase pharactedstic as given by
Eq. (4-68). Notice that for values of the forcing frequency w much less
than the reciprocal of the time constants the total phase shift of the
transfer function is zero. As w increases, the numerator factor is the
first to become significant, causing the characteristic in Fig. 4-23 to
become initially positive. However, as w continues to increase, the two
denominator factors add negative angular contributions, with the phase
characteristic finally approaching - 90 0 for large values of w. The
numerator factor contributes +90 0 and the two denominator factors
give 1800 total, thus resulting in a net _90 0 at high frequencies.
Having more real distinct factors in the numerator and denominator
than in the transfer function originally proposed by Eq. (4-66) would
simply mean more terms would have to be included in the magnitude
characteristic of Fig. 4-21 and the angle characteristic of Fig. 4-23.
However, both of these plots are formed by adding various quantities,
so the inclusion of more terms does not change the form of the individual
contributions. Adding asymptotes or phase contributions for a more
complicated transfer function is therefore just a simple extension of the
ideas presented for the simple transfer function of Eq. (4-66).
As a second consideration, let us assume the transfer function to
contain a complex conjugate factor. Since the effect of the other real
90r----r------~--~----,------,----~
60
~
30~--1-----_b~---~--1-----~--~
Fig. 4-22 Normalized phase contribution for the real distinct factor 1 + ja as 8.
function of a.
The analysis of linear systems 183
30
f.--- f,.-- 1\
o
~-30
\
-60 ~ ""
- ~
-90 i'-..
lIT, 11r2
W, rod /sec
Fig. 4-23 Total phase-shift characteristics for the transfer function Xk/Q. =
K(j7',,,, + 1)/(j7'2'" + l)(jr .... + 1) with 7'3 > 7'2 ? 7',.
distinct factor can easily be added to the complex conjugate factor, let
us restrict our considerations to the 1l1mple transfer function
XA: 1
(4-73)
Q, = (jw)2/WI2 + 2rl(jW)/WI + 1
containing a single complex conjugate factor. The magnitude of the
factor, in decibels, is given by
after taking the square root of the sum of the squares of the real and
imaginary parts and putting the 1- exponent as a coefficient of the loga
rithm. The phase angle of the transfer function (4-73) is obtained from
(4-75)
Notice that in both the magnitude and angle functions the driving
frequency W is always divided by the undamped natural frequency WI
of the complex factor. Therefore, by drawing the Bode plot as a function
of the normalized frequency W/WI, we shall have characteristics useful
for any value of WI.
Considering W/WI to be very small, the low-frequency asymptote
is obtained. From (4-74), we have
for W » WI
(4-77)
20
(,Lo..LreqJency LympJe
o
"" '"
r<.High-frequenc y asymptote
Slope -40 db/decade
:g -20
.
a
"'
.;(-40
-60
'""" "
-80
0.01 0.1 2
Normalized frequency
5
W/Wl
10
'"~ 100
Fig. 4-24 Asymptotic ma.gnitude plots for the tra.lIBfer function Xk/Qi =
1/[1 - (W/W,)2 + j(2(,w/w,)].
The analysis of linear systems 18S
201~.----~---,--~,-._~~------,_--r_,_-~I~~
I ~,=0.05
18r-~~~+-~-r--T--r-+-r~~~-----+--~+--+~+-+-~-+1
5 6 7 8 10
Normalized frequency w/w.
Fig. 4-25 Decibel correction quantities to be added to the asymptotic magnitude
plot of Xk/Qi = 1/[1 - (W/WI)' + i(2 sl w/Wl)\ for various values of the damping
ratio SI ~ 1.0.
that /ilmall values of the damping ratio require large corrections, particu
larly near W/WI 1.0. With no damping, a steady-state response of
infinite magnitude occurs when the driving frequency W equals the
undamped natural frequency WI. Only values of r 1 equal to unity or
less are shown, because for damping ratios greater than unity the complex
conjugate term can be factored into two real distinct factors.
Figure 4-26 shows the phase characteristics for Eq. (4-75), again
plotted as a function of the normalized frequency W/Wl. For each value
of rIa different phase characteristic is shown. Notice, however, that
all these characteristics start at 0 0 for W/Wl « 1.0 and approach -180 0
for W/Wl» 1.0. The phase characteristic for small damping ratios
changes very abruptly in the vicinity of W/WI 1.0, becoming more
gradual as r 1 increases.
The third type of factor which we might have in a transfer function
is a repeated factor either real or complex conjugate in form. The con
tribution of such a term is the same as that of the single non!,"epeated
factor, except that it is included the same number of times as the factor
is repeated. Suppose the transfer function contains a real factor repeated
186 Principles of electromechanical-energy conversion
(4-78)
Equation (4-79) is identical with that for a real distinct factor except
for the factor of 3 preceding the braces. The low-frequency asymptote
is still at 0 db up to the corner frequency w = l/T. Beyond the corner
frequency the slope of the high-frequency asymptote is -60 db/frequency
decade, or three times the slope for the distinct factor. The corrections
also are multiplied by 3. Thus at W = l/T we should subtract 9 db
(or 3 times 3 db) from the asymptotes. Also at w = 0.5/T and at w = 2/T,
3 db (or 3 times 1 db) should be subtracted.
The phase angle of Eq. (4-78) is given by
-3 tan- 1 TW (4-80)
or simply 3 times the phase shift of the real distinct factor. Thus, at
each corresponding frequency, the angular contribution is 3 times that
of the same nonrepeated factor. Figure 4-27 shows the complete Bode
plot of the transfer function (4-78). For a repeated complex conjugate
factor the same ideas apply. Simply find the magnitude and angle
contributions of the factor if it is not repeated and multiply these decibel
O~~~~~~~~UL~~~~~~IITTn
-20
-401---
i-601---
:t: -801-,-----1:-
.c:: i
.
.. i
... -1001-'- - - + -
o
if. -120 t----t---t--tf--+-+--t-+-+-w~~-.;;;:-+---t--
-1401-----t--+--tf-t-+-+_
-160 t----t---t--tf--t
-180'- _ _--'_ _...1.---''--.1....
0.1 0.2 0.3 0.4 0.6 0.8 1.0 :3 4 10
w/Wj
Normalized frequency
Fig. 4-26 Phase-ahift characteristics for the transfer function Xk/Q. =
1/[1 - ("'/"'1)' + j(2ilw/"'I)) for various values of the damping ratio il :::; 1.0.
\ The analysis of linear systems 187
-20
o
r-
-- ~
"
-I'--
l""
~
,
o
-30
-60
90 ~
...
~ -40
'c "" ",,
~
"
120£
:;::
150 :
"
V>
0
o ._ -180,f
::0
-60
~ ~ -210
~ I~
-80
O.51r 1/r
W, rod /sec
2/r
--- .~
-240
270
Fig. 4-27 Magnitude and phase characteristics ofthe transferfunction 1/ (irw + 1)3.
Example 4-7
A particular electromechanical system has only two ports, one being electrical and the
other mechanical. For a certain operating range, the following linear differential
equation in operator form describes the motion of the system:
(pI + 6.5p2 + I03p + 50)x = (1.5p + 3)v
where x is taken as the position response of the mechanical member, in meters, and v
is the voltage applied to the electrical port, in volts. Determine the frequency response
of the system in the form of a Bode plot.
Solution: The frequency response requires that the driving function v(t) be
sinusoidal in form. The magnitude and phase angle of the sinusoidal steady-state
response xCt) are to be determined as the frequency w of the driving function is varied.
By using operator notation, the transfer function between the response and the forcing
function can be put into the form
:: = 1.5p + 3
v pi + 6.5p2 + 103p + 50
which can be factored and manipulated to become
x O.06(O.5p + 1)
Factoring the third-order polynomial into the form shown can often be a difficult job.
The second-order factor in the denominator factors into two complex conjugate
factors, so we shall leave it as shown.
188 Principles of electromechanical-energy conversion
Substituting jw for p gives us the complex transfer function relating the sinor
forcing function V to the sinor response X. Thus, upon making this substitution, we
have
x (4-81)
y
The composite of the asymptotes for the magnitude of this transfer function is
shown in Fig. 4-28. For frequencies below j-, or 0.5, rad/sec the asymptote has a
magnitude equal to 20 loglo 0.06, or - 24.4 db. At w = 0.5 the real denominator factor
contributes a -20 db/decade slope. Then at w = 2.0 (which is 1/0.5) the real
numerator factor contributes a +20 db/decade slope, thus resulting in a combined
slope of zero. At w = 10, the complex conjugate factor contributes a -40 db/decade
slope, as shown in Fig. 4-28.
The corrections to the asymptotes are summarized in Table 4-2. For the real
factors, the corrections are 3 db at the corner frequencies and 1 db at half and twice the
corner frequency. Corrections for the complex factor are taken from Fig. 4-25, using a
damping ratio r = 0.3. By using the normalized phase characteristics shown in
Fig. 4-22 for real factors and in Fig. 4-26 for complex conjugate factors, we can con
struct Table 4-3. The total phase characteristic is also shown in Fig. 4-28 referred to
the right-hand scale.
By using these two plots, the magnitude and phase shift of the response for any
sinusoidal forcing function can be taken by inspection. For example, suppose the
forcing voltage, applied to the electromechanical device, is given by
11(0 = 15 sin (4t + 60°) volts (4-82)
At a frequency of 4 rad/sec the magnitude and phase angle of the transfer func
tion are seen from Fig. 4-28 to be -35 db and -36°, respectively. The magnitude
-20
- ~
~ ChL I
0
40
1 y
,
.,'"
.0 ~ -Corrected "
"" ~ magnitude ;;
~ -40 -80 x
-
x
o
...
'0
Campo~ite of
the asymptotes
\
\
~
120
----
'0
~
'"
<=
i! -60
'"
'c
""
'"
::lE
\ ' "~ ~
'"
.s::
a.
-80 160
.",
r--
-100 200
0.1 0.2 0.5 1.0 2 5 10 20 50 100 200
w. rad/sec
Fig. 4-28 Magnitude and phase characteristics for the transfer function
X 0.06(jO.Sw + 1)
V .. (j2w + 1)[1 - (w/lO)2 + .i2(0.3)w/l01
The analysis of linear systems 189
(j",) 2 2 (0.3)j",
'" j2", + 1 jO.5w + 1
100 + -----w- + 1
Total
0.1 -11 0 0 11
0.25 -27 +7 0 -20
0.5 -45 +14 0 -31
1.0 -63 +27 -4 -40
2.0 -76 +45 -7 -38
4.0 -83 +63 -16 -36
5.0 -84 +68 -22 -38
10 -90 +79 -90 -101
20 -90 +84 -157 -163
450 -90 +90 -173 -173
1000 -90 +90 -175 -175
of - 35 db corresponds to
1O-3~. = 0.018 m/volt
Therefore, the response of the system to Eq. (4-82) is given by
x(t) = 0.27 sin (4t + 24°) meters
Example 4-8
As a second example, determine the Bode plot for the transfer function
T(P) = 5(0.011' + 1)
1'(0.002p + 1)1
T( ' ) = 5(jO.01", + 1)
(4-83)
3'" j",ljO.OO2w 1)1+
190 PFinciples of electrolr1echanica/-energy conversio"
o
'< Slope =1- 20 db/decode JJ I
0
-20
~ Corrected magnitude
40
~k
com~osite of
0>
'"
-0
~
--..........
3-40
~ r- i'-..
the asymptotes
80 ~
'"'~"
I I
....
o
....
0
'"
-0
.-Ec: -60
Phose)
~ '"c:
120 0>
'"
0
0>
o ~
::;;; 0
.J:;
"'r-"
Cl.
-80
'" "-.......
160
-100 -200
10 100 200 500 10 3 10 4
w, radlsec
Fig. 4-29 Magnitude and phase characteristics for the transfer function
T(jw)
5(j0.Olw +1)
jw(j0.002w +1)2
The factor of j", in the denominator requires a slight extension of the techniques we have
developed for obtaining the frequency-response plots. The phase contribution of the
jw factor is -90 0 even at very low frequencies. Expressing the magnitude of the
transfer function' T(jw) in decibels, the jw factor contributes ~20 log,o w db to the
total magnitudj). Therefore, for each decade in frequency the factor subtracts 20 db,
or it has a slope of -20 db/decade even at very low frequencies. Remember, how
ever, that the decibel contribution of this term is zero only when w = 1 rad/sec. If
the w axis starts at 0.1 rad/sec, then this jw factor contributes +20 db to the total
magnitude plot. With w starting at 10 rad/sec, the contribution is -20 db.
Figure 4-29 shows the magnitude and phase plots of Eq. (4-83). At w = 10 rad/
sec the magnitude of T{,iw) is given by
IT(jl0)1 = 20 loglo 5 - 20 loglo 10 = -6 db
Observe that the magnitude plot starts at -6 db with a slope of -20 db/decade.
The repeated real factor has its corner frequency at w = 500 rad/sec (which is 1/0.002).
Past this frequency the factor contributes a slope of -40 db/decade, since it is raised
to the second power. Also, all its decibel corrections are double those of a nonrepeated
factor. The phase characteristic at w = 10 rad/sec starts at -84 0 because of a +6 0
contribution from the numerator factor. Also, all phase-angle contributions from
the repeated factor are double those of a nonrepeated real factor.
f(t)
au) =Iim f(tl
a-O
11--_---.
a
11 kl(t) kF(s)
12 klNt) + k.j.(t) k 1F I (s) + k,F.(8)
13 I(t - bluet - b) e""F(s)
-
14 f I(t) dt = I~)
F(s)
-+-
s
p-1/(0+)
s
15 ff I(t) dt = I~t;
F(s)
-
S2
+ - 2- + -
1'-1/(0+)
8
p- 2/(0+)
8
16 II··· /../(t) dt = :.
F(s)
8"
p-1f(O+)
- + - - + ... + -
8"
p-nf(O+)
8
The analysis of linear systems 193
the time when the argument is zero; thus oCt - b) would be a unit
impulse at time t b. On substituting the unit impulse oCt) into
Eq. (4-84), we have
The only time when the integrand 0(t)e-B1 is not zero is at the time t 0+,
as seen from Fig. 4-30. At this ~ime instant the integrand, reduces to
simply oCt), which has unit area. In fact for any finite f(t) we can say
The second transform pair in Table 4--4 is for the unit step function
u(t) pictured in Fig. 4-31. The function u(t) is zero for all values of its
argument which are less than zero and positive unity for all positive
values of its argument. Therefore, u(t - b) is a unit step function, with
the step occurring at time t = b.
Transform pairs 3 and 4 are actually just special cases of transform
pair 5, with n = 1 and n 2, respectively. Notice that a u(t) is included
with' the time function to ensure our remembering that the transform
is defined only in the time interval t = 0+ to t = 00 and gives no infor
mation for t < O. Thus, if we are given the function 1/(8 + a) and are
asked for the corresponding time function, the correct reply is e-"'u(t),
and not just e-"'. The two answers are equivalent only for t > O. By
letting the constant a in transform pair 5 be equal to zero, w.e have the
unit step for n = 1, the unit ramp for n = 2, the unit parabola t2u(t)
for n = 3, and so forth for higher powers of n.
Transform pairs 6 and 7 are exponentially damped sinusoids.
Notice that the transforms of these time functions lead to functions of 8
which have complex conjugate denominator factors of the form 8 + a + jw
and s + a - jw.
In transform pairs 8 to 10 we have the transform of the derivative
operation. With regard to the general nth derivative given by pair 10,
notice that the Laplace transform is obtained by substituting F(s) for
f(t) and 8" for the differential operator p". In addition, the transform
contains a series of initial conditions multiplied by s raised to various
194 Principles of electromechanical-energy conversion
t
Fig. 4-32 A shifted
(al (bl time function.
(4-94)
(4-95)
Example 4-9
At time t = 0 a mass of M kilograms is moving with a velocity Va m/sec in the positive
x direction. A viscous force having a coefficient of D newton-sec/m opposes this
motion. Determine the velocity of the mass valid for all time t ~ O.
Solution: In operator notation, the equilibrium equation for the system valid for
all time t ~ 0 is given by
(Mp + D)v = 0
where we have used v = x as the velocity of the mass in meters per second. By
Laplace-transforming the equilibrium equation, we have
M[sV(s) - v(O+)j + DV(s) = 0
From Table 4-4 we have used pair 12, which equates the transform of the sum of two
time functions to the sum of the individual transforms. The capital V(s) stands for
the .e[v(t)j, and v(O+) is the value of v(t) at time t = 0+.
On solving for V(s), we have
Va
V(s) = 13 + D/M (4-96)
where we have substituted Va for v(O+) as the initial velocity. The time function
corresponding to Eq. (4-96) can be easily obtained by scanning the right-hand column
of Table 4-4 for an identical function of s. Pair 3 with a = D / M matches this
requirement, and therefore we have
(4-97)
as the x velocity for all time t ~ O. Notice that Eq. (4-97) says that v(t) is zero for all
time t < O. This mayor may not be true. However, the Laplace transform is not
valid in the time interval t < 0, so we do not expect a valid solution for values of
t < O. The inclusion of the unit step function u(t) helps us remember this fact.
Figure 4-33 shows a block-diagram representation for the transformed equation
(4-96). The input to the block is the initial momentum Mv(O+), and the output is the
transformed velocity V(s). The transfer function contained within the block is
obtained by substituting the Laplacian variable 8 for the differential operator p.
Remember, however, that the appropriate initial-condition inputs must be included in
the transformed block diagram as shown in Fig. 4-33.
Example 4-10
Figure 4-34a shows two rotating inertias 11 and 12 kg-m 2 coupled by means of a clutch
and a shaft of compliance K rad/newton-m. Each inertia has a windage or rotating
viscous friction coefficient of DI and D2 newton-m-sec/rad. At time t = 0 the inertia
I s is stationary and inertia I I is rotating at an angular velocity of ("'.}o rad/sec. The
clutch is then closed. Find the angular velocity of 12 for all time t ~ 0 after the
clutch is closed.
The analysis of linear systems 197
...
Solution: The equilibrium equations valid for all time t 2: 0 can be written by
using Lagrange's equation or, for a case that is this simple, directly by iUflpection.
Another technique also is found to be of some use. Consider the electrical network
shown in Fig. 4-34b with capacitors of J 1 and J 2 farads, admittances of D 1 and D. mhos,
a.nd a.n inducta.nce of K henrys. The node volta.ges are ta.ken to be WI and Wz with
respect to the datum node. Writing the nodal equilibrium equations for this network
would yield equa.tions identica.l with those we would write for the original system
shown in Fig. 4-34a. Such a network is often'called the electrical analog of the mechani
cal system. Mechanica.l velocities become node voltages; forces and torques are
analogs of electrical currents. Mass and inertial elements are capacitors drawn from
their respective nodes to the datum node. Inertial velocities are always with respect
to a stationary reference, which in the a.nalog is the da.tum node. Viscous coefficients
wl(tl K
0,
Datum
(bl
p-l(W,-W2)(O+)
K$
~ Dotum
(e)
Fig. 4-34 (a) A lumped-linear rotational system. (b) The analog electrical net
work. (c) The La.placia.n equivalent circuit.
198 Principles of electromechanical-energy conversion
become conductances, and compliances are drawn as inductances. Notice how the
same connections are used in the analog as in the original system.
Another way to show the equivalence of the analog circuit is to write the state
functions for the two diagrams. These state functions are seen to be absolutely
identical; thus the same equilibrium equations must result once Lagrange's equation is
employed. Writing the equilibrium equations for the system, from Fig. 4-34a or b, we
have
- ill 1 0 (8) + (J21l + D. + is) fh(s) J.w.(O+) - ill [/ (w. - wtl dt ]'_0+
(4-98)
The transformed equilibrium equations show the substitution of s for p, with the
initial-condition terms appearing as sources on the right-hand side of the equations.
The term [ / ("" - "'.) dt 1-0+ is simply the angular twist of the coupling shaft at
time t 0+. This quantity is analogous to the initial flux linkages in the inductor of
Fig. 4-34b. With the clutch open the angular twist can be taken to zero. The twist
cannot change instantaneously, because the angular positions of the two inertias can
not change instantaneously without an infinite angular velocity. Therefore, these
two integral initial conditions are zero, as is the term ",,(0+). The initial condition
"" (0+) is given as (WI)..
From Eqs. (4-98) we can draw an additional analog circuit known as the Laplacian
equ.ivalent circu.it, as shown in Fig. 4-34c. Here the node variables are the transform of
the angular velocities. The initial conditions become Laplacian current sources.
All elements involving a p operator in the original network must have an initial
condition source in the Laplacian circuit representation. By writing the nodal equa
tions of Fig. 4-34c, we obtain Eqs. (4-98), thus verifying this circuital representation of
the equations.
The transform of the angular velocity of J. can be obtained from Eqs. (4-98) by
using determinants as follows,
1
J,Il + DI + Ks
1
- Ks o
where
1 1
Jls + D, + Ks
1 1
- Ks J 2B + D. + KG
After expanding these two determinants, the result can be manipulated into the form
(w,)o
KJ.
(4-99)
The Qnolysis of linear systems 199
The only technique we are going to use for finding inverse transforms (i.e., the
time functions corresponding to functions of 8) is that of identifying the particular F(s)
in a table similar to Table 4-4. If the particular F(s) is found in the table, then the
corresponding f(t) is obtained by inspection. Very often, however, the transform
functions of interest become quite complicated, and an enormous table would be
necessary to include all possibilities. Equation (4-99) giving [l.(s) has a third-order
polynomial in II as its denominator. An examination of Table 4-4 shows that it con
tains no such function. Generally, if nume~ical values for aU the parameters are
known, then a mathematical manipulation known as a partial-fraction expansion can be
used to separate the function into parts which are always individually found in a table
as simple as Table 4-4. Even if values of the parameters are not known, the general
character of the response can be determined from the transformed function. In the
next section we shall study techniques which provide pieces of information concerning
the time function from the s function even though the inverse transform cannot
formally be determined.
Suppose, for a particular set of parameter values, Eq. (4-99) reduces to
(4-100)
(8 + 2)[(8 + 2)2 + I)
This expression for [l2(8) is now rearranged into the sum of three fractions,
A = [(8+2)°2(8)]
("'1)0 0 __ 2
= (8
1
+ 2)2 +1 I
a=-2
=1
B [ (8+2+;1)02(8)]
(W1)0 .--2-i1
= 1
(8+2)(8+2 ;1)
I0=-2-;1
1 1
-2
and C is given by
C [(8+2_;1)°2(8)] =B* 1
(W1)o '--2+i1 - '2
where B* means the complex conjugate of B. To conjugate a complex number,
replace every j by -j.
On substituting A, B, and C into Eq. (4-101) and combining the last two fractions
we have
200 Principles of electromechanical-energy conversion
Transform pairs 3 and 7 in Table 4-4 are recognized to correspond to these two partial
fractions. Pair 12 tells us that the inverse transform of the sum of two 8 functions is
equal to the sum of their individual inverse transforms. Thus the angular velocity of
the second inertia is given by
W2(t) ,.. (wl).[e-"(l cos t)]u(t)
The term 1 - cos t oscillates between a value of zero and +2, with the coefficient e-2<
causing these oscillations to exponentially decay to zero as t goes to infinity. When I
reaches 2 sec, r2< has a value of rt "" 0.02; thus the angular velocity w. is very close
to being zero after four time constants, or 2 sec.
Linear systems with coefficients whose values are independent of time can
always be described by a relationship in the form
Q(t) (4-102)
where p = dldt is the differential operator and all a .. , an-I, . . • , au
coefficients are constants. The forcing function Q(t) is usually a specified
function of time, and the system response x(t) is desired. If the actual
system has many forcing functions, and correspondingly many response
coordinates, then we must first do some manipUlating to obtain a single
equation involving just one response, as given by Eq. (4-102). Section
4-3 gives a complete discussion of these manipulations.
Laplace-transforming Eq. (4-102) term by term gives us
-r als -r ao»)[(s)
with all initial conditions moved to the right side of the equation. The
response of the system is therefore given, by
)[(s) = Q(s) -r initial-condition terms (4-104)
ans" -r an_IS" -r . . . -r al8 -r ao
I
tion from the Xes) function directly without having to actually obtain the
corresponding x(t).
The most general form that Eq. (4-104) will take, for our purposes,
can always be expressed by
(4-105)
Equation (4-105) indicates that any response can always be given as the
ratio of two polynomials in s. Further, we can always have the order of
the numeratQr polynomial, in this case m, at least one integer unit less
than the order of the denominator n. If this were initially not true
(that is, n were not greater than m), then the denominator could be divided
into the numerator to get Xes) as a polynomial in s plus a remainder in
the form of Eq. (4-105). The inverse transform. of each of the polynomial
terms can easily be found, and only the remainder of the long-division
process, given by (4-105), is of interest.
On factoring b", and an from the numerator and denominator,
respectively, we have
Xes) = b", s'" + (b",_I/b.,.)S_l + ... + (bI/b",,)s + bo/b"" (4-106)
a" s" + (an-I/a,,)s" 1 + ... + (at/an)s + ao/an
In theory Eq. (4-106) can be factored into the form
Xes) = b.,. (s + {Jl)(S + (J2) ••. (s + (Jm) (4-107)
a" (s + al)(s + at) . . . (8 + an)
where the numerator has m factors and the denominator has n factors.
As discussed in Sec. 4-7 in connection with the Bode plot, these factors are
real or complex conjugate in form and are either distinct or repeated.
If Eq. (4-106) can be factored into the form of (4-107), then a partial
fraction expansion can be performed and Eq. (4-107) is split into a sUm of
terms given generally by
for i = 1, 2, . . . ,r (4-111)
If r = 1, meaning the factor is not actually repeated, then Eq. (4-111)
reduces to Eq. (4-109), since 1/0! equals 1.
The important thing to notice is that all of the terms in the expansions
of Eqs. (4-108) and (4-110) are given in Table 4-4. We have started with
a general response, Eq. (4-105), from which the complete time response
can be obtained if we can factor the polynomials in s. Names are com
monly given to these all-important numerator and denominator factors
in Eq. (4-107). A value of s which makes Xes) equal to zero is liter
ally called a zero of the function. This Xes) has m zeros located at
s -1'1, -1'2, ... ,-I'm' Similarly, a value of s which makes Xes)
have an infinite value is termed a pole of the function. The function
Xes) has n poles which are located at s -aI, -a2, . . . , -an. At
values of s near infinity we have
X( co) = lim Xes)
(4-112)
as shown in Fig. 4-35. The fourth pole is taken with a4 equal to a nega
tive real number, thus placing this pole at a point on the positive real axis.
In a similar fashion all the remaining poles and zeros of X(s) can be
plotted in the s plane.
Referring to Table 4-4 and Eq. (4-108), we see that the poles of the
F(s) functions determine the character of the response. For real distinct
or repeated poles transform pairs 3 to 5 summarize the relationships
between the terms in F(s) and their corresponding time functions f(t).
Thus;if the partial-fraction expansion of Xes), as given by (4-108) and
(4-110), contains poles locate~ on the real axis in the s plane, meaning
that a particular a, is a real number, then the transform pairs 3 to 5 give
the corresponding time function. On examining these transform rela
tionships we see that poles located on the negative real axis (that is, ai is a
positive real number) contain a damped exponential in their correspond
ing time function. Using L'Hospital's rule, we can show that!
tive real axis, then the time response will contain a term of the form e'"
which grows exponentially to infinity as t goes to infinity.
For complex conjugate poles, transform pairs 6 and 7 give us a
similar result. For poles located in the left half of the s plane the time
response contains a decaying exponentiaL The total response oscillates
sinusoidally with an exponentially decreasing amplitude. On the other
hand, complex conjugate poles in the right half of the s plane (RHP) oscil
.late sinusoidally with an exponentially increasing amplitude as t increases.
We can summarize these conclusions by two simple statements:
1. Poles in the LHP correspond to exponentially decaying time
responses.
2. Poles in the RHP correspond to exponentially growing time
responses.
Poles located on the boundary, or jw axis in Fig. 4-35, between the
RHP and LHP, give time responses which neither grow nor decay. If
a in transform pairs 6 L \d 7 in Table 4-4 were zero, then we would have
complex conjugate poles located on the jw axis. The corresponding time
responses are sinusoidal with amplitudes neither growing nor decaying.
For a distinct pole located at the origin, we have a step response u(t)
as the corresponding time function. This response neither grows nor
decays, but simply remains constant in time. Thus with an s-plane
plot the form of the corresponding time function can be completely
predicted simply from the pole locations. From Eqs. (4-109) and (4-111)
we see that the zeros enter into the calculation of the partial-fraction
expansion coefficients. Thus the zeros of the function serve only to
influence the magnitude of the responses.
In general, the transform of any time response can be expressed
as the ratio of two polynomials in s in the form of (4-105), A pole-zero
plot for Eq. (4-105) cannot be obtained unless these polynomials can
be factored. Any information that we can obtain concerning the factors
or, equivalently, the location of the poles and zeros without actually
factoring will provide information about the form of the time response.
For example, if we can show that all poles are located in the LHP, then
we know that the time response must decay with increasing time. For
a linear system this decay property will serve as a definition of stability.
A stable linear system has a transfer function of s containing only LHP
poles.
pes) = a"s" +
Second row
(4-115)
First we must formulate an array of numbers known as the Routh array.
The complete array is as follows:
a" a..~2 a_4 a,,~6
The first two rows of the array are composed of the coefficients in the
polynomial P(s). The first row starts with the coefficient of the highest
power of s, in this case an. The second term in the first row is the (n - 2)
coefficient, and so forth taking every other term in P(s). In the second
row of the array are the remaining coefficients in the polynomial (4-115).
The third row of the Routh array is formed in the following manner.
For the first term in the third row bi we have
_, I
Ia"'x
a
hi = a..-I a-. = a..-Ia..-, - a"a..-6
a..
_t a.._t
Only the second column of the determinant changes to the terms in the
two rows above b2 located one column to the right of b2 • Similarly,
for ba we have
a,,_la,,_6 - a"a,,_l
an-I
The same rules are used for all remaining rows of the array. Thus,
for the fourth row we have
and
Always use the first two terms in the two rows above for the first column
of the determinant· and the two terms in the two rows above and one
column to the right for the second column of the determinant. Divide
the product of the solid-arrow terms minus the dashed-arrow terms by
the second-row first-column term in the determinant. Zero elements
do not have to be entered into the array.
The number of sign changes in the first column of the Routh array equals
the number of roots of P(s) with positive real parts. Thus; given any
denominator polynomial, we can formulate the Routh array. By count
ing sign changes in the first column we know the number of RHP poles.
A stable system has no RHP poles 1
Example 4- J J
Use the Routh criterion to show that
x _ lO(s + 1)
(s) - (s + 2)(s - 1 + j3)(s - 1 - j3)(s + 3)
has two poles in the RHP.
Solution: The denominator of X(s) is given by
+ 1 6 60
+ 3 38
-20/3 60
+ 65
+ 60
The terms are obtained a8 follows:
I. First row consists of a" aI, and ao in P(a).
2. Second row consists of a, and al in P(a).
3. Third row:
Seoond term = \3
1 ><,:60 I
3 '" 0 - 60
.
Third term, etc. = I!><:~I
3 = 0
4. Fourth row:
First t
erm
= I 3 .... ><':'381
-20/3 '"60
'-20/3
= 38 + 27
.
= 65
. t -2gl;'<~1
Second term, etc. = -20/3 = 0
5. Fifth row:
-20/;.x 60 I
65 01
etc.
Example 4-12
As another example of the Routh criterion consider the angular-velocity response
function for the inertia J 2 in Example 4-10. This angular-velocity response is given
by Eq. (4-99), which for convenience is repeated as
("'1)0
KJ.
Use the Routh criterion to see if there are any real values of the parameters which
would yield an exponentia.lly growing response for ",.(t).
Solu.tion: The Routh array for the denominator is given by
D1 +Ds
J, J.
01
Dl +D.
KJ 1J.
where
01 = 1.. (.!. + 1..) + D D.
1 _ (D, + D.)/KJIJ.
K J1 J. J,J. D,/J, +D./J.
Notice that for positive values of all coefficients every first-column term in the array
must be positive. Thus we have no sign changes and no RHP .poles in O.(s). For
non real negative coefficients we could have sign changes and an unstable response.
For example, if the shaft compliance K were negative, 0, could be negative and the
first column would have one sign change and hence one RHP pole. A negative K is
not a physically realistic parameter.
4-13 SUMMARY
In this chapter we have reviewed certain properties and techniques
applicable to a linear system. The homogeneity and superposition
The ollOlysis of linear systems 209
PROBLEMS
4-1 The equilibrium equations for a group of physical systems, where x(t) is the
system response, are given by the following equations. Determine whether these
systems are linear or nonlinear by applying both the homogeneity and superposition
tests. Take Q(t) to be a known forcing function.
a. (5p2 + 6p + 2)x(t) = Q(t)
b. (16t2ps + 4t~~ sin 6t p + 3) xO) = Q(t)
c. pep + l)x(t) = Q(t)
d. [5p' + (px)(6p) + 2]x(t) = Q(t)
4-3 Given the differential operators L,(p) = 6tp, L.(p) = 3p' + 2p, and
L3(P) 4pl.
a. Check the commutative and associative laws of addition.
b. Check the distributive and associative laws of multiplication.
c. Find the two operators which obey the commutative laws of multiplication.
Prove your answer.
d. Which of these operators are linear differential operators and which are classed
as constant-coefficient operators?
4-4 a. Explain why the differential operator p can be manipulated as an algebraic
coefficient in the solution of simultaneous constant-coefficient linear differential equa
tions by means of Cramer's rule.
b. If n ... 4 in Eqs. (4-24), find the 412 and 422 cofactors required for the solution
of x•.
4-5 Reduce the block diagram shown in Fig. P4-5 to the form shown in Fig.
4-14a. Find expressions for G(p) and H(p).
The analysis of linear systems 211
Fig. P4-5
4-6 a. For the linear electrical network shown in Fig. P4-6 write the loop equa
tions in the form of Eqs. (4-24).
b. From these loop equations determine the linear differential operator relating
i.(t) to vet). If i.(t) = L(p)v(t), find L(p).
C. Draw a block diagram representing the two loop equations.
d. Reduce the diagram of part c to a single block with vet) as the input and i.(t)
as the response output. Compare the operator in this single block with L(p) from
part b.
Rj
vI!)
Fig. P4-6
.....7 a. Suppose vet) in Prob. 4-6 is a d-c voltage source of magnitude v. volts.
Reduce the general loop equations to their doc steady-state form by replacing p by
zero. Do the same thing with the general block diagram for Fig. P4-6.
b. If vet) = V2 V sin wt, rewrite the equilibrium equations and the block dia
gram for a steady-state description of the network. What is p replaced by, arid in
what form are the variables expressed?
C. Briefly justify the two different substitutions for p to obtain the d-c and
sinusoidal steady-state responses.
4-8 If the denominator polynomial in Eq. (4-65) is of seventh order, thus n = 7,
make a list of all possible combinations of factors. How many real factors and how
many complex conjugate factors could the polYll,omial contain?
4-9 For the following linear differential equations in operator form determine
the frequency response in the form of a Bode plot. Assume vet) to be a sinusoidal
input function and x(l) to be the response function. For the magnitude plots, draw
the asymptotes and then add the correction quantities.
a. (p8 + 12p' + 20p)x = (1000p)v
b. (p' + 16p s + 100p)x = (lOp + 50)v
C. (p' + 16p· + 64 p 2)X (15p + 30)v
212 Prim:iples of electromechanical-energy conversion
4-10 For the asymptotic magnitude plot shown in Fig. P4-l0 determine the
transfer fUJlction GU",). Be Bure to evaluate the magnitude of the constant multiplier.
-40 db/decode
5 20 40
liJ. rod/sec
Fig. P4-10
4-12 By using the Laplace transform method, solve the following system
equations subject to the given initial conditions:
a. (pi + 4p)x = 6u(t) X(O+) = OJ x(O+) - 0
b. (p' + 4p)x = 66-"'14(0 x(O+) == OJ x(O+) ... 0
c. (p' + 4)x = 3 sin 2t '14(1) x(O+) ... OJ x(O+) = 0
d. (3p' + 15p + 18)x = 0 x(O+) = 2; x(O+) = 5
4-13 Expand the following fUJlctions in partial fractions and find the inverse
transforms:
5(s + 1)(s + 2)
a. G(s) = s(s + 3)(8 + 4)(s + 5)
10(s + 0.5)
b. G(8) == (48 + 1)(23 + l)(s2 + 48 + 5)
0.2(s + 3)"
3+1
0.2 1'0(5)
(lOs +1)(15s + 1)
Fig. P4-16
4-17 For the following transfer function list the conditions on the coefficients
such that the system is stable.
G(s) _ K(b.8 2 + b,s + bo)
. - a,8' +
a38' + a,8' + alS + ao
5
OF ELECTROMECHANICAL SYSTEMS
Iron Slotor
Moss of iron
~Gro"itY plunger" M
voltage source vet) is connected to the N-turn coil. The source has a
resistance R. ohms, and the coil resistance is R. ohms. The motion of
the iron plunger, of mass M kilograms, is restricted by a spring of com
pliance K m/newton and a viscous damper with coefficient D newton
sec/m. A brass pole face, with an area of A square meters and a thick
ness of d meters, prevents the plunger from forming a complete iron
path. The force of gravity is assumed to be acting in the downward
direction as shown.
In Example 3-5 we determined a Lagrangian for this system to be
where a ;= p.oN 2A/2 and b is the position of the plunger when the spring
is at its free length. When x = b, the spring does not exert a force
either up or down. The Rayleigh dissipa.tion function and the external
force terms are respectively given by .
x - bai'
M:f + Di + + (d + X)I Mg
(5-5)
2a , . 2aiX
The details in deriving these equations are given in Example 3-5. The
current symbol i has replaced Ii simply for convenience in the two equilib
rium equations (5-5) and (5-6). For a specified voltage function vet) we
are interested in the position of the plunger given by x(t) and the current
in the coil i(t). However, the equilibrium equations are nonlinear, since
the unknown dependent variables x and i appear raised to powers other
than unity. Also appearing are products of the variables with their
derivatives.
In order to obtain some information concerning the response of the
system, let us assume that the applied voltage is held constant at some
value Vo. Since the system contains both electrical- and mechanical
energy-loss elements, namely, R., R., and D, the steady-state response
Respona oleledromechonicol systems 217
II = Mg -
Xo - b
---x- (5-13)
and
avo! 1
II = (R. + R.)1 (d + xo)1 (5-14)
Now if we plot It and It as functions of Xo for various fixed values of Vo,
the intersection of the two curves represents a solution of Eq. (5-12).
I In all of our work the aubeeript zero will be used to designate conat&nt values of the
variables.
218 Principles of electromechanical· energy conversion
Increasing Ivol
-d 0
In Fig. 5-2 the two functions are plotted. The first function Ii> defined
by Eq. (5-13), is a straight line with slope -1/K and intercepts of
Mg + b/K and KMg + b for the vertical and horizontal axes, as shown.
The second function h, defined by Eq. (5-14), is a set of equilateral
hyperbolas about the Xo = -d line. Positive and negative values of Va
give the same two hyperbolaa, since Vo appears raised to the second power
in I:. Four typical plots of h, shown in Fig. 5-2, are drawn with four
specific values for Vo. The direction of increasing the magnitude of Vo
is also noted.
For the curves labeled with l's, we have three intersections of the h
function with the II function. These three intersections are denoted by
dots. Increasing Ivol still further to the curves labeled with 4's, we have
only the one intersection indicated by a diamond. For each value of
Ivol the corresponding values of Xo, which m~ke It = h and therefore
satisfy Eq. (5-12), can be taken directly from Fig. 5-2. Plotting these
values of Xo as a function of the magnitude of Vo gives the curve shown
in Fig. 5-3. The intersection markings shown in Fig. 5-2 are also included
in Fig. 5-3. Notice that there is a dot at each of three values of Xo:
slightly less than -d, slightly greater than -d, and just less than
K M g + I). The triangle intersections corresponding to curve 3 show
only two values of Xo, as seen in Fig. 5-3. The value of Ivol at this limiting
point is designated as v::'u.
The value of v::,ax can be found by setting dvo/dxo equal to zero.
Equivalently, we can set dvo 2 /dxo equal to zero and thus get
Xo = -32 (K M g + b) d
- -
3
(5-16)
(KMr;+bl r--'----__
."..,...", lIotn
---- -
"" Fig. 5-4 Summary of the
stable and unstable static operat
-------------- ing points showing the jump
phenomenon.
Figure &.3 is a plot of all the possible d-c or static steady-state operating
conditions for the electromechanical system pictured in Fig. &.1. If a
d-c voltage 110 is impre8l!led on the N-tum winding, then the plunger will
finally come to rest at a position Xo as specified by Fig. 5-3 or, equivalently.
by Eq. (5-12), from which the curve was drawn. Values of Xo less than
zero are not physically realizable, since the plunger is stopped by the
brass pole face as shown in Fig. &.1.
The time-dependent operation of the nonlinear electromechanical
system can be studied by using the linear-system techniques discull8ed in
Chap. 4 if the motion of the system is limited to the neighborhood of a
static operating point. The equilibrium equations for the electro
222 Principles of electromechanical-energy conversion
mechaniral devire shown in Fig. 5-1 are given by (1i-5) and (5-6), which
for cOllvenience are repeated as
b 2
lift + D:i: + x ai
+ -----
(d + X)2
= Mg (5-19)
x(1) (5-22)
and
(5-23)
where the subscripts 0 and 1 have the significance described with Eq.
(1i-21). Here we arc just guessing that the response of the nonlinear
system to a constant forcing furwtion plus an inercmental time-varying
forcing function will be in the form of a (~onstant plus an incremental
time variation. As VI(t) is made smaller and smaller, the assumptions
of Eqs. (5-22) and (5-23) become increasingly more accurate.
Not only will earh of the response coordinates x(t) and i(t) be com
posed of a d-c component plus a small variation but each of the terms
in the equilibrium equations will have the same general form. For
example, the mechanical force of electrical origin given by
f. (5-24)
f( X,l.) = f( Xo,lo
.) + af(x,i)
!> I (-
X Xo) + af(x,i)
!> . I (.l - to
.)
uX ro,i'o ut. zo,io
2 2
+ ! ' a f(x,i)
!> 2
I . (-
X Xo
)2 +! a f(x,i) I (- )(. - .)
2'!>!> . . X Xo t to
2. uX XO,IO • uX u1. ZO,IO
The first term is j'(Xo,io), or the force of electrical origin evaluated at the
operating point. The coefficient in the second term is the aj./ax evalu
ated at X = Xo and i = io. All terms involving XI and i l as products
or raised to the second and higher powers are neglected.
The first term in Eq. (5-20) can similarly be approximated by
realizing that i = io + i l ; therefore, i = 0 + i l and thus the constant
operating level for i = di/dt is io = O. The term therefore is expanded as
2ai 2aio 2aio 2a:
d +X "'" d + Xo - (d + xo)2 XI + d + Xo II
which reduces to
2ai 2a,
(5-27)
d + x"'" d + Xo II
since io = o.
224 Principles of ./ectromechanica/.energy conversion
All of the terms in Eqs. (5-19) and (5-20) can similarly be expanded
to yield the following set of approximate equilibrium equations:
- b Xl aio' 2aiot
(d
2aio .
+ (d + Xo)! ~l Mg (5-28)
2aio
2
[ Mp
2
+ Dp + K1 - (d + xo)' .]
Xl + [2aio] .
(d + xo)! ~l = 0 (5-30)
- [ (d
2aio
+ xo)! P ] Xl + [2a (
d + Xo P + R. + Re)] ~l. = VI(t
) (5-31)
(5-36)
2aD
at + + (R. + Re) M
= d Xo (5-37)
2a
al = K(d + xo) + (R. + R.)D
(5-38)
2ai o2
aa = (R. + Re) [ K1 - (d + xo)' ]
(5-39)
226 Principles of electromechanical-energy conversion
ao
The coefficients a; and a2 are always positive numbers, since Xo can never
be negative. Therefore, we must have
al -
aaao
--
a2
>0 (5-41)
and
ao> 0 (5-42)
for all of the poles of 1/4.(11) to be located in the LHP. When we sub
stitute Eqs. (5-36) to (.'>-39) into the stability condition given by (5-41),
the following inequality, after some simplification, results:
4a D
2
+ 2aD2(R. + R.) + DM(R. + ReP
K (Xo + d)2 d + Xo
Inequality (5-43) is satisfied for all positive values of Xo and for positive
values of the parameters.
Stability of the operating points now depends only on satisfying
condition (5-42). From Eq. (5-39) this condition is equivalent to
2aKvo2 ]~' (5-44)
Xo > [ (R. + ReP - d
upon substituting io = vo/(R. + Rc) according to (5-11). The steady
state characteristic shown in Fig. ,1)..,3 was originalty plotted from Eq.
(.1)-12) for vo 2, and 8ubst.ituting the result ill the stability condition (5-44)
On examining Fig. 5-3, we see that the Routh criterion verifies the fact
that only the upper portion of the characteristic represents stable operat
ing points as indicated in Fig. 1)-4. If the system were somehow posi
tioned at an unstable operating point along the dashed A-B curve in
Fig. 5-4, then any incremental change in Vo would cause the system to
move from that position. Every practical voltage source has some
slight magnitude variation; therefore, the system could never remain
at an operating point on the dashed A-B curve.
With a constant or d-c voltage excit,ation applied to the electrical
port of the system shown in Fig. 5-1, the mechanical member moves to a
steady-state position. In Sec. 5-1 these steady-state positions or static
operating points were determined from the equilibrium equations as
functions of the applied d-c voltage. Figure 5-3 represents a plot of
the d-c steady-state plunger position as a function of the d-c applied
voltage. However, certain operating points are found to be physically
unstable. Moving the system very slightly away from such an unstable
operating point results in a gross motion to an entirely new operating
point. In this section we mathematically formulate a test for stability
or instability of the operating points. Assuming that the system remains
in the neighborhood of the point being tested, a set of incremental linear
differential equations is derived from t.he original set of nonlinear equilib
rium equations. The incremental equations are a reasonably good
description of the physical system so long as the system remains very
near the operating point. If the linear incremental equations predict a
bounded responscr for an incremental change in the system excitation,
then we can predict that the operating point is stable. On the other
hand, if the incremental equations predict an exponentially growing
response, then the operating point is unstable. The incremental equa
tions serve as a tool for studying stability or instability of operating
points.
f'm=a--Brass pole
face
Nonma\jnetic Iran
support plunger
tJ
y
Stationory
reference Fig. 5-5 A mechanical-ro-electrical
transducer.
Response of e/ectromecltonicol systems 229
Vertical motion
Transducer
Machine
The equilibrium equations for the system are identical with those
obtained when the stator of the device is stationary with the addition
of an external force on the mechanical equation. With the stator struc
ture stationary the only external force acting in the vertical direction
is due to the acceleration of gravity given by g. With the stat.or position
given by y, the total external acceleration acting on the plunger becomes
g + 'Ii and the total external mechanical force is M (g + ii) acting in the
downward direction. The equilibrium equations, given by (5-5) and
(5-6), are therefore modified to
.... x-b ai 2 _
Mx + Dx + ---g- ~ (d + X)2 = M(g + y) (5-46)
2a, • . 2aix
d + x ~ + (R. + R.)~ - (d + x)2 = v(t) (5-47)
where a = ".ON 2Aj2, in which A is the pole face area, N is the number of
coil turns, and "'0 = 4r X 10-7 hjm is the permeability of air. The con
stant b is the value of x when the spring is at its free length. All the
remaining parameters are shown in Fig. 5-5.
A linear transducer can be obtained by having the nonlinear electro
mechanical system restrained in its motion to the neighborhood of an
operating point. All time-dependent forcing functions are made to
have the form of a constant plus an incremental time-dependent varia
tion. Therefore, in Eqs. (5-46) and (5-47) we specify that
[ d-~
+ II + (R. + Rc)i
Xo
l - ~a~_o
(d + xo)2
Xl] + [(R. + Rc)io]
= [Vl(l)] + [vol
The equilibrium equations must be valid wit'h the system resting at
the operating point, which means that all variables with SUbscript 1 are
equal to zero. Therefore, at the operating point the first bracketed
terms on both sides of Eqs. (5-53) and (.5-54) are equal to zero and the
ResponM of electromechanical systems 231
'+ (R •
2a Xo 11 + R c) h. - +-Xo 0)2 XI• =
(d 2ai VI
( )
t (5-58)
amp-sec'/m (5-63)
(rad/sec)% (5-64)
(5-65)
2a Lo
+ xo)(R. + Re) (5-66)
'T ". (d (d + xo)(R. + R.) R. + R.
The pteady-state operating-point equation (5-55) has been used to
eliminate io in Eqs. (5-63) and (5-64). The quality X;:'h' is defined as
I - - Velocity Position
I transducer i - - t r a n s d u c e r -
I
I
I
I
1fT
(al
Acceleration 1.
.._ _ _, Pasition
tronsducer I transducer
I
I
I
I
40 db/deCOde
I
I I
I I
1 I
I I
60 db/decode I
I I
1 I
1 I
I/T &J
(bl
Fig. 5-7 '.Asymptotic magnitude plot for the transfer function
I, k(j... jI
Y, - [(jw)'/.....• + 2r(j"')/..... + l](fr", + 1)
(a) lI.IlIIuming ..... «!,
T
(b) lI.IlIIuming ! « "''''
T
i2 ... Ap (5-71)
YI
234 Principles of e/edromechanical·energy conversion
Contour
electri('al transdueer, the out.put of the device iii! the ineremental change
ill the winding eurrcnt. The static magneti(~ field (~an be established in a
more convenient fashion by using a permanent magnet in the magnetic
cireuit, thus eliminating the need for an external d-c voltage source.
The system shown in Fig. »-8 is a ('ross-sertional view of a cylindrical
iron frame. The cross-hatched portion of the center post contains a
eylindri('al permanent. magnet, of le'ngth 1 and cross-seetional area A.
Magneti<, flux emanates from t.he nort.h pole of the magnet and circulates
around the iron path as shown in Fig. ii-8. A small air gap of length d is
cut in the iron path. The 1('II,,d,h of the air gap is nHwh smaller than the
radius of the gap a.
With the (~ontour shown in Fig. »-8 Ampere's cireuital law can be
used to determine the magneti(~-Rux densit.y in the air gap. Applying
Ampere's circuital law to the contour shown in Fig . .5-8 gives
Had + H",l o (5-72)
where 11 car is the magnetic-intensity veetor in the air gap and H ",a. is the
magnetic-intensity veetor in the permanent magnet. The permeability
of the remaining iron ('ircuit is assumed to be large enough that H can be
taken as zero along this part of t.he' (·olltour. Neglecting fringing fields,
the total magnetic flux in the air gap will equal the total magnetic flux in
the magnet. Therefore, we have
(5-73)
wheT(' En i~ the map:llcti(·-flux dpl)sity ill the air ~aJl and E", is the mag
netic-flux density in the magnet. The quantity 211'ab is the nominal sur
236 Principles of electromechanical-energy conver,ion
Shearing line
dA
face area of the air gap through which B. flows, and A is the cross-sec
tional area of the magnet through which B ... flows. I
The flux density in the air gap Bo. is related to the magnetic intensity
in the air gap Ho. by the relationship
(5-74)
where 11-0 = 4r X 10- 7 him is the permeability of air. On substituting
Eqs. (5-73) and (5-74) into Eq. (5-72) we have
21f'II-()Clbl
Bm - -""-<1X" H... (5-75)
8"
Shearin\! line
Slope" _ 21Tjl.pabl
4A
By usink the basic magnetic structure discussed in Sec. 5-5 and shown in
Fig. ,1-8, a wide variety of useful electromechanical devices can be con
struded. One such device is shown in Fig. 5-11. A concentrated circu
lar N-turn eoil of radius a is wound 011 a eylindrical coil form. The eoil
form and the N-turn ('oil are centered in the air gap by two springs each
having a ('omplialwe 2K. The ('ombined mass of the coil form and the
N -turn ('oil is M. The position of the coil with resped to the iron frame
is designated by the coordinate x, and the position of the frame with
respeet to a stationary reference is given by the coordinate y, as shown in
Fig. ii-II. If y is a specified time-dependent fUllction, the entire frame
wil! move with respc<,t to the stationary reference. In turn, the coil
form and I'oil could have some motion with respect to the iron frame.
Thus, the coil would be moving in the air-gap magnetic field established
by the permanent magnet, and a voltage would be induced in the coil.
The induced voltage bears some relationship to the specified coordinate
yet), and therefore the device can serve as a mechanical motion transducer.
The equilibrium equations for the device shown in Fig. 5-11 can be
determined by using a Lagrange formulation. Since there is magnetic
coupling between the permanent magnet and the coil, a loop formulation
is more convenient than a nodal formulation. Thus the magnetic
coenergy-state function is required. If a current q in the coil is defined to
be positive as shown by the dot and cross in Fig. 5-11, the total coil flux
linkages are as follows:! '
Since y is to be a known time function, only the mass M of the coil form
and N-turn coil is included in the Lagrangian. The velocity of the
mass M with respect to the stationary reference is iI +:t. The two
springs each of compliance 2K are assumed to be at their free lengths
when x = O. The x = 0 point could be taken with the coil centered
axially in the air gap.
Assume a mechanical viscous damping to exist between the coil form
and the iron frame, as shown in Fig. 5-11. The source of the viscous
damping will be discussed later in this section. Taking the resistance of
I The dot in the top circle indicates current coming out or the paper, while the cr088
Notice that the mechanical force of electrical origin f. is simply the total
length of the coil (21I'aN) times the coil current q times the air-gap field
B". Since f. has a negative sign on the left side of (5-84), the force is in
the positive x direction. With B" directed radially outward in the air
gap, the crOHS produet q X B is a veetor to the left in Fig. !i-1l, which is
the positive x direction.
Whcn it is used as a mechanical position transducer, the coil is
normally open-circuited and the induced voltage is measured with a high
impedanee device. The indueed coil voltage, which is also the voltage
of meehanical origin, is given by
(5-85)
(5-86)
as the open-eircuit induced voltage in the coil. Observe that this voltage
is the total length of the eoil (211'aN) times the relative velocity :i; multi
plied by the air-gap field B".
By taking q = 0, Eqs. (!i-84) and (5-86) can be used to solve for the
indueed voltage e as a function of the forcing position y(t). On eliminat
ing x between these two expressions we have
e =- M p2 + Dp + 1/K y
(5-87)
Response of electromechanical systems 241
r=~~ (5-91)
feoil = 27raBo x
The ohmic resistance of the coil form is simply 27rap/hb, where p is
the resistivity of the coil-form material. The force opposing the motion
of the coil form is similar to f. in Eq. (5-84); thus we have
(.5-93)
(5-94)
By altering the physical parameters of the coil form, the damping coeffi
cient D can be adjusted to obtain the desired optimum da.mping ra.tio
r = 0.6.
Response 01 electromechanical systems 243
( M p2 + Dp + k) x 21raB Ni a 0 (5-98)
(21raB a Np)x + (Lp + R)i = vet)
The current symbol i has been substituted for q. By solving for i as a
function of the forcing voltage vet), we have
Since the loudspeaker is used for generating audio tones, the fre
quency-response characteristics are of great interest. Therefore, assume
that I!(t) is a sinusoidal time funetion represented as sinor V. The ratio
of sinor V to the responding sinor I is the complex dynamic input imped
Re$ponse 01 electromechanical $ydems 245
The length of the coill has been substituted for the quantity 2raN in Eqs.
(5-102) and (5-103).
The reactance portion of the dynamic input impedance Z given by
(5-101) is usually much smaller than the resistance portion. Thus, as a
practical approximation the current input for a given applied voltage can
be represented as
(5-104)
The current i(t) is therefore in phase with the sinusoidal voltage v(t).
For a constant V the magnitude of I as a function of frequency has the
chara(~teristi(~ shown in Fig. 5-15. Th(~ millimum current occurs at the
mechanical resonant frequency
(5-105)
which makes the motional resistance R... have its maximum value. The
current at low and high frequencies has a magnitude roughly 3 times the
where I is the rms magnitude of the coil current. The quantity RJ2
contributes to the ohmic heating of the coil. The remammg average
power input is approximately equal to the radiated acoustical power.
Thus we have
p.;:OOUI!I (5-109)
t=~~ (5-112)
(Vl~.K)
2
k = D watt-sec2 (5-113)
i I i
/'"
./'
: ['-Corrected
curve ~
I
1 I ""'-..
1 r---...
I , I
I , I ~
!
I ! I ............
i I i
Wit
Frequency, rod / sec
Fig. 5-16 The output acoustical power as a. function of frequency with the rms coil
voltage held constant.
unit for a tra.nsfer relationship as defined in Cha.p. 4. Quite often the power, in
decibels, is ta.ken with respect to some reference power level P e. ThUll we would take
10 log" (P fPc) - 10 loglo P - 10 loglo Po and say that P is a certain number of
decibels above the decibels of the reference power level.
248 Principles of electromechanical-energy conversion
term contributes -40 db/decade, whieh combines with the first two
terms to give a net slope of -20 db/decade. Notice that Eqs. (.5-115)
and (.'i-llfi) give jdentical results at W Wn•
the structure shown in Fig. 5-8, placed end to end about the center line
of the device shown in Fig. 5-17. The two permanent magnets establish
a radial magnetic field B" in the air gap. Since fringing fields were
neglected in the calculation of the air-gap field given in Sec. 5-5 and
approximated by Eq. (5-77), we have the same radial field in the con
figuration of Fig. 5-17. Each magnet is establishing B" over half of the
air-gap length, which is designated as b. A cylindrical conducting coil
form with an attached N-turn coil is mounted in the air gap, as shown in
Fig. 5-17. The co~l form is supported by four identical springs each
having a compliance equal to 4K m/newton. The coil form is eon
strained to move only in the axial direction. The position of the coil
form with respect to the center of the air gap is designated by the coordi
nate x in .Fig. 5-17. When x = 0 the coil form is exactly centered an4
each of the four springs is stretched a distance c meters from its free
length. A positive value for x means that the coil form has moved to
the left with respect to the magnetic structure.
The mass of the coil form and the attached N-turn coil is equal to M
kilograms. The position of the magnetic structure with respect to a
stationary reference is designated by the coordinate y in Fig. 5-17. If
the magnetic structure is accelerating to the left (that is, Ii is positive),
then the coil-form mass M will be displaced to the right with respect to
the magnetic structure, thus making x negative. The displacement of the
coil form is related to the acceleration of the transducer. This displace
250 Principles of electromechanical-energy conversion
ment is sensed by two capacitance rings mounted axially with the coil
form. The current in the N-turn coil is then automatically controlled
to return the coil form to the x = 0 position. The current required to
restore the coil form to the x = 0 position is related to the imposed
acceleration; thus, by measuring the coil current, the acceleration ii can
be measured. Let us now study the static and dynamic operation of the
accelerometer in detail.
The conducting coil form and the two axially mounted conducting
rings are illustrated in Fig. 5-18. The coil form is supported by the four
identical springs. The two conducting rings mounted on each end of the
coil form are rigidly attached to the magnetic structure by insulated
supports, as shown in Fig. 5-17. With x = 0 the coil form is exactly
centered between the two rings and the capacitances between the coil
form ahd each of the two rings are exactly equal. If x increases in the
positive direction, the coil form moves toward the left-hand ring. The
capacitance between the left-hand ring and the coil form is therefore
greater than the capacitance between the right-hand ring and the coil
form. For small displacements of the coil form, neglecting fringing,
these two capacitances can be written in the form
(5-117)
Fig. 5-18 A detailed view of the spring-supported coil form and the two
capa.cita.nce rings.
Response 01 electromechanical system, 251
where Co is the value of the two capacitances with the coil form centered
at x = 0 and a is a constant giving the farads per meter displacement of
the coil form.
The two capacitances formed by the coil form and the a~ial rings can
be put into the bridge circuit shown in Fig. 5-19. Two fixed capacitors
with a value of Co farads make up the remaining two legs of the bridge.
The frequency of the sinusoidal voltage source used to excite the bridge
is selected to be much greater than the highest significant frequency
component in x(t). The sinor output voltage Eo for a given sinor driving
voltage Ei for the circuit shown in Fig. 5-19 can be put into the form
Eo COWl - C2)
(5-118)
Ei = (C 1 + Co)(C 2 + Co)
Notice that Eo equals zero if C l C2 , or the bridge is balanced. On
substituting Eqs. (5-117) for C l and C2 into (5-118), we have
Eo a
-=-x (5-119)
Ei 2C o
assuming that (ax)2 is negligible compared with C o2.
Equation (5-119) shows that the output of the bridge is a sinusoidal
time function with the same frequency as the input and with a magnitude
determined by the position x of the coil form. For positive values of x
the output is in phase with the input, and for negative values of x the
output is 1800 out of phase with the input. The output of the bridge is an
amplitude-modulated, or a-m, time function. The signal information can
be recovered by supplying the time function to a phase-sensitive demodu
lation circuit, which supplies a d-c output proportional to magnitude of
Eo; it is positive when Eo is in phase with Ei and negative when Eo is
1800 out of phase with E i . t The combination of the capacitance bridge,
t See, for example, C. J. Savant, Jr., "Control System Design," 2d ed., pp. 345--347.
McGraw-Hill Book Company, New York, 1964.
252 Principles of electromechanical-energy conversion
1 M( y-
£ 1= 2 + x')2 + 21 L'2q + 21raB N'
axq - ----,rr -
(x - c)2 +
(x 4Kc)2
(5-121)
where a is the radius of the coil, L is the self-inductance of the coil, and c
is the distance each spring is stretched when x = O.
The Rayleigh funetion for the system is given by
(5-122)
where h is the thickness of the coil form and p is the resistivity, in ohm
meters, of the coil-form material. Notice in Fig. 5-17 that b is defined as
half the length of the coil form.
The externally applied forces are given by
(5-124)
where e(t) is as given by Eq. (5-120). The external force in the electrical
equation Qq must have a negative sign so that the coil current tends to
restore the coil form to the x = 0 position. Equation (5-120) has e(t)
positive if x goes positive. By examining Fig. 5-17, we can see that a
negative current (opposite to the dots and crosses) causes a force of
electrical origin in the negative x direction, and thus e(t) must be negative.
Response of electromechanical systems 253
M
kl = amp-sec 2 /m (5-126)
lBo
k2 = lB.KfJ (5-127)
R
lBo
TI = If seconds (5-128)
L
T2 = Ii seconds (5-129)
2 _ 1
w" - MK (rad/sec) 2 (5-130)
r=Q~
2 M
(5-131)
(5-132)
~- ------- - - - - - - - -
254 Principles of electromechanical-energy conversion
la)
(0)
Fig. .')...20 (a) Overall block diagram for the closed-loop accelerometer transducer.
(0) Simplified block diagram.
By substituting Eqs. (5-126) and (5-127) into (5-132), the d-c steady
state transfer function can be put into the form
in MfJ
(p2y)O - R/K + lB.fJ (5-133)
If the quantity lB,.{J can be made much greater than R/K, then Eq.
(5-133) reduces to
in M (5-134)
(p2y)o "'" lB..
t Remember that fJ is the net coefficient for the bridge, phase-sensitive demodulating
circuit, and an output amplifier.
Response of electromechanical systems 255
(5-137)
(5-138)
(5-139)
(5-140)
All the quantities in Eqs. (fi-137) to (fi-140) are defined by Eqs. (5-127)
to (5-131).
The Routh array for a third-order polynomial is given by (5-40),
Since all the polynomial coefficients are positive, the only criterion for
stability is
2f"'n"2
- k2 - ( "2 + -w,..
2f + - -12 + k2'1'1 ) >
'TsWn
("2 - "1
)
Notice that for n :> T2 the right side of (5·142) is negative and the
inequality is always satisfied, since all parameters are positive. From
Eqs. (5-128) and (5-129), this sufficient, but not necessary, requirement
256 Principles of electromechanical-energy conversion
means that
lBa L
~> sufficient for stability (5-143)
fJ R
Increasing the magnitude of the air-gap field Ba tends to stabilize the
system. Increasing the coil length 1 by adding more turns will not help
stability, since the self-inductance L (which is proportional to the square
of the number of turns) divided by the resistance R (which is proportional
to the number of turns) increases at the same rate as l. A higher
capacitance bridge coefficient fJ causes instability. Using a smaller
diameter coil wire increases the coil resistance R and thereby aids stability.
By examining (.')-142), we can see that increasing the damping ratio
r makes tfe left side larger and helps satisfy the inequality. Perhaps the
best. way to stabilize the system is to increase the eddy-current damping
in the coil form. From Eq. (5-123) we have the factors influencing the
damping coefficient D appearing in the expression (5-131) for the damping
ratio r. Remember that the static operation becomes independent of all
system parameters which are likely to change value if the quantity
lBafJ» R/K, as discussed with Eq. (.')-133). Increasing the air-gap
field Ba not only satisfies this condition but also increases the damping
coefficient D and thus aids stability. The design of a stable system, the
parameters of which adequately satisfy (5-142), is an involved process
requiring a large amount of judgment and experience.
Assuming that the system is stable, let us examine the closed-loop
frequeney response. A substitution of jw for 8 in the transfer function of
Eq. (5-135) yields the phasor relationship between sinor I(jw) and sinor
"i(jw) as given by
I(jw)
Y(jw) = [1 - (w/w n}2 + j(nw/wn)](jT2w + 1) + k2(jTIW + 1)
(5-144)
A Bode plot for Eq. (5-144) cannot easily be obtained without factoring
the third-order denominator. In general, the factoring of a third-order
polynomial is a difficult job. The subject of feedback systems develops
numerous techniques for adjusting the closed-loop system response by
dealing primarily with the open-loop transfer functions. The open-loop
transfer functions usually appear in factored form, so the conventional
Bode plots and polar plots can easily be drawn. Also, the locations of the
open-loop poles and zeros are known.
An adequate discussion of these feedback-system techniques is
beyond the scope of this chapter. The accelerometer transducer serves
to illustrate the very difficult and intriguing problems faced by the system
designer. The static system equations show us that, by making one
Response of electromechanical system, 251
group of parameters much larger than another group, the system becomes
insensitive to parameter variations. On the other hand, the dynamic
equations show us that the system can become unstable if only the static
requirements are met. In addition to these two factors, a good overall
closed-loop response must be obtained by the designer, who can work
conveniently only with the open-loop transfer "functions. The subject of
feedback theory or control theory develops detailed techniques for solving
these problems.
5-9 SUMMARY
PR08LEMS
5-1 A system is described, for 11.\1 t <?: 0, by the following two nonlinear dif
ferential eqUl!.tions:
5x + 8x + x1y' - A(t) l2g + 20y + 4yx - B(t)
where
A(t) - 128 B(t) - 32
G. Find the steadY-f!tate values of x and y for the constant forcing functions A
and B.
b. The term x'y' ca.n be linearized into the following form:
Response 01 electromechanical systems 259
Fig. P5-4
5-5 The simple pendulum shown in Fig. P5-5 consists of a stiff massless rod
of length 1connected to a point mass M. The pivot point is frictionless, but Ii viscous
friction torque proportional to the angular velocity 0 of the pendulum opposes the
motion.
a. By using Lagrange's equation show that the equilibrium equation for the
system is MI.(J + DO + MgI sin 8 - 0, where D is the viscous friction coefficient
and g is the acceleration of gravity.
260 Principle. of electromechanical· energy conyer.ion
b. By letting 6(1) -
(10, find the system's two static equilibrium positions.
c. Find a linear incremental equation involving (I,(t) which is valid about any
operating point (10_
d. By using the incremental equation found in part c, prove that one of the static
operating points found in part b is stable and the other is unstable.
I
I
I
I
I
/
/
/
Fig. P5-5
5-6 The three-port torque transducer shown in Fig. P5-6 has the following
parameter values:
Rotor self-inductance Li" - 25 mh
Rotor resistance R;r - 5 ohms
Rotor-to-stator mutual inductance M" - 0 cos'" henrys
Rotor moment of inertia J - 0.01 kg-m*
Rotor viscous friction coefficient D, - 0.5 newton-m-sec/rad
Spring compliance K, - 0.02 rad/newton-m
gr J
~-1
\ '
Fig. P5-6
Response of electromechanical systems 261
The spring is at its free length when fI' .. O. The stator winding is excited by a 5-amp
constant-current source. The rotor winding is excited by a voltage source such that
v'O) .. vo + v~(t)
and the mechanical port has an applied torque
Tr(t) = T~ + T~(t)
a. Write a Lagrangian for the system using a loop formulation on the rotor
winding.
b. By using Lagrange's equation, find the equilibrium equations for the rotor
winding and the mechanical port.
c. Separate the two equations in part b into a set of d-c steady-state equations
and a set of linear incremental different.ial equl\tions.
d. If the d-c rotor voltage v; .. 50 volts and the constant applied rotor torque
T~ "" 89.4 newton-m, find the operating-point values for the rotor current i; and the
rotor position 1/;.
e. By substituting the operating-point values from part d into the linear incre
mental equations found in part c, draw a linear incremental block diagram for the
system. Show vW) and Tr(!) as inputs and 11';(1) as the response of interest. Reduce
the diagram to its simplest form.
f. Draw a Bode diagram for Tj(r) "" A sin wt and v;(t) .. O. The IT;(t)1 « T;.
g. Repeat part f for v;(t) = B sin wt and T~(t) = O. The Iv;(t)1 «v;.
h.. If v;(t) .. 5u(t) and T~(t) .. 0, find 1I'(t) for all t ;::: O.
i. If T;( t) - 5u(t) and v;W ... 0, find fir( t) for all t ;::: O.
5-7 The permanent-magnet vibration transducer shown in Fig. 5-11 has been
designed with the following specifications:
b. Calculate the total m&S8 of the moving system consisting of the coil form and
the 100-turn winding. AB8ume the springs to be m&S8IeB8.
c. Find the viscous damping coefficient D resulting from motion of the copper
coil form in the air-gap magnetic field.
d. Determine the complete transfer function between the open-circuit coil
voltage and the velocity of the stator frame with respect to a stationary reference.
See Eq. (5-88).
e. Plot the frequency response of the device and indicate the lowest frequency
over which it can be used &8 a velocity transducer.
,. What would be the effect of using an aluminum rather than a copper coil form?
5-8 The &ccelerometer transducer shown in Fig. 5-17 and discuB8ed in Sec. 5-8
haa the following design parameters:
Coillorm
Materi41 is aluminum
Radius - 1 em
Length - 1 em
ThickneB8 - 1 mm
Spring.
Coil
Material is 4O-gauge copper wire with a diameter of 0.08 mm
Self-inductance - 20 jl.h
Pomw1'/. /ramduur
Gain constant of cap&citance bridge, demodulating circuit, and amplifier
(J - 6 X 10' volts/m
The two permanent magnets establish a radial air-gap !lux density of 10,000
gausa.
a. Calculate the resistance of the tOO-tum coil and the total m&S8 of the moving
system consisting of the coil form and the tOO-turn coil.
b. Draw the complete block diagram for the system in the form of Fig. 5-20a.
Substitute parameter values. Reduce this diagram to the form of Fig. 5-20b.
c. For a stea.dy-state &cceleration of 109, where g is the acceleration of gravity,
find the steady-state coil current.
d. Plot the open-loop frequency response between 1(j",) and k.V(j",) in the form of
a Bode plot.
e. On the sa.me graph, plot the same frequency response, taking TJ and Tl to be
negligible. DiscU88 the proposal that this second plot re8.llOnably approximates the
first.
I. Taking T, and Tl to be equal to zero, plot the closed-loop frequency response
between l(j",) and k,Y(j",) on the same graph. What is the bandwidth of the closed
loop &ccelerometer? (Note: The bandwidth is the range of frequencies over which the
device approximates a constant-gain &ccelerometer by not attenuating its low-frequency
gain by 1eB8 than 0.7.)
g. Use the approximations in part 1 to find the closed-loop response i(t) to a 109
step change in the input &cceleration.
11.. Prove that the complete system (with Tl and TI not zero) is stable.
Response 01 electromechanical qstemll 263
5-9 A magnetic relay is designed as shown in Fig. P5-9. The stack height of the
iron laminations is 0.25 in. The other physical dimensions are as shown. Each of
the two series-connected coils has 12,020 turns of wire having a resistance of 2250 ohms.
The linear massle88 spring has a compliance K - 2.26 X lO-1 in./lb, and it is at
its free length when % - 0.011 in. Assume that the iron has infinite permeability
and neglect all leakage effects. The closed air-gap length is 0.001 in. owing to
imperfections in the pole faces and armature surfaces.
a. Derive the d-e steady-state equilibrium equations for the relay.
b. Sketch the steady-state position of the armature %0 as a function of the d-e
applied voltage Vo. Be sure to label the pickup and dropout voltages. What is the
minimum air-gap length with the relay contacts still open?
c. For Vo - 20 volts calculate the two steady-state air-gap lengths.
d. Develop a set of incremental linear differential equilibrium equations valid
about each of the operating points in part c and thereby prove one of the points to be
stable and the other unstable.
/l' Contocts
Fig. P5-9 vI t I
5-10 The purI10se of this problem is to demonstrate the effect of considering the
iron structure in Prob. 5-9 to have a noninfinite permeability. A steady-atate volt
age of 32 volts is applied to the two series windings of the relay in Fig. P5-9.
a. Calculate the stea.dy-state current in the windings.
b. With the armature held at its free position (i.e., the air-gap length is 0.0l1 in.),
find the force of electrical origin, assuming the iron to have infinite permeability.
c. Repeat part b with the armature in its closed position and the air-gap length
equal to 0.001 in.
d. Repeat the calculations of parts band c with the assumption that the iron hall
the following magnetization curve:
H., amp-tums/m o 80 249 478 800 2490 4780
B, webers/m. o 0.76 1.14 1.27 1.33 1.46 1.56
Hint: ExpreSB the force of electrical origin in terms of the flux density B. The
flux density can be determined by using Ampere's circuital law, which for a llon
infinite permeability in the iron becomes
!!. 2%0 + ZH. - 2Nio
fIoO
264 Principle. of electromechanical-energy conve,..;on
where l is the mean path length through the iron circuit, Hi is the magnetic intensity
in the iron, and N is the number of turns on each coil. By Bolving this equation for B,
an appropriate load line can be plotted on the B versus Hi magnetization curve, with
the intersection giving the flux density B. The corresponding force of electrical origin
can now be found.
e. Compare the answers in part d with those obtained in parts band c. When is
the aasumption that the iron has infinite permeability valid?
5-11 Figure P5-11a shows the essential features of a circular electrostatic audio
loudspeaker. The speaker consists of two conducting plates separated by an elastic
dielectric material. The left plate in Fig. P5-l1a is stationary. When a voltage is
applied across the two plates, the right plate moves toward the stationary plate. If
the motion of the right plate follows the applied signal voltage, the device will act as a
loudspeaker converting the electrical signal to a corresponding acoustical output.
In order for the device to be a linear transducer, the signal voltage Vl(t) must be much
lC88 than a.p. applied d-c bias voltage Vo. The speaker has the following design
parameters:
Radius of the plates and dielectric - 10 cm
Frequency range: 3000 to 12,000 cps
Mechanical viscous damping coefficient D - 40 newton-sec/m
Permittivity of the dielectric material .... 2 X 10-1 farad/m
Young's modulus for the dielectric material Y - 5 X 10' newtons/m'
Effective maas per unit area of the right plate and the dielectric - 18.1 g/m'
Thickness of the dielectric with zero applied voltage d - 5 X 10-' m
Voltage-eource resistance R - 6000 ohms
a. Indicate the physieal origin of each of the lumped elements in the l\1mped
linear model shown in Fig. P5-11b.
b. Find the effective mass M of the moving system. Show that the compliance
K - d/ A Y, where A is the area of the elastic dielectric. Find the value of K.
j)
R R
+ +
v( t) vi tl
(al Ibl
Fig. P5-11
Response 01 electromechanical sys'ems 265
c. Use Lagrange's equation to find the two equilibrium equations for the system
using a loop formulation for the electrical portion of the system. Separate these
equations into a set of static equations and a set of incremental linear differential
equations.
d. From the static equilibrium equations sketch the position of the system Xo as a
function of the d-c applied voltage v.. This curve is similar to Fig. 5-3. Calculate
v::'"' and the value of Xo at v:;'''. Indicate stable and unstable operating regions on the
X.,..Vo curve.
6. From the linear incremental equations, about an operating bias voltage
Vo - 3750 volts, find the sinusoidal steady-state input impedance Z - V1(j... )/II (j...),
where il(t) - q,(t). Arrange Z into the form
1 1
Z - R + R .. +.,-C
J"'.
+.,--C
J"' ..
where C• .. fA/xo. Find numerical values for R.., COl and C.. and thereby ehow that
Z "" R over the speaker's frequency range. .
f. From the results of part e show that the radiated acoustical power is given
approximately by P '" V,'R../R'. Find the radiated acoustical power for an rms
signal voltage V I - 250 volts.
g. With VI - 250 volts find the total input audio power, and UIIe the results of
part f to find the efficiency of the speaker.
chaptar
CONSTRUCTION OF THE
6 PRIMITIVE MACHINE
. Quadrature oxis
Direct axis
Slotor
in Fig. 6-1. The horizontal axis is denoted as the direct axis, and the
vertical axis is termed the quadrature axis. The symbols d and q are
respectively to be used for designating these two axes. Owing to the
machine geometry, a cylindrical-coordinate system is found to be more
convenient. For this purpose the polar angle (J is defined in Fig. 6-1 as
being zero along the positive direct axis and (J is taken positive in the
counterclockwise direction. The radial cylindrical coordinate r is
measured from the center of the machine, and the axial coordinate z is
positive in the direction out of the paper in Fig. 6-1. Using a right-hand
cylindrical-coordinate system, we have
ar X a, = a. (6-1)
Rotor
4,
~O~~--~~~==~==~~~~===f--oo
Stotor
Fig. 6-2 A developed view of the air gap of the salient two-pole machine.
Only cosine terms appear because of the even symmetry. The argu
ments of the cosine terms contain 28 owing to the fact that g(8) completes
two cycles as 8 proceeds from zero to 2'1f'. For air gaps which are not
square wave in form we can still develop a Fourier-series expansion in the
general form of (6-3). This conclusion is based on the physical fact that
the air gap must at least be periodic and progress through an even number
of cycles as 8 goes from zero to 2'1f'.
Retaining only the fundamental term in the Fourier-series expansion,
we have
g(8) "" go - gl cos 28 (6-4)
where for the square-wave variation expanded in Eq. (6-3) we have defined
gq + gd
go = --2- the average gap length (6-5)
and
(6-6)
Since gl is usually much less than go, retaining only the first term in the
series of (6-3) is a reasonably good approximation. The higher harmonics
contain a coefficient lin, where n is the order of the harmonic; therefore,
for larger values of n these terms are less significant.
For a periodic air gap differing from a square wave, go and gl will not
be given by (6-5) and (6-6). However, an expression with the form of
Eq. (6-4) can always be obtained. The actual equations for go and gl, in a
particular case, depend on the exact air-gap geometry. For the structure
of our primitive machine Eq. (6-4) will be assumed to be exact. No
practical'machine is generally constructed with such an air gap. How
ever, when it is used as a model, the primitive machine poses as a reasona
ble first-order approximation for the salient machine. Notice that gl
becomes zero and Eq. (6-4) is exact for the case when no saliency exists,
or when gq = gd.
~--~~~------~----~--~--~----.d
Fig. 6-3 Cross-sectional view of the magnetie structure of the primitive machine
containing a quadrature-stator winding.
o = 11' in Fig. 6-3 and vcry few coil sides near 0 = 11'/2 and 0 = 311'/2, a
sinusoidal distribution of current can be approximated. For the winding
shown in Fig. 6-3 a magnetic field in the positive quadrature axis direc
tion will be found in the rotor structure for the current direetions shown.
Therefore, this windillJ!: is known as the quadratu.l'e-slalor winding. The
current in this winding is denoted by i:.
where the subscript q denotes the
quadrature axis an!il the superscript s indicates the stator.
The eurrent sheet set up by this quadrature-stator winding is given by
amp/m
The factor K; cos 0 is a distribution term giving the number of conductors
per meter (in the 0 direction) each carrying a current of i; amperes.
Assuming that the air-gap length is negligible compared with the radius of
the rotor a, the quantity of eUlTcnt 011 the stator surface running axially
along the machine in a peripheral distance a dO is given by
r
G
r
~.
0
r-------- -,
I r(Z
- +-_-_--_--
-®--..;.--r--
-...
11"!
... (!.....!)...
. f-.. .
Rotor LOirected contour
3",/2 T Z11"
-_-:r"itr----'... -=t-l--~-tL->----"':8
L
9(8)=90-9, c0528
Stotor
Fig. 6-4 A developed view of the air gap of the primitive machine showing the
currents in a quadrature-stator winding.
g(tJ) gi\lCn by (6-4) will be used. Equation (6-7) gives the number of
amperes per unit length along the tJ axis established on the st.ator surface
by the current i:.
In the two quadrants 0 < tJ < 11"/2 and &r/2 < 8 < 211",
the current-density vector J:
is into the paper, or in the negative z-axis
j
direction, as verified by Eq. (6-7). For the other two quadrant.s J; is in
the positive a. direction.
I A function f(x) possesses odd symmetry if f(x) - -f( -x). A sine function is an
example of odd symmetry. By the term "odd" we do not mean "peculiar."
Construction of the primitive machine 273
where we have used ll. as the symbol for the radial component of H. The
minus sign preceding the second term on the left side of (6-11) results
from the fact that at IJ + .,. the contour direction is opposite to the positive
radial direction specified by aT' The surface integral on the right side of
(6-11) represents the total current passing through allY surface capping
the contour.
For the contour direction shown in Fig. 6-4, current coming out of
the paper must be taken as positive according to the right-hand rule, and
therefore dS = d8 a.. By \ISC of F:qs. (fi-7) and (6-8), the total current
is given by
(B+ ..
f J . dS
S = J8 - K; cos 8 i;a d8
which reduces to
with g(8) g(8 + .,.), as seen from Eq. (6-4). The constant K: is a
distribution factor with units of turns per meter, the rotor radius is a
meters, a.nd the air-p;ap length is g meters and is a function of (J. With the
coil current i:.
in amperes, t.he units of H; are ampere-turns per rrl:eter.
The magnetic-flux-density vector can be obtained simply by multiplying
274 Principles of electromechanical·energy conversion
Quadrature axis
Direct axis
with 10'0 = 4n- X 10- 7 him and B: having units of webers per square meter.
In the air gap, Eq. (6-14) shows that the magnetic-flux density is also
radial and varies in magnitude as the sine of the angle e. Figure 6-5
shows only the rotor structure of the machine with the flux lines drawn
radially around its periphery. For a machine with no saliency (that is,
g( e) = go) and an isotropic, cylindrical rotor of a high-permeability material,
the flux density within the rotor member will be uniform and directed as
illustrated in Fig. 6-5. This uniform flux density, within the rotor
structure, can be expressed in cylindrical coordinates by
Io'oK'ai' .
--g-g
go
(sm e a, + cos e a,) in rotor (6-15)
Taking the direct axis as the x axis and the quadrature axis as the y
axis in a cartesian-coordinate system, Eq. (6-15) is equivalent to
in rotor (6-16)
Quadrature axis
•
.,
4----1 q
upper half of the cylinder carry currents out of the paper, as do all con
ductors on the inside of the lower half of the cylinder. Similarly, all
conductors on the outside of the lower half and inside the upper half of
the cylinder have currents going into the paper.
While holding the brushes in a fixed position, consider rotating the
cylinder. The two brushes will each slide off one turn of the winding and
onto an adjacent turn. Notine, however, that the current distribution
would not be altered. Conductors located on the outside upper half and
inside lower half still have currents out of the paper, and all conductors
located on the outside lower half and inside upper half have currents into
the paper. If the brushes are held fixed, the current distribution is inde
pendent of the rotary motion of the cylinder.
The winding shown in Fig. 6-7 is very inconvenient and impractical
for a number of reasons. One of the major objections would be the lack
of iron in much of the rotor structure, since Fig. 6-7 requires a hollow
cylinder. The magnitude of the magnetic field B available for a given
field intensity H would be seriously reduced. For this reason the solid
(or, in practice, laminated) rotor cylinder shown in Figs. 6-1 and 6-3 is
preferred. A solid cylinder can be used if all the conductors inside the
cylinder are moved to the outside periphery without changing any of the
relative connections. Figure 6-8 shows how one such conductor is moved.
Notice that the conductor I' is moved to a position 11" rad around to the
outside of the cylinder without any change in electrical connections. If
all conductors on the inside walls of the cylinder are similarly moved to the
outer periphery of the rotor, in a position 11" rad from their original location,
then the cylinder need no longer be hollow. Since no change has been
made in any electrical connections, the currents in each respective con
ductor are the same as they originally were in Fig. 6-7. Therefore, all of
the currents on the outside of the lower half of the cylinder are into the
paper and all currents on the outside of the upper half are out of the paper.
278 Principle, of electromechanical-energy conve",ion
1'~~------------+-r-
Fig. 6-8 Repositioning the coil sides loca.ted inside the hollow rotor cylinder to the
outside or the cylinder.
w'
\ I I 1 I 1 I 1 1 1
8'1 1'1 6'1 5'1 4'1 3'1 2'1 11'1 10'1
, , ,
9'1
11"
j ,I t
I I I
I I
I 1
I
1
I
I
I
I I
I1r
I
I
I
I
1
1
I ,
1
I
I
I
I
1
I
1
I
I
I
I
1
1
I
I
1 I I I
1
I I I I I I I 1 1
I
I 1 I I I 1 1 1 I
1
1 1 1 1 I I I 1 I
1
I I I I I I 1 I I
I
I 1 I I I I I I
I
9 I I + • 5 • 4
+ 12 I 11 I 10
Fig. 6-9 A developed view or the origina.l configuration or the rotor conductors.
Comtrvdion of the primit;". machine 279
Fig. 6-10 Rotor conductors rearra.nged to be loca.ted on the outer periphery of the
rotor cylinder.
Fig. 6-7 are duplicated in Fig. 6-9 to aid in verifying the similarity of the
two drawings. Also, notice the brush positions in the two figures are
identical. Current directions into the paper in Fig. 6-7 correspond to
currents up the page or away from the brushes in Fig. 6-9. Rotation of
the rotor counterclockwise with respect to the stationary brushes corre
sponds to motion of the coil sides to the left in the developed view of the
same rotor.
The movement of all primed conductors 'It" rad around the rotor and to
the outside walls may easily be seen in this developed view. For example,
coil side I' will be moved above coil side 7, coil side 2' goes above 8, and so
forth without changing any electrical connections. This rearrangement
of the conductors is shown in Fig. 6-10. Again for clarity, where one coil
side is actually on top of another, it is displaced slightly and drawn adja
cent to the original coil side. The developed view in Fig. 6-10 looks very
involved, but if you remember that no change in connections has been
made between Figs. 6-9 and 6-10, the tracing of the circuit is not difficult.
Observe in Fig. 6-10 that all conductors in the shaded area have currents
toward the brushes. The shaded area corresponds to the upper half of
280 Principles of electromechanical-energy conversion
the cylinder, and the nonshaded area corresponds to the lower half of the
cylinder.
We have now developed a mechanism for maintaining a stationary
current distribution on the surface of the rotor that is independent of the
position of the rotor structure. The sliding contact between the brushes
and the winding can no longer be made by having the brush slide over the
end connections, as shown in Fig. 6-7, since two coil sides are now located
in each slot on the rotor periphery in Fig. 6-10. This sliding connection is
usually made by attaching the conductor ends to copper segments
insulated from each other and mounted on the surface of a small cylinder
on the shaft of the rotor. The brushes then ride on these segments and
make a low-resistance sliding connection to the rotor coils. This struc "
ture composed of insulated copper segments is known as a commutator.
A devJloped view of the commutator is also included in Fig. 6-10.
The winding shown in Fig. 6-10 is only one configuration out of a vast
number of possible arrangements that will result in a stationary current
distribution that is independent of the rotor position. The purpose of
this discussion is only to show that the desired result is possible, and not
to give the vast details in the art of rotor winding.
Rotor
Stator
Fig. 6-11 A developed view of the air gap showing the uniform surface-eurrent
distribution on the rotor periphery.
Con.truction of the primm.,. machine 281
axially and very close together on the rotor surface, the nearly uniform
current distribution shown in Fig. 6-11 can be approximated. Analyti
('ally the rotor surface current can be expressed as
= 4Kr'i
r
J~ __01_01 (sin 0 + ! sin 30 + i sin flO + ...) a. (6-22)
'If'
This series is similar to Eq. (6-3) for the air-gap length, except that here
we have odd symmetry giving only sinc terms and the total period is 2'1f'
rad rather than 'If' rad as in the air-gap expansion. In the analysis of our
primitive machine only the first term in the series is to be retained.
When we define a slightly modified distribution factor
K~
4Kr'
= _ _
01 turns/m (6-23)
'If'
Quadrature axis
Direct
axis
Quadrature axis
Direct
axis
In general, the total winding flux linkages would then be given in matrix
form by
O
X9!]
X _ M~'
[LOqd
(6-31)
Xd - Md'a
X; M;'a
The L coefficients are the respective self-inductances of the four windings.
The remaining M coefficients are mutual inductances. The first set of
sub- and superscripts on the M coefficients designate the windings in
which the linkages exist, and the second set of sub- and superscripts
indicate the windings carrying the current causing these linkages. Thus
M~~i: are flux linkages in the direct-stator winding reSUlting from a current
in the quadrature-stator winding. From our previous work we know
that the inductance matrix in (6-31) must exhibit principal-diagona.l
symmetry in order that the energy- a.nd coenergy-state functions be
independent of the particular manner in which the final state of the sys
tem is reached. This means that M~~ = M;, M'.tt = M'da, and so forth
for all other mutual coeffieients. Also, we shall see that many of these
mutual eoefficients are zero.
The magnetic coenergy of the primitive-machine system can be
obtained from
(6-32)
where Band H are the total fields at each point in space. The iron por
tions of the primitive machine have a permeability which has been assumed
infinite compared with the permeability of the air gap. For this reason
the I1tagnetic-vector intensity H in the stator and rotor iron must be zero.
The magnetic field B outside the stator structure is also zero because of
this assumed permeability. All of the magnetic energy stored is therefore
found only in the air gap of the machine, since everywhere else the energy
density U", is zero becauseB and/or H are zero. In the air gap H is related
to B by the permeability of air ~o = 4Ir X 10-7 him, and therefore Eq.
(6-34) reduces to
B2
U", = - joules/m a (6-35)
2~o
B
where is the magnitude of the total air-gap magnetic field.
With the (our windings of the primitive machine carrying the respec
tive currents shown in Fig. 6-13, the total air-gap magnetic-field vector is
given by
B = Bd + B; + B~ + B; (6-36)
On SUbstituting Eqs. (6-14), (6-20), (6-29), and (6-30) into Eq. (6-36), we
have
~oa
B = g( e) (K'dtd'. cos 8 + K' '. sm
qti . 8 + Krd~dOr cos 8 + Kr Or sm 8) ~
iti °
(6-37)
with all components in the air gap radially directed, The magnitude of B
is not a function of the cylindrical coordinates rand Zo Thus the total
energy stored in the air gap is simply
( 2.,
10 U ...(e)g(e)la de (6-38)
The air-gap length g(e) times the length of the rotor cylinder l is the area
of a radially oriented rectangle in the air gap at an angle e. Assuming
that 9 is much smaller than the radius of the rotor, given by a, our
incremental volume is the g(e)l rectangle times the arc length (a de).
On mUltiplying by the energy density U"" we have the energy in this
incremental volume. Summing over e from zero to 21f gives Eq. (6-38),
which is the total energy stored in the air gap by the four windings.
Using the value of B from (6-37) in Eq. (6-35) and substituting into
Eq. (6-38), we have for the energy stored
W... (id,i;,id,i~)
= ~oa3l (2., (K~i~ COS 8 + K:i: sin 8 + K~i~ cos 0 + Kgi; sin 8)2 do
2 10 g(~
(6-39)
COIMtrvction of the primitive mochine 287
The eight mutual-inductance terms which are all zero in Eq. (6-48) are
mutuals between two windings in quadrature or at 1r'/2 rad from each
other. The mutual inductance between either winding on the direct axis
and either winding on the quadrature axis is zero. The mutual induct
ance between two windings on the same axis is not zero, as indicated by
Eqs. (6-46) and (6-47).
On substituting these results into the linear inductance matrix given
in (6-31), we have
o
X~] [L~0 o ] ~:.r]
M';{
X' L~ 0 M~'
X~ = M,;{ o L~ o i:; (6-49)
X; I 0 M~' 0 L; $~
with Eqs. (6-42) to (6-47) as expressions for these parameters in terms of
the primitive-machine dimensions and the winding-distribution factors.
Notice from these parameter equations that
M~r = (L4L:;) >i (6-50)
and
(6-51)
as must be the case when all leakage magnetic flux is neglected. Also,
ob8erve that all of these inductance parameters are independent of the angular
p08ition or angular velocity of the rotor structure. This angular independ
ence should be expected, since both the rotor- and stator-current dis
tributions are independent of the rotor angle.
eiDd = - fs aB.
at dS +,f..
'f' c
v X B • dl (6-52)
where v is the velocity vector for a vector increment of length dl and the
surface S is bounded by the closed contour C. The integrals on the right
hand side of Eq. (6-52) are a general form of Faraday's law, which states
that a voltage is induced in a closed circuit if the circuit flux linkages are
changing with respect to time. The change in the surface integral occurs
owing to the flux-density vector B being nonstationary, or a function of
time. Thus the iJB/iJt would be nonzero. The closed-contour integral in
Eq. (6-52) has the flux linkages changing because of motion of the closed
circuit with respect to the magnetic field. Thus Eq. (6-52) expresses the
Construction of the primitive m<Jchine 289
general law that changing the flux linkages in a closed circuit causes an
induced voltage regardless of how the flux linkages are changed.
The air-gap magnetic fields in our primitive machine can vary with
time only if the winding currents setting up these fields are themselves
time-varying. The contribution to the voltage induced in anyone of
the windings owing to a time-changing field is taken into account by the
inductance parameters developed in Sec. 6-6. Thus the first integral on
the right side of (6-52) reduces to
r aB. dS -+ ax (6-53)
J8 at at
with the respeetive flux linkages X given by Eqs. (60049) in terms of the
self- and mutual-inductance paramet.ers. Since all of the winding currents
are functions only of time and are not functions of any position coordinates,
the ax/at reduces to a sum of terms with the form L di/dt and/or M di/dt,
using the total instead of the partial deriva.tive of the current. The next
chapter is concerned with equilibrium equations for the primitive machine,
and the sense or direction of these induced voltages. At this point let us
just realize that the sense of the L di/dt voltage terms must be such as to
oppose an increase in the winding current i.
In this section we wish to examine the effect of the second integral in
Eq. (6-52). The total radial air-gap field due to currents in all four
windings on the primitive machine is given by Eq. (6-37). For conven
ience, let us repeat this equation as
B I/ofIJ (K'"
= g(8) dtd cos 8 + K ~l.,'. sm
. 8 + Kr'r
dId cos 8 + Kr'r . 8) a
.,t" sm r
(6-54)
where a is the radius of the rotor cylinder, g(8) is the air-gap length as a
function of the polar-coordinate angle 8, and the respective K's represent
the sinusoidal-distribution factors with units of turns per meter.
The quadrature-rotor winding conducting a current i; is assumed to
establish a current density on the surface of the rotor as given by
on the current lines give the direction. Carefully verify that Eq. (6-55)
does indeed represent the current distribution pictured in "Fig. 6-14.
The quadrature-rotor winding is constructed from a certain number
of coil turns wound on the rotor cylinder. One such turn is shown in
Fig. 6-14. Notice that this single turn is composed of four sides in a
rectangular form. The two sides running axially on the curved surface
of the rotor are often called the active sides, because the currents in these
two sides of the coil establish the air-gap fields. The two sides along the
front and back of the rotor cylinder simply serve to conduct the winding
current from one active side to another. The front and back sides are
termed end connections. The effects of these end connections are being
neglected in the formulation of the primitive-machine parameters. The
magnetic fields in the vicinity of the end connections are usually quite
small, since the mt'diur..... urrounding the ends of the rotor is air.
If the rotor is revolving a.t a.n angular frequency w', as shown in
Fig. 6-14, then the velocity of the active sides is given by
v = aw'a, (6-56)
where a is the rotor radius. Using Eq. (6-54), we can form the vector
product v X B and get as a result
1J~2W'"
v xB = -
, .,
g(8) {Kd~d cos 8 + Ktt, *sm
t
.."
8 + KdZd cos 8
r .,.
The voltage induced in the single coil shown in Fig. 6-14 can be
determined from the second term in (6-52), which is
(6-58)
Selecting the contour direction shown by the arrows on the coil in Fig.
6-14 gives us for vector length dl
dl = -dl a. for the active side at tJ
and (6-59)
dl = dl a. for the active side at tJ + ,..
However, since B(tJ) = -B(tJ + ,..), owing to the odd symmetry of each
current distribution, Eq. (6-58) reduces simply to
1 (I
Cind = -2 10 [v X B(tJ)] • dl a. (6-60)
where l is the length of the rotor cylinder or the length of each active side.
Since v X B, as given by (6-57), has only a z component, the scalar or dot
product under the integral in Eq. (6-60) is easily taken. By substituting
(6-57) into Eq. (6-60) and integrating, we have for the voltage induced in
this single coil
,,/2 1
e~ = / -0'/2 eiJ\dK~ cos fJ a dfJ (6-62)
Using the approximate expression (6-63) for the reciprocal of the air-gap
length in Eq. (6-61) and then substituting into Eq, (6-62) will give us four
integrals to be evaluated. After some amount of manipulation these four
integral expressions can be combined and reduced, finally, to give us as the
total induced voltage!
(6-64)
(6-65)
(6-66)
.
The G coefficients have units of henrys. They are usually termed
rotational inductances with the induced voltage being a function of the
,,/2 cost
/ -,,/2
6(1 +!l.! cos 26) d6 "':':2 (1 +
(10
2(11)
(10
and
f ,,12 cos
-./2
6sin 6(I +!l.! 26) dB
(10
COil 0
Construction of the primitive machine 293
angular velocity of the rotor w'. The first sub- and superscripts denote
that the rotational voltage is induced in the quadrature-rotor winding.
The second sub- and superscript.s denote the winding establishing the
field which causes this indu('ed voltage. Thus G';a refers to the quadrature
rotor rotationally induced voIt.age due to current in the direct-stator
winding. From Eq. (6-64) it can be seen that the quadrature':'rotor rota
tionally induced voltage e; results from the two windings on the direct axis
located at 11'/2 rad to the quadrature-rotor winding. Remember that in
See. 6-6 we found that the so-called transformer voltages induced through
the mutual or AI coefficients, owing to a winding current changing with
time, occurred between windings on the same axis. Rotationally induced
voltages occur only in rotor windings owing to currents in rotor or stator
windings in quadrature, while mutual or transformer voltages are between
windings located on the same axis. For positive w' and positive i~ or i:l,
Eq. (6-64) shows that the induced voltage e; is also positive, indicating
that the rotational voltage is aiding the winding current t~ to be positive.
The rotationally induced voltage in the direct-rotor winding due to
currents in all four primitive-machine windings can be obtained by a
similar development. Equation (6-61) for the voltage induced in a
single turn is still valid, since this voltage is due only to the four com
ponents of the total air-gap field. The direet-rotor current-density
vector from (6-24) is given by
The limits on the integral again span 11' rad. The direct-rotor surface
current dhtribution given by Eq. (6-67) is in the positive z-axis direction
for 0 < 8.< 11'. Referring to Fig. 6-14, we can see that the contour
direction for the single coil is opposite to the direction of Jd as the angle e
varies from one brush at () = 0 to the other brush located at e 11'.
Therefore, in order to have ed positive, when directed to aid i:i, a negative
sign must be included in Eq. (6-68). In Eq. (6-62), for the quadrature
rotor rotationally induced voltage, the contour direction matched the
direction of J; over the angles -11'/2 < 8 < 11'/2 between the two brushes,
so the integral has a positive coefficient.
When Eq. (6-61) is substituted into (6-68), again using the approxima
tion (6-63) for the reciprocal of the air-gap length, four integrals must be
294 Principles of elecfromechanical-energy conversion
(6-70)
and
Grr = p.oa3l1rKqK'd
dq go
(1 - 2go
~) (6-71)
Rotor
.. 1
,
..12" ,
r/l! ,
3..kl! 2",In
I ..0
C-t-' Stator
(al
Rotor
"0
[i[:/~~
r
CL
I Stator
(ol
Fig. 6-15 (0) A developed view of the air gap of the salient n-pole-pair machine.
(b) The surface-current distribution for an n-po\e-pair winding.
For a nonsalient air gap g(8) becomes equal to go, since the saliency factor
gl is zero.
In Sees. 6-2 and 6-3 the surface currents established by the primitive
machine stator windings were assumed to have a periodicity of 211" rad.
After expanding the surface currents in a Fourier series and retaining only
296 PrillCip/e, of .Iectromechanical.energy conve",ion
For a nonsalient air gap, Eq. (6-16) and Fig. 6-!} show us that a magnetic
field in the air gap with the form of Eq. (6-77) results in a uniform field
inside the rotor cylinder in the positive quadrature-axis direction. The
cross-sectional view of the rotor, as shown in Fig. 6-5, resembles a bar
magnet with a single north and south pole. The quadrature-stator wind
ing establishes a two-pole field inside the rotor. In a similar fashion the
direct-stator winding establishes a two-pole field along the direct axis.
Thus current distributions with a periodicity of 2'lr rad establish a so-called
two-pole field inside the rotor.
Machines having a salient stator structure with an air-gap periodicity
of 1I'/n radians have corresponding surface-current distributions with a
periodicity of 211'/n radians, as shown in Fig. 6-1.')b. Therefore, the funda
mental terms in the quadrature- and direct-stator surface-current Fourier
series expansions are given by
(6-81)
(6-82)
ConstlVct ion of the primitive machine '197
r Quadrature axis
Oirect axis
The magnetic field inside the rotor cylinder with the quadrature
stator air-gap field, given by (6-81), would resemble a magnet with a
total of 2n poles around its periphery. Figure 6-16 represents the case
when n = a, showing only the maximum (both positive and negativej
values of the air-gap field speeified by Eq. (6-81). By convention, nux
emanates from the north poles and enters the south poles, as shown by the
arrows in Fig. 6-16. The integer n thus specifies the number of pole pairs
established by the current dist,ribution. For n = 3 we have six poles, or
three pole pairs, as shown in Fig. 6-16.
For stator-surfaee-current distributions having the general form of
(6-79) and (6-80), the primitive machine has rotor-surface-current dis
tributions with the same form. Because of the commutator-and-brush
switching, the rotor-surface currents are actually uniform with a periodicity
of 2r/n radians. As usual, only the fundamental term in the Fourier
series expansion is retained, giving
w'
\
GI/
'3'1
I
81
I
7~
I 6"1 5"
I
4'1
1
3"
I
2~
I
1~
1 11"I 10~
1
I I I t t t I I 1 t
-L~ I
I
1
,,
I I
I
, ,, ,
I
I
1 I
I
I
I
I
1
1
,
I
I
I
t
I
1
I I I I I I
,,
I I
I ,
I I
I I
I
I
I
I ,
I I
1 ,
I I
I
, I
,
I I I I 1 1 I
,I 1
•, ,
I I I I
l~
I I
I I
I '3 I 8 I
I I
I
I
I I
I
,
I
, I
, 5 , 4 I 3 I 2 I 1
I
I
I
I
7 I 6 12 I 11 I 10
ut ~2i
Conductors 2, 3, 4 = + 2i a.
~
Conductors 5, 6, 7 = - 2 a.
(6-85)
Conductors 8, 9, 10 = + 2i a.
~
Conductors 11, 12, 1 = - 2 a.
I
Now moving all of the conductors, located on the inside of the rotor,
r/2 rad around to the outside gives a stationary current distribution
having a periodicity of r rad. With four brushes on the same commutator,
we have established a four-pole (i.e., two-pole-pair or n = 2) direct-rotor
winding. By using additional pairs of brushes, multi pole windings of
higher orders can be established. The conductors located on the inside of
the rotor cylinder are moved r/n radians around to the outside of the rotor,
where n is the number of pole pairs established by the winding. Brushes
Construdion of the primitive machine 299
are '/fIn radians apart on the commutator, with alternate brushes con
nected together to make n parallel rotor circuits. A multipole winding
constructed in this manner is known as a lap winding.
By rearranging the active coil sides o'i'ilj()th, the inside and outside of
the rotor cylinder shown in Fig. 6-7, multipole windings using only two
brushes can be established. This alternate winding scheme contains only
two parallel paths and is known as a wave winding. Since the active coil
sides for a wave winding are connected in series in only two parallel paths,
this winding can generate higher voltages than an equivalent lap winding.
However, the lap winding with its n parallel paths between the terminals
is capable of handling larger port currents. l
Notice in Fig. 6-17 that the currents in the active sides of the winding
arc i/2 amperes, whereas the terminal winding current is 2i amperes. In
Eqs. (6-83) and (6-84) the currents i; and i:i are the terminal or port
currents for the respective windings. -~ 'The winding distribution factors
K; and K:i must be calculated such that J~ and Jd represent the funda
mental term in the Fourier expansion for the actual current distribution
on the rotor surface. l"or a specific rotor winding the surface-current
distribution should be expressed in terms of the terminal current, and not
the '!urrents in the actual active coil sides. A Fourier-series expansion for
the surface-eurrent distribution as a function of the terminal current will
yield the required expressions for the distribution factor. This formula
tion of the distribution factors allows us to assume that the priluitive
maehine mulLipolc winding is a single continuous winding actually carrying
the terminal current in the active coil sides. Thus the winding" establish
the respective n-pole-pair sinusoidal surfa{~e-current distributions. 2
The air-gap magnet.ic fields resulting from the rntor-surface distribu
tions given by (6-~3) and (6-84) are found by Ampt:.e's circuital law to be
Assuming that gl go
« and by using Eq. (6-74), the reciprocal of the air ?
gap length can be approximated by
-1-
g(8)
~ ]
-,
go
(gl
1 + -- cos 2n8)
go
(6-89)
n2yo
3
U = l"oa 11l'(K;)2 (1 gl) henrys (6-91)
q 2go
L~ = lJ.oa l1l'(Kd)2
3
n2go
(1 + Jl!.
2go
) henrys (6-92)
Lr =
q
l"oailr(Krp
n 2go
q
(
1 Yl)
2go
henrys (6-93)
M~r = lJ.oaalrK~Kd
2
(1 + Jl!. ) henrys (6-94)
n yo 2yo
Mor = l"oa311l'K~K; (1 _ _g~ ) henrys (6-95)
q n 2go 2go
The self- and mutual-inductance parameters for the multi pole wind
ings are redu{'ed to the expressions found in Sec. 6-6 for the two-pole case
simply by setting n = 1. By using a derivation paralleling Sec. 6-7, the
rotational inductances for the multipole windings are found to be
(6-96)
(6-97)
(6-98)
(6-99)
ConS#rvction of the primitive machine 301
Again notice that these expressions all reduce to the one-pole-pair case
originally considered in Sec. 6-7 when n is set equal to L
In deriving the self- and mutual-inductance parameters given by Eqs.
(6-90) to (6-95) the following three definite-integral forms are used
repeatedly:
{2r
10 cos n6 sm n6 1
. (
+ gl cosgo 2n6) d6 = 0
The rotational inductances require the evaluation of similar integrals over
appropriate limits spanning only 11" instead of 211" rad. Dividing the integral
values by 2 provides the results for these reduced limits.
The resistance of the stator winding can easily be determined by using the
relationship
R = pL (6-100)
A
where R is the winding resistance, in ohms; p is the resistivity of the con
ductor material, in ohm-meters; L is the total length of the winding
conductor; and A is the cross-sectional area of the conductor. For a
sinusoidally distributed direct-stator winding the number of turns in an
increment of arc l~ngth a d6 is given by Kd sin n6 a d6; therefore, the total
number of direct-stator turns in an angle spanning 11" In radians is
equaling the rotor diameter. The assumption that the rotor radius is
much greater than the air gap is again being used. Using Eq. (6-100),
the resistance of the winding is given by
' _ 4aK~(l
Rd - A
+ 2a) ohms (6-103)
1 The mutual inductance between two identical windings with unity coupling equal8
the self-inductance of either winding. These two self-inductances are also equal to
each other.
304 Principles of electromechanical-energy conversion
rature-rotor winding with its brushes relocated to the direct axis. Equa
tion (6-110) shows that the rotational coefficient G';d is equal to n times
the mutual induetanr:e hetween the direr:t-rotor winding and the relocated
quadrature-rotor wiuding, with this mutual inductance becoming a self
inductance because of unity coupling. Equations (6-107) and (6-109) pose
a similar relatioIlship upon rotation of the direct-rotor brushes 1r/2n
radians in the counterclockwise direction.
6-11 SUMMARY
A primitive-machine model is to be used as an analysis tool for investigat
ing a wide class of electric machinery. The stator and rotor windings of
this machine are constructed in a fashion which is sufficiently general to
allow for an analysis of all machines commonly encountered. The rotor
structure is a smooth cylinder of high relative permeability with windings
placed in slots around its periphery. Connections to points between the
various loops of the rotor winding are made by means of a commutator
"lrudurc and stationary pairs of brushes. The purpose of this arrange
ment is to create a magnetic-flux-density veetor which is stationary in
space and completely independent of the rotor position. The stator
windings, being placed on the stator structure, create' a similar vee tor
field completely independent of the rotor position.
The analytic form of all magnetic-flu x-density vectors is taken to be
radially directed in the air gap between the stator and rotor. The form
of the space dependency is assumed to be sinusoidal with a maximum
value equal to the actual maximum of the fundamental component in a
Fourier-series expansion of the periodic magnetic fields. In the designing
of many actual machines an attempt is made, as nearly as possible, to
achieve this result. If the air-gap fields are radial in direction with a
magnitude varying as a sinusoidal function of the polar angle, then the
magnetic-flux-density vector inside the rotor cylinder of high-relative
permeability material has a uniform magnitUde directed along either the
direct or quadrature axis.
Neglecting leakage fluxes and saturation effects, expressions for the
various self-, mutual, and rotational inductances are developed. Certain
interesting relationships between these coefficients, given by Eqs. (6-107)
to (6-110), exist under the assumption of sinusoidally distributed, radially
directed air-gap fields and identical rotor-winding-distribution factors.
In the next chapter these coefficients are used in formulating the equilib
rium equations for the complete primitive machine. The approximations
used in the construction of the n-pole-pair primitive machine and the
derivation of its parameter are as follows:
Construction of the primitive machine 305
Also, the radius of the rotor a is mueh larger than the air-gap length; thus
a + g(8) "" a (6-113)
a. The direel- and quadraturc-stator windings are assumed to have
their active eoil sidcs sinusoidally distributed around the inner periphery of
the stator. The cffects of the slots in the stator structure holding the coil
sides arc ncglccted. The sinusoidally distributed stator winding estab
lishes a surface eurrcnt which is axially diredcd with a sinusoidal angUlar
variation. The dircet- and t[uadrature-stator windings carrying currents
ij and i;, respc(·tivcly, ('stahlish eurrent shcets as follows:
where the distribution factors K~ and K; have units of turns per meter.
4. The direct- and quadrature-rotor windings are assumed to estab
lish similar current sheets on the rotor surface. These rotor windings are
supplied through a commutator-and-hrush mc('hallism which maintains
the rotor-eurrent shects stationary in spaee completely independently of
the rotor angular position or velocity. Thus we have taken
PROBLEMS
Rotor
d, j "4 2 9, t ~ 3"{2 1- •e
C-i fl ~2
Slolor
Fig. P6-1
1
6-2 For the stator-eurrent distrihutions shown in Fig. P6-2 sketch the surfaee
current distribution and the air-gap magnetic-flux density determined graphically by
using Arnp/,zoe'l! eircuital law. Since the atator has eight slots, a!!8ume the current
density to be uniform over each 21</8 rad or ± (2,../16)a meters (where a is the radius
of the stator) around the center of each slot. Assume that each of the active coil
sides shown hy a dot or cross contains N conductors. Each conductor carries the
terminal current of i amperes in the directions indicated. List all of the approxima
tions you use in solving for the air-gap flux density vector.
Fig. P6-2
Construction of the primitive machine 307
6-3 Figure P6-3a shows a. developed view of a stator with a four-pole double
layer full-pilch wave winding. The stator strueture has 12 equally spaced slots with
slot 1 at 6 = 0, slot 2 at 6 "./6, etc., for the remaining slots. The coils are so wound
that cach active side contains 15 turns connected in series. The coils are con
'nected into two separate circuits labeled A-A' and R-B'. The term double layer
means that each slot contains two active coil sides. 'Full pilch indicates that the
active sides for each coil are 'lrln radians apart, where n is the number of pole pairs
developed by the winding. In this case n = 2 and thus the coils with sides in slot
1 have their other sides in slot 4, which is 11'/2 rad away. The term "wave" refers to
the way the end connections are formed. See Fig. 1>6-00 for an illustration of a lap
winding.
a. For the connections shown in Fig. P6-3b show that the stator-surface current
can be approximated by
I!lOi ,.. 3".
J = - -".a a. for II < 6 < 2 and ... < 6 < '2
J = 18~ a. f or ".
2 < 6
a.r <
< 11' an d '2 6 < 211'
'Ira
where a is the mean radius of the stator.
11' 211'
, • IJ
A
8 (a)
-,-...
I
-,-.
I
li/z
..i
0
- i
TT/Z TT 2."
I I I .0
c
o (a) 0'
-i
(d)
(b) (e)
(e)
Fig. P6-5
The series of arrows around the windings indicate motion of the coil sides with respect
to the stationary brushes for positive angular rotation of the rotor.
a. Carefully examine Fig. P6-7 and verify its oorrespondence to Fig. P6-OO.
Make 0. drawing similar to Fig. 6-10 for this rotor winding.
b. Indicate the brush positions for a r/6-rad rotation of the rotor, and show that
the rotor-surface-eurrent distribution is stationary with respect to the stator.
-Brush II
\
8'
Fig. P6-7
c. Determine the distribution factor for the winding; assume that each coil
contains 10 turns in series. (Hint; See Prob. 6-5, particularly Fig. P6-5e, for an
outline on finding K.)
d. Repeat part c with the rotor turned through 1</2 rad.
6-8 Repeat Prob. 6-4 for the rotor winding shown in Fig. P6-7.
6-9 a. By expanding the integral given by Eq. (6-39), develop the self- and
mutual-inductance expressions given by Eqs. (6-42) through (6-48).
b. Why can Eq. (6-39) for the total magnetic energy stored in the air gap be
equated to Eq. (6-33), which represents the coenergY-Btate function?
c. Give a physical explanation showing that all the mutual inductances in Eq.
(6-48) are zero.
6-10 Derive Eqs. (6-96) to (6-99) for the rotational inductance coefficients for a
multi pole winding. In this derivation explain why the integrals are evaluated over
limits spanning 1< rad. (Nok; These limits are equivalent to using a span of ... /n.
radians and then multiplying the result by n.)
6-11 The rotor structure of a nonsalient primitive machine has 56 slots. A two
pole double-layer full-pitch lap winding consisting of 56 coils is placed in the rotor slots
and connected to a 56-segment commutator. (Note: This rotor winding is identical
with the winding in Fig. 6-10.except that 56 coils are used instead of 12.) Each rotor
coil contains five turns. The stator contains two identical sinusoidally distributed
two-pole windings in quadrature. The stator distribution factor is K' - 2900
turns/m. The physical dimensions of the machine are as follows: rotor core aD
4.585 in., stator ID (bore) - 4.631 in., rotor core length - 3.0 in., stator core length
3.0 in. Both the rotor and stator are wound with double-stranded No. 20 copper
wire having a diameter of 0.032 in. per strand. One pair of brushes ill placed on the
commutator ... rad apart, and they are so aligned that the rotor circuit has an axis in
Construction of the primitive mochine 311
line with one of the stator windings. A second pair of brushes is placed in quadrature
with the first pair, thus establishing a second rotor circuit in line with the other stator
winding. The total primitive machine can thus be represented by Fig. 6--13.
Owing to the slots used to hold the windings on the rotor and stator structures,
the air-gap length must be increased. By using a flux plot, the reluctance between
the actual rotor and stator surfaces can be determined for the actual slot configura
tion. An effective air-gap length is defined as the length between two smooth rotor
and stator structures giving the same reluctance. For the primitive machine in this
problem the effective air-gap length is 2 times the distance between the rotor and
stator teeth.
a. Calculate the resistallce R' of each of the rotor windings. The resistivity of
copper is PCu - 1.72 X 1O-8 ohm-m. Hint: Between either pair of brushes there are
two parallel paths of 28 coils each.
b. Calculate the resistance R' of each of the stator windings by using (6--103) or
(6--104).
c. Determine the distribution factor K- for a sinusoidal approximation for the
rotor winding. Hint: Follow the steps outlined in parts a and b of Prob. 6--3.
d. Determine the self-inductances of all four primitive-machine windings.
e. Determine the mutual inductances between each pair of primitive-machine
windings.
f. Determine all four rotational inductances for the machine.
g. With a 2-amp current in each of the stator windings, find the magnitude of
the open-circuit voltage across each pair of brushes when the rotor is stationary and
when the rotor revolves at 126 rad/sec. Note: The parameters used in this problem
have been taken from the Westinghouse generalized machine. If the actual machine
is available, your answers could be checked experimentally.
chapter
EQUILIBRIUM EQUATIONS
7 FOR THE PRIMITIVE MACHINE
Quadrature axis q
Oirect
axis d
tion that the magnetic field in each stator winding is in the positive axial
direction for positive winding currents. For example, id, the direct-axis
stator current, if actually flowing as indicated ill Fig. 7-1, creates a field
directed to the right in the rotor of the machine. The dots originally used
on the windings in Chap. 6 are being omitted with the understanding that
reference currents are always to be chosen into the aots.
The two rotor windings are of the commutator-and-brush variety
discussed in Sec. 6-4. These windings create magnetic fields which are
stationary in space and eompletely independent of the rotor position.
The rotor fields are assumed to depend only on the currents in· the respec
tive rotor windings. As before for the stator windings, the direction of
the rotor currents is chosen to create a uniform magnetic field in the rotor
cylinder oriented in the positive direction for the respective axes. Thus
i;, the quadrature-rotor current, if flowing in the assumed direction, creates
a uniform field in the upward direction in Fig. 7-1.
The terminals of the four electrical windings and the mechanical
shaft of the machine represent places where energy can flow in or out of
the primitive machine. The term port is commonly used to denote these
energy entrance and exit points. The assumed positive sense for all
winding terminal voltages is 80 chosen that energy is flowing into the
winding or port if the respective port power is positive. Thus if the
winding current is actually flowing in the assumed direction, then a posi
tive terminal voltage indicates an instantaneous energy flow into the
winding. For an electric motor, which converts electrical energy to
mechanical energy, we would expect that average port power would
actually be positive at the electrical ports of the device. However, with a.
generator, converting mechanical energy to electrical energy, average
power will
be negative at certain electrical ports.
314 Principles of elec:tromechonicol-energy conversion
For the single mechanical port, physically the shaft of the machine,
the assumed positive direction for angular velocity wr is taken to be
counterclockwise, as shown in Fig. 7-1. This direction corresponds with
the usual direction for positive angle_ In order to again have a positive
mechanical-port power, corresponding to energy flow into the mechanical
port, the assumed direction for the externally applied torque Tr is also
taken to be counterclockwise. If energy, in reality, is flowing out of the
mechanical port, then either wr or Tr will turn out to be negative, or
opposite to the assumed positive direction, and the mechanical port
power will be a negative quantity.
If all of the terminal quantities, namely, Tr and w' for the mechanical
port, and all the v's and i's for the electrical ports are positive, then all
port powers are positive and energy is flowing into all ports of the prim
itive mJchine. This would mean that the device must be either storing
or dissipating (in the form of heat) all this accumulating energy. For an
energy-conversion device such operation does not seem very satisfying;
consequently, we shall be expecting energy actually to flow out of at least
one of the five ports.
Direct-stator Winding
The equation for this winding is given by
• - R'"
Vd - dId
+V d
di~
dt + M" de
d
did (7-1)
field in the same direction. If the conventional dot notation were included
on the two windings, both currents would be into the dots, resulting in a
positive sign on the mutual term. The positive sign preceding the mutual
term indicates a voltage opposing the flow of the direct-stator current i~.
No further terms are needed in Eq. (7-1), since all other windings are in
quadrature with the direct-stator winding.
Quadrature-stator winding
Following the reasoning for the direct-stator winding, the equation for
this winding is given by
(7-2)
Quadrature-rotor winding
The equation for the rotor windings becomes slightly more complicated,
since rotational voltages must be included. The loop equation for the
quadrature-rotor winding is given by
d" d"
v'q = R'i'
q q + U::::!.J.
q dt + M"~
dt -
q Grow'i'
qd d - G"w'i'
qd d (7-3)
q Direction of induced
in the +d direction
Assumed
di rection of i;
~~~---4------ ~------~d
voltage iends to aid or hinder the flow of l~. If the induced voitage
hinders the flow of t~, then the sign of the term should correspond to that
for the resistance and self-inductance terms in Eq. (7-3), which certainly
oppose the flow of i~. Figure 7-2 shows the distribution of currents on the
periphery of the rotor set up by a two-pole quadrature-rotor winding and
i;. t As shown, the left half of the rotor has currents out of the paper and
-the right half has c:urrents into the paper. With both the direct-rotor and
direct-stator windings creating a field directed to the right, the induced
currents in the rotor conductors, moving in the assumed direction for the
rotor velocity, would flow as shown by the inner circle of dots and crosses.
This direction is determined by forming the vector cross product of the
velocity of the conductors with the direction of the direct-axis field.
SiIlce the induced currents are in the same direction. as the assumed
quadrature-rotor current, the induced-voltage term is aiding the flow of
~~. Consequently, the rotational terms are negative on the right-hand
side of Eq. (7-3). Equation (6-64) indicates that the voltage induced in
the quadrature-rotor winding from the direct-stator and direct-rotor wind
ings has a positive sign. The reference sense on this induced voltage
was originally chosen in the development of Sec. 6-7 to be positive in a
direction aiding the conventional direction of t~. Thus the intuitive
argument presented here is consistent with the original derivation of the
terms in Sec. 6-7. ...
Direct-rotor winding
t For the purposes of this qualitative discussion neglect the iact that the surface
current density has a. sinusoidal variation.
Equilibtium equations for the primitive machine 317
Directions ot induced
currents due to a field
in the +q direction
Assumed
direction of i/
~~+---+------e~-----.d
where the symbol p is used for the differential operator d/dt. t By using
an abbreviated matrix notation, Eq. (7-5) can be written more con
veniently as
~r = v~
dq ":J
v~
vq
r
(7-7)
td
.,
·.r _ 'to'
i;'J
Idq - 'r (7-8)
td
6l =
[R: 0
0
R'q
0
0
(7-9)
0 0 R~
. 0 0 0 iJ
~~]
0 M';
[ L:
~ = ~';
L'q 0
(7-10)
0 L~
M"q 0 Lrq
~]
0 0
~= [ 0
0 0 0
(7-11)
Gd~
-~;d
0
0 -
Grrqd
Equation (7-6) is a very general form representing the loop equations for
the electrical ports. If more than four ports were present, the form of
this equation would not change. The number of terms in each of the
individual matrices would increase, but no additional matrix quantities
would be required.
Electricol Mechonicol
ports port
r-------,
p;
pi Loss
pi' If
{p;
dq Storoge Fig. 7-5 The first law of thermo
p'It dynamics applied to the d,..q primitive
machine.
mechanical ports have been included. A heat port has been neglected,
because with electric machines energy flow, in the form of heat, is only
outwardly directed.' The quantity of energy leaving the primitive
machine system, in the form of heat, must be supplied from either the
mechanical or electrical ports or from the stored-energy supply. There
fore, for lour system, the first law can be expressed as follows:
The algebraic sum of the energy flow into the electrical and mechanical ports
of the primitive machine is accountable in either energy storage or energy
loss in the form of heat.
in which p~t"l is the total instantaneous power at all ports of the machine,
Pd~ is the portion at the electrical ports, and P;" is the portion at the
mechanical port. When Eq. (7-6) for v~~ and Eq. (7-12) for TT are sub
stituted into Eq. (7-17), the total power is expressed by .
p~~tAl = (ido)tlR(id~) + (id~)t£p(i~~) + (id;)twT(J(id~)
+ wTJpwT + wTDwT + wTT. (7-18)
The various terms in Eq. (7-18) can readily be identified as follows:
Power loss = P~":' + P:;:·· (7-19)
where (ido)llR(id"q) is ohmic or electrical power loss and wTDw T is viscous
mechanical or windage loss.
Power into storage = P'd~()red + p~or.d = (idq)'£p(i do ) + w'Jpw'
(7-20)
where (ido)t£p(i::q) is power going into storage in the magnetic fields and
w'Jpw' is storage in the form of kinetic energy in the rotating inertia. A
compliance torque would result in mechanical storage in the form of
potential energy in a twisted shaft. I
The remaining two terms ill Eq. (7-18), since they are not contained
as contributors to either loss or storage, must equate to zero. Therefore
(i::.)tw'(J(id~) + w 7'. = 0
r (7-21)
and thus the torque of electrical origin is given by
T. = - \
w
(id~)twr(J(i:rq) = - (i~~)I(J(i~~) (7-22)
When Eq. (7-2~) is expanded in terms of the matrix expressions for i~~ and
(J, given by expressions (7-8) and (7-11), the torque of electrical origin for
the primitive machine becomes
T. =
o
o
0]
0 "i
G~~:~
d
]
-G';a 0 i;
(7-23)
For the special case when the two rotor windings have the same turns
distribution factors (that is, K~ = K~), Eqs. (6-107) to (6-110) give
equivalent relationships for the rotational inductances in terms of the
I Capacita.nce effects in the windings would a.ccount for electric-field energy storage.
On the basis that this quantity of energy is negligible, these electric-field effects have
also been neglected.
322 Principles of electromechanical· energy conversion
T = - p. (7-26)
• W,.
_
.. - - _.. ......- .---_._..._ - •....... - - - _ .
i2 = - ~ i
N2 l
(7-28)
(7-30)
v. = Ni;VI
and
(7-31)
If energy flow is into port 1, then Eqs. (7-30) and (7-31) give
" (7-32)
324 Principles of electromechanical-energy conversion
Nz
[~}
+{:Ji.."
"'1
,
t} :
I
F.}'-,
Ideol Fig. 7-7 A multiwinding ideal transformer.
showing that the power is again invariant. All of the energy supplied to
port 1 is found emerging from the remaining ports. The device is ideal
in the sense that it loses no energy and simply transforms the magnitudes
of the voltages, aceording to Eqs. (7-30), and imposes the current con
straint of Eq. (7-31). Notice that only one port voltage can be inde
pendcnt.ly specified. However, (k - 1) currents could be arbitrarily
selected.
One further point should be mentioned with regard to the constraint
definitions given by Eqs. (7-30) and (7-31). These relationships are
taken to apply without regard to the form of the voltage and current time
dependent functions. Thus, for example, if VI(t) were a constant d-c
value, then Vk would still be given by Nk times this value and would also
be a constant provided Nk was a constant. As we shall use this multi
winding transformer in following sections, the number of turns on certain
windings will be found to be functions of certain port currents.
'l]
u'q
Ud
-
-
[R: +0 L;p
lW;p
R~
0
+
G~~u{
L;p
R~
M~rp
0
+ D'dp
on
M';p
G'daw'
i~
i'd
(7-3:!)
v~ -G;'aw r M;'p - G"
qd W r R; + L;p i;
T' (Jp + D)w' + (G;~ - G'da)i'di; - Gd~i;~~ + G;~id~~ (7-34)
Direct
stator port
Direct
I':i rotor port
Rotor
port
w:{ J
o---..I--...i-----l
+
Quadrature
rotor port
Ideol . 1.1",/
f f
~
~
}:! ,
Quadrature
stator po,.t
R:
Fig. 7-8 The complete equivalent circuit for the five-port do{] primitive maehine.
326 Principles of electromechanical-energy conversion
where the subscript 0 refers to the cOllstant quiescent or d-c portion and
the subscript 1 indicates the time-dependent incremental portion.
These five sets of equations can now be substituted into the port
equilibrium equations (7-33) and (7-34). Since the resulting equations
must be valid at the quiescent points, with all increments equal to zero,
two sets of port equations result. The first set of five equilibrium equa
tions involve only the steady-state quiescent portions of the port vari
ables. The second set of five equations involve the incremental portions
of the variables as unknowns with the quiescent values appearing in
certain coefficients. If all products of incremental quantities are neg
lected, the resulting sets of equations are as follows:
o
o (7-40)
R~
-G~dw~
(7-41)
o Mdrp
R~ + L~p o
G~;w~ R~ + L~p
Jlf~rp -G~dw~
-G~;i~o (G;~ - G~~}i~o - G~;i;o
o
M";p
G~~wo
(7-42)
R~ + L~p
7-7 SUMMARY
A general analytic model machine has been formulated to aid in the
analysis of a wide class of electromechanical energy-conversion devices.
This model, named the primitive machine, consists of two magnetic
axes in space quadrature. On the stator of the model machine two
windings are placed, one on each axis. The rotor structure also has one
winding on each axis. The rotor windings are connected to their external
Equilibrium equation. for the primitive machine 329
PROBLEMS
7.1 For each of the following conditions on the primitive machine shown in
Fig. 7-1 describe the vector-magnetic field inside the rotor cylinder. Assume the
stator structure to be nonsalient. Assume also that all four windings have identical
distribution factors. All currents are in am peres, and the speed is in radians per
second.
.,
Condition i:' i~ t. i"
• ",'
4 1 1 1 1 0
b 1 1 I 1 100
c 1 1 -1 -1 100
Ii 0 0 C08 t sin t 100
e I cos t 0 0 sin t -100
7-2 The primitive machine shown in Fig. 7-1 has four windings with distribu
tion factors K~ = K: and K~ - K; and the following parameter values:
Rd = R; = 10 ohms R~ = R; = 1 ohm
J = 20 kg-m t G,:.=G;'d=3h
D = 0.5 newton-m-sec/rad
The rotor speed is maintained constant at "'~ = 4.0 rad/sec. With a short circuit on
the quadrature-rotor port and the other two electrical ports open-(lircuited:
a. Find all of the steady-state port currents.
b. Determine all of the steady-state port voltages.
c. Find th.e average torque of electrical origin and the average torque required
to maintain the rotor speed at 4.0 rad/sec.
d. Calculate the average steady-state power at each of the five ports of the
machine and prepare a complete tabulation of the power transferred through the
machine. Include in this tabulation the input or output at each port, all losses, and
the converted power.
7-7 a. By expanding the matrices in Eqs. (7-19) and (7-20), show that these
equations correctly represent the power into storage in the primitive machine.
b. (Optional) For sinusoidal excitation on the electrical ports of the primitive
machine show that the steady-state ohmic power loss is given by
where I::.is the column matrix for the port currents in sinor form and the asterisk
indicates conjugation of the matrix.
7-8 The two totor windings on the primitive machine, with the rotor revolving
at constant speed, can serve as a transformer capable of handling direct voltages
and currents.
a. Write the d-(l steady-state equations for the two rotor windings.
b. Derive an expression for the resistance of a load RL connected to the quad
rature-rotor port as seen from the direct-rotor port under d-(l steady-state conditions.
c. For the connections of part b derive an expression for the efficiency of the
d-(l transformer. Neglect rotational windage losses, and assume that R'd and R;
are very small.
7-9 The following parts are to be answered for' an ideal transformer having
five windings.
a. How many port variables (voltages and currents) describe the state of the
transformer?
b. How many constraint equations between port voltages does the transformer
impose?
c. How many constraint equations between currents does the transformer
impose?
332 Principles of e/ectromechonico/-energy conversion
2n
Gqf
v;=-.f2 Vsin wt
RL
Fig. P7-12 Fig. P7-13
Equilibrium equations for the primitive machine 333
Tf =1'1
- seconds (9-36)
R,
(9-37)
seconds (9-38)
ohms (9-39)
By carefully designing the compensating winding, the feedback and
feed-forward paths in Fig. 9-16 can be eliminated. Following the form
of Eq. (6-94) the mutual inductances between the control field f and the
compensating and armature windings, for a two-pole machine (n = 1),
are given by
G,,< = p.oa'l1l'K"K c
go
(1 + 1l.!.)
2go
(9-42)
3l
where
K. = fiq!!h7~:'o)2 volts/volt (9-48)
Example 9·3
The Ward Leonard system discussed in Example 9-1 is to be controlled by adjusting
a. small potentiometer. Since the generator field winding requires more power than
•I Molor
v~
!
Molor rI field
~--.~-~ \ ~II~
230vol15
w.:;..:v
J(£ r,
Fig. 9-17 A Wa.rd Leonard speed-control system using an llmplidyne for an addi
tional stage of amplification.
Dired-current commUfator machines (II) 403
By 8uhRtitutinjl; param('ter valucR, with the !!pced "'~. in radians per sC(,ond, we have
Thc generator fi('111 winding is in a volt.ajl;e-divider ('ireuit "dth the amplidyne armature
!'ir!'uit. r!'1'i8tanc'('. TheTpfore. the st,('ndY-fltate jl;pnerator field voltage is
II:.....R, + R'
R: + R: K ".
~.----."----.- vI
II~. -
2:~() X (7i)
-"
+ 2.0)
71) X 11.6
t The suo.~pript a rcfcrs to the amplidyne, and tht' :mpt'Tseripts follow the notation
used in this !!('('tion. Thifl notation is therefore I'onsistent with that of Example 9-1.
where suhscripts g and", refer to the gf'nerator lind motor.
404 Prirn;iple' 01 eledromechanicol.energy conver,ion
I,. If the generator field voltage is to he 120 per cent of the rateel value, thp
battery V in Fig. 9-17 must have a value not Ipss than
V ~ 1.2 X 20.4 ... 24 ..'> volts
The potentiomt'ter R, for a linear voltagf' adjustmf'nt, shoulel be approximately
10 pt'r eent of the {'ontrol ficld resistan!'!'. Thprefore
R - 0.1 X 1200 == 120 ohms
With the potentiometer tap slightly down from its maximum setting, the current in
the pot!'ntiomeh>r above thp tap is approximately
v
I - -
R
+ -R!t" ~
24.5
-120 24.5
+ -1200 == 0.22 amp
The potentiometer wattage rating eorr""pondinp: to thi" WOTllt ron <lit ion ill thprf'forl'
Watt,Jge == (O.ZZ): X IZO "" 5Jl watt!!
Thl' powl'r gain of the amplidyne can be caleulated from
. output power (v!o)'IR~
(75 + 2)'
Notice that the voltage gain of the amplidyne is only 11.6, but the power gain for
this circuit is 2040.
c. The overall sinusoidal steady-state transfer function for the complete system
shown in Fig. 9·17 is Bimply a cascading of the functions given by Eq. (9-10) for the
generator and motor and Eq. (9-47) for the amplidyne. However, the generator
field resistance R;
must include the total resistance of the amplidyne armature
generator field loop. Thp generator field inductance I!! not altered owing to thf' unity
coupling between the d and c windings of the amplidynp. ThuB for R: in Eq. (I)- 11) Wf'
should use 77 rather than 75 ohms.
The overall transfer function, using the parameter values in Example 901, is
therefore
Q;'(j",) 11.6 0.80
---- = -~-- X .. .-----,-
V!(j",) (0.167j", + 1)(0.6j... + 1) (0.117}", + 1)(0.44j", + I)
For the amplidyne we have substituted
L~ 200
T'• - -R! = - - == 0.167 sec T' = L! = 0.6 '" 0.6 sec
1200 • R! 1.0
of -360·, The amplidyne has added a significant power gain to the system. How
ever, the bandwidth has been reduced from 2.3 to 1.67 rad/see. Notice that the
amplidync's quadrature-winding time constant governs the bandwidth figure. In
order to have II. large quadrature field, Il! is made as small as possible by an external
short circuit. Thus the quadrature-winding time constant,.: = IJ:/R! becomes large.
The ('omhination of large gains and large time constants tends to make the system
unstahle ir reedback is included.
(9-49)
ItO
Fig. 9-18 D-c steady-atate block diagram representing the bBllic metadyne.
406 Principle. of electromechanical-energy conver$ion
Example 9-,4
A constant-('urrent metadyne jl;f'nE'rator is constructpd hy removing the compensating
winejin!!: from thE' amplidynl' in Example 9-a, with all other parameters rl'maininjl;
f'onstant.
a. Assumin!!: that th!' machine is supplying field current to the same jl;enerator
as in Example 9-:3, ('hpC'k the approximation of Eq. (9-50). Compute the gain K.
of thp metadynf'.
h. Calf'ulate thp powpr lI:ain of the metadyne and make a comparison with the
amplidynp.
Soiulion: Il. Thp ampli<iynp IIn!1 nJ!'tadyne are constructed with a nonaali!'nt
!;tator anll two pairs of hrushps on the Hame commutator. Therefore, for the machine
in EXRJUI>h' !I-a WI' havl'
R, = H. = I.n ohm G•• = G.. 0.26 h
Equation !!l-IlO) ill ...hpckpd as follows:
]("iU.( + /( tJ = 1.0 X 11.0 + 75) = 76 ohms
r;.,I;".lw~)· = m.2G X lK!J)2 24200hmo;
Rim'e 76 is mudl 11'88 than 2420, Eq. (\)-.')0) is certainly reasonahle for this particular
pxampll'.
The metndyne gain from (9-1>1) is
K G,/ 1.5 4 / I
i = "(:•• Iij = n.2!i X 120() = .R rna vo t
Hat!'d field ('urrent for the generator load is \~ = 3.1 amp. With a gain K. "" 4.8
rna/volt the metadyne would require If/o = 650 vo\t1'l to supply rated load current,
as opposed to approximately V/O 20 volts in the <'ase of the amplidyne. The con
stant-current feature of the metadyne is achieved by large <,urrent feedback at the
expense of rorward gain.
b. The power gain is calculated rrom
The self-excited shunt genera.tor shown in Fig. 9-22 is one of the most
commonly used of all doC machines. Until the introduction of the alter
Dired-eulT'lln' commutator machine, (II) .co9
(9-54)
Fig. 9-23 Block-diagram repreaentation of the equilibrium equations for the self
excited shunt generator.
410 Principles of electromechanical-energy conver.ion
I
I
I
1
I
I RaRf
Slope = R, + Ra + -n
I L
I
I
I
I
I
Equation (9-55) is nonlinear because G' and Ga , are functions of the field
current if.
The static equilibrium points are obtained by assuming i, = i,o,
where i/o is a constant or d-c value. Equation (9-5.1) thus reduces to
The operating points for the machine are obtained when the left-hand
side of Eq. (9-t>6) equals the right-hand side, or when the straight line
in Fig. 9-24 intersects the open-circuit characteristic. For example,
when RL = R LI , the intersection occurs at point A and the field current
has a value i/o = i A • The steady-state terminal voltage of the machine
is simply
(9-57)
so if i/o is known, the d-c load voltage VLo can be calculated. Now by
reducing RL to a value Ru in Fig. 9-24 we see two intersections at points
Band C with corresponding field currents i8 and i e . By multiplying
iB and ic by R" the corresponding load voltages VLO are obtained.
Figure 9-2.1 shows a plot of the load voltages as a function of the
load currents iLO = vLo/R L for the same values of RL as used in Fig. 9-24.
Thus for RL = RLI the load voltage is R,iA , as shown in Fig. 9-25. By
decreasing RL to RL2, we have two possible values of VLO, as shown. In
a similar fashion the remaining load values Ru and RL4 are transferred
from Fig. 9-24 to Fig. 9-25. Notice in Fig. 9-25 that for a load ~L
between Ru and RL2 three load voltages seem possible. We shall show
that the intermediate voltage value, between points C and D, is unstable
and physically never occurs.
In order for the self-excited machine to build up to an appreciable
output voltage, the load RL must be greater than R L2 . Figure 9-25
shows the output voltage to remain fairly constant for all loads greater
than R L2 • If the load is less than IlL2, then the voltage builds up only
to the low value at point C. Further decrease of RL to RLI and RL4
~~ -------- ---
~ia
i£Ol s.c.
causes the load voltage to reduce even further to points E and P, respe(~
tively. For RI- = 0 the field current reduees to zero, the load voltage
is zero, and the load current is given by
.I
tLO =k.wo
_. (9-58)
•• r. R"
as shown in Fig. 9-25. The seif-exeited shunt machine is thus current
limited and tends to protect itself from a short-circuited load.
The stability of the static operating characteristic show II in Fig.
9-2.5 can be checked by obtaining the linear incremental differential
equation for the machine. Subtracting the steady-state equation (9-56)
from the general equation (9-55) gives us
where ifl is the increment in the field current. Notice in Eq. (9-,19)
that Gal is to be evaluated at the operating point.1 Now R(I. (9-f>9)
gives rise to a stable incremental response provided all the coefficients
are positive. Thus stable operating points are obtained for
The inequality of (9-60) is satisfied when the slopes of the straight lines
in Fig. 9-24 are greater than the slope of the open-circuit characteristic
where the two intersect. Thus, for intersections between points C and D,
(9-60) is not satisfied and the operating points are unstable. For all
other intersections, (9-60) is satisfied and we have stable operating points.
v(i,) .. v(i,o) + d~ I
d~, iI.
i,1 + ...
The term v(i,p) hall been used in the steady-state equation (9-56), The coefficient
dl1/di, evaluated at if .. if. is simply ",'Il., (i,p) , and therefore only the term
",'Il.Ai,.)i,1 appears in Eq. (9-59).
Direct-current commutator machine, (II) 413
are shown in Fig. 9-2;'). For a load resistalll'c greater than R1.2 the output
voltage reillains relativt'ly (·onstan!.. Bclow a eriti('ai load resist,allce
the machine call not huild up to its rated voltage and remains at a low
output voltage. In this section we shall examine a graphical technique
for solving the nonlillcar equilibrium equation for the sclf-exdted machine.
For small excursions about an operating condition the linear incremental
equation (9-59) can be used to predict transient. changes from one operat
ing point to another. However, for the initial buildup of the generator
voltage, from zero to its final operating value, an incremental technique is
very tediou!<.
With most self-excited shunt generators the armature inductance
is very small and can he neglected. Therefore, with L.. "" 0 t.he equilib
rium equation (9-5.''» for the machine becomes
After sollie slight rearrangement Eq. (U-B1) ('an he put. int,o the form
(9-62)
From Fig. 9-24 we call see that the right,-hand side of Eq. (9-62) is simply
t.he vertical distance from the straight line at, a slope RI + Ra + R.RdR,.
to the open-drcuit. l'haracteristic curve. Referring to Fig. 9-26a, the
right side of (9-62) is the voltage increment t.v.
By separation'of variables, Eq. (9-62) can be rewritten in the form
(
1.1 + I:..IRa) (il(/) di'. = t (9-64)
R ,. 10 t.1I('/)
if
(al
Rotor
a, I,
0
+ l
1712 1r
I
¥
31r/2
I
+
I
I
~1r
I •
k r 1 [ ]
9
I Stotor
lal
, ..
(b)
,I
Rolor 2."
field
9
(c)
Totol
field
9
Fig. 9-27 (a) Developed view of the air gap for a practical two-pole d.c machine.
(b) The radial air-gap magnetic field established by currents in the stator field windinp;.
(c) The radial air-gap magnetic field established by the armature currents. (d) The
total radial air-gap magnetic field due to both field and a.rmature currents.
Direct-evmmt commutator machine. (II) 417
For currents in both the stator and rotor windings, the total air-gap
field is obtained by adding Fig. 9-27b and c, as shown in Fig. 9-27d. The
effect of the armat.ure or rotor currents in distorting the total air-gap
field is known as armature reaction. If the armature current is zero, then
the total field reduces to Fig. 9-27b with zero field at 8 = .../2 and 8 = 3.../2.
With currents in the armature, the armature reaction shifts the magnetic
neutral, or zero field point, to 8 = 8" and 8 = 8" + ... in Fig. 9-27d. The
value of 8. depends on the magnitude of the armature current.
i/Z
ti ti
(al (bl tel
Fig. 9-28 A detailed view of the commutation procel!8 (0) just previous to the
commutation interval, (/I) at the center of the interval with the brush equally spaced
between adjacent commutator segments, and (c) at the concluaion of the commutation
interval.
418 Principle. of eleclromechanical-energy conver,;on
tl/2
v
. -
ie
Lc Re
8
fi
1
i2 t (l-{JRb
te
(9-65)
9~28a shows the right-hand coil side just previous to the commutation
period carrying i/2 amperes in the - a, direction. In Fig. 9-27a the a.
direction is out of the paper alld the rotor moves in the a, direction.
Thus the right-hand coil side is located at an angle (J just less than 1r/2.
Similarly, t he left-hand eoil side is located at all angle (J just lel'll'! than 31r/2.
Figure 9-27d shows, the total air-gap field at (J = 1r/2 to be in the a.
direction. Similarly, at (J = 31r/2 the total field is in the -a. direction.
With the rotor revolving at an angular velocity of w' rad/sec a voltaga is
induced ill the dosed loop as given by
where a is the rotor radius and B is tlw total air-gap field. By evaluating
Eq. (9~6(), we have
whereT = L./Rbseconds, and the assumption has been made that Re« Rb,
which for carbon brushes is usually a valid approximation.
Figure 9-30 shows characteristic plots of Eq. (9-69) for various values
of v and T. At t = 0 t,he coil current it starts at i/2 amperes. When
t = ie, t.he current is forced to be. -i/2 amperes as the brush leaves the
left-hand commutator segment in Fig. 9-28. The instantaneous change
in ie at the end of the commutation period t ie is equal to
(9-70)
420 Principles 01 electromechonicol·energy conversion
Fig. 9-30 I The coil current as a function of time during the commutation interval.
If Ai. is not equal to zero, then a spark occurs between the brush and the
left-hand segment as they separate. The intensity of the spark is a func
tion of tli, and the magnitude of the coil inductance L •.
One method of eliminating the current discontinuity tli. is to shift the
brushes on the commutator. For the armature current assumed in con
structing Fig. 9-27d, the total air-gap field is zero at 8 = 89 and 8 = 8.. + 'If'.
Shifting the brushes so that the coil undergoing commutation has its
active coil sides located at some angle slightly greater than 89 and 8a + 'If'
would cause BN in Eq. (9-67) to be negative. Thus the rotationally
induced voltage v is also negative. By properly adjusting the new brush
position, tli., given by (9-70), can be made equal to zero and sparkless
commutation can be achieved. In the early days of d-c machinery,
brush-shifting mechanisms were incorporated on the machines. The
major difficulty with brush shifting as a means of achieving good commu
tation is the dependency on the magnitude of the armature current. AI'!
the armature current increase!', the shape of the air-gap field in Fig. 9-27d
changes shape. The magnitude of v therefore ehanges alld a new brush
pOl'!ition mURt be obtained to make tli. again equal to zero.
On modern d-c machines nearly perfect commut.ation is achieved by
adding small stator poles between the main stator poles, as shown in Fig.
9-31. These additional stator poles, usually ealled interpoles because of
their location between t he main field poles, have windings eonnected ill
series with the armature winding. For the armature-<:urrent direction
used in Fig. 9-27 the interpole currents would be as shown in Fig. 9-31.
The interpoles establish a localized air-gap field at 8 = 'If'/2 and 8 3'1f'/2.
The brushes are so plaeed 011 the machine that the coil undergoing com
mutation is sweeping past the interpole. An additional rot,ational voltage
due to the interpole field is thus induced in the coil being commutated.
Examining FigR. 9-27 and 9-31, the additional voltage /Jr due to the inter
Direct-current commutator machines (II) 421
Rolor
[jJ •
•
•
•
+
+
+
+
Slolor
[jJ +
+
+
+
•
•
•
•
Fig. 9-31 The inclusion of interpole windings on the d-e machine stator.
poles opposes the original rotational voltage v given by Eq. (9-67). The
coil-current di!'!eontinuity at the end of t.he (~Olllll\ut.at,ion period t = t. is
now given by
!J.i, = - (..!::..
llb
+ !t, i - ..!!)
Rb
(1 - e-I"r) (9-71)
the voltage hetween segments exeeeds the breakdown voltage, then a low
resistanee are will form between segments. The total voltage between the
hrushes now appears acro!';s the remaining segments, thus continuing the
arc formal iOIl unt ill he hruflhes arc l<hort-cireuit cd hy a cOlllplete flashover
or "ring of fire" at the COlHlllutator. The large di,,/dt can thus trigger a
('haill reo.<'tioll of arcing hetween adjacent eOlllmutator seglhents until the
entire commutator is short-ein'uited hy olle ('olltinuoul'l arc.
The possibility of commutator flashover can be greatly minimized by
the addition of the compensating winding shown in Fig. 9-33. The
compensat iug winding is a dist ributed winding wound on the stator t.o
establish a eurrenl. distribution on the st.ator surface that is as dose as
possible to the ('urrent distribution established by the armature. Now
the armature and eOlllpensating winding arc ('onllceted ill series in such
a fashion that the two ('urrent distrihutions on either side of the air gap
are cqual but opposite in direction. Therefore, the voltage appearing
aeross the brushes owing to self-induetion is now given by
(9-73)
}<'ig.9-32 The series connection of the individual armature coils at the commutator
segments.
Direct-cum'"t commutator machine. (II) "23
COmpensating
winding
'·13 SUMMARY
The Ward Leonard system is used where precise control of large amounts
of mechanical power is required. The source of this power can be an a-c
electrical supply, which is commonly available. The combination of
these two factors (i.e., precise control and readily available power) make
the Ward Leonard system a widely used industrial drive. In formulating
the equilibrium equations for the Ward Leonard system the formalism of
the primitive-machine notation was modified for greater convenience.
By inspection of the system, the interconnected equivalent circuit, shown
in Fig. 9-1, was drawn. If the general energy-flow direction is known, the
reference sense of all current and voltage variables as well as the trans
former dots can be assigned for a more compact formulation. From the
equivalent circuit, both the doc steady-state operation and the transient
operation are readily obtained. The response capabilities of the Ward
Leonard system are studied by using step- and frequency-response char
acteristics. Even though the system equations appear to be linear, an
incremental analysis is used to emphasize the fact that parameter values
are constant only in the neighborhood of an operating point.
424 Principles of .Iedromechonicol·energy conversion
The compound d-c generator has two stator fields in quadrature with
the armature. One of the stator fields is constructed of many turns of a
light-gauge wire and is separately excited. The other winding is made up "
induction in the armature coils. Since these coils are connected in series
at closely spaced commutator segments, a voltage breakdown or flashover
between segments can occur, particularly on large d-c machines. Com
mutator flashover is eliminated by adding a distributed compensating
winding in series with the armature. The compensating winding induces
a voltage equal and opposite to the self-induced voltage and thus tends
to cancel the voltage between commutator segments caused by rapid
changes in armature currents. The compensating winding is closely
coupled to the armature winding, making the net inductance of the series
combination practically zero. .
In the analysis of the various d-c machines in Chaps. 8 and 9 an
ideal commutator was assumed. Also, we very often assumed the anua
ture inductance to be negligible. Both the ideal commutator and zero
armature inductance are reasonable assumptions because interpole and
compensating windings are included in all machines in which brush
sparking and flashover are encountered.
PROBLEMS
'.1 A Ward Leonard speed-oontrol system is shown in Fig. P9-l. The ampli
fier input voltage is the difference between a reference voltage and the tachometer
voltage. Assume the voltage amplifier to be linear with a very high input impedance
and negligihle output impedance. If the motor speed decreases, then the tachometer
output voltage decreases and the net input voltage into the amplifier increases.
The amplified volta.ge applied to the generator field increll.lles, and 80 the genera.tor
Ra
9
Amplifier
Fig. P9-1
,(26 Principles of electromechanical-energy conversion
output voltage increases, thus raising the motor armature voltage and tending to
increase the motor speed. The generator is heing driven at constant speed, and the
field current of the motor is constant. Neglect the armature inductances of the two
machines. Other system parameters arc as follows:
G
1 >
-
Amplifier
----~
~
0",
\
\~
I
Fig. P9-2
Direct-current commutator machines (II) 477
c. From the complete block diagram of part a find the overall transfer function
8~(8)/VR(8). Use the Routh criteria to find the number of RHP poles. Is the
system stable?
d. Sketch a frequency response for the overall transfer function.
e. What would be the elTect of including the tachometer in the position-control
system, thereby producing a voltage subtracting from VB for positive motor speed?
9-3 A lO-kw 125-volt 1750-rpm compound generator has an armature resistance
of 0.06 ohm. With the shunt field separately excited at rated voltage, the generator
produces rated voltage when driven at rated speed with the series field short-circuited
and the armature open-circuited. The scries field has a resistance of 0.04 ohm and
induces 0.1 volt in the armature per ampere of series field current at rated speed.
The shunt field resistance is 75 ohms.
a. Find the two rotational inductances Ga , and G....
b. If the shunt field is excited at rated voltall:e and the machine is driven at rated
speed, find the output voltage when the machine delivers rated load current first
with the series field cumulatively compounded and then with the field differentially
compounded.
c. Find the overall efficiency for each case in part b.
d. Repeat parts band c for a generator speed of 2500 rpm.
9-4 A 220-volt 5O-kw 120ft-rpm compound d-c generator when running at rated
speed produces rated no-load voltall:e with a shunt field current of 5.0 amp. At full
load current 7.2 amp is required in the shunt field to produce rated output voltage
at the same speed.
a. Calculate the rotational inductance G., and the armature resistance R. for
the machine.
b. Assuming that G•• = 0.02R., where G•• is the series field rotational inductance,
in henrys, and R. is the series field resistance, in ohms, find the values of Ga. and R. to
flat-compound the machine at rated speed.
c. If a 0.2-ohm resistance is placed in parallel with the series field, find the output
voltage at full-load current with 5.0 amp in the shunt field.
9-5 A fully compensated amplidyne being driven at constant speed has the
following parameters:
The machine has an open-circuit output voltage of 250 volts with 0.1 amp in the
control field.
a. Write the open-circuit transfer function between the armature voltage and
the control field voltage.
b. If the control field voltage is suddenly changed from 4 to 6 volts, find the
open-circuit voltage as a function of time.
c. For a total armature load of 12 ohms in series with 25 h, find the steady-atate
load current with 4.0 volts applied to the control field.
d. For the conditions in part c find the load current as a function of time following:
a change in the control field voltage from 4.0 to 6.0 volts.
e. Find the d-c power gain of the amplidyne.
....28 Principles of electromechanical· energy converlion
9-61 Industrial applications requiring very accurate speed control of large drive
motors often employ the extended Ward Leonard system shown in Fig. P9-6. The
voltage at the input terminals of the electronic amplifier is the difference between
the preset reference voltage VII and the tachometer output voltage. If the motor
speed decreases, owing to an increased load, the tachometer output voltage decreases
and the amplifier input voltage increases. Tire amplifier supplies increased current
to the control field of the fully compensated amplidyne. The amplidyne supplies a
greater current to the field of the doC generator being driven at a constant speed
Amplidyne
~er
I ...r--~~/
~
/
/'--w;o
/
Tachomder
Gain K, .. 1 volt-see/rad
Motor ahaft load
Inertia J.. - 6000 kg-m'
Windage is negligible
Constant load torque TL - 40,000 newton-m
a. Draw a complete block diagram for the system. Reduce the diagram showing
VB and TL 8.8 inputs and the motor speed ...:;. as the response.
b. Find the d-c steady-state relationship between two inputs VR. and TLO and the
steady-state speed ...:;... What reference voltage is required for a motor speed of
15 rad/sec? With this reference voltage find the change in speed for a 10 per cent
increase in load torque.
c. The system is operating under the steady-state conditions described in part b.
Draw an incremental Laplacian block diagram showing operation about the steady
state conditions. Determine the transfer function relating 0:;'.(8) to V R.(8). Is the
system stable?
9-1 By running a lIO-volt d-c shunt motor 8.8 a separately excited generator
at a speed of 1800 rpm, the following data are obtained:
Field current 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70
Open-circuit armature
voltage 70.0 78.0 84.5 91.0 96.0 101.5 106.6 111.0 115.0
The armature resistance is 0.64 ohm, and the field resistance is 192 ohms.
a. If the shunt motor has 115 volts impressed across its field and armature, find
the no-load speed of the machine.
b. Determine the rotational inductance G./ for an incremental linear model
valid around the operating conditions of part a.
9-8 A d-e generator being driven at its rated speed of 1200 rpm has the following
saturation curve:
Open-circuit armature voltage 4 21 40 58 72 89 105 115
Field current o 0.1 0.2 0.3 0.4 0.6 1.0 1.5
The field resistance is 75 ohms, and the armature resistance is 0.8 ohm. If the
generator is driven at rated speed find:
a. The open-circuit armature voltage with the field separately excited across a
nO-volt supply.
b. The open-circuit armature voltage with the field shunt-connected. Is the
sense of the field connections important?
c. Repeat parts a and b for half rated speed.
d. Find the short-circuit armature current for the connections of part.s a and b
for rated speed and half rated speed.
9-9 A 5O-kw 220-volt d-c generator, being driven at its rated speed, has the
following saturation curve:
Open-circuit armature voltage 11 61 110 145 170 190 206 220 228 233
Field curren t O O . 5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.4
The armature resistance is 0.04 ohm, and the field rheostat is 80 adjusted that the
machine, operating as a shunt generator, produces 233 volts at no load.
430 Principles of .tecfromechonicol.energy conversion
10
II
from the rotor shaft, are so mounted that t.hey revolve with the rotor
structure. Leads are connected from the proper two points 011 the rotor
winding to these two disks, whieh are known as slip rings. Brushes
sliding on the slip rings provide the required ('onnection to the rotor
circuit.
Slip-ring windings on the maehine rotor ean be represented as shown
in Fig. 1()"2. The two axes are designated as the aT and bT axes, symbol
izing the a and b rotor axes on the a-b ma(~hine. Since the two windings
are on the rotor, they are drawn inside the cil'cle, as shown in Fig. 10-2.
The a axis is drawn at an angle a to the d axis of our fixed d-q axis frame.
As the rotor turns, the aT and fir llxes llnd the slip-ring winding turn right.
along wit h it. Therefore, the a-fl rotor axes fraille is essentially "glued"
to the rotor structure. Rotor windings supplied through slip rings are
drawn on the a-b rotor frame (Le., on the aT Ilnd br axes) and revolve wit.h
the rotor. Rotor windings on the d-q frame have their coil sides physically
moving with t.h& rotor, but because of the commutator-brush switching
mechanism the magnetic fields est ablished by currents in these windings
remain stationary in space and are completely independent of the rotor
position. For positive rotor currents i:
and fb the respectiv"l magnetic
fields are established in the positive axis directed as indicated by the
arrows in Fig. 10-2.
In Chap. 6 the current on the rotor surface is assumed to be sinus
oidally distributed. For the commutator-and-brush rotor windings of
our original d-q primitive machine the assumed sinusoidal distribution is
only the first term in a Fourier-series expansion of the actual square-wave
distribution of surface currents. The direct- and quadrature-rotor
winding surface ('urrents for all n-pole-pair machine, as given by Eqs.
(6-83) and (6-84), are repeated for ('ollvenience as
with K~ sin n8 and K; cos n8 being the respective turns per meter at an
angle 8 for each of the two windings. Equations (10-1) and (10-2) are
independent of the angular position or velocity of the rotor. The current
distributions have a periodicity of 27(/n radians.
The slip-ring windings shown in Fig. 10-2 are also going to establish
a sinusoidal current distribution OIl the rotor surface. Since no switching
is involved with the slip-ring windings, the active coil sides could actually
be grouped to establish a sinusoidal surface current in the same manner as
for the stator circuits. If the a-b rotor windings are uniformly distributed,
then a Fourier-series expansion is made and only the fundamental term
for the current distribution is retained. For the a-b rotor windings the
rotor-surface currents for an n-pole-pair machine are given by
J; :oJ K; sin n(8 - a) i;a. amp/m (10-3)
J~ = - K~ cos n(8 - a) iba. amp/ill (10-4)
where a is the angle measured counterclockwise between the stationary
d axis and the movable a axis or, equivalently, from the q axis to the b
axis, as shown in Fig. 10-2. The (:onstants K~ and Kb are the respective
distribution factors. The a-b rotor windings establish (mrrent sheets
identical with those of the d-q rotor windings upon substituting 8 - a for
8 in Eqs. (10-1) and (10-2).
For a changing a, Eqs. (10-3) and (10-4) will indicate that the (mrrent
sheet distributions move with respect to the stator. If the rotor is
revolving at a constant angular frequeney w~, then a w~ and
(10-9)
where 8max is the value of 8 where the maximum surface current occurs.
The value of 8max changes linearly with time at an angular rate of w~
rad/sec. Therefore, the two slip-ring surface-('urrent distributions are
The a-b primitive machine 435
B br /.loa Kr'r'
= ng(8) btb Sill n
(0 - a
) ar (10-11)
where a is the radius of the rotor and g(8) is t,he air-gap length having a
periodicity of "KIn radians as given by Eq. (6-74). If a = w~t, then these
air-gap fields will revolve in a counterclockwise direction at an angular
frequency of w~ rad/sec.
Suppose the t.wo d-q rotor windings are excited by sinusoidal currents with
t,he following phase relationship:
i~ V2 I~ cos (wt + "') (10-12)
where I~ and I; are the rms values of the currents, w is the radian fre
quency of the excitations, and", is a constant arbitrary phase angle. By
substitution of these two d-q rotor current.s into Eqs. (10-1) and (10-2) the
d-q surface-current distributions (~an be put. into the form
J~ = Ji [sin
Krlr
(n8 + wt + "') + sin (nO - wt - ",)]a. (10-13)
by using the trigonometric identity for the product of the sine and cosine
of two angles.
When Eqs. (10-13) and (10-14) are combined, the total d-q rotor
surface current is given by
In all of our work with the a-b and d-q primitive machines we shall find it
convenient to have all winding distribution factors equal. Thus, when
I Refer to Sec. 6-5, where a similar development is performed for the d-q rotor wind
ings having sinusoidal current distributions, with n - 1.
436 Principles of e/edromechanical-energy conversion
Kr
Notice that Eq. (10-16) is composed of two main terms. The first
term represents a current distribution rotating around the rotor in the
positive angular direction (countercloekwisc) at an angular frequeney of
win rad/sec. The second term in 00-16) is a similar surface-eurrent
distribution rotating in the clockwise or positive angular direction at the
same radian frequency win rad/sec. For the special ease when I~ = I;,
only a c~untercIockwise rotating current distribution exists. Similarly,
when -ld = I;.
only a elockwise rotating distribution is found on the
rotor surfaee. Notice again that the angular position or velocity of the
rotor does not afi'eet t he surface currents for the d-q windings.
A sinusoidally distributed current sheet rotating in the positive or
counterclockwise dire(~t,ion at an angular velodty of win rad/sec can also
be established by using the slip-ring windings. By supplying a direct
current to either t,he a- or b-axis rotor winding, a sinusoidal surface current
is established on the rotor. Revolving the rotor at an angular velocity
of win rad/sec will ('RUSe the current, whieh remains stationary with
respect to the rotor strueture, to revolve at the same angular frequency
in the same direction. The technique of establishing identical current
sheet distributions with either fOl'llI of rotor winding is considered in
detail in the next se(~tion.
Figure 10-3 schematically shows the d-q and a-b rotor windings and
their respective axis frames. Let us now examine what currents should
/
be supplied to the d-q rotor windings ill order to establish the same current
distribution on the rotor surface as that set up by a given set, of currents
in the a-b rotor windings. Thus, if i: and ''b are specified, we wish t.o find
values for i~ and t-; which result in idenHcal rotor-surface currents. The
total surface current due to i~ and i; is obtained by summing Eqs. (10-17) .
and (10-18), thereby gettingl
J~Q = (K~i~ sin n6 - K~i; cos n6)a. (10-21)
Similarly, the total rotor-surface current resulting from i: and ib is t,he
sum of (10-19) and (10-20), which becomes
J~ = [K:i: sin n(6 - a) - K/,ii: cos n(6 - a}Ja. (l0-22)
By using the trigonometric identities for t.he sine and cosine of the
difference of two angles, Eq. (10-22) can be expanded and rearranged to
the form
J~ = [(K:i: cos na - K;i; sin na) sin n6
- (K:i: sin na + Ki:i~ cos na) cos n61a. (10-23)
By comparing Eqs. (10-21) and (10-23), we can see that the two total cur
rent distributions J'aIJ and J'oJ, will be identical provided
(10-24)
and
(10-25)
I The double SUbscript and one superscript on J:I. indicate that both the d- and q-axis
rotor windings are being considered.
438 Principles of electromechanical-energy conversion
This relationship between the d-q and a-b currents can be expressed in
matrix form as follows:
(10-28)
where
(10-29)
(10-30)
a~,dq
(a r )-1 = [ c~s na sin na] (10-33)
do,Gb -sm na cos na
tws the a-b currents are being transformed into the equivalent d-q currents. The
single superscript indicates that only the rotor currents are being transformed.
"
revolving in the positive angular direction and the other revolving in the
negative angular direction. Let us now examine what would be the
equivalent currents in t.he a-b reference frame. For convenience, Eqs.
(10-12) specifying the direct- and quadrature-axis rotor currents are
repeated as follows:
'r
1""
= i:]i~ = V2 [ cos na sin na] I~ cos (wt
-sin ncr cos na I; sin (wt
+ <p)]
+ <p) 00-35)
The angle a, as illustrated in Fig. 10-3, defines the angular rotation between
the two coordinate systems. Suppose we allow the a-b coordinate frame
to rotate in the positive angular direction with a constant angular fre
quency of win rad/sec. This condition is expressed by setting '"
aCt} = ~ t
n
+ f;3 (10-40)
where f;3 is a constant phase angle defining the angular displacement of the
two coordinate frames at time t O. Substitution of this expression for
440 Prindp!e, of electromechanical-energy conver,ion
a into Eqs. (10-39) gives the currents in this rotating a-b reference frame as
~: = V2 r cos ('" - fJ) i~ = V2 r sin ('" - fJ) (10-41)
with both", and fJ as constant phase angles. Equations (10-41) show
that the currents in the a and b coils on this rotating coordinate system
are not time-dependent but are actually constant or direct currents.
Direct currents in stationary coils set up a constant or static surface
current distribution in space. By physically rotating the coils at a con
stant angular frequency, this curren~ distribution is caused to rotate in
space. Thus, a rotating current distribution can be generated by station
ary windings, such as the d-q windings, with time-varying current,s or,
equivalently, by rotating a-b windings with constant currents.
SUllpose that the original d-q windings set up a surface-current dis
tribution in the negative-angular direction. As shown in the preceding
section, this can be accomplished by having balanced excitation with
-14 = 1; = r (10-42)
Substituting these currents into Eqs. (10-36) and (10-37) gives the a-b
axis coil currents as
i: = - v'2 Ir cos (wt + '" + na)
(10-43)
i~ = V2I r sin (wt + '" + na)
As before, consider the a-b axes to be rotating in a positive-angular direc
tion with angular frequency win rad/sec. Therefore, substituting Eq.
(10-40) for a into (10-43) gives
i; = -V2Ir cos (2wt + '" + fJ) (10-44)
i~ = V2 Ir sin (2wt + '" + fJ)
Equations (10-44) indicate that the equivalent a-b axis currents are now
time-dependent at twice the radian frequency of the original d-q currents.
In fact, the currents, defined by Eqs. (10-44), are setting up a current
distribution rotating in the negative-angular direction at an angular
frequency of 2w/n rad/sec. Since the a-b axes are rotating at win rad/sec
in the positive-angular direction, the net angular frequency of the field is
win rad/sec in the negative direction, as it must be if this transformation
procedure is valid.
1
If'
i{
the four electrical ports and by Eq. (7-34) for the mechanical port. In
forming a d-q machine equivalent to the a-b primitive machine we shall
assume that t,he two d-q stator windings are identical. Also, the two
d-q rotor windings are identicaL Thus we have
K~ = K; = K' turns/m (10-45)
K~ = K; = Kr turns/m (10-46)
By use of the d-q parameters developed in Sees. 6-8 and 6-9, Eq.
(10-4.') or (10-46) can be substituted to give the following results:
Rd = R'
q
= R' = 4aK'p(lA + 2a) ohms (10-47)
G~~ =
q ngo
3
nMor ;. Jl.oa l1fK'Kr (1 _~)
2go
henrys 00-49)
In Eqs. (10-51) and (10-52) the inductance terms 11 and L. are defined by
8
L ' -_ Jl.oa 11r(Kr)2 henrys (10-53)
n2yo
Jl.oa 3l1f(Kr) 2g1
L. = 2 n 2go 2 henrys (10-34)
(10-56)
(10-57)
a'
as
(10-60)
I-CO-:~O 'in~
n -a-- ""]
-sin na cos na
(10-61)
and i~~ is defined by (l0-57). Expand Eq. (10-59) and convince yourself
that this equation is correct. The matrix given by Eq. (10-61) is referred
to as the complete d-q to a-b transformation matrix. Sometimes it is also
called the stationary to rotating transformation matrix. Notice that the
matrix a~.dq given by (10-33) is only a part of the complete transformation
matrix.
Premultiplying Eq. (10-55) by the d-q to a-b transformation matrix
yields
a·'
aI>.dqv·' - a·'
dt -
7..'j"
'f'.dq-drdv (10-62)
where a~.dV is defined by expression (10-61). The d-q axis currents
can be expressed in terms of equivalent a-b axis currents by using
(10-63)
Substituting these a-b currents into Eq. (10-62) gives us
(10-64)
For convenience the following two definitions are made:
I The order of the SUbscripts indicates that the d-q currents are being transformed
into the equivalent a-b currents. The double superscript means that botA the stator
and rotor currents are being transformed.
444 Principles 0' e/ecfromecMmicQI.energy conversion
(10-67)
and has precisely the same form as Eq. (10-515) in terms of the d-q coordi
nates. The transformed voltage matrix v''; defined by Eq. (10-65) can
be considered as a set of coil voltages in the a-b reference frame equivalent
to the original coil voltages in the d-q coordinate frame; therefore, we have
~'"" = :iv:]
Vb
(10-68)
I
If the d-q voltages are given, then, by using (10-65), all four of the a-b
voltages can be found.
The transformed impedance matrix Z~ gives the relationship between
the coil currents and voltages in this new a-b frame. After a great
amount of manipulation Eq. (10-66), usingZ~~ defined by Eq. (10-58)
and a~.dq defined by Eq. (10-61), can be expanded to the following form: 1
In obtaining Eq. (10-69) the assumption that 0: = w' has been made.
Thus the rotor of the a-b machine has an angular velocity equal to that
of the rotor of the d-q machine.
The various terms in the impedance matrix for the a-b primitive
machine can easily be identified. For the two a-b axis stator windings
(10-70)
as we would expect, since the stator windings on both machines' are
identical. Between adjacent stator and rotor windings of the a-b machine
we have the following mutual inductances:
M:;'(a) = M"; cos na M~(a) = - M"; sin na
(10-71)
Mb;;(a) M:r cos na M~(a) = M"; sin na
Notice in each case that the mutual inductances are functions of
the angle a. The a-b machine has its parameters as functions of the
rotor coordinate, while the d-q parameters are always independent of the
I Remember that or is a function of time and therefore, for example, the first..row
third-column term M';p cos nor is equivalent to M'; cos nor p - M';na sin nor.
The a-b primitn,e machine 445
for the a-b machine can be written, then the equilibrium equations can
be obtained very systematically. Lagrange's equation applies to the
a-b machine because the slip rings represent a holonomic coordinate
constraint. As discussed in Sec. 2-4, holonomic constraints between a
set of coordinates allow us to obtain a set of generalized coordinates
sufficient to describe the system completely. The number of generalized
coordinates is equal to the number of degrees of freedom possessed by
the system. The slip ring represents a holonomic constraint because
t.he actual currents ill each of the rotor coils can be defined in terms of
t,he brush or port (·urrents. The brush eurrent. can then serve as a
generalized eoordinatc for the slip-ring winding. Once the brush current
is specified, each of the <'oil currents is then also specified.
T\le ('olllll1utator-brush switehing me(~hanism on the rotor of the
d-q primitive machine represents a nonholonomic constraint. If the port
current for a cOlHlllutator-brush rotor winding is specified, then eaeh
of t he individual rotor eoil currents is not determinable until the position
of the rotor is also specified. Thus the constraint equations for a com
mutator-brush winding give the individual coil eurrents as a nonanalytic i
function of the port or brush eurrent and t,he rotor position. The func- ~
tion is 1I0nanalytic because the coil currents jump from one value to
another as the coil is cOllllllutated. The rotor position is often an
unknown coordinate whieh we are interested in finding. Since the
constraint equations for the d-q ma('hine are nonholonomic, a set of
generalized coordinates cannot be obtained and Lagrange's equation
cannot be used for obtaining the d-q equilibrium equations. In Chaps.
6 and 7 the d-q equilibrium equations were obtained by resorting directly
to Faraday's law.
Heferring to Fig. 10-;,), a loop formulation must be used because we
have ll1utual-induetive coupling. The flux linkages, !IS functions of the
coil ('urrents, are as follows:
o M~:(a}
L: Mb:(a}
(10-75)
It;:(a) L:(a}
lIf;;(a} lIf:;;(a}
It''('''}
.. tab,a =
\' Jo
L" (i."(,,
"k f ,a
}d"fk (10-76)
i-\
where J is the polar moment of inertia of the rotor. Neglecting the sha.ft
complia.nce, the system energy function 'l) is zero. The system's La.gra.n
gian is therefore
(10-80)
448 Principles of electromechanical-energy conversion
for k = 1, . . . ,5 (10-83)
where E! = i:, ~2 = i~, ~3 i:, E4 i b, and h a. Using k = 1, 2, 3,
and 4 in Eq. (10-83) will give Eq. (10-67) as a result. The mechanical
equilibrium equation for the a-b machine is obtained by setting k = 5,
thus giving
~ (:~) Ja (10-85)
o.c
oa
+ 111-; sin na i:ib + L. sin 2na (i:)2 + 2L. cos 2na i~i;;
- L. sin 2na (iiYl (10-86)
05 = Da (10-87)
oa
Substituting Eqs. (10-85) to (10-87) with Q.. = Tr into Eq. (10-84) gives
us for the mechanical-equilibrium equation of the a-b machine
, Ja + Da + T. = Tr (10-88)
'. where the torque of electrical origin T. is given by Eq. (10-86).
If equivalent d-q and a-b currents are respectively applied to the two
equivalent primitive machines, then the same torque of electrical origin
450 Principle. of electromechanical-energy conver.ion
By use of the d-q to a-b transformation given by Eq. (10-63), the equiv
where 'U stands for the unit matrix, with l's on the principal diagonal and
O's in every other position. If a transformation matrix premultiplied by
the transpose of the transformation matrix equals the unit matrix, then the
The a-b primitive machine 449
should be generated in each case. Equation (7-24) expresses the d-q torque
of electrical origin for the case of identical quadrature- and direct-axis
rotor windings. For convenience (7-24) is repeated as
(10-89)
with the substitution L~ - 1..; = 21... from Eqs. (10-51) and (10-52).
The d-q currents equivalent to the a-b currents can be obtained from the
matrix (10-59) by premultiplying both sides by (1~~."" = «(1~.tl'l)-l. Upon
expanding this equation, we have
i~ = i; cos na - i;; sin na (10-90)
i; = i; sin na + i;; cos na
Substituting the equivalent currents of (10-90) into the d-q machine torque
of electrical origin gives us precisely Eq. (10-86), which is the a-b machine
torque of electrical origin. Thus, as we would expect, both primitive
machines generate the same torques for equivalent currents. In Sec. 7-3, we
obtained T. for the d-q machine by using the first law of thermodynamics.
Here we have obtained T. for the a-b machine by using Lagrange's equa
tion. The equivalence of these different formulations is certainly evident.
~. ~. 1]
R'
iR = 0 (10-92)
[ o
o
The a·b primitive machine 451
[1oo 0
-L.]
0 0 0
1
0
0
cos na
oo ]
sin na
X [ 00 1
0
0
cosna
o 0 -sin na cosna 0 0 sin na cosna
[1o 00 0]
1 0 0
(10-103)
o0 1 0
000 1
Since Eq. (10-101) is satisfied for the a-b to d-q transformation matrix
a 4 ."", the total electrical powers supplied to each machine are equal; thus
we can write
(10-104)
Using the first law of thermodynamics, these total powers can only con
tribut.e to energy lost or stored. Therefore at each instant in time
(10-105)
Equation (10-97) shows the power-loss terms to be equal, and Eq. (10-104)
equates t.he total powers. Therefore, the portions of the total power
supplied to each machine adding to t.he stored energy are also equal.
Thus we have shown that the total power is invariant under the a-b to d-q
transformation. Also, each component of the power is invariant.
The commutator of the d-q machine has been replaced by slip rings
to obtain the a-b ~)achine. As a compensation for this constructional
change, transformed currents are supplied to the a-b rotor windings in such
a manner that all current distributions remain the same. Since the
windings are the same and the currents are the same between the two
machines, it is not too surprising that the powers should be invariant.
K! = K~ = K:O (10-107)
I The symbol a is al80 used to designate the angular position of the a-b machine rotor.
The three symhols a, {3, and "I used as phll.8e-winding designations will usually appear
8.8 subscripts, 80 no confusion should result.
K'b
J•')<- . (8
- K'')<sm 211') 1,''.)<a.
n -3 amp/m
For three identical windings the distribution factors are all equal; there
fore we may write
(10-109)
The total surfa('c-current. distrihution duc to the three-phase stator
willdings ean be found by simply adding the individual distributions.
Thus we have as the total three-phase stator distribution
J~#')< = J~ + J~ + J~ (10-110)
By subst.ituting Eqs, (10-108), the t.otal distribution J~#')< can be put into
the form
,
J all')< = K'a(h [(.t~ + cos 3211' 1,.& + cos 211'
3 1,;.), sm n 8
- ( - sm 3 t~ + Sill
,211'. . 3211'.)
t; cos n 8] a. (10-111)
where we have used the trigonometric identities for the sine and cosine for
the sum of two angles.
In a similar fashion the total surface-current distribution established
by the a and b axis windings is given by
J~ = J~ + Jt (10-112)
454 Principle. 01 electromechanical-energy conversion
(10-118)
and
a'""'.,.{joy -- i
[
1
0 -']
~~ (10-119)
I This transformation will he used only for balanced three-phase excitations and ill
therefore sufficient for our needs. For Ii more complete discU88ion see A. E. Fitz
gerald and C. Kingsley, "Electric Machinery," 2d cd., pp. 229-231, McGraw-Hill
Book Company, New York, 1961.
The o-b primilive mochine 455
.
V_L = V!] (10-121)
- Vb
(10-128)
(10-129)
Notice that the two-phase voltages are also balanced, but they are only
90° out of phase. Since balanced two-phase voltages are applied to the
two identical a-b windings, the average power supplied to each phase is
I Equations (10-127) can easily be checked by first rorming the sinor equivalents of
the three sinusoidal voltages and then adding the appropriate parts or the complex
numbers either analytically or graphically.
The a·b primitive machine 457
(10-131)
Taking the average value of both sides of Eq. (10-128) and dividing
by 2 gives us
(10-132)
Equation (10-1:J3) tells us that the average power supplied to each phase
of the two-phase system is equal to the average power supplied to each
phase of the three-phase system.
By using the transformation given by Eq. (10-119) and the distribu
tion factors given by (10-114), a set of two-phase windings can be con
structed and excited in such a fashion that current sheets whieh are
equivalent to but not identical with those established by a balanced three
phase system are established. Equations (10-130) show us that v~ = v~,
and therefore we ('an easily show that i~ i!. Thus the a phase of the
two-phase system has port variahles ident.ietll with those of the a phase
of the thl'ee-phase system. A trallsforllllttioll b:wk to three-phase quan
t.ities from the t.wo-phllSe quantities is re!tdily aceolllplished, since the
/J- and 'V-phase quulltit.ies arc simply displaced by -2'11'/3 and 2'11'/3 rad,
respect.i vely.
All power quantit.ies eal(~uhtted in terms of the two-phase system can
be written on a per-phase basis. For the three-phase system a multiplica
tion by 3 yields the corresponding total t.hree-phase power quantity. By
using the results of this section, all symmetrical three-phase a-e machines
can be reduced to equivalent. t.wo-phase machines and thus fit the a-b
primitive-machine model.
By comparing Eq. (10-128) with Eq. (10-124), we can see that for
the special ease of balanced three-phase excitations the matrix product
1 0 0]
(a:u,.a~.,)'a:b.a~., = i
[00 01 10 = i'U (10-134)
(10-135)
(10-136)
By expanding (10-136) with the use of (10-119), we can again see that the
a-phase quantities are identieal with the a-phase quantities. A similar
expression can be written in terms of the eurrents; thus
(10-137)
For the three-phase stator windings shown in Fig. 10-7 assume that
we can write a set of equilibrium equations in the form
(10-138)
By using Eq. (10-137) for i!~., and premultiplying by the ab-aI'J'Y matrix,
we have
(10-139)
(10-140)
any leakage effects, the equilibrium equations for the three windings are
given by Eq. (10-138), where
L' L'
R' + Vp -2 P -2 P
L' p L'
Z!/I.., = R' + Vp -2 P (10-141)
V V
-2 P -2 P R' + Vp
The mutual inductance between any two of the windings is equal to
La cos (211'/3), or -V/2 wit.h no leakage fields.
By using Eq. (10-140), the equivalent a-b impedance matrix can be
found with the ab-a{fJ'Y transformation matrix defined by (10-119). After
a good deal of manipUlation, we obtain
· - [R' +0 lLap
Zoo - R' +0 iL'p J (10-142)
(10-143)
where
(10-144)
460 Principles of electromechanical-energy conversion
yS
)L----j--.pC¥::q--L-_ _ (x/
/3'
(a)
bS
10-9 SUMMARY
The study of a-c machines requires a model primitive machine having
rotor windings supplied currents through slip rings rather than a com
mutator-and-brush meehanism. The a-b model machine provides the
desired new primitive ma(~hine. The iron struct,ure of the a-b machine is
identical with that of the original d-q primitive machine. The two stator
windings of each machine are also identical with those of the other.
Since most a-c machines are c01lstrll(~ted with identieal windings on each
stator axis, the winding distribution faetors of the two stator windings on
the a-b machille are taken to be identical with each other, and they are
also identical with the two stator windings on the equivalent d-q machine.
In addition, the two a-b rotor windings are identical with the two d-q
rotor windings. However, t.he rotor and stator windings are, in general,
different.
The only constructional differe1lce between the two machines is the
mechanism by which current is supplied to the identical rotor windirigs.
The a-b machine uses slip rings, whereas the d-q primitive machine
employs the commutator-brush mechanism. In Chap. 6 we found the
commutator-and-brush switching on the d-q machine maintained the
rotor-surface currents stationary in space irrespective of the angular posi
tion or veloeity of the rotor. No switching takes place with the slip rings
on the a-b machine, and therefore the rotor-surface currents move with
the rotor.
By using a matrix transformation, a-b and d-q currents which main
tain identical rotor and stator surface-current distributions on each of the
two primitive machines can be found. The form of the transformation
giving the a-b currents in terms of the d-q currents is
<l:.b.dQ = [~o !
0
11 La 'in no]
-sin na
g
eos na
(10-145)
The subseript notation indicates that the transformation gives the first
two subscripts from the second two subscripts, or that we obtain the a-b
462 Principles of e/ectromechanical-etlftrgy conversion
currents from the d-q currents. The two superscripts indicate both the
stator and rotor currents to be involved in the transformation. The
inverse of matrix (10-145) is written as ad~ ...I" giving the d-q currents in
terms of the a-b currents.
The electrical-equilibrium equa.tions for the a-b machine are obtained
by transforming the original equilibrium equations for the d-q machine.
The a-b parameters can be identified in terms of the d-q parameters. In
Chap. 6 all the d-q parameters were determined as functions of the
primitive-machine geometry. By this identification process, the a-b
parameters are thus related to machine geometry.
For equivalent currents the torques of electrical origin for the two
machines arc found to be identical. Also because of the parameter rela
tionship~, the components of power (or the two machines are the same.
The (ollbwing list of powers can be equated:
Total power supplied p~~~al = p:~al
Total electrical power supplied Pd~ = P:;'
Total ohmic loss p~~.. = P~~"
Total electrical storage Pd~o'ed = p~ored
Total mechanical power supplied P:" = w'r'
Total viscous or windage loss P!:'" = w'Dw'
Total mechanical storage p:ored = wrJpw'
All rotor-port quantities of one primitive machine are identical with those
of the other. The power that each machine converts from electrical to
mechanical or mecha.nical to electrical is also identical with that converted
by the other.
::\lost practical polyphase a-c machines have three-phase stator or
rotor windings rather than the two-phase windings on the a-b primitive
machine model. An equivalent set of two-phase windings with the same
number of poles can be constructed with the same distribution factors as
are used for the three-phase windings. By using the transformation
~l (10-146)
at the same time. In writing the impedance matrix for either the a-b or
d-q machine a factor of j must be inserted before each inductance param
eter involving a term containing a current related through transformations
to the three-phase currents.
PROBLEMS
omp/m
10 ,-
00 Trln 2Trln 9
-10
(a)
amp/m
20
00 9
-20
(0)
amp/m
15.0
7.5
J = 10 kg-m' U = 3h
a. Are the stator and rotor structures of this primitive maehine salient or non
salient? Explain your answer.
b. Write the five equilibrium equations for the machine, including an expression
for the torque of electrical origin.
c. Formulate an equivalent d-q primitive machine. List al1 d-q parameter
values.
d- 10-7 A 2-amp direct-current source is connected to the a-axis stator winding
of the primitive machine in Prob. 10-6. The speed of the rotor is maintained con
stant at 100 rad/sec in the positive angular direction. All other ports are
open-()ircuited.
a. Calculate the steady-state voltage across the current source.
b. Find the open-()ircuit voltage on each of the other three electrical ports.
c. Find the corresponding voltages on the four electrical ports of the equivalent
d-q primitive machine.
d. How much' external torque is required to maintain the speed at 100 rad/seef
10-8 The rotor or the primitive machine in Prob. 10-6 is held stationary at an
angle a 10°. The following constraints are plaeed on the four electrical portli:
Vb = V2 50 sin 377t volts II: "" lib = 0 volts
a. Formulate the three electrical-equilibrium equations in complex or sinor form
necessary to yield the steady-state port currents.
The a-b primitive machine 465
b. Solve the equations of part a for the three steady-state port currents. Express
your answers as functions of time.
c. Find the set of port constraints for the equivalent d-q primitive machine.
d. Formulate the equilibrium equations for the equivalent d-q machine and solve
for all port currents.
e. Transform the d-q currents, found in part b, to equivalent a-b currents. Com~
pare your results with the currents found in part b. .
,. Find the magnitude and direction of the external torque required to maintain
the rotor at a fixed angular position.
~,10-9 The rotor of the primitive machine in Prob. 10-6 ill now being driven in
the positive angular direction at a constant speed of 377/3 rad/sec. The two stator
windings are open~ircuited, and the two rotor windings have the following constraints:
Assume that ",(0) = 0, where ",(t) is the angle between the rotor and stator axes.
a. Derive an expression for <>(/).
b. Find the open-circuit port voltages for the two stator windings.
c. List the set of port constraintl! for the equivalent d-q primitive machine.
Explain these constraints qualitatively.
d. Solve for the open-circuit port voltages for the two stator windings on the
d-q primitive machine. Compare thCS(l rC8ultll with the results in part b.
e. Hepeat parts b to d for v~ = - V2 50 cos 377/, with all other conditions as
previously given.
10-10 A reluctance synchronous motor can he constructed as shown in Fig.
PlO-lO. The single rotor winding is supplied through slip rings. The stator is salient
with two pole projections as shown, and it has no electrical windings. The rotor
has an inertia J and a windage coefficient D. The resistance of the winding is Rr,
and the self-inductance ('an be approximatffi hy Lr L. cos 2a. No external torque
is applied to the rotor shaft, and a voltage source v(/) is ('onnected through the slip
rings to the rotor winding.
a. Draw a primitive-machine model of the machine in the form of Fig. 10-5.
b. Formulate a Lagrangian, a Rayldgh dissipation function, and the external
forces for the machine.
c. By using Lagrange's equation, formulate the two equilibrium equations for
the machine.
Fig. PIO-IO
.c66 Principles of electromechanical.energy conversion
d. If v(t) = v'2 V sin wt, determine the speed at which the rotor must revolve
so that the average torque of electrical origin is nonzero. (Hint: IC v is sinusoidal,
then in the steady state the rotor current must also be sinusoidal with the same
frequency.)
e. If the stator structure has 2n poles (where n is the number of pole pairs)
reformulate an approximation for the winding inductance and repeat parts b to d.
Note: Common electric clock motors are reluctance synchronous machines. Usually,
the single winding is placed on the stator and the rotor structure is made to have
salient poles. Since our primitive-machine model can only have a sa.lient stator with
a nonsalient rotor, the model of the actual machine must be inverted. A similar
procedure is followed in Chap. 11 in discussing synchronous machines with d-c fields.
10-11 a. Prove the matrix identity (10-95) by expanding both sides of the
equation.
b. Verify that Eq. (10-96) is valid and thus prove that the total ohmic losses in
the two ertfivalent primitive machines are identical.
10-12 o. For the following set of three-phase currents circulating in three
identical stator windings, arranged as shown in Fig. 10-7, find an equivalent set of
a..o stator currents which result in two-thirds the current sheet established by the
three-phase curren ts.
i~ = v'2 120 sin (wi + 30) i~ = v'2 120 sin (wi - 90)
i~ v'2 120 sin (wt + 150)
Hint: Sinor notation is very useful in performing this transformation.
b. What is the relationship between the magnetic fields established by the two
phl\S(> windings and those established by the three-phase windings?
10-13 Show that the transformation given by (10-140), when applied to (10-141),
yields Eq. (10-142).
10-14 Each of the three identical stator phases shown in Fig. 10-7 has a resist
ance of 3 ohms and a self-inductance of 2 h. Sinusoidal three-phase currents at a
frequency of 2 rad/sec and with an rms magnitude of 5 amp are established in these
stator windings.
a. By using sinor notation, determine the port voltages on each of the three
phases. Take i~ as reference; thus, i~ v'25 sin 2t.
b. Find the avcrage power supplied per phase.
c. Determine the equivalent set of two-phase stator currents and use (10-142)
to find the appropriate two-phase port voltages. Compare the a-phase and a-phase
quantities.
d. Find the average power per phase being supplied to the two-phase windings.
Compare this resulting per-phase power with that obtained in part b.
coMpte.
of a sinusoidally distrihuted field are revolving in the air gap of the devi-ce.
The rotor of the machine consists of a winding to which a d-c source is
connected, thus establishing a constant field depicted in Fig. 11-1 by the
north (N) and south (8) on the rotor structure.
The rotor north pole is attracted to the stator south pole, and in a
similar fashion the I'Otor sout.h pole is aUraded t.o the stator north pole.
This magnetic bond between the stator and rotor fields causes the rotor
to revolve in synchronism with the stator field. The stator field is
revolving at a ("onstant speed and thus the rotor, being magnetically
locked to this field, revolves at the same constant speed. The position
of the rotor with respect to the stator field is given by the angle 0 in Fig.
11-1. If the machine is operating at a constant speed, the angle 0 remains
constant. If the mechanical load on the shaft of the machine is increased,
the rotor will tend to lag farther behind the stator field and thus increase
the angle o. A simple, but useful, model of the magnetic bond is obtained
by considering the bonding or locking force between the stator and rotor
fields to be provided by rubber bands between the two fields. Loading of
the rotor causes an increase in the angle 0 that stretches these rubber
bands, thus providing an increased torque output to drive the increased
load. Extreme loading of the rotor shaft will finally cause these rubber
bands to break and thus stall the machine. With this simple moGel, much
of the steady-state machine performance can be predicted.
If mechanical power is put into the mechanical port of the synchro
nous machine in Fig. 11-1 and the direct current is maintained in the
rotor structure, an alternating voltage will be induced in each of the
470 Principles of electromechanical-energy conversion
st.ator coils. The relative phasing of these induced stator voltages will
be different in each coil owing to their physical displaeement on the stator
structure. With the stator windings mechanically separated hy 211"/3 rad, .
the respective induced stator voltages will have a phase relationship of
this same magnitude if the rotor has the single north and south poles
shown in Fig. 11 -1. If electrical loads are connected to the stator
windings, electric power may be extracted from the electrical ports of the
machine.
revolving eountewlockwise at. a speed w' with respect to the stator field
winding. A synchronous machine is not usually constructed as shown
in Fig. 11-4. However, t he machine could be built in this fashion with
the two a-b rotor windings supplied power through slip rings. The chief
drawback tot he inverted arrangement lies in the relative magnitudes of
the armature power and the d-c field winding power. The field winding
operates at a much lower power level and thus is more easily supplied
through the slip-ring mechanism.
The analysis of the a-b primitive-machine model shown in Fig. 11-4
can be readily formulated by first developing an equivalent d-q primitive
machine. Equations (10-70) to (10-74) summarize the relationships
oetween t he parameters of the two machines. The electrical-port vari
ables are related by the d-q to a-b transformation given by Eq. (10-145).
,
The mechanical-port variables for the two primitive machines are
identicaL J.<'igure 11-5 shows the d-q model of the inverted two-phase
synchronous machine.
Let us briefly review how we have modeled the three-phase syn
chronous lIlachine shown in Fig. 11-1 and schematically in Fig. 11-2.
The three-phase set of stator windings is first converted to an equivalent
set of two-phase stator windings. Because the field is often constructed
on a salient strudure, our view of the machine is inverted with the two
phase armature placed on the rotor and the d-c field put on the stator.
We now have a ('oflventional a-b model which is transformed to an
equivalent d-q model machine. Equations for the d-q primitive-machine
model ('an be readily written and solutions can be obtained for whatever
constraints are placed upon the ports of the primitive machine. Con
straints pla{'cd upon the ports of the actual synchronous machine must
first be transformed to correspond to equivalent constraints on the d-q
primitive machine. When the primitive-machine response has been
The synchronous machine 473
v~l
v'd
Hf~~p
R'd + iLdP (11-1)
v'Q -~G;dw'
.-
T - - P.
w'
(11-3)
....
where V is used as the rms value of the applied phase voltages. The
three-phase to equivalent two-phase transformation matrix is given by
(10-119) as being
•
(!ab.G<lh -
_ 2
1r
[
1
0
f]
By applying this transformation to the balanced three-phase bus voltages
(11-6)
given by
v: = y'2 V cos wt v: = -y'2 V sin wt (11-7)
Since the stator and rotor of the machine are inverted, the two-phase
armature voltage constraints given by (11-7) are the rotor-port voltages
in Fig. 11-4. Therefore, the rotor-port variables are specified as
v: = y'2 VII cos wt vi; = -y'2 VI> sin wt (11-8)
where VII and Vb are the rms or effective values of the respective rotor
voltages. The corresponding constraints for the d-q primitive machine
can be found quite simply by using the linear transformation introduced
in Sec. 10-3. Thus, the equivalent d-q rotor voltages are given by the
matrix equation
(11-9)
v:; = y'2 [Vo -; Vb cos (wt _ ncr) + V.. ; Vo cos (wt + ncr)]
V~ .. [V.. +
= - v_'2 2 Vb sm
. (wt - ncr) - VII -2 Vb'
sm (wt + ncr)]
(11-12)
With balanced phase voltages we have
V.. = Vb = V (11-13)
and the d-q primitive-machine equivalent rotor-port voltages are given by
v~ = y'2 V cos (wt - ncr) v~ = -0 V sin (wt - ncr) (11-14)
The third constraint on the synchronous machine is the specification
of the field winding excitation. In the usual operation of this type of
J,76 Principles of electromechanical-energy conversion
(11-15)
If the field winding current remains constant, then whether a d-c current
or voltage source is chosen becomes equivalent. However, a time-depend
ent voltage induced in the field circuit would give an additional field
current component if a voltage source were used on the field winding.
A fourth constraint is available by limiting our considerations to
the steady-state operation of the machine. Under steady-state condi
tions, tre rotor angular velocity remains constant and is equal to the
angular velocity of the rotating magnetic field. Equations (11-7) define
the original stator voltages to be sinusoidal with an angular frequency
of w rad/sec. Therefore, in the sinusoidal steady state, the original
stator currents must also be of this same frequency. In Sec. 10-2 ba.l
anced stationary currents of frequency ware shown to produce a magnetic
field rotating in space at an angular velocity equal to win, where n is
the number of pole pairs established by the winding. The angular speed
of the rotating magnetic field win is known as the synchronous speed.
Notice that the synchronous speed is dependent only on the excitation
frequency and the number of pole pairs established by the windings. In
the steady state the rotor must rotate in synchronism with this revolving
field; thus the rotor of the actual machine must rotate at an angular
velocity of win rad/sec. This constraint can be expressed by
w' = ~ (11-16)
n
Example J 1- J
A IO-hp 230-volt four-pole 6O-cps three-phase wye-connected synchronou8 machine has
the line-to-line voltages shown in Fig. 11-00 applied to the armature windings. In
The .ynchrol'lOll' machine 477
V'Z
. .
V23
V31
.
3
(al
V31
where w = 2,..1 is the excitation frequency and n = t is the number of pole pairs
established hy the armature windings. For this example we have
?,77
(jJ"=- 188.5 rad/sec = lROO rpm
2
Figure 11-60 shows a sinor diagram of the three-phase line-to-line and line-to-neutral
voltages. The diagram can be checked by noting, for example, that
VI< - Vu - Vu + V., - Va
The corrrponding time forms of the three-phase stator voltages are
The expansion of the matrix transformation can be performed most easily if all time
quantities are expressed in sinor form. Thus we have
V' ,.
•
i X 133 (II-----
- ~ -!I-----
6
- ~ - ~/~)
2 6 2 2
---
V'~T
- III X 133 ( v'3 I s... + -v'3 I"')
- -2- - -6 2 2- I'" = 133 3
--- ---
The corresponding time forms for the equivalent two-phase voltages are
If the speed of the machine is equal to synchronous speed, then a - .lI.¥ and thus
a(t) .. .lp t + a(O)
The expressions for v:i and v; are then d-c quantities, since they are independent of
the time t.
a Wt_...!.+~ (II-IS)
n 2n n
since the synchronous speed w' = win according to Eq. (11-16). The
constant angle (l/n)(o - 71'/2) defines the value of a at time t O.
The physieal significance of the phase angle a is considered in Sec. l1-S,
Substitution of Eq. (11-18) for a into Eqs, (11-14) leaves the direct
and quadrature-rotor port voltages as
(11-20)
(11-21)
(11-22)
(11-23)
The quantities X 9 and X d , defined by Eqs. (11-21) and (11-22), are com
monly given the names quadrature and direct-axis synchronous reactance,
respeetivcly. For identical rotor windings on the primitive-machine
model, Eqs. 00-.')1) and (l0-.''j2) give the rotational inductances involved
in the synchronous reactances as being
where Lr and L. are defined by (10-53) and (10-54). When Eqs. (11-24)
are substituted into Eqs. (11-21) and (11-22), we have
Xd = IUl(l] + L.) (11~25)
For a nonsalient-pole machine the air gap between the stator and
rotor is uniform and L. = O. If this is the case, then the quadrature
and direct-axis synchronous reactances are equal and
XII. = X. nonsalient machine (11-26)
and X. is simply called the synchronous reactance of the machine.
In terms of the three defined quantities given by Eqs. (11-21) to
(11-23), Eqs. (11-20) become
(11-27)
The voltage terms involving R~ and R; al'e usually much smaller than
the other terms in these equations. Thus, without introducing a sig
nificant error in a practical calculation, these terms can be neglected.
Eliminating the ohmic terms, Eqs. (11-27) are now simply
(11-28)
. = -------;X"""d---
~dO
V2 V cos 6 - E I (11-29)
(11-30)
482 Principle. of electromechonieol-energy conver.ion
Motor ---1
Generator
operation operation I
TotoI torque curve
nVEf .
V2wX Sl~ 8
d
Fig. 11·7 Synchronous torque of electrical origin per phase for a salient·pole
machine aa a function of the torque angle.
By substituting Eqs. (11-29) for the two currents ~~o and ~~o and dividing
by 2, this torque expression can be manipulated into the following form:
n [VEl'
T .. = w V2 Xd Sill "
V2
+ 2XdX (X
d -
X)'
q Sill
2]
"
(11 31)
q
The model machine used for this analysis has two phases, so dividing
the total torque, given by (11-30), by 2 makes Eq. (11-31) the synchro
nous torque of electrical origin per phase. Equation (11-31) multiplied
by 3 would be the total generated torque for a three-phase machine.
Figure 11-7 shows a typical plot of the per-phase torque of electrical
origin T.. as a function of the angle". From Eqs. (11-25) the quantity
Xd - X q is equal to 3wL, and is positive. This assumption is made in
the plot of Fig. 11-7. For a nonsalient-pole machine, Eq. (11-31)
simplifies to
Example J J·2
The 10-hp 230-volt four-pole 6O-cps three-phase wye-connected 8ynchronou8 motor
in Example 11-1 hll.8 its armature excited at rated voltage, 11.8 discuesed in
Example 11·1. The d-c field current is I!O adju8ted that E, .. 192 volts. The
windage coefficient at 8ynchronou8 speed i8 D .. 0.01 newton-m-sec/rad. Assume
the armature resistance to be negligible. The machine hll.8 II. nonsalient structure
with II. synchronous reactance X ... 1.5 ohms, and it is delivering rated power to a
constant load torque.
a. Determine the total power converted from electrieal to mechanical form.
All!O find the total electric power input.
b. Find the synchronous torque of electrical origin per phll.8e.
c. Determine the torque angle &.
d. Calculate the'magnitude of the line currents and the power factor of the
machine.
8olutmn: a. The total converted power is the output power plus the windage
losses; thus we have
p. == '(10 hp X 746) + D(",')'
p. ""-1460 + 0.01 X (.lP)t "" 1816 watts
Since there are no electrical losses, the total electric power input is also 7816 watts.
b. The total synchronous torque of electrical origin is the converted power
divided by the speed; therefore
P. 18\6
T • .. -;;. == - 377/2 == -41.5 newton-m
Since the machine hll.8 three phll.8C8, the synchronous torque per phll.8e is simply
The torque of electrical origin can also be expreesed in units of synchrOfllJU4 watts.
These units 8imply use the corresponding power quantities 11.8 torque, without dividing
484 Principles of electromechanical-energy conversion
hy the synrhronous speed. Thull the total gen!'rat!'d torque would he 7H16 syn
chronous watts, and the torque of I'le('tri('al origin per phase is 2605 syn('hronous watts.
c. The torque angle 0 for the nonsalil'nt machine can he det..rmin..rl from Eq.
(11-32). Hepeating (1l-:~2) for convenience, we hav..
nt'E, .
r.d> ----sm 0
V2wX.
as the g!'nerated torque per phase for a nonsalient machine. The voltage quantity V
is the rms phase voltage applied to the armature of the two-phas(, model; however,
the three-phase to two-phase transformation maintains the same phase voltage.
Thus V 2aO/ y:l and we have
X a77 X 1.5 X (
2 X l:J:-l X 192
= -0.216
o -12.5 0
I
as the torque angle o. Since the maehine is operating as a motor, 0 should he a
negative angle according to our sign convention.
fl. The magnitude nf the line ('urrents can he determined hy first calculating the
transformed d..fJ rotor currents. From Eqs. (l1-2!1) we have
V sin 0
X.
for a nonsalient machine where X d X. = X.. Huhstituting numerical values
gives us
X l:.l:J cos (-12.5) 192
-5.:.13
1.5
= -27.2
From ScI'. to-a the d-q to a-b transformation for a four-pole mac,hine (i.e., two pole
pairs) gives UK for th(, a phase
1/:]
"r [
('o~ na
- Bill na
sin na] ll~]
cos na I';
(11-34)
b
The subscript zero indicates constant or d-c values which result from
the spccial constraints being C'onsidered.
By substitution of the constant values for the direct- and quadrature
rotor voltages, t.he a-b rotor voltages of Eqs. (11-3;») can be expressed as
V'
"
(11-40) [
Multiplying the first of Eqs. (11-40) by the operator j gives us
·V) _ VdO + . v;o (11-41)
J"--y'2 Jy'2
(11-43)
as a complex equation for the sinor value of the a-phase rotor-port volt
age. The original constraint on v:
is given by the first of Eqs. (11-8) "
and for convenience is repeated as
v: = y'2 V cos wt (11-44)
,
being considered is defined by Eq. (11-18), and upon slight rearrangement
The synchronous machine 481
Fig. 11-8 Complete sinor diagram for the a phase of the two-phase salient-pole
eynchronoUil machine.
it can be written as
(11-48)
As with the voltage quant.ities, t.he b-phase port current is ident.ical with
the a-phase current upon rotating :t by r/2 rad. The a-phase current
is also plotted in Fig. 11-8 according to the first of Eqs. (11-48). With
488 Principles of electromechanical-energy conversion
E, ~ 8
V2 Projection of ~
I onto V
Fig. 11-9 Sinor diagram for the IOB8less salient-pole synchronous machine.
V cos • _
(J -
E,
y2
+ XdidO
y2 (11-49)
. y2 V cos ~ - E, . y2 V sin ~
(11-50)
tdo =
Xd t;o = Xq
The per-phase electrical power into each of the two stator phases
is given directly by
where PH_ indicates electrical power input per phase. Now the term
I cos e represents the projection of I onto V taken positive if this projec
tion is in the same direction as V, meaning e is an angle in the first or
fourth quadrants. In terms of its two components the projection of I
The rynchronOUfi machine 489
upon V is given by
I cos 8 = .!::2 (~
Xd
- ~)
X"
sin 28 - ---!L-
y2X
sin 0 (11-53)
d
after the trigonometric identit.y for the sin 20 is used. On substitution
of Eq. (11-53) into (ll-fil), the electric power intput per phase becomes
P H. = - V2(I - - - 1 )sin
. 2'
(I - ----
VE,.
sm 0 (11-54)
2 Xd X" y2Xd
With two phases on the actual machine the total electric power input
would be simply t.wice this amount, or
P E• = p •• (11-56)
T.. = - p ..
r
(11-57)
iN
and thus
with the machine speed w' held fixed at win, the excitation frequency
divided by the number of pole pairs. Equation (11-.')8) is exactly
equivalent to the previous expression (11-31) developed with the d-q
primitive machine.
490 Principles of e/ecfromechanical-energy conversion
Example 11-3
A l3,200-volt 3000-kva eight-pole 6O-<:ps three-phase wye-<:onnected salient-pole alter
nator is delivering rated kilovolt-amperes at 0.8 power factor leading and at rated voltage
to a balanced power system. The direct-axis synchronous reactance is X d - 66 ohms,
and the quadrature-axis synchronous reactance is X. - 50 ohms. Windage losses are
90 kw at synchronous speed. Armature resistance can be neglected.
o. Determine the magnitude of the three-phase line currents.
b. Find the total torque required to drive the alternator.
c. Find the torque angle 0 and the voltage E I induced by the field in the armature.
d. Draw a sinor diagram for these operating conditions.
SoluliDn: a. With the machine delivering rated kilovolt-amperes to a balanced
three-phase bus at rated voltage, we can write
= 2490 kw
The total torque required to drive the alternator is equal to the total mechanical
power divided by the synchronous speed of the machine. Thus we have
'2.49 X 10'
T' - 377/4 - 26,400 newton-m
The machine has eight poles or four pole pairs; therefore, synchronous speed is ",/4.
c. By using Fig. 11-9 as a guide, the following two current equations can be
written:
"r
.!!!. = I cos (180 - e- &) -'40 = I .
SIn (180 - IJ - 6)
viz viz
The current I is the magnitude of the two-phase armature currents, which are equal
to the three-phase line currents. The power-factor angle /} is equal to 1800 - 37",
since cos e must be 0.8. The angle e is less than 1800 because the power factor is
leading, which means that -1 must lead Vor, equivalently, -V must lag 1. t Substi
tuting IJ = 143" into the two current equations leads to
Also from Fig. 11-9, with R; = Rd = 0, we can write the following voltage
expressions:
X.l~.
- -= V'
sma
V2
corresponding to Eqll. (11-49). Substituting the exprell8ion for i;Ol V2 in the first
voltage equations gives us
with V = 13,2001 v'3 as the phase voltage. By solving for the tan /; co (sin 8)1
(cos 8), we find
a .. 54.7·
By 8ub8tituting /; = 54.7° into the current equation for idol v"2 and using the
second voltage equation, we can obtain
E, . 13,200
V2 + 66(78.6 cos 54.7 - 104.8 sm 54.7) = V2 cos 54.7
which gives us
E,
V2 = 7050 volta E, = 9970 volta
d. The sinor diagram for the alternator is shown in Fig. 11-10. The current
idol v"2 is equal to -40 amp and therefore must be drawn downward in Fig. 11-10
opposite to the &ll8umed positive direction shown in Fig. 11-9. The sinor labeled
X .ti~oI V2 is also reversed in Fig. 11-10. Remember that the sinor diagram is valid
for the phases or both the two-phase machine and the actual three-phase machine.
jXq 6220
~ =7050
Fig. 11-11 Sinor diagram for the 108111ess nonsalient-pole synchronoUll machine
rotated through an angle 8.
The synchronou. machimt 493
If the machine has two stator windings, or two phases, the total electric
power input is simply
Pr. = 21
)K,. V2 VE, Sill
= - -~X---:--
.
0 (11-60)
Since ohmic losses are being neglected, the expression for the total electric
input power per phase, given by (11-59), is equal to the total converted
power per phase P... Thus, the torque of electrical origin per phase
is readily derived as being
p.. nVE,.
T .. = - - = - - - sm 0 (11-61)
w' V2 wX.
With the machine revolving at synchronous speed, making w' = win,
the torque of electrical origin is related to the electric power input by
the factor -n/w. In many instances, the electrical t,orque is given in
units of synchronous watts, where the quantity p. is used as the torque,
with the understanding that a division by the synchronous speed yields
the magnitude of the actual electrical torque.
The ,constant-power constraint can readily be imposed by examining
the right-hand sitle of Eq. (11-.1)9). Since the machine is operating at a
constant terminal voltage, V is a fixed quantity. Similarly, X., the
synchronous reactance, is a parameter of the particular machine being
used, and it also is a fixed factor. Therefore, for constant power we
can write
E, .
V2 sm 0 = const (11-62)
where E" defined by Eq. (11-23), is directly proportional t.o the direct
field current and 05 is the angle by which the sinor EdV2leads the sinor V
in Fig. 11-11. The constant-power constraint, given by expression
(11-62), can be illustrated geometrically, as shown in Fig. 11-12. The
voltage triangle abo indicates a typical operating condition with the
machine handling rated power. The quantity (Edv2) sin 0 is equal to
the dista.nce be, and therefore, according to Eq. (11-62), it must be held
"9" Principler 0' e/ec:tromechaniea/-enerr1Y eOlfYerrion
constant if the machine continues to handle full rated power_ If the direct
field current excitation is changed, thus changing the magnitude of E" the
point b must move along the horizontal rated power line. Reducing
E, from its value in the abo triangle moves point b to position d and the
machine operation is given by the ado triangle. The angle 8, now given
by the angle Md, has increased in value and is maintaining the relation
ship of expression (11-62).
If E, is further reduced to the point where the machine operation
is described by the triangle !leO, then the angle ~ is equal to 1('/2 rad.
This is the minimum value that E, can have with the machine still
delivering rated power. Increasing E, to a value such that EJ!-v'2 is
greater than V causes the operating condition to shift to a point such as f.
For leach of these operating conditions the sinor phase current I
must lag the voltage sinor jX.I by 1('/2 rad. Thus the current sinon!
corresponding to the operating points b to g in Fig. 11-12 are labeled
with primes. Since the machine is delivering constant power, the
projection of the sinor I on the sinor V must be held constant. A line
drawn perpendicular to sinor V, therefore, traverses the tips of each
e'
ower line
Fig. 11-14 Sinor diagram for the lossless nonsalient-pole synchronous motor.
The qnchl'onoull machi,.. 497
(I'
the same lowercase letter with a prime notation. Thus current sinor
o'b' eorresponds to operation at point band (J'd' eorresponds to operation
at point d. Notice that reducing the field excitation causes an increase
in the magnitude of the phase eUrrC'nt, in addition to increasing the power
factor in the lagging direction.
Point g ill lng. 11-15 corresponds to unity-power-factor operation
and the minimulll magnitude for t.he sinor phase current. Increasing
Edv2 beyond point, g to point. f gives rise to the current sinor o'f' at a
leading power-factor angle. As in the ease of the alternator, the syn
chronous motor ('an supply a ('onstant torque at cOllstant speed to a
load while extra('ting power from an infinite bus at an adjustable power
factor. Adjustment of the dire(~t field current controls EI/V2 and
thus the power factor. If EdV2 is reduced below point e, correspond
ing to ~ = -1f/2, then the machine ean no longer supply the required
load power and will lose synchronism. This will cause large phase
current pulsations, and the machine will eventually stalL
V curves similar to those shown in Fig. 11-13 can be drawn for the
motor operation. A succession of constant-power lines at varying
distances from the sinor V could be constructed on Fig. 11-15. The
phase-current magnitude I could then be plotted as a function of the
field excitation Edv2 for each of these constant-power lines. Curves
498 Principles of e/ectromechanical·energy conversion
identical with t hose for the alternator would result wit h the simple
interchange of the words leading and lagging ill Fig. 11-13. For values
of Etlv'2 greater than the unity-power-fador eonditioll the alternator
delivers power at a lagging power-factor angle, whereas the motor would
receive power at a leading power-factor angle. The remainder of Fig.
11-13 is the same for alternator or motor operation.
Example 11.4 1
A 5O-hp 230-volt d-e shunt motor is coupled mechanically to a 5O-hp 230-volt six
pole 60-<,ps three-phase wye-connected nonsalient synchronous machine, as shown in
Fig. 11-16. The field of the do(' machinfl is connected in series with a rheostat R
acroSll a 230-volt doC infinite bus. The a-(' RynchronoUH machine i" connected to a
230-volt tbree-phll.lle a-c infinite bus, and it is running at llyn{·hronou8 speed. The
IIwitch K, in scries with the d-c machinfl armature, is open. The machine parameters
are 8.8 follows:
For the doC maehine:
Field resistance R, = 20 ohms
Rotational inductanee
Armature resistance fl.
G.,
= 0.22 h
0.1 ohm
For the a-e machine:
:;;ynehronous reactance X. = 0.8 ohm
Armature resistance is negligible
The windage loSlles of each machine are 3 kw when the machine is running at syn
chronous speed.
a. With switeh K open the field of the synchronous machine is so adjusted that
the a-e line currents are a minimum. Determine the magnitude of the line currents
I, the torque an!!:Je 0, and the induced voltage Etlv'2.
b. While the a-e machine field current is held constant, the d-e field rheostat
Il is so adiusted that, when switch K ill {'Iosed, the direct armature current remains
at zero. Calculate the required value of Il.
c. While the a-e machine field current is kept at the value set in part a, the d-c
field rheostat is reduced 2.0 ohms from the value set in part b. Calculate the new
I This example makes a very interesting laboratory experiment.
steady-state direct armature current J., torque angle, alternating line current, and
power factor. Which way is energy flowing?
d. With R held at the value set in part cdetermine E,/V2, 0, and J for conditions
of unity power factor and a power factor of lUi leading.
Solution: a. The synchronous machine is acting as a motor supplying the required
windage losses of its own rotor and the armature rotor of the d-c machine. Since
the armature electrical losses of the a-c machine are zero, the total converted power
is equal to the power drawn from the three-phase bus. Therefore, we have
v'3 V Ll cos 6 '= 3000 + 3000
where the phase current I equals the line current, since the machine is wye-eonnected.
For minimum line currents the a-c machine must operate at unity power factor; thus
6000
~ - V -jX.I
V2
E,
fi
iXI/~
Vz230/"tj /=15.1
tal !h)
r
112volts
LiX,1
~V~z-1-33----JL---------~
(e)
which reduces to
Ej 2ao . I
~2 v'3 - }O.S X 15.1 = la3..=§.~
Therefore, the torque angle Ii _6.2° and the induced voltage E,I..;'i "" 133 volts.
Since Ii is less than 5.7°, the sin Ii = tan Ii and E,I..;'i "" V = 13:J volts.
b. In order for the d-e armature current I G to remain zero when the switch K
is closed, the back emf of the d-e machine must equal the d-c line voltage. I Thus
we must have
Go,..,' T, 230 volts
On solving for the field current, we have
230 .
I, 0.22 X (3n/a) = 8..32 amp
The total\ field l'ireuit resistance consists of the rheostnt R in series with the field
resistance R, = 20 ohms. The required value II is ohtained from
2ao
]{~+ /{,
By solving for R, we have
2ao
]{
X.:12 - 20 7.fi ohmtl
- - - --;;-'c;----
9.0 = -
2 00 amp
Since T. is negative, the doc machine must he acting as a generator. Power is bcing
drawn from the a-e line!!, !'onverted to mechanical energy hy the synchronous machine,
transmitted through the shaft to the doc machin/!, rec,onverted to electrical form, and
supplied to the doc bUB.
The total power converted hy the a-c machine is equal to the windage of both
machines pillS the power converted hy the doc machine. Thus we have for the a-c
converted pow('r per phase
P,q, = i(total windage power + G.,..,'l,T.)
p,. i(30()() + 3000 + 0.22 X 126 X 9.0 X 2(0)
P •.,. 1101.7 kw (per phase)
Sim'!' the alternating field current is held ('qual to the current set in part a, the
valu(' of Et! v'2 remains at 133 volts. Thus the per-phase sinor diagram has the
inter{'sting form shown in Fig. 11-17b with E,l V2 and V forming two sides of an
, Capital letters are heing used for the constant d-c variables ra.ther than the more
('umbersome subscript zero used previously for constant values.
Th. Iync:hronoul moehirHt 501
isoscelm. triangle with jX.I as the base. Since the sinor current I is nonnal to jX.I,
we have sinor I bisecting the torque angle a, and thus the power-factor angle (J = '/2.
By using Eq. (11-59), we have
sin 0 X.P••
VE,/v'2
and so
, = -57.7°
The minussign indicates that Ed V2 should lag V as shown in Fig. 11-17& for motoring
operation. The power-factor angle is simply
Hinre the a-c machine has no elm:trical 10l1l'i1',8, the power input from the a-c Jines
equals the power convcrtNI. Thus on a per-phase basis, we have
Vlcoll(J=P..
.. 175 volts
The 0.8 leading-power-factor case is illustrated in Fig. 11-17d. Since the sinor
jX.I is perpendicular to the sinor circuit I, the angle between jX.I and a vertical
line is the power-factor angle (J = 37°. Thus we have X.l cos (J .. 112, and 80
112 112
Notice how the adjustment of the field current of the a-c machine alters the power
factor of the synchronous motor from lagging in Fig. 11-17b to unity in Fig. 11-17c
and on to leading in Fig. 11-17d. The converted power in all three cases has not
changed.
(11-63)
where R' is the resistance of each armature phase and X. is the synchro
nous reactance. The diagram shown in Fig. 11-18 has been rotated
counterclockwise by an angle 6 from that of Fig. 11-8 strictly for conven
/
I
I
I
I
I
/
/
, .,fl I
, rr I
"/2~/
, I
, I
'v
Fig. 11-18 Complete sinor diagram for one phase of a non salient-pole synchro-
DOUB machine.
The synchronous machine 503
R' jX.
ience. The four voltage sinors drawn in heavy lines represent the equation
From Fig. 11-18 notice that 8 + 15 is the angle between Edv'2 and I.
For a cylindrio&l-air-gap, or nonsalient, machine the equivalent cir
cuit shown in Fig. 11-19 is very convenient. For a salient-pole machine
this equivalent circuit gives a reasonable approximation to actual results,
particularly when the machine is running near full-load conditions, by
use of the average value of the direct- and quadrature-axis synchronous
reactances as X,.
Exomple 11-5
For the alternator deecribed in Example 11-3 IIOlve Cor the torque angle 8 and the
induced armature voltace E, by using the nonsalient equivalent circuit of Fig. 11-19.
504 Principle. of electromechanical.energy conversion
Use a synchronous reactance value which is the average of the direct- and quadrature
axi!! reactan!'es.
Solution: The sinor phase voltage V for the wye-eonnected machine can be
specified as
V '" !~!200 /0
V3 . . . ..
For convenience we are taking the phase voltage as reference with an angle of zero.
Since the machine is acting as an alternator at a leading power factor of 0.8, let us
define a sinor phase current I' = -I. Thus from part a of Example 11...3 we have
I' 131/37°
In terms of I' = - I in Fif!:. 11·19, we have the following sinor loop equation:
~ =. V +jX,I'
V2
with Rr = O. 1'iubstituting the values for V and l' gives us
~ = !3,2C!Q /0 + (.')8/90)031/37)
V2 V3 - .~ --
which reduces to
7620 + 7590/127
By solving for E" we have, finally,
E, = 9640L6:l.2
The torque angle is defined as the angle by which E, leads V, and thus .$ '" 63.2°,
The magnitude of E, is 9640 volts. The correct values for a and E, as found in
Example 11-3 are 54.7 0 and 9970 volts, The results of this simple nonsalient analy
sis are a reasonable approximation to the correct results.
The angle l> entered into the analysis in Eq. (11-18), which, for con
venien('e, is repeated as follows:
W
a(t) = - t -
n
11'
-
2n
+ n-l> (11-66)
(11-70)
On substit.uting Eqs. (11-69) and (11-70) into (11-68) we have for ,the
a-axis stator equilibrium equation
v'a (11-71)
IEquations (J I-R) speeify these armature voltages as being applied to rotor windings.
Rempmber that the synchronous machine is analyzed by inverting the stator and
rotor and that the actual two-phase equivalent voltages are applied to the stator.
506 Principles of electromechanical-energy conversion
Synchronous
field oxis
I The three-phase machine is by far the most widely used synchronous machine.
Therefore, all of our results are being carefully referred back to the three-phase
machine.
The synchronous machine 509
(11-80)
I
/
Air-gop line ...._/
/
/
Therefore, the slope of the curve shown in Fig. 11-21 is wM:/10 and the
mutual inductance M d' can be obtained. The straight-line portion of
Fig. 11-21, below the knee of the curve, is called the air-gap line. If we
use the air-gap line as an approximation to the actual open-circuit char
acteristic, M:{ is a constant independent of the field current.
I With both i do and I~O equal to zero the sinor diagram of Fig. 11-8 shows us directly
that the quantity E tI v'2 equals the magnitude of the phase voltage.
The synchronous machine 511
of Salient-pole Machines, Trans. AlEE, 62: 6R4-689 (1943). ,Four tests tor deter
mining X. are discussed in this paper, induding the slip test, the maximum-Iagging
current test, and two other less popular tests known as the reluctance-torque and
negative-excitation tests.
512 Principle, of electromechanical·energy conversion
field reduced to zero. With E, 0 the second term in Eq. (11-89), often
called the reluctance torque, is sufficient to supply the required windage
torque.
Assuming the windage torque to be very small, the torque angle 0
will also be small, particularly with rated voltage on each phase. There
fore, for small 0 we have sin 0 "" 0 and sin 20 "" 20, and Eq. (11-89)
becomes
(11-90)
(11-91)
(11-92)
for very small values of o. The value of I just before the machine pulls
out of step, as the field current is increased in the negative sense, is
The $,nchronou$ machine 513
given by
V - Ej.... /y'2 V
I pO = ·~~-X-;;--- = Xq (11-94)
wh{'r(~ V is I he phase volt a~() and I pO is the armature phase ellrr{'nt. just
prc('('(lillj:/; t he point wh{,l'e the llllll·hin{' loses sym,hronislll as }<;/ is Illude
i n('I'{'asi ngly ne~a t i ve,
The various parameter tests described in this sectioll are briefly sum
lIlul'ized as follows:
1. The open-cireuit test gives liS the relationship between the field
indueed volt a~{' }<;/ and t he field cUI'l'cn\' i d(},
2. The short-eil'('uit test provides t he lIla~nit IIde of the direet-axis
synehronous real'lluH'e X d • If t he windage losses are known, then an
effect ive annal ure rcsist tlnt'e pel' phase R~rr ('an also be obtained frolll I he
short-eir(,lIil test.
:~, The IIInxillllllll-laggillg-('U1'1'{,llt test gives the value for the quad
rature-axis sYIH'hroIlO\lS n~:wt IlIH'e X q.
(11-96)
We are interested in the torque angle 6 as a fUllction of time after the load
on the synchronous machine is changed.
The mechJt.nieal-port equation is given by (I 1-2), which for conven
ience is repeated as
T' = (Jp + D)w' + T. (11-97)
where J is the total inertia of the rotating system, D is the total viscous
friction coefficient, T. is the total torque of electrical origin, and T' is the
total externally applied torque. Usually the inertia of the rotat ing system
J is sufficiently large that any variation in speed, about synchronous
speed, occurs slowly enough that the elcrtrieal-port varil~bles are always
at their steady-state values. If the torque angle is to ehange in value,
then the speed of the machine must differ slight ly from synchronous
speed. However, t.hese speed fluctuations are sufiieiently slow that. the
torque of electrical origin given by Eq. 01-31) is still valid. Hepeating
Eq_ (11-31), we have for the electrical torque per phase for a salient-pole
machine
T ...(6) (11-98)
where a is the angle between the a-stator and a-rotor axes on the a-b
primitive-machine modeL The angular velocity of the rotor w' is there
fore given by
pw' =& n
(11-101)
Tr = ( -Jn p2 + nf»)
p 6 + D w + vT.~(a)
n
(11-102)
The synchronous machine 515
(11-104)
Similarly, all nonlinear terms in Eq. (11-102) are expanded in a Taylor
sCl·jes ahout Ihe d-(~ value of th(' torque angle. The tOlal torque of
c1eetl'ienl origin is I herehy approximated by
(1l-1Q.t»
-1 = dT.(a) I vn [
=-
VEl
-",--cosoo+
V2 (Xd-Xo)cos2a ] ,
o
K d6 A, w y2 Xd XdXq
newton-m/rad (11-106)
Notice that the eompliallce is IargN in the ease of H. nOllsalient maehine.
The saliency, in effect, gives us a stiffer spring constant or smaller COlll
plil1.1H'e fador.
Upon substitution of Eqs. (11-103) to (11-10;')) into Eq. (11-102) the
mechanicl1.l-port equilibrium equation becomes
(11-107)
\VhCll all time-dependent increments nre sct equal to zero, the steady
state equilibriulll cqul1.tion is given by
To = D ~ + T.(oo) (11-108)
n
(11-109)
where
(11-110)
(11-111)
Notice that as long as K is positive (Le., the slope of the T. versus 0 curve
is positive) the solutions of Eq. (11-109) are stable. Therefore all oper
ating values of 00 between the first positive and negative peaks in Fig.
11-7 are stable operating points. !
For a step change in the applied torque Ti(t), the incremental torque
angle Ol(t) has a form similar to one of the curves shown in Fig. 11-22. I
H the damping ratio r is less than unity, the torque angle has a damped I.
~
oscillatory motion about its new final value. These oscillations of the
torque angle are often referred to as hunting.
In order to minimize this undesirable hunting, the damping ratio
must be increased. For a given machine at a particular operating point !
00, the inertia J, number of pole pairs n, and effective compliance K are
all fixed parameters. The only remaining factor influencing the damping
ratio r is the viscous coefficient D. However, if D is increased, then the
steady-state windage losses of the machine will also increase. A more
desirable way to inc-rease the damping ratio is by means of additional
rotor circuits known as damper windings or amortisseur windings. These
circuits consist of conducting bars imbedded axially in the pole faces of the
rotor or around the rotor periphery in a nonsalient machine. At each end
of the rotor a shorting ring connects the ends of all the damper bars.
H the rotor revolves at synchronous speed, then the bars revolve at
the same speed as the rotating air-gap field. However, if the rotor goes
into a torsional oscillation, then currents will be induced in the shorted
dampcr circuits in such a direction that t,hey oppose the oscillations. For
small torsional oscillations the additional damping torque is proportional
!< T~
2 3 4 5 6 7 8 9 10 11 12 WItt
FiJI;. 11-22 In('reml'ntal change in thl' torque an!l;le for a small step change in the
applied rotor torque, plotted as n function of the normalized time ",.t, for three par
ticular damping ratios.
to the time rate of change of the incremental torque angle; thus we can
write .
(11-112)
where Tn is the additional damping torque and D' is the damping coeffi
cient due to damper windings. Adding this additional damping
term into gq. (11-107) increases f) in expression (11-111) for the
damping ratio by an amount D' without changing the steady-state
windage torque Dw/n in Eq. (11-108). The damper windings are also
used as starting windings for synchronous motors, as will be explained in
Chap. 12.
If the increlllent al driving torque T~ (t) should contain a sinusoidal
cOlllponent, then «h(/) will be fOl'ced to oscillate at this same driving fre
quency. The Bode plots described in Chap. 4 are a convenient represen
tation fOl' t he sinusoidal torque-angle response as a functioll of the driving
frequell('Y. For small damping ratios, a forcing frequency near the
natural frequen(,y w. of the rotor as given by (11-110) can cause large
torsional oscillations that result in extrellle shaft torques and lIlechanical
failure.of the system. Pulsa! ing loads such as air compressors or pulsating
prime movers such as internal-combustion engines must never contain a
518 Principle, of electromechanical-energy conversion
(11-113)
TO do = J (p 2 8) do
n
+ vT.~(8) d8 (11-114)
The first term on the right-hand side of (11-114) represents the portion
of the supplied energy going into storage in the form of kinetic energy in
the rotating system. If the torque angle changes from zero to a value
8..., the total change in kinetic energy is given by
,.. J
KE change =
to n - (p 2 8) do (11-115)
" Remember that a positive value of T' means an applied torque in the direction or
rotation correllponding to alternator operation. For a load torque on a synchronous
motor, rr would be negative.
The synchronous machine 519
(a)
(b)
(c)
Fig. 11-2:J A graphical determination for the stability of the synchronous machine
for a large suddenly applied rotor torqueTo.
520 Principles of electromechanical-energy conversion
Figure 11-7 is a general plot for the torque of electrical origin per phase
T,.,. By IIlultiplying T<4> by the nUll1ber of phases II, the last integral in
01-11;'») can be interpreted as ill Fig. 11-23a. The curve T~ - liT'4>(0)
is plot1cd as a funetion of o. The integral frolll 0 = 0 to 0 = 0",
is simply the total area between the eurve and the axis, as is indieated by
the two shaded portions labeled A and B. When the shaft torque T~ is
suddenly applied, the torque angle 0 increases until a lIlaximulll value 0..
is rettehed where area B is equal to ar('a A. At this torque angle the
kinetic energy in the system is zero, whi{'h means that 5 = 0 and the
torque angle has swung to its peak value. After a series of these oseilhL
tions the damping whieh is aduully present in the system will eause the
rotor to seUle at Of in Fig. 11-2aa, where the e1erlri(~al torque equals the
applieq torque.
Suppose that the applied torque T~ were sufficiently large that area
B could not equal area A, as is the ease in Fig. 11-2!ib. The kinet ie energy
would never go to zero and a value for 0.. could not be found. The rotor
would be out of step with the synchronous field and stable synehronous
machine operation would not result. After slipping a numher of poles,
the damping might {:ause synchronism to reo{'('ur and 0 would settle to a
value Of. However, such a mode of operation is usually undesirable.
The p('ak torque which can he suddenly applied wit hout loss of synchro
nism is obtained hy having area ,1 just equal to the complete area B in
Fig. 11-2:{b. A sudden torque larger than this critical value causes loss
of synchronism.
If the initial torque angle just prC('cding sudden application of the
driving torque were 00 instead of zero, then all lower limits in (11-115)
would change to 00. Consider the case when the machine is initially
operating as a synchronous motor at, a torque angle - 00, as shown in Fig.
11-23c. A driving torque T~ is now suddenly applied, eausing the machine
to switch to generator operation. Stahility is maintained as long as area
B is larger than the complete area A shown in Fig. 11-2:k Assuming the
system is stable, the final operating value will he a torque angle Of as
shown in the figure. I
I For a more complete dis('ussion se«' J. J. Stoker, "Nonlinear Vibrations," pp. 70-S0,
I For ('olll[ll(,t!' dis('ussions of short-{'inuit r('spoII!<{>S 1>1'1': (1) n. Adkins, "The General
Throry of EiI'I'trie :'IlllI'hilll's," John Will'Y & :-:ons, 1111'., ~e\\' York, HIS!). (2)
C. Coneoruia, ":-:Yllchrono\ls :'Iim:hin('s," .John Wil"y & Nns, Ine" New York, 1951.
522 Principles of electromechanical-energy conversion
Tlu, oril!;inlll equilibriulll I'l!lwtiom; for the ('Iel'tri('al ports of the d-q
Illmll'l art' gi\'('11 hy Eqs. (11-1). The spc!'d of llw machine is assulllcd to
hI' {'!HIH! ant at sYJII"hrOIlOlls speed; t hils
w
w' =- -- (11-119)
n
Also, from Eq8. (10-;")0) to (10-;")2) we have
,rr
( rqd nJ.J~ (11-120)
ASSllll!il!1!; the t hrl'(, anllat 111"(' willdiuw> of the oril!;inal machine to be
idelltieul wit II resiHI uJ!{'es equal to H', we have
R~ R; = If' (11-121)
131use of the silllplifi{'ations givcn hy Eqs. (11-118) t.o (11-121), the
clcdrieal-porl t'quilihriulll equations redUl'c 10
P/d'P
IT' +JJ.J~p 01-122)
-~L~w
Wit 11 the fic!!l {~x{'it I'd hy a eo list ant -volt :lI!;C souree, we have sct v~ = VdO
ill (lJ-122).
Figure 11-24 shows a eompict e equivalellt eir('uit for the d-q model
of the syn(·hrolloIlH machine. The rotor speed is assullled <'onslant at
synchronolls value, flO a d-e SOUf('C w' win is eOlllledcd to the rotor
port. A d-I' sour!'e I'~o is eonneeted to the field port. The two armature
ports are simul1lllleollsly short-I'irt'uited at time t 0, as indi('ated hy the
t \Yo-pole swit('h.
The solutions for i; and i; will involve t hi I'd-order differential equa
tions; t.herefore, three initial ('ondilions are required for eaeh unknown
variable. The initial values of i~ and i; are
o (11-123)
Binee the armature is open-eircuited for t < O. The initial first derivative
of i~ is ohtained hy evaluating the first two of Eqs. (11-122) at tinw
t = 0+, thus getting
IJ~pi~(O+) + iM~'pi~(O+) (11-124)
M::pi~(O+) + ~/J~pi;(O+)
The quantity t'~" - R,ji~(O+) is equal to zero; therefore, both
pid(O+) o and pi;(O+) = 0 (11-12i)
provided
• nL IIr l·r
'Zlq
'
'- ...,
I
I
I
I
.. l·r I
ztq I
I
t=O I
' I
...... ...J
Ideol
Fig. 11-24 Equiv/lknt C'in'uit for til(' d-q model of the synchronous machine with a
symmetrical short rircuit applied at t .., O.
The parameters L d, Va, and lII~r arc defined in terms of t.he machine
dimensions hy Eqs. (6-90), (H-!)2), and (6-94). If these relationships were
substituted into (1l-120), the expression would be equal to zero. Our
prirnitive-maehine model IlSsumes that. we have unity coupling between
all mutually coupled coils. In pnwtieal synchronous machines unity
coupling is not achieved (owing to fields around the end connections and
slot effeels which have been neglected), and so we shall assume that
(11-120) is not equal to zero.
By differentiating both of the first two of Eqs. (11-122) with respect
to t.ime and again evaluating the results at t = 0+, we have
(11-127)
(11-129)
524 Principles of electromechanical-energy conversion
Let us examine the first term on the right side of (11-130). The resistance
of the field circuit is usually small, so this term can be approximated by
V~O l'ao
"d(t) (11-133)
(11-134)
using (11-132) and the approximation that the field flux linkages remain
constant. Substituting (11-134) into Eq. (11-130) gives us
(11-135)
0= {R: + [I d' - i!
,. ,
.<~!~o.~e.]
/(j + l/'dP p} iT!
'~'.
+ 3[ 'wi'
q
(11-136)
2
_ a [['d' - ( AJ d'L. p_] wid' + (Rr + [Ip)i r
ltd + Ljp (11-137)
.3
~ of. q
(11-138)
Solving Eqs. (11-1:~(i) lllld (11-1:l7) for i~ and i;, by using determinants,
leads to
where
(11-142)
(11-143)
X'd --~w
a [I'
'd -
(Al~r)2]
--r;;- (11-144)
We now substitute Eq. (11-142) into (11-139) and (11-140), and, after
some manipulation, we have
r I Rr (J.,Xd + ..!.)
X
q
(11-149)
Xd ~wL~ the direct-axis synchronous reactance (11-150)
The field circuit time constant Td is often called the direct-axis transient
open-circuit time constant, and the quantity T~ is called the direct-axis
transient short-circuit time constant.
The solution to Eq. (11-145), using i~(O+) = pi:i(O+) = p2i:i(0+) = 0,
after some manipulation, can be put into the form
i:i(t) = - E,
Xd
E~ e-i.. ' cos wt -
+ Xd E, (J.,Xd - ..!.)
Xd
e-*" (11-1.'51)
as one phase of the short-circuit currents. The other two phases i~ and
i~are obtained from (11-156) by adding -2'lr/3 and 2'lr/3 to the arguments
of the sine and cosine terms.
Equations (11-151) and (11-152) are now substituted into Eq.
(11-156), and, after some algebraic manipulation, the a-phase short-circuit
current can be put into the form 1
..
i' =
(11-157)
i~ (d-c term) = EI
2
(1,Xd + ~)
Xq
e-t"'t cos (8 0 _ 2'3lr) (11-158)
.S
1/1
(b)
·s
Iy
(c)
Fig. 11-25 Phase currents resulting from a three-phase symmetrical short circuit
applied when a(O) ,.. (Join = ."./2n.
For (Jo = 1f"/2 the term is positive, and at t = 0 it raises the exponential
envelope to
E, E, ( 1 1) (1f") (11-159)
X~ + 2" X~ + X q cos - 6
The i3-phase short-circuit current is shown in Fig. 11-25b, and it is similar
to i! except for a - 21f" /3 phase shift and the exponential deeay of the d-c
level. The"f phase is obtained by letting 00 in Eq. (ll-Vm become
00 + 21f"/3. The d-c term now shifts the envelope in the negative direc
tion, as shown in Fig. 11-25c, to a magnitude identical with (11-159).
The syrn;hronous machine 529
11-12 SUMMARY
A model for the three-phas(> synchronous ma('hine is formulated in terms
of an equivalent, two-phase a-b llIaehillC'. Hinee t,he aet.ua.! machine often
has salient. rotor poles, the a-I) model Inust, be inverted with the salient
field winding ('onstrtleled on the slator. The windillgs on the a-a model
are identi('al with the windings on the three-phase maehille. The two
phase excitations are equivalent in the sense that, I he total a-b surface
current. distributiolls are two-thirds the m:\gnitude of the total a-{1-'Y
surface-current. distributions. However, all power quant.ities calculated
per phase for t he two-phase model nrC' ident ieal wit h the pC'r-phase power
quantit ies for the aetual maehine. The synchronous-machine equations
can be put in a more (·onvenient. f01"1II hy first transforming the a-b model
into an equivalent d-q model. Remembering that the a-phase variables
of the two-phase system are equivalent to the a-phase variables of the
three-phase syst em, a simple d-q to a-I> t ransformat ion changes the d-q
variables hack to aet.ual-machine variables for the a phnsc. The {1 and
'Y phases are obtained by phase-shifting the a-phase quantities hy -271-/3
and 211"/3 md, respeetively.
The most. commonly elH'ollntered opcl'llting mode for the synchronous
ma(:hilH~ is to have the armature eOlllw(·ted 10 a ('onstnnt-rolyphase-volt
age supply known as an infinile hus. Under sleady-state eonditiolls the
excitations for the d-q model IIl1whine are shown to reduee to simple doc
constraints. The d-q Iluwhine ('urrrnts al'C also ('onsl ants, and the equilib
rium equations are rcdu('ed to simple algebraic (nondifferent ial) equations.
Expressions for the eonverted power Hnd the torque of eleetrical origin are
ohtained for the d-q model. Hilllply dividing the torque and power expres
sions hy 2 redu('es these quantities to ft. per-phase hasis. For the actual
three-phft.Re lIul.('hine, a lIlultipli('ation by 3 yields the total ('onverted
power ft.nd the tolltl gellernted torque.
On examining the d-q to a-b transformation for constant-speed opera
tion, we ('an sec that, the d-q variables appear as coefficients of constant
fr('queney sinusoidal quanl it ies. Therefore, t he I ransforlllatioll from d-g
10 a-b variables ('an most easily he :l('('olllplished by using a sinor diagram
with the d-q variahles appearing as sinor quanlities. We need only draw
the sinO!' diap;ralll for 01\(' phnsp of Ihe lwo-phnse a-Ii llIode'l, silH'e the see
01H1 phase is !'limply di!'lpi:t('('d hy 7r 12 rad. The sinor dia!l;mlll for the a
phase also repr(~R('nls t II<' sinor diagmm for I he' a phase of I he Ihree-phase
syslem. Thus, hy drawing the a-phns(' diagram, we automatieally have
the a-phase' diap;ralll. The' fJ and 'Y phases are' ohtained by phase shifts of
-27r/;~ and 211"/:~ Tad, r('sp<'l'IiV<'ly, wllilp til(> II phase requires ft. 11"/2 rad
phase' shirl. By using the' sinor diagram. I he ('olllplete steady-st.ate
synehrollous-l1Ia('hine operation ('IHl easily he ohtained. Under constant
530 Principles of electromechanical-energy conversion
PROBLEMS
ll-1 Find the speed, in radians per second and rpm, for the following syn
chronous machines:
The $ynchronou$ machine 531
II-" a. The three-phase synchronous machine, shown in Fig. 11-2, has three
identical armature windings. How arc each of the two armature windings in the
inverted fl-q model, shown in Fig. 11-5, related to the three original windings?
b. How are the field windings in Figs. 11-2 and 11-5 related?
c. With the fi('ld structure of the three-phase machine in a suitable position the
maximum s('lf-inductan~e of one armature phasc is 1.2 mho After rotation of the
field hy nn appropriate angle the minimum self-inductance is found to be 0.8 mho
Is the rotor salient or nonsalient? \)('termine the quadrature- and direct-axis syn
chronous reactanc('s for a rated frequefl!'y of GO cps.
d. Through how many mechanical dcgr('cs is the field in part c rotated if the
machine has a rated speed of gOO rpm?
N 11-5 A 'lIIlient-polc three-phase GO-cps four-pole 220-volt 30-hp delta-connected
\1Ynchronous machine is opernting in the steady state such as to have a per-phase
sinor diagram ns shown in Fig. Pl1-5. The direct- and quadrature-axis synchronous
reaetanres per phase are respectively given by X d = 2.0 ohms and X. = 1.5 ohms.
The nrmnturc rellistanrc is R' = 0.10 ohm pcr phase.
a. Calculate the rotor angular velocity in rpm and radians per second.
b. Find the power factor and the magnitude of the phase and line currents.
c. Is the machine operating as a motor or as a generator? Prove your answer.
d. Find the torque angle (I and the induced voltage E,/V2.
e. Find the total converted power and the total torque of electrical origin .
•r
~ =30omp
.[i
.r
~ =40omp
.fi
Fig. PII-5
532 Prillciples of electromechallical-ellergy COli version
Vol
rent
\
~~.1
~
J A thrf'c-phnsf' nonsalif'nt syn('hronous alt('rnator has a diref't field .eur
nmp. At rntt'd spf'(>{l tht' mn('hine generates nn open-('ircuit terlkinal
voltnge of IH6 volts and a short-('ircuit terminal ('urrent of on amp with all· three
tcrminals short-('ircuit('d. Cal('ulate the syndlronous rf'u('tnnt·f' r, if:
a. The Mtntor il'! wYf'-eonneeted.
Armature open-circuited
Direct field curr!'nt, amp 10 20 40
Armature terminal voltng(', volts 131!l 22!}() 3770
Prime-mover driving pow('r, watts JO,lOO (const)
Armature sllort-circuited
Direct field current, amp 30.5
Armature current, amp 39.4 (rat('d current)
Prime-mover driving power, watts 17,100
Use Eq. (11-79) to ('alculate the effective phn!!e rcsistnnc(' for the alternator.
11.
b. Use the slope of the uir-gap line and Eq. (ll-RI;) to !'alculate the mutual
inductanNl parametf'r Atri'.
c. From the short-cireuit test ('ompute the per-phase synchronous rea('tance X •.
d. Draw the per-phl!.8c equivalent ('ir('uit for the mlU'hine.
Armature 8hor/..circuit£d ,
Direct field current, amp 6.2
e. By usinp; the as!!umptions leading to E'l. (l1-!)!), ('ftkulll.te the nep;ative ,lirt'('t
fielcl ('urrent when the motor slips a pol .. in the maximum-llLll:ll:inp;-ellrr('nt t ..st.
f. Write expresRions for the Ilelf-indudanc('s of the three Rtator phusel< as fune
I ionR of I Ill' rotor anp;ular posit inn e.
II-J,i A pertain Hix-pole !lO-epl< Ryne'hronous nltt·mator haH n totnl shaft inr·rtill.
of 75 kp;-m'. The synt·hronous torque' of elt'l'tril'al orip;in genernted P('f flulian change
in torque anp:le is !HOO newton-m. OV('f the nornllLI operating rep;ion ('onsi<ier the
T, VB. 0 characteristic to he linpar. A 1~()-kw electri('al load is suddenly applied to
the machine. All windage, friction, and rotational damping are pra<'ti('ally nt'gligihle.
a. Find t he final 5t eady-state torque anp;le.
b. If the shaft is Iwing observed with a strnhoscnpi(' lamp flashinjl; at line fre
quency, by how many degrees will the shaft appmr to move when tIl<' loftll is applied?
In which direction with respect to the dire(,tion of rotation?
c. Will the rotor oscillate about the final torque angle? With very Il1l1all damp
ing find the frequency of these oscillations if they occur, as obs<'fved hy uHing the
Btroboseo~ic lamp in part h.
d. Find the maximum torque angle reached by the rotor after the electrical
load is applied.
e. If the motor were driving a periodically pulsating load, what would happen
if a component frequency in the load matched the frequl'ney found in part c1
11-15 An amort.isseur winding is now included on the rotor of the alternator
discussed in Proh. 11-14. The effective damping cocfncient for the o.TTlortisReur wind
ing is 2000 newton-m-sec/rad. The 180-kw load is again applied at time t = o.
a, Find the final steady-state torque angle after applying the ISO-kw load.
/" Hketch the torque anp;le vs. time charactcril!ti" for all I > n.
c. Comment on the damping ratio I achievel! with this amorti!lSeur winding
design.
6 A 2000-hp 12-pole three-phase 6fJ-cps wye-conneded synchronous motor
is co ecte a 6600-volt (line-to-line) supply. The direct field current is so adjusted
that the per-phase induced armature voltajl;e E ,I v'2 = :i560 volts. The direct- and
quadrature-axis synchronous reactances arc respeetivcly p;iven by X d 10 ohms per
phase and X. = 7 ohms per' phase. The armature resistance can be neglected.
Windage, friction, and damping effects are also very small. The motor is initially
running with no shaft load, and full-load torque is suddenly applied to the shaft.
a. By using the equal-area method discussed in Sec. 11-10 determine the maxi
mum torque angle reached by the motor on the first swing after the load is applied.
b. With full load on the motor, find the steady-state torque angle.
c. How much additional load torque can suddenly be applied without the motor
slipping a pole?
Il-17 This problem is a review of the more complicated steps in the develop
ment in Sec. 11-11.
a. Starting with Eqs. (11-136) and (11-137), show that Eqs. (11-145) and (11-146)
are valid.
b. Carry out the solutions of the third-order differential equations in part a, and
arrange the results in the form of (11-151) and (11-152).
c. Derive the a-phase short-circuit current in the form of (11-157).
d. Write expressions for i~ and i; that are valid for all t ~ O.
11-18 1 A three-phase IO,OOO-kva wye-connected 60-cps 12,OOO-volt synchronous
machine has the following parameters:
I Answers to this problem are listed at the end of the book.
The synchronous machine 535
11-19 The 1O,OOO-kva machine described in Prob. 11-18 is so excited that its
steariy-state Hhort-circuit current is 5 times rated current. A symmetrical three-phue
°
short circuit is applied to the armaturc of the machine at time t = when the field
axis is in line with the a stator axis.
a. Find the magnitude of the stator induced voltage E,/0.
h. }I'ind tho a-stator short-circuit current for all time t ~ 0,
c. Prepare a careful plot of i~ as a function of time similar to Fig. 11-25.
11-20 A nonsalient-pole Hynchronous motor is being supplied by a suitable
nonsalient-pole synehronous generator being driven by a prime mover at synchronous
speed. The synehronous reactanc-es of the motor and generator are respectively given
hy X ... and .Y.. Armature re!!istanl'c in hoth 1lIILehines can bc neglected.
a. Draw a pcr-phnse equivalent cireuit for the two machines. Use E,./0
and E,.. /V2 as the stator indm'ed voltages per phase for the generator and motor,
resp<'ctivcly, and IT and I as the phlk~c terminal voltage and current.
h. Draw a sinor diagram for the unity-power-factor case.
c. Show that the developed power p<'r phase by the motor is given by
p BloB,... (
,<f, =2(X.-+X;;') Sill ~. + ~..)
where ~. and ~ .. are .the torque an!l;les of the generator and motor, respectively.
d, For constant ficld excitations on hoth machines draw a sinor diagram for the
condition whereby the 1lI0tor is developing maximum torque.
chaptn
12
Fig. 12-2 The induced voltage in a single loop of thc stationary squirrcl-caRc rotor
structure.
e·In d = -
Jrs ~~.
at
dS (12-2)
where a and l are the radius and length of the rotor cylinder, respectively,
and thus 2afis the cross-sectional area of the loop formed by the two rotor
bars. The angle th is the pia~e angle for the kth loop, as shown in Fig.
12-2.
The result expressed by Eq. (12-3) can be summarized for all of the
rotor bars, as shown in Fig. 12-3. The small dots and crosses indicate the
direction of the induced voltage in each of the rotor bars. For the
particular bar considered in Fig. 12-2, fh appears to be approximately
1f/3 rad. Thus at time t = 0 the induced voltage is negative, or opposite
to the assumed positivc dire(·tion for the indueed voltage shown in Fig.
12-2. Thus, the dots and crosses are as shown in Fig. 12-3 for this par
ticular loop. The rule that a voltage is so induced as to cause a current
in a'direction to (~ancel the d\ll.llging flux linkages can aid in checking
these results. For the loop at 1f/3 rad the flux linkages are directed to the
right in Fig. 12-3 and nre decreasing at time t = O. Thus the induced
voltage has a direetion which would cause a current creating a field also
directed to the right to maintain these flux linkages, The rotor bar at
1f/3 has current coming out of the paper, while the bar at 1f + 1f/3 has
current, going into the paper. By the right-hand rule, flux is directed to
the righ t as desired.
The induced voltages, given by Eq. (12-3), are sinusoidal and have
the same angular frequency as the applied stator currents. With the
rotor stationary each of these rotor loops acts as a secondary of an ordinary
transformer; thus, the secondary and primary frequencies should be
identicl:l!. In the steady state the current-in each oft-he shorted loops
depends on the nature of the (~olllplex irnpedulwc of the loop's circuit.
This illlpedulH'c has a resistive eOlllpolwllt, and lUI illduetive reactive
eomponcnt. III general, the induced rotor currcnts in each loop, made
up of two bars Oli opposite sides of the rotor, can be expressed in sinor
Zero induced
voltage axis
form bt
R lIr ~.=_.9k
(12-4)
~R+J~L·
where E = -B.. (2al)w/V2 from Eq. (12-3) and Rand D are the resist
ance and self-inductance of each of the rotor loops. Since the rotor bars
are relatively large in cross-sectional area, their resistance is small com
pared with the inductive reactance wD. Consequently, the induced sinor
currents will lag the induced voltage by an angle approaching '11"/2 rad.
Figure 12-4 shows the direction of the resulting currents in each of the
rotor bars at time t = O. Notice that t he axis of zero indueed currents
Jags the zero-induced-voltage axis, shown in Fig. 12-3, by an angle
approaching '11"/2 rad.
The force on each of these current-carrying conducting bars can be
determined from
F=lixB
where 1 is the length of each bar, i is thf1 inst,antaneous-current vector,
and B is the magnetic-field vector at the conduding bar. The directions
of these forces are up for currents coming out of t.he paper and down for
{'urrents going into the paper, as shown in Fig. 12-4. The net sum of
these forcf1s causes It counterelockwise torque on the rotor strudure. The
rnagnit ude of this torque of eleetrieal origin can be shown to be propor
tional to sin 0, where 0 is the angle between the direetion of the stator
field and the axis of zero induced currents, as shown in Fig. 12-4. As
the stator field rotates, the axis of zero induced currents also rotates and
maintains a ('onstant dlsplacement angle 0 and thus a constant torque in
the sallle direction as the direction of stator field rotat ion.
If the rotor turns at an angular velocity less than that of the rotating
stator field, the voltage induced in the rotor bars has the same form as
The induction machine 541
Eq. (12-3), except that the frequency will be the difference between the
field and rotor angular velocities. The angle a will still remain constant
at some value between zero and 11"/2 rad; thus a constant torque is still
generated. Since a lower-frequency voltage is induced in the rotor, the
impedance of the rotor loops becomes smaller and more resistive in char
acter. The inductive reactance wI- is reduced because w is decreased.
Thus, the rotor currents are more in phase with the induced rotor voltages.
Referring to Fig. 12-4, this means that a would increase, approaching
11"/2 rad. The sin a also increases as the value 11"/2 is reached. However,
the magnitude of the induced voltage decreases and the generated torque
may actually increase or decrease depending on the exact values of the
parameters. In a later section these considerations are discussed
analytically in great detail.
If the rotor revolves faster than the stator rotating field, induction
generator action results and energy flows out of the stator ports. The
wound-rotor construction is introduced in Sec. 12-2, and the analysis
proceeds from this mechanistic model. This initial section is intended
only as an introduction to give some physical feeling for how the induc
tion machine operates.
Sec. 6-8.
542 Principles of electromechanical-energy conversion
y'
t';,
1 -i -j 0 0 0
0 va Va
-2 0 0 0
2
(l~ ...~y = i 0 0 0 1 -i _t
(12-9)
11"
0 0 0 0
va
--2 2
va
The voltage variables are also transformed hy a similar matrix equation;
thus
(12-10)
where I
v~
v~
r v~ (12-11 )
v!~y = V'
a
v~
v~
(12-15)
_.[~l_ ~
a"
Iif.ob - 0 0 cos na
o 0 sin na
-'i~ no]
cos na
(12-16)
The d-q and a-b stator quantities are identical. However, the rotor
quantities are transformed in such a fashion that the d-q machine using a
commutator-and-brush mechanism has a total rotor-surface-current den
sity that is identical with the total rotor-surface-current density of the
a-b machine using slip-ring connections.
In construction the stator and rotor windings of the d-q machine are
respectively idcntical with the stator and rotor windings of the a-b
machine (which in t urn are rcspcetivcly idcnt ieal with those of the original
three-phase machine). The equilibrium equations for the d-q primitive
S46 Principle. of electromechanical-energy conver.ion
+
0
L~p
T tI -- - P.
wr
(12-19)
(12-21)
J The rrlationllhips giV('n by 02-20) are interpreted by using only one of the suh
scripts. Thus, the term Uj. is a shorthand notation for Uj, = U; and U'ap-y stands
for fl~ = flo R;.
The induction machine 547
Vti]
I"
g =
[HI +0
Up
H'
o
+ Up
A/"p
o
l'~ A/"p nAI"w' +
Fl' Up
l'~ -nAl"w" A/"p -nlJ'w'
where the expression for 7'. from (12-19) has heen substituted into the
rotor-port torque cquation.
Equations (12-2a) and (12-24) represent a set of nonlinear simul
tancous differential equut ions describing t he opera! ion of the d-q model of
the induction machinc. \Ve huv(> been very careful in the formulation of
this model to keep traek of thc relationship between t,he d-q parameters and
the paramcters of hoth the two-phase a-II model and the three-pha,se a-{J-"{
induction machine.
(12-26)
(12-27)
where V' is the rms value of the applied phase voltages. By using the
stator portion of the a-{3-'Y to a-b transformation given hy (12-9), the
equivalent two-phase stator excitations are found to be
lJ~ = V2 1" cos wi l'~ = y2 V' sin wt (12-29)
The stator portion of the a-b to d-q transformation, given hy (12-16),
shows 1l' that
t'~ = 1': = y2 P cos wi (12-30)
V~] _ o
V'9
o -
[R< +0 jwL'
jwM"
R' + jwV
n,~J"wr
jwM"
o
RT + jwJI
o ]
jWM".
n11w' I;;
1'I~ ]
o -nJ~I"w' jwl\J" -nI1w' R' + jwlI I;
(12-31)
The induction machine 549
The si nor rcprcsen (at ions for the sl at or excitations, as given by Eqs.
(12-30), arC' 10 he taken as
Vd = jVd (12-32)
where v~ and VZ a\"(' tim rlltH vahl!'!' of thl' t wo-phai\(' sl al or-port voltages.
If a balalH'cu t.wo-phase supply is used, then \'d = ~.; = V'.
If balanccd phase voltagcs, in the forl\l of (12-30), arc applied to the
two-phasc stator windings, then Ihe stator ('urrcnts will also form a bal
am'ed two-phase set with the salllc frcquency of w rad/scc. In Sec. 10-2
hal anced stational'y C'urrcnts of fl'cqueney ware shown to produce R
!Ilagncti(~ field rot at ing in t h(' air ~ap at all allgular frcquency equal to
win, where n is the nUllIhcr of pole pail'S estahlished by the windings.
The angular spced of the rot at illg Illngnetit, field win is termed the syn
chronous speed of thc mat'hinc. Let us define this synchronous speed as
w
w, = (12-33)
n
In Icrms of the sYfH'hronolls ~p('('d tilt' ,~lilJ of the mlwhine is defined as
(12-34)
whcrc w' is the aetual rotor angular vcloeity and 8 is the dimensionless
quantily tcrmcd "slip." Notil'e that if the rotor angular velocity equals
w., thcn thc slip equals zcro. Also, if the 1'01.01' isstandingstill, then the slip
cquals unity. Thc rotor angular veloeity is given in terms of the slip by
V~]
V'q
[ R' +0 jwl,' R'
o
+ jwl,'
jwilf"
o o
jwM" ]I~l
I:
o -- jwllf" wM"(1 - 8) R' + jwU wU(1 - S) I~
o -wl\f"(l - 8) .iwM" -wU(1 - S) Fl' + jwU I;
(12-36)
(12-37)
If all eurrents are sinusoidal time funetiolls, then the average value of T.
is given by
(12-38)
where He indieat!'s thai only Ih!' real part of the ('ollJplex expression is
retained and th!' • III ("a liS to conjugate the iudi('atcd siuor-eurrenl term.
Rpuwlllbpr that Eq. (l2-:~8) g;ivl's ouly th(" aI'pra(le I'af/l(~ of Ihp torque of
clef'tri('a! origin, Pulsating {'olllpon("nls with zero llveral1:e vllluc {'ould
very ~\'ell exist, hut t h("y nr(" not ine\ tided ill expression (12-:18).
(12-39)
v.
dq =
Vd]
V; (12-40)
V.
+
= V~]
V' (12-41)
Notice that both thed-q voltage matrix and the (+ -) voltage matrix are
composcd of sinor quant ities. The transformation is defined in terms of
complex quantities and is ('ommonly callcd a complex transformation.
The a-b to d-q transformation, defined in See, 10-3, involves real-time
functions for the variables, and hence it is called a real transformation.
The symmetricai-romponent transformation matrix for the stator
quantities is defined by·
.
«+-.d9 = 1[11
V'2 (12-42)
1When used with the d-q variable!!, thill transformation is often ealled 8. jQrvlard
backward (I-b) ·t.ransformation.
SCI' D. C, White and H. H. Woodson, "Electro
mechanical Energy Conversion," ('hap. 4, John Wiley & Sons, Inc., ~cw York, 1959.
The induc:tion machine 551
The complex power supplied to the sl alor ports of the d-q 1ll11ehine, in
terms of 1he si nor quail t Hies, is j!;iven by
P~q = [(Idq)t]*Vaq (12-43)
where V dq is defined by (12-40) and
[(I~q) 'J * = (I~) * (I;) * (12-44)
1--_
For both complex power expressions, (12-43) and (12-45), the real part
represents the watts and the illlnginnry part t,he l'ars being supplied to the
stator ports.
When 1·:q. (12-39) is subst.ituted into Eq. (12-4!i), t,he (+ -) power
beeomes
(12-47)
The eurrents undergo I ransforll1!1.t.ioll in the sallie fUHhioll as the volt.nges;
thus
(12-48)
and b y transposin~, we have
(I~j = (a~_.dqIdq)1 = (Idq)t(a~_.,fq)t (12-49)
where we have used the mat.rix identit.y for the transpose of t,he produet of
two matriees. By conjugation of Eq. (12-49) and substit,ution into
Eq. (12-47) the (+ -) complex power is given, finally, by
P'+_ = [(Idq)ll*[(a~_.dq)'J*a~_.dqV~q (12-50)
By comparing Eq. (12-50) with Eq. (12-43), we ean see that the power
will be invariant under tmnsforlllnt ion, meaning P'+_ = Pdq, if
(12-51)
where 'U is the unit matrix with l's on the main diagonal and O's in all
other positions in the matrix. For the ease of a real-transformation
matrix, such as the a-b to d-q transformal ion, all terms are real and tJ:e
conjugate operation is not required. Thus, for a real transformation,
552 Principles of electromechanical-energy conversion
Eq. (12-51) reduces to the form of Eq. (10-101), which does not have the
conjllgale operation.
Let us now t esl the symmetrical-eolllponent. transformation III at rix,
defined by Eq. (12-42), for power invariance.
,.
[(a:_.&q)'] a~_.dq =
1
y2 [1_j 1]
j X
1
y2 [1
1
·1[1+1 j-j]
= 2 -j+j -p-p
= [~ ~J = ~
Thus the symlllet ri("al-('olllponent I ransforlllal ion III at rix keeps the lotal
complex power invariant under I nlnsforlllat iOIl.
Given I he (+ -) voltage quant ities, the d-q voltage quantit.ies can
readily be found simply by inverting Eq. (12-39), which yields
V~. (12-52)
where
(12-53)
V~ = jVd (12-54)
V·- = ~
V2 (V'd - V!)
•
(12-56)
The induction machine 553
thml
V~ = 0 (12-58)
V~]
V'
q =
[ R' +0jwV o
R' + jwL!
jw.U"
o o
jwM" ] I:Id]
V~ jwM" wMor(l - S) Rr +jwU wII(1 - S) I,j
V; -wM"(l - 8) jwM" -wl./(l - S) R' + jwU I;
(12-59)
where V.j and V; have been reinserted for darity in the following develop
ment. Let us now apply t.he sYlllllletri('al-component transformation,
introduced ill SC('. 12-4, to Eqs. (12-!'iH). The two-phase symmetrical
component transformatioll matrix for the two stator windings is given by
554 Principles of electromechanical-energy conversion
(12-42). The same transformation CI\n also be applied to t.he two rotor
windings; and so, in general, the complete two-phase symmetrical-com
ponent transformation for both stator and rotor windings is given by
.,.
a+_. dq =
1
-\12 ['
1
~
j
-j
0 1
0 1
lJl j
-j
(12-60)
(12--61a)
(12-61b)
where
Rather than just write the results of Eqs. (12-61) to (12-64), let us carry
out these matrix operations in a more detailed manner. By expanding
Eq. (12-62), we have
1 1
V"+ v'2 (V~ + jV;) V: = V'2 (V~ - jV;) (12-66)
Substituting the equations for Vd and V; from (12-.')9) into Eqs. (12-65)
gives us, after some slight rearrangement of terms,
(12-67)
The induction machine 555
1~ =
1
V2 (14 + .11;) 1~ = J2 (14 - .11:) (12-68)
(12-71)
(12-73)
556 Principles of e/ectromechanical.energy conversion
The forcing fundiollsllre zero ill both cuses, since short circuits are assumed
on the rotor ports of the induetion machine aI'! originally proposed in Sec.
12-3.
By c'olllhining Eqll. (12-70) and (12-7a) into matrix form, we have
Eq. (I2-Hlb), where
R'
R' jw(L'-M'") 5 jw(L'-M")
jwM"
jwM'"
(12-75)
where Rc means to retain only the real part. of the complex expression in
square brackets. By using the inverse symmetrical-component trans
(ormatioIl, given by
(12-76)
we can eliminate the direct and quadrature currents in favor of the posi
tive- and negative-se<tuel1(~e currents. Expanding Eq. (12-76) for stator
and rotor quantities gives us
r=~I--=(r+r)
d V2 + (12-78)
(12-80)
(12-81)
iwM"
- ~-'--'-'--'--'----- I'
1-:.. (12-83)
R'I (2 - 8) + jwl.!
When E<)s. (12-82) and (l2-8~) arc solved for 1't. and I:' and the results are
substituted into (12-81), the total average cleetrieal torque becomes
T•.• v • (12-8;"})
Notice that the total average torque of eleetrieal origin is made up of two
terllls. The first term depends on the positive-scquenee rotor curren t,
and t he second term depends on t he negative-sequence rotor eurrent.
Equations (12-;')7) and (12-;)8) give the sequenee voltages for balanced
st at or ex!·it at ions on t he induction Illa!'hine. These equat ions show I,hal
the negative-sequence 8t ator voltage is zero if a UllifOl"lII posit ively rotating
field result s frOlll the d-q stat or voltages. Referring to the eif(~uit of Fig.
12-8, if V'- = 0, then I~ O. Thus, only the first term ill Eq. (12-8.'»
remains and T ...... is a negative quantity. By the sign eonventions we
have been using, a negative eleetrical torque means that the machine is
generat iug a torque in the positive-angular direetion. Thus, the positive
sequence portion of the circuit generates a torque in the po.~itive-angular
direction. An entirely similar argument follows for the negative-sequence
portion of the equivalent. cireuit, wit h the se!'ond term in (12-85) generating
a torque in the negative-angular direction.
(12-86)
and (12-89)
The equivalent, !'ireuit, ('all 1I0W he I'cdrnwn, as shown in Fig. 12-9, with
these resistors divided as indi('ated hy Eqs. (12-89). If the positive
and negative-sequence stator exeit,ations V+ and V'- in Fig. 12-8 are
H'
r l-S
-H 2-S
H'
Fig. 12-9 Complete per-phase sinusoidal steady-state equivalent circuit for the
induction machine.
560 Principles of electromechonical·energy conversion
divided by V2, the four loop currents will also be reduced by a factor of
I/v2, as shown in Fig. 12-9. With the equivalent circuit in this form,
the power dissipated in the two resistors, which are functions of the slip S,
is preeisely the eonverted power per phase, as given by Eq. (12-88). For
balanced stator excitations, Eq. (12-!>7) or Eq. (12-.'i8) indicates that
V~/V2 or V'-/V2 has a magnitude equal to the actual voltage v- applied
to eaeh stator phase for both the two- and three-phase model machines.
Referring to the original sinusoidal steady-state relations for the
induction maehine, given by Eqs. (12-36) in terms of t.hc d-q model, we
can identify the per-phase average power lost in ohmic heating of the
stator willdings as being given by
Pert-phase power loss in stator = !R'(I~I~* + 1;1;*) (12-90)
This POWN quantily is usually given the name stator copper loss. The
real-part operator fie l:3.n be omitted in (12-90), since no imaginary terms
appear. This results from the fa(~t that any sinor times its conjugate is a
real number equal to the magnitude of the sinor squared. If Eqs. (12-77)
are substituted for the direct- and quadrature-stator currents in Eq.
(12-90), the stator loss becomes
In other words, referring t{) the equivalent circuit of Fig. 12-9, the machine
stator copper loss per phase is simply the power dissipated in the two R
resistors by the positive- and negative-sequence stator currents. In an
identical fashion the per-phase rotor copper loss can be shown to be given
by
Example 12-1
A 23lJ1l-volt 75-hp four-pole flH-cps thrl'e-phase wye-connected wound-rotor induction
motor ill <lrivinl/: Il ('Olllltant lond torque at It spe{'(\ of 1741l rpm. Windage losses are
5 per <'ent of tI", lotal input 1){)\V('r. Th" parltlneter~ of the three-phase machine are
as follows:
Find:
a. The machine slip
h. The complete per-phase equivalent circuit
c. The stator line currents and power fal'lor
d. The total stat or nnd rotor ('opper IOllsell
c. The tolal ('onverted po\V('r and the value of th" load torque
f. The eflieiency of th!' motor
Solution: a. Th!' synchronous SPI'C(] of !I. four-pol!' (or two-pole-pn.ir) machine is
b. Thl' complete per-phase equivn.lent cireuit for the induction motor is shown
in Fig. 12-9. The excitations applied to this cir!'uit are the positive- and negative
sequence stator voJtap;es divided hy 0, as shown in Fig. 12-9. With balanced three
phallI' voltages applied to the stator, the equivalent-circuit excitations, according to
Eqs. (l2-57), reduce to
IV'I
~ = V' "" vf:i
2300 = 1330 volts I~\ =0
562 Principles of electromechanical· energy conversion
~
1330[9 j324 .f2 71.B
Fill;. 12-10 Per-phase equivalent circuit for thl' machine in Example 12-1.
Since the negative-sl'qu€ ' nce excitation is zero, for halanced ex('itations, only the
upper half, or positive-sequence half, of Fill:. 12-9 needs to he considered.
The inductance qunntities in the per-phase I'quivalent eircuit are J times the
inductan('e quantities for the three-phase machine. Thus we have
],r =. il'~/lY
= ~ X 0.5"7 = O.RRO h
1 - S J - 0.03 _
R, ";<," = 2.22 X-1fli:f'- = ,1.8 ohms
Thus the ('omplete per-phase equivalent cir('uit is as shown in Fill:. 12-10. For con
velli"Il('I', tlu, phuHIl afll/:le of V·,! 0 i!l tILkr'fi all 7,I'TO.
r. The actual Htator line eurrent is equal to 1-:,1 V2, sirll'e the rwgativc-sequf'nr'e
stator ('urn'nt I ' j 0 is zero. Let us review why this is true. Equations (12-77)
show us that
I~
with I:' = O. Hince the d-q and a-b ma"hine stator quantitil's are id("ntical, we also
have
I~ I:' = I;
No\\' tho:> a phase of the three-phase machine is identi('al with the a phas!' of the t\\'o
phase equivnlcnt machino:>; thus
I'
I'a = I'a = ....±..
y'2
and so the actual stator phase curTo:>nts, or lin!' ('UTrents for the wye connection,
(!'1 ual 1'.1 0.
'I'll" "I1Trents in the o:>quivaient dn'uit or Fig. 12-10 ('an he ohtll.in(>d by a loop
formulation, hut a simpler proredure is to solve for the impc;lan('e seen hy the \'oltage
source. This drivinll:-point impedanee is
The induction machine 563
The line ~urrentl'! nrc 17.4 nmp at a power fadOT of cos 2:1.1 = (U)2 lagging.
d. The per-phase (·opper los.~el! arc simply the power dissipated in the 2.90- and
2.22-ohm r('sistors in Fig. 12-10. The positive-sequence rotor current is given by
r =_
._J~~
r+
V2 (71.8 + 2.22) + j(7.54 + ;)24) ~2
Thus we have
Total stator copper losses =:1 ('0)'/' U· ax (17.4)' x 2.90 = 2630 watts
e. The converted power per phllse is the power dissipated in the output resistor;
therefore, the total converted power is
p=a
• ( r)' ) s
v'2
+ fl'
S
:\ X (lG.6)' X 71." 5!l,aOO wntts
The windage lo...~('l' ar(' I} p('r ('('nt of til(' lotal input power. The total three-phase
input IlOwer i..
P F. = v'3 V 1.1 I, ('Os (J = v':l x 2:100 X 17.4 X 0.92 = sa,800 watts
where VI, nnd h ure the rcspc(,tive line quantities. The total input power can also
be obtuincd by adding the total stator and rotor copper losses to the converted power,
thus giving
PI-' 26ao + IHall + 50,:100 = Ga,766 watts
Taking 5 per cent o( the input power, we have for the windage losses
and so the net output power absorbed by the constant load torque is
PI...d 56,110 8
The rotor speed w' = a, and therefore, hy suhstitulinl!; i~ and i;. we have
- V2 X 16.6 sin (wi nw"t + '1') (12-97)
Sw = (w - nw') (12-98)
Thus I he a-phase rotor rurrelll, which is ident j<'al wit h t hc a-phase rotor
('11 rren t on the three-phase machine, is
X ot ice I hat the [l('t lIal fre<jueney of the indtl('cd rotor currents is
Sw, whieh is comll1only known as slip frequency. At sl andstill the slip
S = 1.0 and the rotor frequeney equals the sl ator applied frequency.
As the rolor speed approa('hes synchrollous speed the frequency of the
:trtllal rotor ('urrents heeomes smaller and approaehes zero as S goes to
zero. The d-q primitive machine provides a ('ollvellient simplificution
hecause. for the d-q model, both the rotor and stator currents are Itt the
Ralllp ('xc·it at ion frPCIIICIl('Y w, t hUB allowinl!; t he usc of simple sinor notation
ill O(l(' ('olllplete p<jllivalelli. (:ir('lIit. If Wp worked direetly from the a-I)
model. we would have' 10 deal with hoth the exeitatioll frequency w und
the rotor slip frequen('y Sw.
The induction machine 565
[HI .up
+ IjiP
R2
JlIp
+LP 2
]il] '~]
i2
(12-100)
Rj N,: N2 R2
v,
EJ~ (al
R, L,-II( R2 L z -II(
v,
9'~F~l
(bl
v,
E;{~'M~
(el
j.. ,,1
'3 1
'1>11
Fig. 12-12 Component equivalent fluxes for the two couplcd coils.
For linear inductances, the total flux linkages in coils 1 and 2 are
given respectively by
(12-102)
I The term "equivalent flux" means that if the flux quantity is multiplied by the
number of turns which it links, the correct flux linkages are obtained.
The induction machine 567
From Fig. 12-11c let us now ('xnmille ! he 1hree indtH't nll('e quantities. \ (,"
By usillJ.!: Eqs. (12-IO;~) and (12-100), \\'(' lmv('
l P"
J. (
1'1 - aM =
N (
1 <Pt!,
+ <P12) _____
N N
! 2.<p12 =
N
~<P11
\ (Jv<\'-:"~
(12-107)
t\ N2 t\ t\
The induetmwe quantity d('lincd hy (12-107) results frolll flux lines whieh
do 110\ link hoth ('oils. These Ilonlilikitlp; lines are (·all(:\rit.~i.:~g~}l.:ux
lines, and 1he ('OlT('spondillp; indu(,t !\Il('C is known as 1he Iral,age in(/.uctance,
which will be designated by n low('f('nse I. Thus the leakagc'Tildtj('tance
of ('oil 1 is defined as
Similarly, by using 1<:<)s. (12-104) and (12-IO;i), we have for the leakage
inductance of ('oil 2
(12-109)
(12-110)
upon using the leakage indu('tall('e definition of (12-108), The sum of the
two equivalent fluxes ill the last IeI'm of Eq, (12-110), ('omlllonly termed
I he mutual flux, is denoted by
~-.~-~
From Fig. 12-12 we ('an see that the Illutual flux <Pm is the total equivalent
flux in the iron ('ore aetually linking hoth eoils. The last term in Eq.
'-(12-1 iO), which can be written as e = N 1 d<Pm/dt, is the voltage induced in
568 Principles of electromechanical-energy conversion
coil 1 by the total flux in the eore_ This voltage p, appears aeross the
inductance aJl in Fig. 12-11c. The current 111 the inductance a:l! is
equal to i l + i 2/a, which ('an be written as
. . ( ai2)
to:' = h - - r.... "') f'\ ~ .1! J
16. )-- - (
(r' / ( : - (12-112)
1 For a more extensive discussion of this form of the transformer equivalent circuit
see, (or I"xample, A. S. Langsdorf, "Theory o( Altcrnating-current Machinery," 2d
ed., chap. I, McGraw-Hili Book Company, New York, HI5S. ,
-,
v aM
R' jwl'
V!
J2
tot al machine ('ore loss per phos". Heartanre quantities are shown for
all of the iuJu!"! ivc clcnwnts. Tlw ,'e:\('tall('('s wZ', wZ', and wall!" are
~enerally known :\s t he stator and rotor leakage reactances and the mag
nf~tizi7lg rl'lldau('(', I"{'spe('t ivC'ly. Xot i{'C' I hal all rotor parameters ('ontain
the t ransfol"llHlt ion ml io squa.reu and :\I"e I hus referred to the stator. Since
th(' positive- and negalive-sequC'IH"C rotor ('urrents are divided by a, all
rolor pOWN!'! nrC' the salllc a,,, with the Ol'igillal drruit of Fig. 12-9.
In !'!UIII nllU'y , t he per-phase <,quivalcnt cireuit. of Fig. 12-14 is more
useful than the dr('uit shown ill Fig. 12-H because the core-loss effects
eall be takC'1l illto ac('oullt. Pl'lwtieally nil of the literature discussing the
induct ion nllwhine uses the leakage-indu(·tancc-lIIutual-ftux equivalent
('ircmit shown in Fig. 12-14 and not the sC'lf-induetance-mutunl-inductance
circuit of Fig. 12-9.
.1.L
120.
V! jX". a
2R,I-S
S
Jf'
jx'
Fig. J2-J!j Approximate per-phase equivaJl'nt ('ireuit for the induction machine
referred to the stator.
The induction machine 571
quantities are zero, and only the upper half of Fig. 12-15 needs to be con
§idered. If ~r.; .. ""0, the output resistor R'(l - 8)/8 becomes very large.
Effectively, l~ is very nearly zel'O, and nil of the st ator current goes
through the no-load branch. The per-phase equivalent circuit therefore
reduces to Fig. 12-lUa.
By measuring t he phase voltage V., phase current I., and per-phase
input power p., t he elements of the no-load branch are computed as
follows. The input power under no-load conditions is mainly absorbed
by windage and fridion and by hysteresis and eddy-current losses; thus
we have
(12-113)
where p.,+! is the combined per-phase windage and friction loss and Ph+< is
the hysteresis and eddy-current loss per phase. l The sinor phase current
I. lags the sinor phase voltage by the angle (J, where
(J
Ph+<
= cos- l _ ,- (12-114)
l.I.
The driving-point admittance of t he no-load equivalent circuit IS given by
(12-115)
I For a simple technique for separating these losses see A. F. Puchstein, T. C. Lloyd,
and A. G. Conrad, "Alternating Current Machinery," 3d ed., chap. 22, John Wiley
.& Sons, Inc., New York, 1954.
The inc/uction machin. 573
No-load lest
Stator voltage (Iine-to-Iine) = 220 volts
No-load stator line current = 16.6 amp
Total power input = 964 watts
Windage and friction losses at synl'hronous speed = 450 watts
Rlocked-rotor test
Stator voltage (Iine-to-line) = 35 volts
Rtator line current = 61.2 amp
Total power input '" 1638 watts
D-c resistance (at rated temperature) oetween any two stator terminals =
0.126 ohm
With rated voltage applied to the stator the op!'n-circuit induced rotor voltage is
measured as 110 volts line to lin!'. Determine the per-phasc approximate equivalent
circuit for the machine. Also, finrl the actual resistance, self-inductance, and the
stator-rotor peak mutual inrluctance for t.he windings of the actual three-phase
machine.
Solution: The pcr-phaRc ('quival{'nt circ'uit for the no-load t{,.8t is shown in Fig.
12-1&. The per-phase power absoroed hy the core-loss resistor RAH is given by
P A+. = 1(964 450) = 171 watts
where we have used Eq. (12-11;». The power-factor angle for Fig. 12-1&, accordill'g
to (12-114), is
171 •
II - coS-I---...---~-- ... 90
(220/0)(16.6)
(220/0)'
RAH --171- '" 94.3 ohms
and for the magnetizing reactance
220/0
X .. - ",aM" = ---.-~--
16.6 sm 90
= 7.65 ohms
The per-phase equivalent circuit for the blocked-rotor test is shown in Fig.
12-16b. The power absorbed oy the circuit per phase is given by
1638
p. - --- = 546 watts
3
which reduces to
1 I. Ph+e
Rh+e = V. cos (J = (V.)2 (12-116)
If the ,otor of the machine is blocked or restrained from rotating, the slip
is equal to unity. Under blocked-rotor conditions, the output resistor in
the positive-sequence half of Fig. 12-15 is equal to zero. The stator and
rotor resistances and leakage reactances have a much lower impedance
than the no-load branch, so under blocked-rotor conditions practically
none of the stator current goes into the no-load branch. The positive
sequence equivalent circuit thus reduces to Fig. 12-16b, where x' and x'
are the stator and rotor leakage reactances equal to wl' and wl', respec
tively. By using a reduced phase voltage adjusted to obtain rated phase
current, we again record the phase voltage V~, phase current I., and per
phase input power p.. The parameters are then obtained-as follows.
The angle by whiclii. lags V. is given by
(J = cos- 1 p. (12-118)
V.I.
From the driving-point impedance of Fig. 12-16b we have
By using a direct current, R' can easily be measured, and aIR' can be
obtained by using (12-119). The two leakage reactances x' and a 2x'
appear only in series in the approximate equivalent circuit, so we do not
have to know how the total leakage reactance is actually distributed__
between the stator and rotor.
Example J 2.2
A 30-hp 22()..volt four-pole three-phase 6O-cps 1725-rpm wye-connected wound-rotor
induction machine gives the following no-load and blocked-rotor test data:
The induction machine 515
Mor 102
The coefficient of coupling hetw(len the stator and rotor when the windings are coaxial
is defined a8
M" 6.8
k - _ r-' - _I - .. 1111.0 per cent
v UTI v 13.5 X a.5
and we can scc that the stator and rotor circuits are very closely coupled.
v R
Fig. 12-11'1 An R-I, circuit in th!' sinusoidal steady
state.
574 Principle, of electromechanical-energy conver5ion
0.083 '~S
Fig. 12-Ji' Approximate ppr-phase equivalent ~ircuit r!'ferred to the sta.tor for
Exampln 12-2.
As an approximation, let us divide the total leakage reactance equally between the
stator and rotor; thus
x' = 0.148 ohm a'x' = 1l.148 ohm
The approxima.te per-phase balanced excitation equivalent circuit is shown in Fig.
12-17, where all resistances and reactances are in ohms.
Assuming that the wound rotor has the same number of poles as the stator, the
transformation ratio is very nearly given by the ratio of the applied stator voltage to
the open-circuit rotor voltage; thus
220
a = ~.~ = 2.0
IlO
Since the excitation frequency is OJ = 377 rad/sec, the stator-rotor peak mutual
inductance for the two-phase model is
The stator and rotor leakage inductances for the two-phase model machine are
respectively given by
From Eqs. (12-108) and (12-109) we can lICe how to convert the mutual inductanee
/If'" and the leakage inductances t, and I' into the stator and rotor self-inductances
576 Principles of electromechanical-energy conversion
Increosing
R
R=-X
Fig. 12-19 A sinor diagram for an R-I, circuit &II R is varied from -.. to + GO.
The indtx:tion machine 577
lmoginorv
axis
v! Reol
-J2 oxis
J
Fig. 12-20 Positive-sequence pt>r-phase eirel!' diagram for t,he indu!'tion machine.
negative-resistance l·egioll. Noti<'<l t hat. the eireuit. absorbs power only at.
a lagging power factor and supplies power only at a leading power factor.
The steady-state approximat.e per-phase equivalent eireuit. for the
induction maehine, shown in Fig. 12-1!i, can he analyzed by a similar
procedure. The sinor cirde diagram of Fig. 12-20 represents the positive
sequence or upper half of the circuit of Fig. 12-11), The positive-sequence
stator voltage is taken as referenee, and it is drawn to a convenient voltage
scale as sinor OJ. The no-load sinor current. I~, drawn to a convenient
current scale, is shown as OA, lagging V~/V2 by angle AOJ. The com
ponent OK, in phase with V~/V2, is the current through Rh+.; the com
ponent 01... at 1f/2 rad behind V~/V2 is the eurrent through jX.. in Fig.
12-15. The remaining portion of the cir<,uit in Fig. 12-15 is similar to the
example circuit shown in Fig. 12-18.
As the slip S varies from minus to plus infinity, the total resistance
will take on all values from minus to plus infinit.y. Thus, the tip of the
positive-sequence rotor current I~ must traverse a circle with t.he slip as a
parameter. The diameter of this drde equals V~/V2 (x' + a'xr) and is
578 Principles of electromechonicol-energy conversion
(12-122)
moving the circle to the tip of I~ allows the addition of currents to be made
in a convenient fashion. Point B represents a particular value of slip
between zero and positive unity. Sinor -I~/v2 a is given by AB, and
the stator current I~/V2 is given by OB. The power-factor angle is
BOJ and is lagging, since I~/V2lags VVV2.
When the slip equals zero, operation is at point A. Values of slip
between 0 and + 1 result in operation on the sector ABH of the circle. For
values of slip hetween 1 and 00 operation takes place between Hand M.
Thus, nil positive values of slip are Joeated 011 t.he sedor ABHM. All
negative values for the slip are located Oil the remainder of the circle,
eoverillg the sedor A P!If going frolll 0 to - 00. Once the slip has been
determined, the solution for the equivalent-cireuit currents is a simple
mat ter of using the circle diagram.
A good deal more information than just t.he current solutions is
available from the circle diagram of Fig. 12-20. The component of the
no-load current which is in phase with V~/V2 is given by OK. When
this component is multiplied hy the magnitude of the stator voltage
1!~/v2, the product equals the watts dissipated in the positive-sequence
resistor Rh+< in Fig. 12-1.1. We could, therefore, set up a wattage scale
which, in effect, would perform the multiplication by the constant magni
tude V~/V2. Applying this scale to the in-phase components (i.e., in
phase with the voltage V~/V2) would yield various powers directly.
As an example, suppose the rotor is blocked or restrained from turn
ing. Therefore, operation is at point H, where S 1.0. The sinor OH
represents the stator current IVV2. The in-phase component of OH,
given by FH, is proportional to the per-phase power supplied to the
positive-sequence portion of the equivalent eircuit. ~\feasuring the length
FH with our wattage ruler would give this per-phase power directly in
watts. Since w' = 0, none of this power can be coming out of the mechani
cal port. Also, FG = OK is the power absorbed by Rh+<. Therefore GH
represents the per-phase power absorbed by R' and a 2R'. The power
dissipated in R' and a 2R' is the per-phase positive-sequence contribution
to the stator and rotor copper losses, respectively. Thus, GH is the
positive-sequence contribution to the per-phase losses in the stator and
rotor" windings due to ohmic heating. If point Q is so positioned on GH
that
GQ R' (12-123)
Gil = R' + a2R'
The induction machine 579
then GQ represents the stator copper loss and QH would be the rotor
copper 1088. This statement is readily proved, since
Then
GQ ex: ( vi;l' )2 R
a (12-12.'j)
(12-127)
where
Zl = [(R- + a2Rr) + {x' + a 2.rr)2Jh
2
(12-128)
(12-129)
580 Principles of electromechanical-energy conversion
(l2-1:l0)
Thus, the ]Pllgth /)l~' is proportional to the stator pIllS rotor eoppcr losscs.
II owpvpr, si II!'€'
DY DE
(12-1:H)
flQ (/H
we ('un writ€', using Eq. (12-12:;),
I
/)}' OQ U'
(12-1:~2)
f) E (; Jl = /(7+" oHF
Ttwfefore, D Y rcprespnts 1he posit ive-sequcrwe per-phase contrihut ion
1081 utor I'opppr 10001S(,S and Y E f(>preRenls thc <,ontrihul ion to rotor copper
lOHHl's.
For t hc opcrat ion at I hc valuc of slip represented hy point B on the
eire-Ie diagram of Fig. 12-20, the foIlowing positive-sequenee quantities
{'an he summarized. All lengths 011 the diagram are to he measured with
a suitahle scale calihratpd in volts, ampcres, or watts depending on what
the lellgt h represents.
Th!' quantity Y B is often terllled til(' power acro,ss the air gap.
Let us consid('1' one other operatinl!; eondition on the diagram of Fig.
12-20. Suppose the machine is opcmtinl!; at a negative slip such as at
point P. Rememh('r I hat a Il('gativ(' slip occurs whcn CUT > CU" or the
rotor is turni ng at a gl'CI\1 er anl!;ular velori t y than t he synchronous speed
for the presrribed cxdtation. The positive-sequence stator current. is
now given hy 01". Angle PO.! is t he positive-sequence powcr-fact.or
an/l:le, and for this value of slip it is grel\ter than1f'/2 rad. Thus, energy
flow is out of the posilive-scqllcnee st alor t.erminals. The magnitude
of Ihis genernt('d output po\\'('r is giv(,H hy I'll, the eOlllponent of the
stator ('urrellt in phase with the stator voltage. An induction machine
/l:enemt i ng eled ric power is known as an induction generator.
The power fador at which \.hi8 power is heinl!; delivered ('un he deter
mill('d hy taking I he lllagnit.\Ide of IIIC! eosine of angle POJ. However,
since the power factor is llsually a positive (!llantity, t.he sense of either
I~/V2 or V~/V2 !lIlIst he reversed. In either ease, -I~/V2 leads
V~/V2 or I~/V2 I('ads -V~/V2 hy angle POX. The cosine of angle
POX cquals Illinus the ('osinc of anglc POJ and represents the leading
power fador at whi('h power PH is delivered to the infinite bus. Froll1
the diagralll of Fig. 12-20, we see 1hal POWN ('an he delivered only at a
leading power faetor.
The distance NT represents t he power absorbed by RH •. Equation
(12-130) shows, for operat.ion nt point. B 011 the circle, Ihat the distance
measured 011 a line p'amllcl to t hc applied voltage V~/V2 from t.he diam
eter A N to t hc 8 = 1.0 line A Il is equal to the st.ator-plus-rotor eopper
losscs. The proof leading 10 thc result of Eq. (12-1:m) is entirely general
for allY ot.her opemting poin1 011 I he eirde. Thus, for the induction
generator ('ase we arc 1I0W ('ollsidcring, with operation at point P, the
dist ance 'I'll', measured hy the wal t age scale, gives t.he stator-plus-rotor
('opper losscs. By again using Eq. (12-12::\), the following relation can be
developed:
TlI all R' (12-133)
1'W = il/J iV+ a2 f{r
Thereby, the copper losses ar.:' divid!'d with Tll being the stator copper
loss and F I\' t he rotor loss.
A get\('ml rule in lIsing 1he drde diagram ean be lIlade corwerning the
S = 1.0 line, given hy A II. The dis1 :1nc'(' along a Jille parallel to V~/V'2
582 Principles of electromechanical-energy conversion
from All to any operating point on the cirde represents t,he posiHve
sequence converted power per phase P:",. For values of slip between
zero and unity, corresponding to rotor angular velocities from synchrollous
speed to standstill, the circle lies to the right of the line All and the con
v"rteu power iH froll! deetrieal to Illechanir~al. Vor all other vl~lues of
slip the eirele lies to t he left of line A H and the converted power is from
mechanical to elce! rieal. For operat ing conditions 011 the portion of the
circle below the horizontal line FH the ('opper losses can still be divided
between the stator and rotor by extending the lines AQ and All. The
intersection of these extended lines with a horizontal line drawn through
the operating slip on the eirde gives these eopper losses as distances
measured with the wattage seale.
~Il currents derived by using the circle diagram of Fig. 12-20 are for
the positive-sequence portion of the equivalent eircuit of Fig. 12-15.
The power quantit ies taken from this circle diagram represent positive
sequCIH'e eontributions to the per-phase quantity. For example, the
Imaginary
axis
Reol
oxis
J'
Direction of
increa~inq slip
Fig. 12-21 Negative-sequence per-phase ('ircle diagram for the induction machine.
The induction machine 583
Vd = jVd v; = V; (12-136)
V·- =L
y2 ('"d - V')
If
(12-137)
T,~ = - Sw 1'2)2 +
nRr ( J' nRr
(2 - S)w
( J' )'
y72 (12-140)
T.~ =
2w[ (R- + a R' I S)
2 --:+---T(x'-c:-·-+:---a7"2x---")C;:;2]
2
(12-144)
and
(12-145)
allgular diredion. Similarly, Eq. (12-14;), for T-;~, represents the per
phaf;e portion trying to drive the rotor in the negative-angular direction.
Let us begin our investigation into the chll.raeteristics of these gen
erated torques by assuming a balanced stator excitation producing a
positively revolving uniform lIJagnetic field. Sueh all excitatioll corre
sponds to the statement that
,..
q (12-146)
Substituting this balanced condition into Eqs. (12-144) and (12-145)
gives us
na 2 R'(V·)2jSw
(12-147)
(H'+ a 2H' ls) 2 + (x' +~x~')2
1J;~ = 0 (12-148)
Thus, for balaneed excitation in the positive direction an electrical
torque only in the positive direction results. A typical plot of Eq.
(12-147) is shown ill Fig. 12-22, where rt~ is plotted as a funetion of the
slip.
For values of slip between zero and unity the rotor angular velocity
is positive and less than the sYIH'hrollous speed w,. Since r:..
also is
positivr~ for these values of slip, the machine is operating as an induetion
lIlotor eonverting eleetriC' energy to meehanical energy, as indicated in
Fig. 12-22. If the slip is greater than unity, the rotor angular velocity
must be negative. The electrical torque, however, is still positive;
hence power is being supplied to the mechanical port. An examination of
1.0
s
the circle diagram of Fig. 12-20 shows t.hat with V~ = 0 and t he slip
greater than unity, power is also being supplied to the eleetrieal ports.
Thus, with energy flowing into all ports, the maehine is operating asa
brake (~onverting both 1lIC1'luUlieal and elccl rical energy into hCltt ene~gy.
For negative values of slip the rotor angular velocity is positive and greater
than synehronous speed. With the electrieal t.orque negative, power is
again being supplied to the mechanical port. However, with the slip
being negative, energy is flowing out of the eleetrical ports and" the
machine is operating as an indu<:tioH generator. The circle diagram of
Fig. 12-20 shows that for values ~f slIp just siightly negative no power
will flow out of the machine terminals owing to absorption by the core
loss resistor. Also, for negative slips beyond point Z in Fig. 12-20, no
power aetually flows out the electrieal ports. However, conversion is
still from mechanical to electrical form.
With the balaneed positive exeitation, proposed by Eq. (12-146), Q1~.
starting torque ('an easily he found silllply hy setting S = 1.0 in Eq.
(12-"147), which gives us
G f )~ 10 (12-149)
Since T;. is zero for the positive balan('ed excitation, Eq. (12-149)
represents the complete per-phase average starting torque generated by
the indudion ma(,hine.
The maxilllum torque the llIaehine ('an generate can be found by
first findiJ;ith~ "Slip ;t" ~hich this nHtXillllllll torque OC(~urs. This slip (,an
he dcic~;;~illed i)y setting the slope of the torque-slip ('urve equal to zero
and solving for the slip that satisfies the zero-slope ('oudition. Thus,
upon setting dT:./dS = 0, the slip at whieh maximum torque O('ClirS is
found to he
(12-150)
Equation (12-HiO) indi('ates two valu('s, one positive and the other
negative, for t he slip Itt whieh IllaxillHlIll torque is developed. These two
peaks are shown ill Fig. 12-22 at !"qual distauees from the torque axis.
Upon suhstituting the two values for the slip at maximum torque into
Rq. (12-147), we have as thE' vahl(,s for the maximulI\ torques
(12-151)
588 Principle, of e/edromec:hanical·energy conversion
Notice that the negative peak is larger than the positive peak, since the
denominator of the second expression is smaller than that of the first
expression.
Substituting 8 m ., from Eq. (12-1.'i0), into Eqs. (12-151) gives the two
torque peaks as
Since Eqs. (12-1!i2) do not contain the rotor resistance HT, the peak torque
must be independent of this quantity. Equation (12-1.'>0) shows that the
faetor b,2HT/S.., must remain constant for fixed values of H', X', and xTi
therefore, varying the rotor resistance can only cause the slip at which the
maximum torque occurs to change. However, th~y:aluELQft!!~~rl!\x}nw~
torque remains constant. Figure 12-23 shows a set of torque-slip curves
for various values of rotor resistance. The magnitudes of the two peaks
are constant" but the slip at whi(~h they occur increases in magnitude as
the rotor resistance increases.
For eurve c the peak occurs at a slip of 1.0 or at starting. Thus, for
one partieular value of rotor resistance the machine generates maximum
torque at start. Selting 8 m,1.0 in Eq. (12-1.')0) gives this particular
rotor resistance as
iw'=o
i
a
b
c
s l
Fig. 12·2:! (lcm'rated "Iectrical torque for various values of the rotor rcsistance 1\8
a function of slip.
The indudion mcxhine 589
v; = - V: = V· (12-154)
Referring to Eqs. (12-144) and (12-141}), the electrically generated torque
per phase becomes
T:. = 0 (12-155)
T _ == _....~-~..._---_._,,_.--
na 2Rr(V.p/w(2
.....----~-.-~-
S)
(12-156)
.. [fl· + a 2 Rr/(2 - 8)]' + (x' + a 2;cr)2
.
Now the positive-sequence component. of t.he torque is zero and the nega
tive-sequence component is, in general, nonzero. Remember that posi
tive values of the negative-sequence torque are directed in the negative
angular direction. Equation (12-1;)6) is identical with Eq. (12-147) for
the balanced positively revolving condit.ion if 2 - S is substituted for S.
By the definition of slip, given by (12-34),
w. - (-w r )
2 - S = ---~-- (12-157)
w.
Thus the examinat.ion of the negative case will give results identical with
those for the positively revolving case if t.he sense of the rotor angular
velocity wr is reversed. Equation (12-157) shows us that reversing the
assumed direetion of wr would require the substitution of 2 - S for S in
all of the previous rfsults. ConsequenUy, all conclusions for the balanced
positive case carry over for the balanced-negative case.
590 'rinciplea of electromechanical·energy conyersion
Exomple 12-3
Thf' ~O-hp 220-volt four-polf' t,hree-phal!f' 6O-(:ps 1725-rpm wye-connected wound
rotnr inrludion machine rliscuRiled in Exampl" 12-2 has th" following per-phase
"'luivlll"nt-rirruit pararnet"rs all r.,r"rr"rl to the Rtator:
which reduces to
(220/V%LQ
~Di6:f"-f-(j.lls:$f+ R6~I~fif+-o~f4S)
(220/'\1'3)/0 (22Ul'\.i:i)/U
Therefore, the total stator inrush current in each phase, which iM also the line current
in a wye connection, is simply the 8um of these two currents, as is seen from Fig.
12-17. Thus the inrush current at start is
n(l;»a'R'
T~
Sw
na'/l'( l~)' 2 X U.Ol~:l X (;J.'!5)'
05.2 newton-Ill /phMC
Sw 1 X 377
The induction machine 591
S O.OR:J 0275
d. Using Eq. (12-IS:J) the maximum torque of a51 newton-m will occur at start
provided
with T;. = O. The normal operating range for the induction machine is
usually at small values of slip, or near synchronous speed. On examining
Fig. 12-22, we can see that for small positive slip values the machine is
very nearly a (~onstant-speed device over a large range of generated torque.
For small values of slip Eq. (12-l!iS) is given approximately by
T .. = n(Vep
wa 2 Rr
S = -k8 for small slip values (12-159)
wr = (1 - 8) w (12-161 )
n
The induction machine 593
and so
W
pw r = - -n p8 (12-162)
- ~ .IpS
n
+ ~n D(1 - S) - IlkS = Tr (12-163)
~
n D(1 - So) + To = "kSo (12-166)
(12-168)
where
.I
l' = -~~-- seconds (12-169)
D + nllk/w
594 Principles of e/eclromechonical·energy conversion
..
(a)
I~
V2
.', - -,--r-
-n-rr-n-TTII£'
-' -
(bl
Fill:. 12-24 (al Slip as a function of time following an incremental step increase in
the load torque on an induction machine. (b) The stator phase current following the
step increase in load torque.
Figure 12~24a shows the tDtal slip as a function of time. For l < 0
the slip 8 = 8 0 • After the increased load torque is applied the slip
exponentially increases to a slightly greater value. From the positive
sequence portion of the equivalent circuit shown in Fig. 12-15, we can see
that as 8 increases, the output resistor a 2R'(1 - 8)/8 decreases in value.
The value of the stator~phase current IVv'2 will therefore increase
exponentially, as shown in Fig. 12~24b. Since the slip is a known function
of time, the magnitude of the currcnts in the equivalent circuit can be
calculated at each time instant. The frequency of the current in Fig.
12·24b has been rcdu('cd for clarity ill drawing the figure. Actually, the
current would go through many more eydes during the transient time
intervaL
Equation (12·11)9) is a linear approximation to the tDrque-slip char
acteristic valid in the normal operating region near zero slip. If the
machine is operating in a different incremental region (i.e., where the slip
is not close to zero), then the constant kin (12-11)9) can be suitably altered
to correspond to the slope of the torque-slip {'urve around the new oper
The incluction machine 595
where T; is the total generated torque tending to drive the machine in the
direction of positive speed and Ti. is a passive load torque. The quantity
~ T is therefore the difference between the generated torque and the torque
absorbed by the load. Figure 12-25a shows the generated tcrque plotted
--('
Load
torque
(T[)
o
(a)
w'
(b)
w'
wr(<o) ----- ?-'---
r
W,
(12-t71)
na 2RrS(V R + V cP/2w
T. = - T; + T; = - (SF+' a2Rr)2+ S2(x'+a%xr)2
na 2 R'(2 - S)(V R - VeF/2w
+ [(2 - sTJ{' + a2R'p +(2 ":::-syt(X~~+-~(i2x'2) (12-175)
-r. = r;- r.
Positive -sequence torque ( r;J
Fill:. 12-27 Total servomotor torque of electrical orill:in as a function of slip for a
positive value of the rontrol-phas(, voltall:c less than the reference-phase voltage.
charaet crist iI· plot ted as a funel ion of slip. Thc IIcgal ivc-sequcnee
chara<'1eristie is idclltieal ill form with two exccptions:
Figure 12-27 Hhows a typi('al plol of Ihc tol al posit ivc- and negative
scqllmll'e clc('tri('altol'Clll<!R for a se'rVO!ll{)tor. The' rolor resishuwe is large
enough 10 haVf~ Ihe' peak positivc-seqllelwe torquc o('(~ur at a slip greater
t han tinily. The reason for slH'h a de'sign wiII he darified latcr in I.his
sel'lioll. The' ('\Irve'!, in Fig. 12-27 arc drawn on t he assumption I hal
the ('onl rol-phasc volt age V c is less I han the referclH'c-phasc voltage V H.
Thus at ,wro spI'cd, 01' ,"; = 1.0, II\(' posit ive-seqllcl\('c torquc is greater
than the ncgative-sequell!'c torq\le and the' lotal steady-state torqlle is
ill I he posit ive-allgular dire(~tioll.
The servomotor is usually used as H positioning devi('c, so the char
acter of t he torque ('urves arOUlld zero Ilpef'd is of greatest interest. For
vahlf's of slip very Ileal' unity, the denominators of t.he two terms in
(12-17;i) are hoth very nearly given hy th(' per-phase blocked-rotor
impedance of the machine. From Fig. 12-161> the per-phase hlocked
rotor il1lpedan('e has It Illagnitude equal to
(12-171)
For Inrgf' 1'0101' rl'sislalH'e'1'I a 2 U' is the· dominant term in the denominators
of (l2-17.'i); so as long as the slip is fairly near to unity, Eq. (12-176) is a.
The induct ion machine 599
Figure 12-28 is a plot of Eq. (12-177) for vnriollS valu('s of ttl(' eontrol
phage voltage V(~. The lleglLtive of the total torque of elC(~tri(·:tl Ql·jgin
'It.
- 7'. is ploH ed !I."I a flllletiOll of speed to ('orrespolld to Fig. 12-27. Remem
ber that a negative torq\w of c\('("\ rit'al origin means that the machine is
-T"
FiR. 12-28 Servomotor torque-speed ",hara",teristies for various values of the con
trol-phase volta~c, MIIullling the motor speed is much less than synchronouB speed, or
S ... 1.0.
600 Principles of electromechanical-energy conversion
gen('rat ing (t torque ill t he posit ive 8pc('d direct Ion. The torque generated
hy Ihe lIl(t('hine is linearly relotN} to the eontrol-phase voltage at w' O.
The 1110101' r:ieclricfll dmnpiny cor.flicicnl J)"" given hy (12-178), defining
t h(' slope of I h(' t orqu('-8pc('d ('urves, i8 prael.ieally independen t, of the
{'ontrol-phase voltage as long as ~·c« V R. However, for larger valll(,s of
Vc the slopes ilH'rease roughly as shown ill Fig. 12-28.
The trunsi('nt operation of the servomolor ('an he studied by using
I he aSstllllplioll I hat the eieetrieal transiellls arc very much faster than
Ihe lIl('('hanical t mllsients. Th(' rotor-port equalioll for t he servomotor
is given by
where
sgn w' +1 for w' > 0
(12-182)
sgn w' -1 for w' < 0
Substituting Eq. (12-177) for the st('ariy-state generated torque into
(12-180) gi ve8 IlS
(l2-18~)
__ ~pt
or
/{_........ ..
Tp+l -~.-~
....
Output
indicating
device
+
P. .=... E
S~ide
wire
L -_ _ _ _ _ _ . _ _ _ _ _ _ _ _ _
~
~
FiR. 12-30 Hchemntic repres('ntlttion for D. nnsic closed-loop servo-instrumentation
system.
and or is the shaft angular position, in radians. For small values of the
control-phase voltage, the slight. dependence of D ... upon V c can be
neglected.
One of the most frequent uses for the two-phase servomotor is in the
system of Fig. 12-30. In the figure are shown the essential features of a
closed-loop serm)-instrumentation system. The output voltage of the
transducer I'. is relatcd 10 the physiea\ va.riablc being instrumented. The
potcnl iomct('r PI, whieh is usually known ItS a slide-wire, is connected to a
constant d-c voltap;e source B. The slidc-wire output voltage Vow is sub
tracted from t he transducer output 111, and the difference is supplied to the
d-c to a-c (~Ollverter. This converter is a synchronously driven vibrator
or choppcr, which is csscntially a double-polc double-throw switch. The
choppcr output, is very nearly a square wave with a magnitude propor
tional to its d-c input voltage. The frequency of the square wave
depcnds on the line frequency applied to the chopper driving coil, which
controls thc rate at which the switch vibrates.
The chopper output is supplied through a transformer to a high-gain
a-c amplifier. The amplifier output is ('onnected to the control phase of a
two-phase servomotor. The rcference phase of the motor is usually con
nected in series with a capacitor across the same a-c line that drives the
The induction machi,.. 603
where
wn 2 = liQ~ . KKfh (rad/secp (12-187)
T
(12-188)
For a step change in the transducer voltage 1'1, the optimum syst.em
response is achieved with a damping ratio r = 0.6. The bandwidth of
t,he system is limited by the und:ullped natural frequClwy Wn. Notice
that increasing the amplifier gain K" improves t.he bandwidth but
decreases the damping ratio. For too small a damping ratio t.he system
has excessive overshoot and an oseillatory response.
or
relates the slide-wire output voltage v.w to the radian position of the motor
shaft 9". The transfer function of the two-phase motor is taken from
Fig. 12-29 with Tr = O. The closed-loop transfer function, from Fig.
12-31, can be put into the form
(12-186)
VI
604 Principle, of elecfromechonical·energy conversion
I ,+ =
I'
- = VI~2 (12-192)
(12-193)
Thus, the sum of the two sequence-stator sources is simply the sinor
voltage applied to the single stator winding of the actual single-phase
machine. The per-phase equivalent circuit of the single-phase machine
(built from a two-phase model) is shown in Fig. 12-33b. Using Eq.
(12-192), we have
I'+
(12-194)
v2
which is half of the actual current in the single-phase stator winding.
The induction machine 60S
R' lx'
v! (if ~
.f2 R,..• jX", J2a
.f2
v~ (if
.f2 Rh+# jX", ~
.f2a
.f2
jX", ~
J2a
v;
jX",
~
J2a
R' jx'
(1) )
Fiji;. 12-:J:3 (al Per-phase equivalent circuit rOT the two-phlUle induction machine.
(II) Two-phR.se equivR.lent. dreuit ,",odifi"d to irwludl' thl' Miflll;ll'-phlUlC ('onstraint.
The equivalellt. ('in'uil shown ill Fig. 12-:3;~1) contains the eOllstraint.s
for 1he single-phase ·maehinc. However, we IllUst remember t hat, the
eir('uit was ('ollstrut'led from a I wo-phase model. Therefore, all powel'
(\uantities eal('ulatcd by using Fig. 12-a:~b arc still per phase and must be
multipli~d by 2 to obtain the total quantity for the actual siilgle-phase
machine. For example, the ele('\rie power supplied to the circuit of Fig.
12-33b is given hy
(12-195)
Y , = 0.024IL::37
Zr = -YI , + -Y,
1 1
= ---------
O.024t/. -37
I
+-----------
(l.1R7L::-!iIl...:'>
= 46.,;/39.2
-
111\/11
2 X46.:'\L:~9~2 4.1I4/-:l1l.2
The magnitude of the stator C'urrent is 4.!)4 amp at a power fa!'tor eflual to ('Os 39.2
n.771l lagJl;ing.
h. The voltages aeross the upper an.1 lower halves of thl' efluivalent C'ircuit in
FiJI;. 12·34 arc givl'n re,spectively hy
V, =-y-;--:'y;IIIl/O = 102/ 1$
= 16.4 watts
The total stator ('opper IORSes arc ohtainC'd hy douhlinJl; the per.phase loases; thuR we
have
r. The per-phase 1'0nvI'I'ted powl'r is simply the power dissipated in the two
output resistors of the equivalent circuit shown in Fig. 12-34. The total converted
606 Principles of electromechanico/.energy converrion
Example 12-4
A t-hp lIS-volt four-pole single-phase 6O-cps induction motor is driving a constant
load torque at rated voltage with a slip S = 0.05. The per-phase equivalent-eircuit
parameters referred to the stator are as follows:
X .. = 69.0 ohms RA+. 508 ohms
R' = 1.97 ohms a'R' = 2.8 ohms
x' = 2.3 ohms atz' = 2.3 ohms
The tota~ friction and windage losses at this speed are 15 watts. Determine the fol
lowing qJantities:
a. The fltator current and power factor
b. The total stator and rotor copper losses
c. The total converted power and the total torque of electrical origin
d. The total core loss
e. The output power and overall efficiency
Solution: a. At a slip S 0.05 the two output resistors in the equivalent circuit
of Fig. 12-33b are calculated to be
1 - S I - 0.05
a'R' - S - = 2.8 X 0.05 = 53.2 ohms
r;*1
~
2 V 508 j69 53.2
I'
115LQ
T--
Vz
I
508 j69
~ -1.36
1 -
p. = 2P•• - 2 [ l,'a'R'-- - II"a"R'-
8
S
2-8
1 - S]
2[(1.76)1 X 53.2 - (2.29)1 X 1.36) = 316 watts
Notice that the negative-sequence converted power subtracts from the positive
sequence converted power. The torque of electrical origin has a magnitude equal to
the converted power divided by the rotor speed, or
P 316
d. The per-phase core loss is equal to the power dissipated in the two R~+.
resistors. Thus the total core 1088 is
t. The output power delivered hy the motor is equal to the converted power
minus the windage and friction 10Mes. Thus we have
P ou • 301
'I - .. -~ - 68 5 per cent
Pi. 440 .
As a check, we can add the core loss, total copper los5, and the windage and friction
1088es to the output power and thus get
which is within slide-rule accuracy of the 440 watts previously calculated as the
input power.
From Eq. (12-86), the torque of elect.rical origin generated per phase
is given by
nRr nRr
- 2Sw (I~->2 + 2(2 _ S)w (I~.Y (12-196)
direetion, so that the slip is no longer unity, the output resistances are no
longer equal and I: "F I:. A net. torque is generated and the machine
can operate successfully as a motor.
Single-phase motors can be made to develop a starting torque by
placing a second winding on the stator in quadrature with the main
single-phase winding. A machine of this type is known as a split-phase
motor. If the impedance of the auxiliary, or starting, winding is different
from that of the main winding, then the two winding currents will be out
of phase when the windings are placed across the same single-phase
supply. The phase displacement in the winding currents roughly sim
ulates thc operation of a two-phasc mot.or connected to an unbalanced
two-phase supply. A starting t.orque is thereby generated, and after a
sufficient speed has been reached t.he starting winding can be open-cir
cuited and the machine operates in a true single-phase mode. The
starting winding is usually disconnected by a centrifugal switch set to
operate when t.he rotor is revolving at nearly synchronous speed.
The phase displacement of the current in the starting winding can
also be aehieved by plaeing a eapacitor in series with the starting winding.
Such a ma<:hine is called !\ capacitor-start induction motor. Usually the
starting winding and series ('apadtor are open-circuited again by means
of a centrifugal switeh set to the proper speed. In some single-phase
machines t.he capacitor and starting winding are not open-circuited.
Such a machine is known as a permanent-split-capacitor induction motor.
For small, low-efficiency, inexpensive single-phase motors, such as
those used in small household fans, a salient-pole stator is used. A small
shorted copper loop is placed around a portion of the stator poles. The
resulting dissymmetry in the magnetic circuit gives the machine a small
starting t.orque. The copper loop is known as a shading coil, and so the
machine is called a ~haded-pole motor.l
All of these starting arrangements obtain a rotating field from a
single-phase supply either by unbalancing the winding impedances, as in
the split-field and capacitor-start machines, or by unbalancing the mag
netic circuit as in the shaded-pole machine.
12-15 SUMMARY
A d-q primitive-machine model has been used as the basis for analyzing
the operation of the induct.ion machine. The d-q model relates to an.
equivalent a-b model representing a two-phase induction machine. With
I For a more extensive discussion of starting arrangements see A. F. Puchlltein,
F. C. Lloyd, and A. G. Conrad, "Alternating Current Machines," chap. 28, John
Wiley &: Sons, Inc., New York, 1954.
610 Principle. of electromechonical-energy convenion
PROBLEMS
12-1 For the following in<ludinn machine~ fin,l the synchronou~ Ilpeed in
radianll per second and rpm:
a. I6-pole 60-cps three-phase
b. 16-pole GO-cps two-phase
c. 6-pble 50-cps three-phase
d. B-pole 25-cps three-phase
12-2 What ill the highest synchronous "peed that 25- and 6O-cps induction
machines have in common? How many poles must eaeh machine have?
12-4 Perform the matrix transformation indicated by Eq. (12-64). The d-q
impedance matrix is given in 02-5Q), and the complete symmetri(~al-componcnt
transformation appears in 02-60). The (+ -) impedance matrix is given by
Eq. 02-74}.
12-rf(A I ~hp
IIO-volt two-phase two-pole wound-rotor induction motor has
two identl II. s tor windings and two identical rotor windings. The machine h8.8 a
smooth nonsali!' air gap. The winding parameters are as follows; R' 6 ohms,
I,' = 1.22 h, fl' = f# ohms, and 11 = 1.23 h. The mutual inductance between the
a-axis stator windings and a-axis rotor windings is 1.20 cos a henry, where a is the
angle between these two axes.
The two rotor windings are short..;)ircuited, and the stator windings are excited
by voltage sources as follows:
torque for the machine. Is the converted enerl/:y Howing out of or into the mechanical
rl
.
port? What are the total copper losses with this new ('xcitation?
12-6 A IS-hp 440-volt 1i0-eps four-pole three-phase wound-rotor wye-connected
'-'i~uetion ma!'hine at standstill has a line-to-Iine open-circuit rotor terminal voltage
which is twiC'e the applied line-to-line stat.or voltal/:e. With the rotor circuit open,
the rotor shaft is driven at 1600 rpm in the direction of the rotating field. Rated
voltage at rated frequency is being applied to the stator.
a. Write suitable expressions for the open-!'ircuit voltages appearing across each
pair of slip rings.
b. At what speed should the rotor he driven to deliver a rotor frequency of 80 cps?
c. If the rotor is driven at )R()() rplll in the direction of the rotating field, what
will be the magnitude and frequency of the line-to-Iine induced rotor voltage?
d. If the rotor is driven at ) ROO rpm in the direction opposite to that of the rotat
ing field, what will be the map:nitude and frequency of the induced rotor voltage?
12-7 A 20-hp 60-cps three-phase four-pole induction motor has a full-load
speed of 1723 rpm. Assuming windal/:e and friction 10ssC8 are 5 per cent of the out
put power, determine the total rotor copper losses.
12-8 A commercial power transformer has the following name-plate data:
1200 V /120 V (Primary/Secondary)
96KVA
60 cps
a. With no load on .the secondary windinp: of the transformer find the magnitude
of the primary current with rated primary voltage applied.
b. Under the conditions of part a find the power snpplied to the primary.
c. On the basis of the approximate equivalent circuit, calculate the primary plu8
secondary ohmic losses for a secondary current of SUO amp.
d. On the basis of the approximate equivalent circuit, find the magnitude of the
primary voltage required to deliver 96 kva to an O.8-power-factor-lagging load, with'
rated secondary voltage across the load.
e. Calculate the self-inductance of the primary and secondary windings and the
mutual inductance between the two windings.
No-load test
Stator voltage (line to line) = 220 volts
Blocked-Towr test
Stator voltage (line to line) "" 34.3 volts
Htator line ('urrl'nt 14.5 amp
Total power input = 710 wattll
At rated temperature, the d-e resistance between any two stator terminals is 0.48
ohm. With rated volta!!;e applied to the stator the open-circuit induced rotor voltage
is 220 volts line to line.
a. Determine the per-phase approximate equivalent circuit for the machine.
b. Find the resistance, self-inductance, and stator-rotor peak mutual inductance
for the windin!!;s of the actual three-phase machine.
c. Calculate the starting inrush current with rated voltage applied to the motor
and the rotor terminals short-circuited.
d. How much inductive reactance should be added to each stator phase to limit
the starting inrush current to 65 amp?
~2-1b A 220-volt three-phase 6O-cps wye-connected induction motor has a doc
resistance of o.:n ohm between any two stator terminals. With no external shaft
load the total power input ill 380 watts and the line current is 12 amp at rated stator
excitation. When the applied volta!!;e ill reduced to 40 volts, the total power input
decreases to 90 watts with a line current of 4.5 amp. At this reduced voltage, the
hysteresis and eddy-eurrent losses may be considered negligible. Compute the total
and per-phalle hyst!'r!'sis and I'ddy-current losses (or core losses) and the total windage
and friction losses.
12-11 A 5O-hp 440-volt 6O-cps four-pole three-phase induction motor draws
9.12 amp and 2134 watts at no load when operated at rated voltage and frequency.
With the stator windin!!;s wye-eonnected:
a. Calculate the elements in the no-load portion of the per-phase equivalent
circuit, assumin!!; a resistor and inductor in parallel.
b. Calculate the elements in the no-load branch of the per-phase equivalent cir
cuit, assuming a resistor and inductor to be connected in series.
c. Repeat parts a and b for stator windings which are dclta-connected.
12-12 a. From Eq. (12-147) show that the per-phase developed power with
balanced~phase stator excitation can be put into the form
y~ a'W(l - S)(V'pIS
P,~ = [R,+"(i~R'+aq{'(C~-s)7sr'+-(x'-+-a'i;f'
b. Show that this expression for P,~ can be manipulated into the form
No-load test
Blocked-rotor test
20 5.7 RO
40 11 .5 24U
6U 17.3 610
80 23.0 117U
lUll 2R.7 1881l
The stator resistance gives a voltage drop of 10 volts hetween terminals with 16
amp doc flowing.
a. From the no-load te.8t data plot the line current and total power input as
functions of the line voltage. By extending the watts ve. voltage curve to the zero
voltage axis, the windage.and friction losses ('an be obtained as the watts input at
zero voltage. Find the windage and friction loss('8 and diSl'uss the reason why this
method is valid.
b. From the blocked-rotor test data plot the lin(' "urn'nt and total power input
as functions of the line voltage. CaJ.-ulate the power-fal·tor angle and also plot a
curve of 1 cos 8, the component of the line eurrent in phase with the line voltage, as
a function of the line voltage. Extrapolate th(' two ('urrent ('urves to the rated voltage
axis. The blocked-rotor power factor at rated voltage "an thus be obtained as the
ratio of the in-phase current to the total line current both taken at rated voltage.
c. By using the techniques described in parts a and b calculate all per-phase
approximate equivalent-circuit parameters.
d. Draw the complete circle diagram for the machine operating at rated voltage.
Show values of slip at equai intervals around the circle.
e. From the circle diagram plot a developed torque VB. slip characteristic curve.
f. For the value of slip where the machine is delivering maximum power oper
ating as lln indu('tion generator, find the mal'hine spel'd in rpm, the total power
delivered, the hystl'r~sis and. eddy-current losses, the total stator and rotor copper
losses, the line current, the power factor, and the overall effi('iency. Use the circle
diagram, with current and wattage rulers, for obtaining these data.
616 Principles 01 eledrome<:honico/-energy conversion
12-16 For the motor of Prob. 12-14 operating at rated voltage and frequency
"alculate:
a. The starting inrush current and power factor
b. The starting torque
c. The speed, in rpm, at which maximum torque occurs
d. The maximum torque
f!. The additional rotor resistance (referred to the stator) required to have maxi
mum torque available for starting the machine
12-17 For the three-phase induction machine of Prob. 12-14, find the self
induetanee per phase for the stator and rotor and the peak stator-rotor mutual induct
ance. Find L:'/I.." L;'/I.." and M::/I,,' Draw the self- and mutual inductance per-phase
equivalent circuit similar to Fig. 12-9. Assume x' - a'x' and that the transfor
mation ratio a ... 2.0. Can a hysteresis and eddy-eurrent 1088 resistor R' H be placed
in this form of the equivalent circuit? Explain your answer.
4
12-18 By using the approximate equivalent circuit for the induction machine,
neglecting the stator resistance, show that
where T is the developed torque at any value of slip 8, T ma. is the maximum torque
the machine can generate, and 8 ..., is the slip at which maximum torque occurs.
12-19 A 2300-volt 1000-hp 60-eps 12-pole wye-eonnected wound-rotor induction
motor has the following per-phase parameters: R' = 0.10 ohm, Rr = 0.034 ohm,
l' - 1.14 mh, and I' = 0.35 mho At rated voltage the no-load current is 75 amp and
the power drawn is 75 kw. The transformation ratio a - 2.0.
a. Select convenient. voltage, current, and power Bcales and draw a complete
circle diagram for the machine.
The induction machine 617
b. Locate thE'S - J.O point on the cir('le, and from the diagram determine the
blocked·rotor line current, power ftl.('tor, ('ore 10SII, stator cop))!'r loss, rotor cop))!'r
loss, and developed power.
c. Repeat part b for S = 0.5.
12-20 A polyphase induction motor has a total per-phase leakage reactance
X' + a'x' equal to 1.5 timE's th(' total ))!'r-ph!lll(, relli"tancE' Jl. + a'R'. The rotor
resistance per phase, referred to the stator, is equal to the stator rrsi~t.an('e prr phase.
At what slip does maximum torque occur?
12-21 A 20-hp 220-volt 60..,·ps eight-pole two-phase induction motor has the
following per-phase approximatt' equivalent-cireuit parameters all referred to the
stator:
For each of the following connections draw the pl'r-phase approximate equivalent
rircuit. Rhow all parameter values and voltage exl'itations. Assume the motor
speed to hI' 840 rpm. Cal('ulate the line "urrent in rAch of the stator phaees, the total
dl'veloped pOWE'r, and the total devclo))!'d torque.
a. The two 8tator phascs are connl'cted to a halanced two-phtl.8l' infinite-hull
voltage Bupply delivering rated voltage at rated frequency.
b. One stator phase re('eives rated voltage at rated frl'qlll'n(1Y wit.h the ot.her
stator phasc short-('ireuited.
c. Same conne('tion as part II elWl'pt that th(' 8000nd phase is now open.."ircuited,
thus giving single-phll.8e operation.
12-22 The developed torque VII. slip curve for an induction motor running
near synchronous speed ('an he IlJ)proximat.rd hy a straight line. Thus,
T. '" -kS
where thl' usual sign convention for 1', is used. The ma('hine is driving a constant
load torque of 100 newton-m under steady-state conditions when, at t "'" 0, the load
is suddenly redueed to SO newton-m. The parameters are as follows: k = 200
newton-m, J = 2 kg-m' (shaft plus load inertia), D .. 0 (neglect windage), anfl
"'. = 500 rad/see. Find the slip as a function of time for all t ;:: o.
a. Plot the torque-speed characteristic for the machine. Use mks units of
newton-meters and radians per second. respectively. (Consult Table A-I, Conver
sion factors.)
b. The ma('hine is coupled to a constant load torque of 16.2 Ib-ft. The total
inertia of the load and the machine rotor is 50 Ib-fV. By using the graphical method
discusscd in Sel'. 12-12, obtain a plot of the speed as a function of time when the
motor is started.
c. Write an inl'remental linear differential equation similar to (12-167) that is
valid in the neighborhood of the steady-state operating speed of part b. Assume the
windage coefficient D to be zero. If the load torque is now increased by 2.0 newton-m
at t 0, derive the incremental slip S,(t) and write an expression for the total slip
as a function of time valid for all t ;::: O.
12-24 A 3-hp 440-volt 60-cps four-pole two-phase wye-conneeted wound-rotor
induction machine has the following per-phase equivalent-circuit parameters all
ref erred to fhe stator:
R' 2.2 ohms fl,,, '" 300 ohms
X' = 4.8 ohms X .. = 100 ohms
a'R' = 2.1 ohms a = 0.5
a'x' '" 4.2 ohms
The machine is to be used as a servomotor.
a. Compute thc additional rotor resistance per phase required to make the slip
at maximum torque S ... be equal to 1.5. Why must S .., he set at such a value?
b. For rotor speeds near zero, determine an approximate expression for the total
developed torque of this servomotor as a function of the shaft speed and the control
phase voltage. Assume rated voltage on the reference phase, and the total rotor
resistance 1I~ determined in part a.
12-25 A IllJ-volt 400-cps two-phase servomotor has the torque-speed charac
teristic shown in Fig. P12-25. The rotor moment of inertia is 3 X 10- 1 kg-m'. The
Speed. rpm
Fig. P12-25
The induction machine 619
12-27 A single-phase induction motor is rated at t hp, 110 volts, 60 cps, 4.3
amp, and 3420 rpm. No-load and hlocked-rotor test data are as follows:
N o-loa,J. lesl
BIBLIOGRAPHY
Energy-state-function formulation
Goldstein, H.: "Classical Mechanics," Addison-Wesley
Publishing Company, Inc., Reading, Mass., 1950.
Guillemin, E. A.: "Introductory Circuit Theory," John
Wiley & Sons, Inc., New York, 1953.
McCuskey, S. W.: "An Introduction to Advanced
Dynamics," Addison-Wesley Publishing Company,
Inc., Reading, Mass., 1959.
White, D. C., and H. H. Woodson: "Electromechanical
Energy Conversion," John Wiley & Sons, Inc., New
York, 1959.
Whittaker, E. T.: "A Treatise on the Analytic Dynamics
of Particles and Rigid Bodies," 4th ed., Cambridge
University Press, London, 1937.
Linear-8Yste'TfUJ theory
Cheng, D. K.: "Analysis of Linear Systems," Addison
Wesley Publishing Company, Inc., Reading, Mass.,
1959.
Gardner, M. F., and J. L. Barnes: "Transients in Linear
Systems," John Wiley & Sons, Inc., New York, 1942.
LePage, W. R.: "Complex Variables and the Laplace
Transform for Engineers," McGraw-Hill Book
Company, New York, 1961.
Lynch, W. A., and J. G. Truxal: "Signals and Systems in
Electrical Engineering," McGraw-Hill Book Com
623 pany, New York, 1962.
624 Principles of electromechanical-energy conversion
ANSWERS TO PROBLEMS
d. C ( 2vm) =
gmT
e. W:(v) = gm (-1
a
+ e-'" + av). d. W.(v) gm (1 _ e-'" - ave-.')
a
b. Q. T., Q. = -To, Qa = 0
Aa 2 Ab·
2-11 a. We
I l'
TtClAe2
l'
+ TtC.Ad', W", = 2L. +
b. .c = W; - W... c. No. d. Ac A., Ad = A" ~a A. - ~2' 4 ~2
2-13 b.:J = 975 watts (approx), :J' 625 watts (approx)
c. D8 0.113 newton-m-sec/rad, iDI!W2 1390 watts
3-3 a. A
f.' ON2hr/Ji 2
- 2g
+ f.'oN hr(a
2g
- 0) di
-dt + Rr vet)
i)x
4"<0 G+ x r
A~·
3-II b. W'•. = 6 (x.•
c. 'U
3aA~~ bA2
3-13 a. .c
4
+ -2 (x c)' + iM.t'
x-c
c. AU 2bA'(x c)3 + ~ + D.t -Mg
j-aA -%>, + b>,(x - 0)4 + 4bA(x - C)3.t + GA = i(t)
9 • Xl' . 3X.·X.'h2;:'
3-15 a. W. TtX.X. 2 A.' + x,
A.A. + 4
oW;(~,x)
c. lei = - , same results a.s for b
ox.
X
4-11 a. = 2.5/ -50 0 or 7.94 db, -50 0 phase. b. 12.06 db, -50 0
I --
c. -12.06 db, 130" or -230°. d. 12.06 db, -140°
4-13 a. get) (i + fe-at + J/'r 6e- ')u(t)
= U 5
10-15 a. All poles in LHP, stahle. b. Two poles in RHP, unstahle. c. Two
poles in RHP, unstable
240
6-7 c. turns/m. d. K is independent of the rotor's angular position
,..2a
6-11 a. 0.446 ohm. b. 2.18 ohms. c. 974 turns/m. d. Lr 0.048 henry,
L' 0.573 henry
e. 0.143 henry. f. Gd~ G;d = 0.573 henry, G';d = G'J. = 0.048 henry. g. 0,
144 volts
7-1 a. Uniform at 45°. b. Same as part a. c. Fields cancel
d. Uniform, revolving CCW at l/n rad/sec. e. Same as part d
7-3 a. V~D = v~o 20 volts, VdO 124 volts, v~u -90 volts
f. 1 voltage or 1 current
8-1 a. 119 volts. b. 0.202 henry. c. 109 volts, 77 volts. d. 6980 watts
8-3 a. 75 kw has 30.4 per cent, 150 kw has 69.6 pel' cent, 222 volts
9-3 a. Gal = 0.408 henry, Ga. = 0.545 millihenry. b. 125 volts (cum.),
109 volts (diff.). c. 92.2 per cent (cum.), 91.2 per cent (diff.)
9-5 b. 133 + 66.6[1 + 4.85(0.le-: 10 ' 0.306e- 3 · 21t )]u(t) volts
c. 10.2 amp. e. 5870: 1
9-6 b. 16.4 volts, change 0.027 rad/sec decrease
r1~1(8) 11.2
[Stable]
C. V Rl(8) = 1.38 X 10- 38 3 + 0.04898 2 + 0.4518 + 12.2
10-5 b. L~ L; = t henry
12-11 a. R k .;.. = 90.7 ohms, X", = 29.2 ohms. b. R h +< 8.55 ohms, X", =
26.6 ohms. c. RA+o = 272 ohms, X .. = 87.6 ohms (parallel), R,,;. 25.7 ohms,
79.8 ohms (series) ,
12-13 a. 102 watts. c. RII+< = 203 ohms, X,n = 15.5 ohms, R' = 0.31 ohm,
a~Rr = 0.43 ohm, x' + aZxr = 1.89 ohms
632 Principles of eleclromechanical-energy conversion
'12-15 a. 0.0145, 1183 rpm. b. 81.1 amp. c. 85.3 amp, OJI56 lagging
d. 2480 newton-m. e. 91.7 per cent