An Analysis DC-Machine Commutation: Japan
An Analysis DC-Machine Commutation: Japan
An Analysis DC-Machine Commutation: Japan
5, 1974
TRlrlPkd from Ilhnld Cpldcoi R o n h ~ h i ,Vol. 94B, No. 10, October 1974, pp. 479-486
T . MATSUDA
Muroran Institute of Technology
75
First, let us construct a tree of the commuta-
tion circuit in Fig. 1by picldng up only the volt-
age sources and resistance branches. We repre-
sent the cutset corresponding to this tree by
F=rFVL FVLF0R L ]
] = ~ R L
FCLFcr
FrI 0
which gives tbe following commutation equation: cumnt & d t y ( A / c m ' )
Ng. 3. Dynamic bar-brush contact voltage vs.
I v (f)+FVLiL(f
)=O
I (3)
current characteristics of a brush.
in(f)+FnLIi(f)+FRlIl(f)=O
I r (t)+FrLIL(t)=O I
Ur.(f)-FvrTUv(f)-FRL7UR(f)
-FFrLrDr (1) =O 1 (4) Equation (7) is an (n + 2) x (n + 2) diagonal matrix
I)I(f)-FRIruR (f)=o i and k]n+ll/m if,ag (n + 2)-dimensional vector
with elements gj , j = 1, 2, ...
n + 2. Accord-
For notations V, R, r, L and I, refer to Fig. 1. ing to the present author's experiment, the dynamic
The matrix with symbol T represents the trans- contact voltage vs. current characteristics of the
pose of the matrix. For instance, FVLrepre- brush are not very dependent on the dc component
sents the comction relationship between the volt- of the brush current and K is kept almost constant
age source branch and inductance branch. A simi- [9]. Therefore, we assume that K is expressed in
lar statement holds for other F matrices. iv, iR, terms of the whole cmtact area resistance RB of
ir, i1, vv, VR, V r and VI represent the current the b w h o r the contact resistance for the rated
and voltage vectors of the corresponding elements. brush current IB flowing uniformly over the whole
contact area. The brush contact voltage VB for the
The circuit equations for individual branches rated brush current IB therefore is expressed aa
are described below.
Vs=leRe=27l~ Re
(1) Resistance branch =K ( 2 7 l ~ / A ) " ~ (8)
It is difficult to express accurately the contact where Y = Ig/2 IF and IF is the rated armature
voltage vs. current characteristics of the ca*n current flowing through all the parallel armature
brush. They vary with the brush materials, air circuits. Consequently, we obtain
humidity, commutator segment temperature, con-
tact pressure, commutator segment velocity, etc. ~/A''"=(2ylr)1-"'n)R~ (9)
[?I. According to the experimental results, the
contact voltage is a nonlinear functlon of current Substituting Eqs. (6), ( 7 ) and (9) into Eq. (51, We
density and its static (or dc) characteristics are obtain
different from the dynamic (or ripple) character-
istics. The most typical dynamic characteristics UR (I) =(2 71 # ) l - ( l ' a )[ $Rh~i r ) sts""' (10)
are illustrated in Rg. 2. For the method of mea-
suring these characteristics, refer to section 5.2. which can be used to represent the contact voltage
The brush current density pulsates with time and vs. current characteristics of arbitrary brush
therefore contact voltage VR is expressed as material by assigning suitable values to m, y and
RB. The circuit equation for the resistance branch
,=
U R ( ~ R1) K [g(t)]ntz"'" (5) of the short-circuited coil is represented by
where m is a constant equal to o r greater than Ur (t)=Grir(1) (11)
unity. Although, rigidly speaking, Kin Eq. (5) is
a function of current density, we assume here that where & denotes an n x n diagonal matrix with
it is a constant [8]; g is the (n + 2)-dimensional element l/rj.
contact current density vector and is expressed 88
(2) Inductance brush
0 =LR.t.ir/A (6)
Assuming that both the self- and the mutual
where A is the total c a t a c t area of the brush and inductances of the short-circuited coil are constant
T = t/Tc is time variable normalized to commuta- during each switching interval, we express the cir-
tor segment period Tc and cuit equation for the inductance branch as follows :
*-2
9 8 =hdiag[- 1 - 11 )......, 1
-'-
D L ( f )= L d l r ( t ) / d t (12)
r ' 0.5+s'
- . 1 1 where L is the inductance matrix in each switching
GI (7)
0.5+U-r' interval.
76
(3) Voltap source branch right-hand side of Eq. (17) represents the commu-
tation voltage, the second term the voltage between
The electromotive force induced in the short- commutator segments and the third term, the volt-
circuited coil (or equivalently in the commutated age drop across coil resistance.
cdU is represented by a voltage source. We call
this electromotive force the commutation volt- 3.2 Contact voltage between brush and commu-
age. Assuming further that the magnetic flux tator segment and voltage between commu-
produced by the interpole is proportiad to the tator segments
armature current, we express the commutation
voltage vector as follows: It is reported [lo, 11J that a spark discharge
occurs when the contact voltage between the brush
DV( t )= a [( 1 +k)ec-e.) +ec (13) and commutator segment exceeds about 3V. We
have also confirmed this fact experimentally. It
where (Y = I/IF, ec is the electromotive force in- is very important for design engineers to predict
duced by the interpole flux which in turn is pro- the contact voltage between the brush and commu-
duced by the armature current, Ba is the electro- tator segment. In the conventional design method,
motive force induced by the armature reaction, k only the average reactance voltage has been con-
is the ratio of magnetomotive force induced by sidered in designing the mechanical strength of the
the interpole auxiliary winding to magnetomotive interpole and the type of armature winding. If the
force induced by the interpole main winding and contact voltage between the brush and commutator
eL is the electromotive force induced by the main- segment can be predicted accurately, then it be-
pole leakage flux. comes possible to design the shape of the interpole,
the type of winding, the number of commutator
segments covered by each brush, the magnitude
and waveform of commutation voltage- more accur-
L d ~ ( 1 ) = F v( ~
1) ' ~ +
FRL'OR(t)
v
ately.
dt
+FrL'Dr(t) (14) Contact voltage VR between the brush and comu-
tator segment can be expressed in terms of nor-
Equation (10) and the second expression of Eq. (3) malized coil current x as follows:
give
D R ( f ) = ( 2 r ) - ( ' / " ' V ~ ( % f-FRLX
(
D R ( t ) = (27 I F ) ~ - ( ' / ' " ' R B{[-B
FRR L~L(~) -FRIC)) I t 2 1 / m (18)
- F R / ~)ntzl'"'
I] (15)
similarly, voltage vs(t) between the commutator
Substitution of Eq. (15) into Eq. (14) yields segments is given by
us(t)= (2 r)-('/'")V BF R L ' ( ~ R ( - F R L X
tdM=FvLruv(t)
dt -FRI'C)]n+21~m (19)
+FRL'(27 IF)^-"'"') R E
X (SR{ -FRLi~( t ) -FBI ) It1""' Equations (18)and (19) are useful for calculating
-FrL'%FrLiL(t) (16) the time behavior of a ( t )and v,(t).
77
m 5. Numerical Example and Experimental Results
To test the validity of the theoretical analysis
Read in dr- in the preceding sections, we have measured the
time behavior of the short-circuit current and con-
tact voltage between the brush and commutator
Set UP coanxmt.Eq. segment of a 3-kW test machine. The short-cir-
cuit current is measured using a search coil.
5.1 Test machine
Read in initial
estim.va1.XI A separately excited dc machine was tested.
of hush cur.6 Its ratings are as follows: 3 kW, 100 V, 30 A,
coanmnt. volt. 1500 rpm, 2 poles, 1/2 slot pitch winding, 3
coils/slot, slot-type electric carbon brush (12.5
x 20 x 32 mmq, 2 brushes/pole, p = 2.5. The
search cull for Short-circuit current measurement
is arranged as shownin Ng. 5. The time rate of
short-circuit current variation can be measured
by observing the voltage waveform appearing across
slip rings 2 and 3. An auxiliary brush is fixed to
one of the two parallel connected brushes to mea-
sure the contact voltage between the commutator
segment and tail edge of the brush. The auxiliary
brush is insulated from the main brush by means
of the brush holder so that they'are not short-cir-
cuited. The contact area of the main brush is little
affected by fixing the auxiliary brush. An auxiliary
winding is also fixed to the interpole to adjust the
Fig. 4. Flow chart for solving the differential
magnetic flux intensity produced by the interpole.
commutation equations. Machine constants and test conditions are summar-
ized in Table 1.
5.2 Determination of machine constants
illustrated in Fig. 4. The short-circuit current Most important quantities for setting up com-
produced by the average reactance voltage, which mutation equation are the contact voltage vs. cur-
is used as the first initial value, consists of the rent characteristics of the brush, inductance
steady-state component and the transient compo- matrices of the short-circuited coil and commu-
nent. Since the steady-state component of the tation voltage. In the present study, the contact
short-circuit current is the true initial value being voltage vs. current characteristics of the brush
sought, the initial value used for the first itera- are determined experimentally and other two quan-
tive calculation has an error equal to the transient tities are determined by the numerical calculation.
component of the short-circuit current. Since the
commutation circuit comprises only resistances The measured contact voltage vs. current char-
and inductances, the transient term attenuates acteristics are shown in Ng. 3. The measuring
with time. According to our experience, the device is also shown in Ng. 7 . Solid curves in
transient component attenuates after about three Fig. 3 are measured by superposing a half-cycle
iterations and 5 o r 6 iterations are sufficient to wave of 50-Ha sinusoidal current upon a steady-
reduce the discrepancy between the initial and the state dc current of 8 A/cm2. The short-circuit
final values below 10-5 times the armature cur- current is measured at the negative brush through
rent. Therefore the number of iterations has little which the commutated current flows in the direc-
effect on the total computing time. The number of tion from the commutator segment to the brush.
short-circuited coils or the dimension of commu- A l l the curves in Flg. 3 have also been measured
tation equation is the most important among sev- at the negative brush. The half-cycle wave of the
eral factors affecting the total computing time. 50-HZ sinusoidal current superposed upon the
steady-state component has an amplitude of 34
A/cm2 for curve (a), 20 A/cm2 for curve (b) and
Generally speaking, the commutation circuit
varies from one switching interval to another. .
14 A/cm2 for curve (c) Section 0 s of these curves
is measured by superposing a negative ripple cur-
Therefore it is necessary to change the commu- rent upon the dc current. Those values of K, m
tation equation each time the switching interval and RB which are derived from curve (b) in Fig. 3
changes. The time behavior of the short-circuit using Eqs. (5) and 8 are as follows:
current over major interval is determined ulti-
mately when the Short-circuit current has been Inductance matrix of the short-circuited coil is
calculated for all switching intervals. derived using equations given in Langsdorf's book
78
Machine C O M ~ E U I ~ ~ Values Machine ~ o ~ t a n t s Values
Test speed (rpm) 1.200 Duration of regular switching
Brush pressure (gr/cm2) 260 interval (ma) 0
Commutator temperature ("C) about 65 Duration of major interval (ms) 1.085
Segment pitch (mm) 5 Maximum inductance of short-
Commutation band (mm) 40.9 cimuited Coil Ls (CCS) 8.747
Commutation period T (ms) 1.809 Coil resistance (m) 1.7
Segment period Tc (ma) 0.723 RB (a) 0.023
Duration of singular switch- m 1.41
ing interval (ms) 0.362 P 2.5
To oscilloscope
Commutator se
3 Slfpdnp
5. Measurement of commutating coil current. 8 H :Metallic gmphite brush
Mg.
FYg. 7. Measurement of dynamic bar-brush con-
tact characteristics.
tact area
contact area
3mm of
F&. 6. Auxiliary brush for measuring the voltage
drop between brush and bar.
79
leakage thrx-
Fig. 9. Oscillograms of e.m.f.8 induced in com-
mutated coil.
kr0.5
ZIS
d - r) I
I I
Z0
d "
-15
I
0 1 2 3 4 5
t/ Tc
I I I I I $ Solid curves: me=. , dashed curves: cdc.
Fig. 13. Computed and measured curves of bar to
brush voltage drop at the trailing edge of brush.
80
case, the transient behavior of the interpole flux,
rotor speed and commutation voltage should be
known by some means.
Similarly to the conventional methods, the pro-
posed method is also effective in analyzing the
effects of the stator shape, types of armature
winding, the number of commutator segments is
covered by each brush and the kinds of brush
a5 ms/div. materials on the commutation characteristics.
Top: k = 0.25, Bottom: k = 0.5 For such an analysis, i t is important to analyze
accurately the magnetic flux distribution in the
Fig. 14. Oscillograms of bar to brush voltage interpole air gap and the contact voltage charac-
drop at the trailing edge of brush. teristics of the brush in advance [15, 16 1.
The general commutation equation derived in
this paper is useful for studying the commutation
6. Conclusions phenomena from various standpoints. They have
already been applied successfully to the analysis
~n this paper, we have proposed an iterative of no-spark commutation zone [l, 17 1.
state variable method for analyzing the commuta-
tion characteristics of dc machines. The calcu-
lated results agree very well with the measured A chowledgement
ones. The proposed method has the following ad-
vantages over the conventional methods: The author thanks Prof. M. Naito of Hokkaido
University for his guidance under 1971 research
(1) It has become possible to analyze syste- scholarship of Ministry of Education, and Prof. H.
matically the commutation characteristics of any Fujihara of Hokkaido University and Prof. K.
lap-winding dc machine. Origasa of Muroran Institute of Technology for
their valuable comments on this work. Digital
(2) The contact voltage vs. current characteris- computation presented in this paper has been car-
tics of any brush material can be taken into ac- ried out at Computer Center of Hokkaido Univers-
count. ity.
81
15. A.D. Moore. Trans. h e r . Inst. E h . 17. Matauda. 1974Hokkaido Branch Conv. Four
Engrs., VOl. 69, pt. 11, p. 1615, 1950. Inst. Elec. Engrs., Japan, No. 46.
16. S.V. Ahamed, et al. IEEE Trans. Pwr.
App. Syst., V O ~ . PAS-85, NO. 9, p. 960, Sublnitted June 1, 1973;
1966. resubmitted May 25,1974
T . MATSUDA (Member)
Muroran Institute of Technology
82