11I

FIGURING IT OUT
PART 1 -INTRODUCING NETWORKS
By Owen Bishop

Here is a new series to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We wiU explain the underlying electronic theory and, whenever
appropriate, ren der some insights into the mathematics involved.

A network consists of two or to the contrary, we assume that

1"'"l.more impedances joined to- element branch the conductors themselves have
gether by conductors so as to negligible resistance. A conse-
perform a particular function. In quence of this is that the poten-
the network of Fig. 1 the impe- tial at one end ef a eonductor 1S
dan ces comprise three resistcrs, exactly equal to the potential at
a capacitor, and an inductor.joined its other end. We also assurne
in one ofthe many configurations that there are noincidentalsources
possible with this particular aa- of ernf such as might occur by
sortment of components. 'I'his "
r : electromagnetic induction from
network happene to be a COID- :: -----'-il:~element
t:t..::.: :..:.:.•..:.''':' :..:..:.•..:.:..::.: :...:':" :.:.!.' '.:..:.: :.::..:.:..:. :.: :::.::.:. •..:. :':':".:.:J::
adjacent networks or from ther-
plete circuit so it also includes moelectrie effects at thejunctions
a source of electromotive force of dissimilar conductors. Ir any
mesh
(emf, E). But this is not an es- such effects are thought to oeeur,
loop
sential component of a network. they must be represented on the
---....-- node
The emf may be derived from network diagram as eurrent or
930010·1·11
another connected network. voltage sources ofspeeified mag-
Figure 1 iUustrates the more nitude.
irnportant terms we use for de- We begin the study of ana-
scribing networks: Fig. 1. Terms used 10 describe a nelwork. lytieal techniques by confining
our attention to resistive im-
Elements:the resistors, capac- pedanee. Resistors are the most
itcrs and other itemsjoined into usual origin of this; they are de-
the network. liberately included in the net-
workbecause they have this prop-
Node (orjunction): a point where a erty. Resistance may also be ex-
two 01' more elements arejoined. erted byotherdevicessuch as in-
ductors, diodes and cells.
930010·'·12
Branch: a connection between The more important equa-
one node and another, usually tions relevant to reeistance are
containing one or more elements. i1lustrated in Fig. 2.

Loop: a closed path through The definition of the ohm:

the network. In Fig. I, the dashed b R1
lines indicate one of the many U/I=R, [Eq. 1]
930010·1·13
Ioops in this network.
where U is the potential differ-
Mesh: a loopwhich does not have ence (in volts, V). I is the cur-
any branches linking different rent(in amperes, A),andR is the
parts of it. In Fig. I, the dotted resistance (in ohms, Q).
lines in dicate all 4 meshesofthis c Note that the arrow repre-
network. senting the potential difference
(pd) across the resistance is di-
Network analysis rected towards the left. 'I'his in-
dicates that the leftend ofthe re-
The aim of analysing a rietwerk sistance is at a higher potential
is to predict the potential dif- (more positive) than the right
ferenees (pd) that develop be- end. Note, too, that the current
930010·1·14
tweenits nodes and the currents 1S shown flowing from the end
that flow in its branches under at higher potential to the end
the influenee of internal or ex- at lower potential. In almost an
ternal emfs. Unless it is stated Fig. 2. lIIuslraling Ihe resislance equations. instances in this series eurrent

ELEKTOR ELECTRONICS JANUARY 1993

 FIGURING IT OUT - PART I

is taken to be conventional CUl'- Mesh-star

rent, from positive to negative, transformation
not electron flow from negative
to positive.

...
AJ}C ce...
"cO A)RAB c
In some networks it is useful to
be able to replace one configu-
Resistances in series:
., !Dia" (") ration of resistors with another
configuration thathas the same
[Eq.2] Bo-«........
------' B B resistive properties. A commonly
D D used tactic is torepIace the mesh
93001~1-15
Resistances in parallel: connection by the equivalent star
connection in Fig. 5. Both net-
lfR=lfR1+ lfRi+ lfR3+" .[Eq. 3] works have three terminals and
For two resistances in parallel Fig. 3. Network reduction. we have to replace Fig. 5a with
Eq. 3 simplifies to Fig. 5b while retaining the same
resistances between each pair
[Eq.4] ofterminals.
First of all consider the sim-
plest ca se where a11 three re-
Network reduction A
sistorsofFig. 5ahaveequal value,
If several resistances in a net- RM. and are to be replaced by

• •
work can be represented by a three resistors of equal value
single rcsistance. the equ iva-
lent resistance, subsequent
~
C
~
~
R
AB RS. Using the same equations
as in the previous paragraph, we
calculations are made much sim- C find the equivalent resistance be-
pler. Reducing the complexity 0
~ "oE tween any pair of'terminals, say,
ofa network by substitutingequiv- between A and B. Redrawing
alent resistances wherever pos- B B B the mesh diagram. while keep-
E E
sible is an important technique ingthe connections the same, we
93001~1-16
in network analysis. can see in Fig. 6 that there are
As an example, take the net- two resistors of value RM in se-
work ofFig. 3a. The aim is to find ries. Theirequivalentresistance
a single resistor which, connected Fig. 4. Network reduction - another example. 1S 2RM. This pair 1S in parallel
between terminals A and B, has with a single resistor and equa-
a resistance equivalent to that of tion 4 gives:
the three resistors shown. The
problem is solved in stages. The
first stage is to replace the 4 Q MESH STAR
and 2 Q resistors by a single Ao-----., AO-----,

..
equivalent reaistor, ReD. Equa-
tion 2 gives In the star configuration, the
resistor connected to terminal
RCD=4+2=6 n. C may be ignored when we are
considering the resistance be-
The equivalent network in tween A and B. There are just
Fig. 3b at this stage has two 6 n two resistors in series between
B 0-----"'---,1
resistors in parallel. Equation 4 Aand B:
gives
CO-------------~ CO-------------~ [Eq. 6]
93001~1-17

If Fig. 5b is to be a replace-
Figure 3c shows the final re- Fig. 5. Mesh-star transformation with equal resistances. ment for Fig. 5a, the value of
sult: the equivalent resistance is
3n. For a range of p.d.s. applied
across A and B, tbe correspond-
ing currents flowing from A to
Bare unaffected by whether the
AO---, AO----_,
network conaists of the three AO-----,
original resistors or is replaced
by a single 3 Q resistor.
Another example is shown in
Fig. 4a. RDEis the equivalent of
10 n and 30 n in parallel (Eq. 4):

RDE=lOx30/(10+30)=7.5 n. B 0---4----f:=}----4 B~-"-

This gives Fig. 4b. The equiv- C 0----- __ ----' co----------'

alent resistance of this is 20 Q, Bo---~
Süfz and 7.5 nin series; by Eq. 2: cO-------' 93001~1-19

93001ß.1-18

Fig. 6. Redrawn version of Fig. 5a. Fig. 7. Mesh-star transformation with unequal resistances.

ELEKTOR ELECTRONICS JANUARY 1993

 GENERAL INTEREST

RAB must be the same in both.

Combining equations 5 and 6:

Ao-~t----, Ao-------, AO----,

Since tbe networks ar-e sym-

metrical, the same result is ob-
tained when we take pairs of
terminale A and C or Band C.
Summing up, if the three re-
sistors are reduced to Olle third
their original value, the mesh
of Fig. 5a may be replaeed by
the star of Fig. 5b. Conversely,
a star network can be replaced B 0--...>----' Bo-----...J Bo-----'
bya mesh network in which the 930010-1·20
resistors have three times their
original value.
If the networks consist of re- Fig. 8. Using the mesh-star transformation.
sistors of unequal value, the
calculations becomerather more
involved, though the principle
of using equations 2 and 4 re-
mains the same. It 18 left to the C C
reader to show that, in Fig. 7: A A

The rille for RA is to multiply

together the two resistances eon-
nected to terminal A,then divide
by the surn of all three resis- Bo-~---+-----' Bo----...J BO------'
tances; RB and Re are calculated 930010-1-21
in a similar way.
The reverse transformation,
from starto mesh, uses equations Fig. 9. Another example of the mesh-star transformation.
ofthe form:

Note that the mesh network is u,

sometimes referred to as the delta, +-
or 6, network, and the star net-
work as the wye, er Y, network.
When we are given a network
to analyse, the calculaticns may
often be shortened by substitut-
inga star network for a mesh net-
work, or the other way about.
In this example, we use the 930010-1-22 930010-1-23
mesh-star transformation to ana-
lyse Fig. 8a. The three 12 n re-
sistors are not dt-awn in a tri- Fig. 10. Kirchhoff's Current Law. Fig. 11. Kirchhoff's Voltage Law.
angle, but it is easy to see that
this is really a mesh network
with three equal resistances. In
Fig. 8b this is replaeed bythe cor-
responding star network with re-
sistors of one third the value.
A
Replacing the resistors that are
in series, the network reduces
to Fig. 8e. The 12 n and 8 n re-
sistances are equivalent to
J I u,
12x8/(12+8)=3.6 n,
u
RJ
which, in series with 4.Q, gives
7.6 Q as the equivalent resis-
I u,
BO-4-'---'
tance for the wh ole network. 930010·1-24 930010-1-25
We consider the network of
Fig. 9a as a star network ofthree
6 .Qresistors in parallel with a Fig. 12. Current division. Fig. 13. Potential division.

ELEKTOR ELECTRONICS JANUARY 1993

 FIGURING IT OUT - PART 1

mesh network of three 9 Q re- node D: 12+13+/4-[=0; node p1ication. 01' more resistanees in series have
sistors. Replaee the mesh with E: [1-[,[4=0. a pd aeross them, the surn ofthe
a star of 3 Q (cne third of 9 Q) pds across them equals the total
Current division
resistors as in Fig. 9b. This re- If the val ues of same of these pd across the whole ehain-see
duces to a star of2 Q resistors- eurrents are known, it 1S possi- It follows frorn KCL that, when Fig. 13. The same currentI flows
see Ftg. 9c. In calculating the ble to caleulate same ofthe others. a current arrives ata pointwhere through each resistance (KCL).
resistancefromA toB, weean ig- Kirchhoff's voltage law, two 01' more resistancs are con- Given a chain of two resistors:
nore the arm of the star leading 01' KVL, states that the sum of nected in parallel-see Fig. 12-
to C, so the equivalent resis- the branch voltages around a it divides into two or more cur-
tance of the whole network is loopis zero (1:U=O). In any branch rents. The total currentremains
4Q. of a network, there is a poten- unchanged. For the two resistors,
Both of the examples given tial difference (pd) across the Eq. 1 gives:
are based on networks with re- element. This may be the direct That is, the pds are directly
sistors of equal values, but the result of the element being a [Eq. 7] proportional to the resistanees.
same principle applies ifresistors souree of emf, or may be pro- For example, if U=lO V, RI=2 Q
are unequal. ducedwhena current flowsthrough and and R2=3 Q,
a resistive element. The arrows
in Fig. 11 indicate the polarity [Eq. 8]
Kirchhoff's laws
of each pd, pointing from low to
These two la ws have many ap- high potential. Ifwe travel round Equatingthe right -hand sides [Eq. 12]
plications in network analysis. a loop in one partieular direc- ofEq. 7 and Eq. 8 gives:
Kirchhoff's current law, tion (conventionally, clockwise), ByKVL,
known for short as KCL, states these pds may be taken as rises
that the sum ofthe branch cur- or falls of potential. By KVL, [Eq. 13]
rentsata nodeis zero (I1=0). Put the total of the rises minus the [Eq.9]
another way, it states that the total of the falls is zero for any From Eq. 12 and Eq. 13 we
total current flowinginto anode loop in the network. That is,thecurrents are inversely find that U1=4 V and U2=6 V.
equals the total current flowing Travelling areund the loop proportional to the resistances. Potential dividing networks of
out of it. If currents are vary- ABDE in Fig. 11 in the same di- As an example, suppose that this kind are in frequent use in
ing, KCL applies for every instant rection as the eurrent, there are [=6 A, R1=4 Q and Rj=8 Q. Then, circuits. •
oftime. three fallsofpotential: UJ, U2and by Eq. 9:
In Fig. 10, the currents are U3' As we pass through the volt-
given arbitrary directians; a cur- age source in the same direc- Nh=8/4=2,
rent that .is actually flowing in tion as the current, there is a rise
theopposite direction to its alTOW in potential, E. By KVL, frorn A [Eq. 10]
is regarded asa negative current. back to A:
The current flowing out of the But we are given that 1=6 A, so,
network must be equal to that byKCL:
flowing into it, so we show cur-
rent 1flowing in at A and out at By KVL, fro rn B back to B: [Eq.11]
B.
For the nodes in this figure, From Eq. 10 and Eq. 11, we find
KCL gives the following equali- thatIl=4 A and h=2 A.
ties: We shall cite KCL and KVL Next month: more applications
frequently in future diseussions Potential division ofKCL and KVL, and two pow-
of circuit behaviour. Below are erful theorems which aidnetwork
two simple examples oftheir ap- It follows from KVL that, iftwo analysis.

Testyourself 2. Analyse Fig. 9abyconsidering 3. In Fig. 14e, given that [=2 A, Answers will be given in next
it as two mesh networks in calculate 11 and 12' Then cal- month's instalment.
1. Find theequivalentreeistance parallel, and show that the culate the pds aeross AB and
of each ofthe five networks of result is the same as obtained each of the resistors.
Fig. 14. previously.

]J
A Ao-+---f

6 30 60-.+--I
930010-1-26 930010-1-30
6
A 6 A 6
93001~1-27 9300 10-1-28 930010-1-29

a b c d e
Fig. 14. Ne1we,ks tor TEST YOURSELF.

ELEKTOR ELECTRONICS JANUARY 1993

 FIGURING IT OUT
PART 2 - CAPACITOR MATHS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

Correction This area is the product of the

At the end of last month's in- Answers to Test Yourself - Part 1 height(current) and width (time)
sta1mentit was stated erroneously ofthe line. Theamountofcharge
that this month's issue would con- a 1.875 Q; b 8.2 Q; c 3 Q; d 4 Q on the capacitor at any given
tain more app1ications of KCL instant equals the area beneath
and KVL; these will be discussed, 3 1[=1.5 A; 11=0.5 A; 11 V across AB; 0.5 V across 1 Q; the line.
however, in our March issue, 3 V across 2 Q; 8 V across 4 Q; 2.5 V across 5 Q There is no difficulty in per-
forming such calculations when
the current IS constant. There
he fundamental relationship is certainly no need to draw a
T for capacitors is theequation
that defines the unit of capaci-
1/ A graph. Complications arise, how-
ever, when the current is not
tance, C, which 1Sthe farad:
1=2 constant. Suppose that the cur-
2
rent is zero to begin with, but
C=Q/U. increases stea+ily at a rate given
by the equation:
The capacitance in farad is the 1=0.5
electric charge, Q, expressed in l=at,
coulombs per volt potential dif-
ference, U, between the plates where a 1S a constant express-
of the capacitor. From this, we ing the rate of increase of current
o 0.1 0.2 0.3 0.4 0.5 0.6 1/5 in amperes per second, A 8-1.
obtain the equation for the charge
on a capacitor: 9300 10-2-11 Figure 16 shows the situation
whenceü.a As+. Thegraph shows
Q=Cu. [Eq. 14J Fig. 15. Capaeitor being eharged bya eonstant eurrent. thatI attainsa valueof1.2 Aafter
aperiod of 3 s. Calculating the
If a capacitor begins by being charge on the capacitor after 3 s
uncharged, and a constant cur- ia simply a matter of evaluat-
rent I passes to Olle plate for a ing the area beneath the graph.
1/ A
time L, the amount of charge on We use the formula
1.2
the capa citor at the end of that
period is 1.0

o-u. [Eq. 15J 0.8 to obtain the ans wer which is

1.8C(oulomb).Ifthecapacitance
This equation defines the coulomb. 0.' is 0.01 F as before, the pd between
t= 3 the plates is 1.8/0.01=180 V.
Combining Eq. 14 and Eq. 15
gives: 0.' In the calculation just com-
pleted, we were able to make
U=It/C. [Eq. 16J 0.2 use of elementary geometry to
calculate the area and hence the
Given the capacitance, the charge and pd. This method is
0 2 3 tl s
constant cut-reut and the length not practicable if the current is
930010-2·12
of time for which the current varying in some more compli-
has flowed, we can ca.lculate the cated way. Suppose the current
potential difference between the Fig. 16. Capaeitor being eharged bya ramping eurrent. is varying in an entirely ran-
plates. For example, if a current dom manner as in Fig. 17. The
of 2 A flows for 0.5 s into a ca- area has an irr egular upper
pacitance of 0.01 F(arad), the In Eq. 16, the term It is the the current I in A at each in- boundary and there is 00 sim-
pd between its plates 1S: product of the current and the stant oftime t in s(econds), this ple geometrical way of calculat-
length of time the current has product is represented by the ing it exactly. The best we can
U=2xO.5/001=100 (V). flowed. In Fig. 15, which plots shaded area beneath the graph. do is to record the current at
ELEKTOR ELECTRONICS FEBRUARY 1993
 FIGURING IT OUT· PART 2

very short intervals oftime, short

enough for the current not to
Integration
11.
varysignificantly duringeach in- There is no space here to consider
terval. This may be done by sam ~ the theoretical basis of integra-
pl ing the current at intervals 1= ? tion. Instead, we shall statesome
of say, 0.1 s with the aid of a ofthe more important results and
microcomputer. The result is a show how to use them in solv-
series ofnarrow strips as in Fig. 18. ing practical problems. The most
We calculate the area of each useful ofthe results are referred
strip and sum the areas to obtain to as the standard integrals.
the total charge. This method has 0.5 A few ofthese are listed in Table l.
area=Q=?
an inherent error in that the Rather than expressing them in
upper boundary of each strip 1S .. t= 1 terms ofvariables x andy, as is
horizontal, whereas with a con- usual in maths texts, we have ex-
tinuously varying current it is pressed them in terms of land
sloping. The area beneath the t so that they may be directly
curve is notquite the same as the applied to problems involvingca-
total area ofthe strips. The error pacitors.
is reduced if we make the time o 0.5 11. Before we go on to use the
interval smaller. Ifwe could make 9300tG-2·13 technique, there is the matter
the strips infinitely narrow, the of the symbols that we use. The
error would disappear. Fig. 17. Charging a capacitor with an irregular currenl. idea of integration is that we
are summing the areas of an in-
Mathematical solution finitely large number ofnarrow
stri ps of infinitely small width.
The technique of sampling the The integral expression refers to
11.
current at very short time in- this in symbolic form
tervals has practicallimitations,
but fortunately the integral cal- 1= 7 charge Q=area under curvee
culus comes to our aid. The term
integration means 'putting to-
getherto make a whole', and the
technique allows us to calculate
directly the area beneath any The f,or'longs', is shortfor'sum-
curve thatcan be mathematically 0.5 area = Q = matlon'. Following this comes a
defined. Itcannot,ofcourse, help 0.63 x 0.05 + 0.72 x 0.05 description of what we are to
with irregular curves such as in + 0.78 x 0.05 + ... ... 1= 1 sum: the productofI (the length
Fig. 17. but any curve that can = J11dl ofthe strip) and dt (the width of
be written as a fu nct.ion is o the strip. The'd'is nota variable:
arnenable to such treatment. it means 'infinitely small incre-
When we refer to 'a function' ment', so the term 'de' means 'an
in this context, we mean that o 0.5 infinitelysmall incrementoftime'.
11,
the way in which the current The tl and t2 indicate the period
930011).2·14
varies wi th time can be expressed of time over which summation
in an equation. We have already is to be done: the period start-
met two such functions. In Fig. 15 Fig_ 18. Finding the charge by integration. ing at tl and ending at t2. When
the function is solving a problem, we replace I
in the expression above by the
1=2. function of t which defines how
Standard integrals I varies in time.
Here I is a constant; its value is (a and bare constants) Wewill useintegration tosolve
independent oftime and the vari-
able t does not enter intc the
equation.
Function of 1 f 1 dt
the problem of Fig. 16 in which
the function is I=O.4t:

In Fig. 16, the function is tn+1 Q=area under the curve»

tn (n"-l)
n+l
I=O.4t. )0.41 dt.
Int (n=-1)
As already stated, we can solve According to the rules of in-
problems involving such ele- tegration, a constant coefficient
mentary functions by drawing of t can be brought in front of
(or imagining) the graph and the integral sign. This makes
using simple geometry. But more sense because any multiplier k
complex functions such as simply makes the strips k times
-1 longerand thearea k times greater.
sin(at+ b) -cos(at+b)
a Note that we have not included
tl and t2 at this stage, as we shall
and first obtain an expression that
cos(at+ b) ~sin(at+ b) applies toany length oftime. The
1=2sin( 4+3t) integral to be evaluated now is
Table 1. Some standard integrals expressed in
require the integration technique. terms 01 land t.

ELEKTOR ELECTRONICS FEBRUARY 1993

BI GENERAL INTEREST

t2both have c added to them. The

Now look at Table 1. The entry
constants disappear when one
which helps with this example 1I A is subtracted from the other.
ist'l, where n=L. Since n-el ez, the 2 I, ,. t e z srn St Putting it in terms of charge,
integralis the amount of charge accumu-
lating during a given period 1S
of----r--~--=----,~------~ not affected by whatever charge
happened to be on the capacitor
1
Substitutingthis in the equation
-, r
1
already.
Ifthe current functionis more
1
-2 ~ complex than those in Table 1,
900
phase lag I
,
special integrating techniques
This expression tells us the amount are required, which will be found
of charge at any given Instant, ale
,.. a = 2/3 (-cos 31 + cos 0) in advanced maths texts. How-
t. This 18the indefinite Inte- ever, the table covers most of
gral In the original problem we the situations commonly met in
wanted to know howrnuch charge electroniccircuits.ln particular,
accumulated during the given the sine function is characteris-
charge I charge
period, defined as 11=0to 12=3. increa.slng decreasing
tic of audio signals and is, there-
We fmd this by ca1culating the 1
fore, a feature of many types of
definite integral as fol1ows 1
circuit, including amplifiers, os-

Q=O.4 f: t d r =0.4(t2/2)3 o
930010-2·15
cil1atars and filters, Wedeal with
this in the next section.
Here we have written the be-
ginning and end of the time pe- Fig. 19. At the top a sinusoidal eharging current Sinusoidal currents
riod beside the integral sign. and below the resulting charge on a capacitor.
It often happens that a capaci-
The resultofintegratioD is writ- tor is being charged by a cur-
ten in brackets with the begin- rent that is varied according to
ning and end times written after a sine function. For example:
the last bracket. The amount
R
accurnulating during the period
I=2sin3t.
tl to t2 is the amount present at
t2 minus the amount present at
tlo so we ean say:

Q=0.4(t2/2)30=0.4[32/2H02/2)]
u, =2RS'"3,!n = 010
u
o

930010-2-16
How much charge accumu-
lates during 0.1 s? Integrating
accordingto the formula in Table L

1
=0.4[9/2~0]=0.4x4.5=1.8 C.
Fig, 20. Resistor-capacitor network,
Q= Jr0 2sin3t dz
O
This 18the same result as we
obtained more simply by geom- =2[ ~C~s3tr
etry but serves to demonstrate where c is the initial charge, if Q=j(2t+1) de-c.
thattheintegration methodgives any. In maths, c is referred to
Weintegrate each term ofthe =2{[ ~CO;03J~[ ~c;SOJ}
the correct result. Before we use as the constant ofintegration.
integration to solve problems In the example, c=ü, for this was expression individually,usingthe
more worthy of its capabilities, stated when the problem was standard integrals: = 2U -O.3181~ [-O.333]}
there is one point to be consid- set out. Being negative iroplies =0.03C
ered. that the initial charge is of op- Q=2jl dl+jI dr- c
posite polarity and that the cur-
=2(t2/2)+I+c. When evaluatingthe integrals
rent flowing into the capacitor
Initial charge will be acting to neutralize it oftrigonometric functions, we al-
Note that when integrating ways work in radi ans, not de-
The integral tells us how much and possibly recharge 'it in the
the constant '1' we consider it grees, since the standard inte-
charge has accumulated on the opposite direct.ion.
>

Unless some statement is to be !tO,so that we take n to be grals are derived on that basis.
capacitor. But there is nothingin Given the samecharging cur-
made about the initial charge, zero. The question states that
the equation to take account of rent, what is the charge that ac-
Q=10 wben 1=2. Substituting
any charge that might exist on or we are told the total charge at cumulates in 1 s? Substituting
the capacitor before the CUITent a given instant in time, we can- these values:
12=1:
began toflow. Ifthecapacitor had not take account ofinitial charge
acharge of 2C on its plates at in the calcu1ation-weonly know 1O=(2x4)12+2+c=6+c,
Q=2{[--cos3/3]~[-O.333]}
11=0,this wouldnotaffecttherate the amount that accumulates dur-
c=4. =210.330+0.333}
at which charge accumulates ing the given period. Often, tbe
=1.326 C.
subsequently. At t,=3, the total initial charge is known to be zero,
so it may be ignored. In other The initial charge, or constant
charge would be 2~1.8=3.8 C. N ow find the charge after
cases, additional information is ofintegration, is 4. Now we can
When calculating the indef- 2.094 s:
evaluate the charge after 5 s:
inite integral, we should allow given, as in this example.
for any possible charge by in- Current is flowing inta a ca-
Q=2{[ --cos6.282/3]-[ ~O.333]
cluding it in the equation. In the pacitor ata rategiven by 1=2t+l. =2{[~0.333]-[-O.333]}
example abovc,the modifiedequa- The charge on the capacitor at
time 2 s is 10 C. What is the The constant of integration =0 C.
tion is:
charge at time 5 s? The total is ignored when we are calcu-
lating definite integrals, as the Wh at has happened to the
charge is given by:
integrals at time tl and at time charge? Figure 19 showe that the

ELEKTOR ELECTRONICS FEBRUARY 1993

 FIGURING IT OUT - PART 2

time t2=2.094 s is the point at In the network ofFig. 20, the equations above, developed by charge accumulates on the
which 31 equals 21<.This is the eherging current is provided by integration, show why a phase capacitor? Wh at is the po-
end of one complete cycle of the a sinusoidal voltage U, passing difference exists between input tential difference across it?
sinusoidal waveform. During the through a resistor: and output signals of such a fil- 2. A capacitor is charged by a
first half of this period (up to ter. In a later issue we shalllook current specified by the func-
1.047 s, which is elose to the sec- U;=IR=2Rsin3t. at the inverse situation in the tion 1=0.5+3t2• What charge
ond example above), the current high-pass filter and examine the accumulates on the capaci-
1S charging the capacitor. During This is the input voltage of the inverse mathematical routines tor during the penod 0.8 s to
the second half-cycle, the current network. The output voltage is of differentiation. • 1.0 s?
is reversed, removing a11 the ac- that developed across the plates 3. A currentI=-8sin4t is applied
cumulated charge. of the capacitor: to a capacitor which already
Figure 19 also shows that, has acharge of 2.45 C. At
since the function forthe current U.=Q I C=(-2cos3t)/3C. Test yourself what timeis the charge on the
is a eine function, and the func- 1. A 0.05 F discharged capacitor capacitor firstreduced to zero?
tion for the charge is a negative The input and output vol t- is charged by a constant cur-
cosine function, the curves are age are 90° out of phase, with rent of 1.5 A for 0.4 s. The
outofphase by 90° with the aine U, leading This network is u; current ia then reduced by Answers will be given in next
curve leading. that of a lew-pass filter and the 1 A s-l until it ia zero. What month's instalrnent.

WE HAVE THE WIDEST CHOICE OF USED !'\j'WP$ PMl\;JoOMURI,,,,,,",,,, DMMB - 6.!5 "I" '0''''
OSCILLOSCOPES IN TME COUNTRY GPI9,'lEEE _ 0nIj'450

TEMTI!ONIJl7.l6D\1i11I'ICI!IOOMH'Oe'-'I'_~ •.. uno

_._"00
MICROCONTROLLERS
rWATSU="'_Qlan""'IOO_"""",~
HITACH'VI05(lf!AloII""",(1(lMHl_m"""'~M_ .••. U50
SCHLUMeEIICEA
~218 DooI I...,. 2OOMIi> DeIoj ~_ TnaI.... 137~ H....DHWlMIILTIMITEII. J_,~tDM'O!i 14~ .. oc.2""11
.. OTIIytu
.,.'u'P$ .... ~_Ch_IO/lII!HZI>eJav_,,""T .. _ ..... Mn .. -J2' ..... ..r./OC'o_O"O'/T"""'_T .. ...,r ... _
ITIIr"!II<'~'7S"""'I" .. 2(I(lMKto.loy_p_. .550 .., 0." '32."
TDCT"!II<IX465 Ir""" IIIHlDeI.>t_ U50
Half Eurocard controllers
Dual 100
RA"-'-ljOANA ,,,,""'!lOa<OGl tOOMH.l
TEIITFlON'~ m50u011'_ro .. ~, 0eI0rS-P "' .... RACA'jDANAIIf_M....-gt(14
TEllTIIIINIl zn; Duo< 1_ so.."' ... "'... mMogtr,~ .. __ .:.450 RACA'IDANA l/341 o..tao<,..
Aoo""""""",~o/LC.R.&Q 0"
""WPSI'MJ21701>i111_5OMIIl1lo10j-~
GOULD ClSl100 0-.... ''''''' _
UOII
W.VftE~ERR!ld24=MeIf' lmo,."..,.
"",01\
____ LU5
~ offer the following features:
,~
W'VNEKUt!l41tOu:R __
üOU~D<JS300 ~"',,,,, 2(WdIU..
. . ...• 200
AVOIoC/OC_l~ &_,""' ..... RM215L'2
ÜOOU><JS1!;OII1kniI1,... lS .. H>
HlTACH' "'02'
KIKUWI5030DooI
_I..." 10M!t1
1,,,,,,, 30M'" __ ..... _ .__ .__ •• __ ... [1&0
n~5
<22.
.. ,FtCOHI
1",.lo»rt
r"",24JU
tllGiTAL flil:QUENCY
IOMHI-OO_
IOMHt2O(M!z
METE~S
* 8051 processor family
TliKl CS)1'O~m~2(lMH, Duo! T"".
TIIIIElIS * 8 channel 10 bit A-D
,.
MAFtCOfiI UNWERSAL CGlJJi1ER
IELEqU'PMOOV. [0""'..,.00 _, .... I." 083/~1'
1"""43T OC lOOMfll
!ELEQUIPMEKISv. ...... s...p _ .... "" DB)'D75 1"",24.J8 OC 52CMHl
* 32k data memory (NVRam)
H'Mm:l(l53!luo1T_~o.g,lI""'''.

GOOLO
1421 ""'" I.... MIII> C!wt~-. ""
""
..... NEU PSU. l'I,IQ1.1Q 70'1OA''SN 1(11.. =
racc
!':I~.,....1, .. ~'"
&OYLO il9III S!mJo .'21S
1lfORNI'SUI}40'1;05Q\mo!.
fA~~ELf'SU_!25~
... , ....
I)?ilm "" ~ * 64k program memory (Eprom)
IHIS IS IUST • SAMIU· MAN~ OIIIERS •• AIU8I.E FARHElPSUtJGEO-3O'< 06llm '''...., -=
.~
~IWPS PM519J Programmable 5yl"ltbemer/F~
TruQIJ ..... ENT
MARCON' ml00..,'
CHI""'"
T,_
.......

MARCONIIf2ll7,...,.,_
Ic:R 8 __

•• ,,,.,oIOOI-II:.I"'I/O.Ot
.,.,,_
=. * RS232 or RS485 interface
""
".
..
Gei>!rirtor O.1.MHZ· 50MHZ IEEE -488..... new ...•...... f1500 AVOv_CIi;o><llOlo",,_VC!,!ll1J

MARCONI 2510 I"", RI>oIS1'<1_ l80fI

MOOE, • ,. •
AVO MULTIMETERS
(oNIaI __ I ••_ ..... ud!
* 12C interface
f.RNWsyntho""""-"-II5GI-OOOOIH'-9!l_ U75
*
S._ """a_."'''''' Keyboard scanner
MARCO~I rF2Ol5....,'" I (J~ H.Z 5<1 Ge,,,.\h1F2171 l30II T... Sol Ho 1.&II.!ISJ( _ ""'" [50
aMks.oIIICiv!)mI!;OOo LOD
M.lRCOHIIf2(l15_~Wl111 n"" 1 Mk I "'In CIlrJ}q. c.... .. ._. • 1120
'ßI
*
MARCOI<I Jf2OI5 .... 101'0tlZI"2OIoIHl_mt1J LlIHI ~. Mo! ...
MARCOMlmolllw_~mtI3
MA~COIiI S>'fJ!OIISSIj; s-a...I'Imu!; __
__ •
Co.or'ogOOCllllH!
llSO
Display driver
a~ ~1~
RACAL
~ElTHLEY
"",rumenI,oO""'_SIao
",~_eu.....
T.N _
40.,., SIin 10_ ",,",1SOO
11.00II .....M.G OSCILJ.OSCOPE HM 1005
cw,. IIn
1,.,;0 1_,00MHl
~ .•_U_I7
* Real Time Clock - battery backed
fEllROGFlAF'Ii
COULO
~""
RTs:1_r_t

1U'.""' .........
~5000 '-
rB'.... ~"'2HB
A_
I1ImUOII
15011
.._ U50
..... MEGI>SCIU'O"""P~""'5(I.00uar1_0DMtIZ.
OoIoySloeop ~'53
* 1/0 expansion
RACI.L/l>ANA 1991N_

!".RNW.=..,~s,."""ID52OMtI<
~ Co<ner _..

_ .•
UIOD

'o.DD
IIAM~G IlSC,LJ.OSCOPE HM2!l3,7 0u0I T..." 2OM""-
~T .. ~
IIAM!& OSC'WI!i<:OPE HMm_3 ""'" 11"'" :KlMtIl-
"""'""' aII~ __ I .._
13&2
* E2ROM
1Ji&l"'-'
_ """""AF__ ,"'"""""",,N~_"'''''''''''''-
filllNru.. 115520T"""""",,I""5e<~d"'INec......-1If [65.]

* BASIC interpreter
._ .._. •..._.....•........ _ ..._ _ ...._...._ ...__ .__ ..._1400
SOUl....... ""Oller _ _ .._.__ •• _._.__ .1150
BlACK STAll EQUIPMENlI"""P oll uni!. ßI
APOLJ.GIQ 1OOI.It(lC<ullo!T ..... R.... _IT .... _""

SPECTRUM
T<XTR!II<I~'94M'101<Hz
ANALYSERS
Z!(Ot,(~""'_""""'1 Uli.OOO
lPOll.OlOO 10IlMKl:1A>_.,t>___
METUIR 100 "IEIl'JflCf UM"'III100l0Vll
<222
132.
UM
prices start at just L54.95
H.P.1.1T_B550.0&IFPD.otln1OMll<"_laGHl _ METE<>A ,;00 FRE(jI.'U'CY WJmIR 8OOIofIl __ e ' U3S
H.P.141h"_&!!Ss:1!J.5OOI'Jl; l~' . . LUDD ME1"'>II IClQ(I Fl!EIJUEkCl a:lUI<TER ,GI{l 1.171
1U'.1401"""8!;5o.l&1I5S .... 5OOIII<l
H_P.l.t1w""ao56Alli65'11.12OMHl
H.P.lOOh"'II5SJGlIl5S2A16ll!1
12_
3OO>\H!
UOMHl
.___
~1OOD
1U1OO
UOO
JUPITO~500FlJNClIOI<GUiE.IUTOII(Ufll~
ORIOI'I COI.lII!R!Wj GE_~
""_BIo<I.StI<~
fWJV~
SltioISQ:I"

.... -.-
LUD
!.2i!9 Tel: (0779) 74868
1oIARCO>IITF23/0.JOH!UIIMI<l_
H.P.laxw""8.'i58!I11DI1 ,_
_ _.ll5OO
U500 Fax: (0779) 74444
lJsed Equlpme.1 Guarani ...... Manual •• "pplle~ 11pe.. ,b'e
Thiö '5 a VCll\' SMill SAMPlE ()F SIOCK. SAI: 01 Telep/looe (01 IIsts. PIe .... c1"Ieo. """,lablUty ",,1o", om.'ItIiII:AllRI,o,GE 11I1
VAl t<> be $Ide<l t<> T<>lalcl Gooos arid Qmlage
IIMs i16
B.H. Systems Ud
~ STEWART of READING ~ 32 Queen St, Peterhead
~
Telephone: (0734) 268041 Fa~ (0734j351696
1.1.0 WYKEHAM ROAD, READING, BERKS RG6 IPl
C~ners Welcome 9am--5.30pm Mon.fri (until8pm Thurs)
....... Aberdeenshire AB42 6TS

ELEKTOR ELECTRONICS FEBRUARY 1993

 FIGURING IT OUT
PART 3 - SIMPLlFYING NETWORK ANALYSIS
By Owen Bishop

This se ries is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our ahn is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

T o begin with, we look at a

routine based on the super-
position law. This law applies
Answers to Test Yourself - Part 2
across the 3 Q resistor caused
by the voltage source is

only to circuits composed of Te- 1. 1.725 C; 34.5 V U'=3/(3+2)x12=7.2 V.

sistances and powered by ideal
current sources and ideal volt- The polarity of this pd is indi-
age sourees. There are many cir- cated by the arrow in Fig. 21b.
cuits thatcan be considered to be CurrentJ' through the resistors
ofthis type, provided that the en- caused by the the voltage source
ergy SOUIces da not depart too far T is
from the ideal behaviour, so this 3. Q= 1-8Sin4t dt=(2coS4tlÖ = -2.45 C,
law is useful over a wide range 1'=12/(3+2)=2.4 A.
of problems.
where T is the time in which the charge .isreduced by There is also a currentI" caused
Superposition lew 2.45 C to becornezero. 4T = C08-1-O.225, and T =0.45 s. by the current source but, for
the moment, we are ignoring
Briefly, the law states that, if this.
there are a number of current Step 2. Restore the current
and voltage sources in a circuit, source but remove the voltage

--
each of these contr-ibutes inde- source, replacing it with a short
pendently to the currents flow- a '0 circuit(Fig. 21c). Now thecircuit
ing 01' to the voltages developed.
The action of each source is su-
perposed on that of the other
sources. As a result of this law
t
Us=12V

+
U consists of two resistors in par-
allel across the current SOUTce.
Voltage U" across these caused
by the current source is
we can simplify the analysis of
the circuit by assuming that one U"=-2x3/(2+3)x5=-6 V.
or more of the sources has been
put out of action or eliminated. The negative sign indicates that
This leaves us free to consider b Us=12V the voltage is opposed to U' de-
the effects of the remaining ac-
tive sources, usually Olle at a
time.
t OPEN
CIRCUIT
veloped by the voltage source.
Current J" is calculated from the
current division rule:
Acircuitmay haveone or more
sources, which may be voltage I"=(3/5)x5=3 A.
sources, currentsources, 01' both.
Acurrentsource 1Seliminated by c The current flows in the same di-
re-connecting it (in the imagi- rection as r, so both have a pos-
nation) so that no current can SHORT itive sign,
flow from it. We simply break the CIRCUIT Step 3. Superpose (sum) the
circuit at that point and replace voltages and currents, taking
the curren t source wi th an open account of sign:
circuit. By contrast, a voltage 930010 -11I-11
source is put out of action by U=u +U" =7.2+(-6)=1.2 V;
connecting a wire across its ter-
minals, so that no potential dif- Fig. 21. lIIustrating the superposition technique. [=1'+1"=2.4+3=5.4 A.
ference can develop. We replace
it with a short circuit. Another example appears in
Consider the example in lern is tocaJculatethevoltageand source wi th an open circuit Fig. 22a, where there are two CUT-
Fig. 21a. Thecircuithasonevolt- current shown as U and I in the (Fig. 21b). The circuit now con- rent sourees.
age source, Ug=12 V, and one figure. sists oftwo resistors in series witb Step 1. Eliminating source
current source, Is=5 A. The prob- Step 1. Replace the current the voltage source. The pd, U', Lgives the circuitofFig. 22b. This

ELEKTOR ELECTRONICS MARCH [993

 FIGURING IT OUT - PART 3

has two 1 Q. reaistors inseries, thetical components into a eir-

shunted by a 2 Q. reeistor. This a cuit in order to compensate for
1S equivalent to 2 .0. in paral1el recognized imperfections.
with 2 .0., so current I~divides Controlled sourees must not
equally so that 1'=3 A. be eliminated when a network
Step 2. Eliminate source 2, is being analysed by superposi-
giving 1 n in parallel with 3 n. tion. Consider the circuit of
Current [si divides in the ratio Fig. 25a, which includes a con-
1:3 with the larger current, l ", trolled voltage source. The volt-
flowing through the 1Q resistor. age produced by this source is
Thus, controlled by 11> the amaunt of
b OPEN current flowingthrough resistor
I"=(3/4)x8=6 A. CIRCUIT RI' In this example, the voltage
equals 411 and must be added
By superposition: into the equations at a11stages.
Step 1. Eliminats the CUT-
rent source, but leave UI and
the controlled source acti ve-see
Given this result, it is possi- Fig. 25b. The current now flow-
ble to use KCL (see Part 1) to c OPEN ingis denoted by [I'. Incalculating
calculate a11the other currents CIRCUlT the voltage from the control1ed
in the network (Fig. 23a). We source we use this value. From
then use Ohm's law to ca1culate KVL (see Part 1) we find:
the voltages (Fig. 23b). 930010 - 111- 12

Ideal sources Fig. 22. Another example 01 superposition.

It is irnportant toremember that
the superposition law applies Step 2. Eliminate the voltage
cnly to ideal sources. An ideal BA 1A GA SOUl"Ce,but leave the current
voltage source maintains a spec- a source and the controlled voltage
ified pd across its terminals, 00 GA source active. Indicating CUl'-
matter what current 1S drawn rent and voltage at this stage
from it. The specified pd may be
constant, vary steadily with time
i by a double prime, and using
KCL at node A-see Fig. 26c:
(a ramp), be pulsed, 01' alternate
periodically. Varying pds make
the calculations more compli-
cated, but they do not alter the =U"/2+W" -4[" )/1=
application of' the law. b 6A
Unfortunately, few practical =U"/2+U" -4(-U"/2).
voltage sources are ideal; alm ost
aBhave same internal resistance,
i 7U"/2=3.5, so that U" =1 V.
01' output impedance. The pre-
cision of the analysis depends 930010- 11I·13 From this we find
on the magnitude of this. If at
all feasible, we assurne that de-
partures from ideal behaviour
are small and can be ignored. Ir Fig. 23. Complete analysis 01 Fig. 2a: By superpoaition:
it is known that they are tao (a) currents; (b) voltages.
large to be ignored, we can allow
for them to a certain extent by
adding to the circuit diagram a BASE COLLECTOR This exampleshows how KVL
resistance of suitable value in Ib • • le and KCL are used to solve the
series with the souree. problem when controlled sourees
Similar remarks apply in the are present.
ease of an ideal current source. hie Juout A further example
Controlled sources The network afFig. 26 does not
In certain circuits there may be
EMITTER
t 930010 - 111- 14
contain anycurrent sources, but
voltage or current sources whose there are external sources sup-
output depends on a eurrent or plying 2 A and 1.5 A to nodes A
pd existingin anotherpartofthe Fig. 24. Simplilied equivalent circuit 01 a and B respectively. As an exten-
same circuit. An example is the cornmon-emltter arnplitier, sion of KCL, the total current
collectorcurrentofa bipolar tran- leaving any section of a network
sistor, which is controlled by the equals the total currententering
base eurrent. The equivalentcir- and the transconductance, gm' tor characteristic. It corrects for it. Tbus, there must be 3.5 A
cuit of a common-emitter am- Resistor Rre, of value lIhre (h~ is the fact that this slope means leaving the network at node D.
plifier (Fig. 24) shows the col- the common-emitter output con- that thecontrolled current source The problem is to find the cur-
lector current as the product of ductance), represents the the is not ideal, and 1S an example rent passing along branch AB.
the input voltage,Uin,atthe base effect of the slope of the collec- ofhow we may introduce hypo- We apply the superposition
,
ELEKTOR ELECTRONICS MARCH 1993
 GENERAL lNTEREST

technique by eliminating each frornwhichweobtaintheThevenin so a11 we now have to do is to fore, no pd develops across it.
entering current in turn. With resistance: calculate.Land divide as shown. Ignoring this resistor, the circuit
tbe 1.5 A current eliminated, by The example in Fig. 28 demon- consists ofl0 n in series with the
disconnecting the network from RTh=UWlsc. strates the procedure. The open- source. There is a drop of 1 V
the extern al circuit at B, 2 A circuit val tage ia the pd across across the 2 n resistor, and a
flows in at A and out at D. The Since UThequalstheopen-circuit the voltage source and the 2 n rise of5 V across the source. The
network consists of 15 Q in par- voltage, UOCJ resistor. Note that the 3 Q re- open-circuit voltage is 4 V, so
allel with 15 Q, so the current sistor does not corne into the cal- that Urn=4 v.
divides equally at A, and thus Rrn=U';I •. culation because there is na cur- The basis af calculating the
I' is 1 A, flowing from A to B. rent flowing along it when AB short-circuit current is clearer
Eliminatingthe2Acurrentmeans We have already ca1culated Uoc• is an open circuit and, there- ifwe re-draw the circuit, including
that 1.5 A enters at B and leaves
at D. Now the network consists
of25 Q in parallel with 5 Q. One
a

-
fifth of the current passes along
BCD. Thus, /" is 0.3 A flowing a 40
from B to A, or -0.3 A from A to
B. By superposition:
2V
I=I' +/" =1+(-0.3)=0.7 A. c
• i2V
Thevenin's theorem 5vi
This theorem states that any
L.......------<>-----{IB
2-terminal network consistingof b t
one or more voltage or current y x
sources and one 01' more resis- t
tances may be represented by a
single voltage source in series b
with a single resistance(Fig. 27).
If various external circuits are
connected to the two terminals, c
the behavicur of the complex 5vi A
Isc
network and the Thevenin equiv- SHORT
alent circuit are indistinguish- CIRCUIT
B
able. Replacing a complex net-
work by its Thevenin equiva- 930010 - 11I- 15
lentconsiderably simplifies sub-
sequent calculations.
The essential features of the Fig. 25. A circuil wilh a conlrolled vollage source.
Thevenin equivalent are
• the Thevenin voltage, UTb, and c
• the value of the Thevenin se
ries resistor, RTh.
Methods of calculating Urn and
Rn are as follows.
Thevenin voltage. Given
the Thevenin equivalent circuit
on the right of Fig. 27, let us L.......-:c------j B
930010 - 111- 18
suppose that terminals A and B
are not connected to any exter-
nal circuit. In this candition, no
current flows through the re- Fig. 28. Finding a Thevenin
sistor and there is no voltage equivalent circuit
drop across it, The open-circuit Fig. 26. Superimposed currents.
voltage across AB equals UTh.
Correspondingly, given a net-
work, we find UTh for its Thevenin
equivalent by calculating the
open-circuit voltage of the net-
work. We can use the superpo-
sition method or other methods
to do th.is. COMPLICATED
Thevenin resistance. Ima- NETWORK OF
girre that terminals A and B of RESISTORS "'UTh
the Thevenin equivalent are short-
circuited. A short-circuit cur-
rent, Ise, flows from A to B. The
AND SOUR CES

i
930010· 111- 19
magnitude of'this current is given B
930010 - 111- 17
by:
Fig. 29. Applying Ihe Thevenin
Fig. 27. Thevenin equivalent circuit. equivalenl melhod.

ELEKTOR ELECTRONICS MARCIl1993

 FIGURING IT OUT - PART 3

he short-circuit link AB-see culat.ion 1S to remove the 6 Q attach the 6 Q reeistor to the =4/(4.6+6)=0.38 A.
ig. 28b. The equivalent reeis- resistor. Then find theThevenin Thevenin equivalent at A and
anee of the fcur resistors is equivalent of the network re- Band calculate I: Summingup this method: to
6/11 Q. Current I is m ai n ing . This we di d in the find a voltage or current in a
previous seetion with the re- I=UTJ!(RTh+6}= branch of a network, 'remove'
1=5x11/46=55/46 A. sult shown in Fig. 28c. Fma lly, the rcsistor, reduee the rest of
the network to its Thevenin
By the rule of cur-rent divi- equivalent, 'conneet' the re-
ion, we find: sistor to this and calculate the
voltage or current.
1&=55/46x8/11=20/23 A. The method is applicable to
a +
more intrieate networks. To
Finally, we calculate RTh: find I in Fig. 30a, eonsider the
isv left and right sides of the net-
work separately, ignoring the
branch AB in each ease. It is
Figure 28c shows the The- left to the reader to eonfirm that
2Ai the equivalent of the left side
enin equivalent resulting from
he above calculations. is specified by UTh=6 V and
t RTh=3 Q. The equiva lent ofthe
B right side has UTh=4 V and
Alternative method
A RTh=1.8 Q. We now 'reassemble'
for Rrh t the network, but connect the
Reverting to Fig. 25a, replace equivalents to braneh AB-
the voltage source by a short b see Fig. 30b. Using KCL to cal-
circui t, and then find the equiv- eulate volta ge U at node A rel-
alent resistance of the circuit ative to node B which we eon-
by reduction , A eurrent source sider to be at 0 V:
would be replaced by an open
circuit. The two 4 Q resistors (6-U)/3+( 4-U)/1.8+U/3=0,
reduce to one 8 Q re si s tor in
parallel with 2 .0., which re- whence U=7.6 V.
duces to 1.6 Q. The 3 Q resistor t
B 930010 - 111·20
is in series with this, making From this resu1t, we find
a toal 'I'h everri n resistance of that
4.6 Q as obtained earlier. Fig. 30. A more complex example.
1=7.6/3=2.53 A.
Using the Thevenin The final example-see
equivalent Fig. 31a-has a controlled source,
The most obvious advantage of a but no independent source. Since
the equivalent is that it makes the voltage from the controlled
it very easy to predict how such source depends on current I,
a simple circuit will behave which is 0 when AB 1S an open
when connected to an extern al circuit. UTh=O. To calculate RTh,
circuit. Having calculated how we apply an arbitrary voltage
the equivalent will behave, we to AB. For convenience, make
k now that the original com- this voltage equal to 1 V. Applying
plicated network will behave I----{A KCL to the currents flowing
in exactly the same way. b in and out of node C:
An application ofthe method
is illust.rated by the network 1=1I5+C1-41)/2 A.
of Fig. 29a. It is required to
find current I. Ta shorten tbe The last term on the right ex-
explanation, we use the same presses the fact tha t th e pd
network as in Fig. 28a with a L---,"',OO=1O-.
"',,,-.
,'"',-{
B aeross the 2 Q resistor is the
6 n resistor conneeted aeross difference between the 1 Vap-
AB. The first step in the cal- Fig. 31. Thevenin equivalent of a controlled source. plied externally and the 41 V

L----~~--~-{B
930010·111 • 22

a b c
Fig. 32. Networks for Test Yourself.

ELEKTOR ELECTRONICS MARCH 1993

 Science & Technology
A sinusoidal alternative:
wave sine wave generators a goodbye

By Michael Soper, MA

I nlimited bandwidth circuits, any periodic

repeated waveform can more easily
01'
sinx~rrl4[sqs(x)--sqs(3x)/3-sqs(5x)-- .. ] base generated by logie ean have much
greater aecuracy and far greater 'repeata-
be made with square waves; the reason is and bility' from one manufactured device to an-
that from a clock at twice the frequency of other.
the upper bandwidth limit, and from the eosx~rrl4[sqe(x)-sqe(3x)/3-sqe( 5x)+ .. l, The triangular wave f(x)=rr.-2x, where
fundamental frequency, by the use oflogic n-cxcn, has the expansion
circuits, any hat-monie can be produced. where the coefficient of sqs(nx) is given by
[d(n)-3]ln when n<27 (and similarly for f(x )~rrl4[ sqe(x )-1/2sq e(2x) ]-1/, sq e(4x)
sqc(nx), and the forms for sin2y, sin3y, and -I/,sqe(8x)+ ....
Why move on this?
so forth, can be obtained by substituting
When a waveform is being constructed, for x=2y, x=3y, we know that by a method sim- Some mathematicallaws for these fune-
use in a synthesizer 01' any other use, there ilar to Fourier series any repetitive wave- tions are:
are many reasons for requiring form can be approximated by the requisite
• ease of generation of harmonics; infinite series in sqs and sqc. 1. sqs(nx)sqe(nx)~sqs(2nx);
• ease of modulation of one frequency by For example, the ramp function fix) for 2. [sqs(nx)J2~l;
another. -1t<X<1t can be represented by 3. [sqe(nx)]2~1;
Therefore, by using sqslx) to denote a 4. [sqs(nx)][(sqs(2nx)]~sqe(nx);
squared sine wave (maximum value 1; min- (4 n )f(x)~ sq s(x )-I/,sq s( 2x )--11, sq s(4x ) 5. [sqs(2nx)][sqe(nx)]~sqs(nx);
imum value-l),forexample, througha com- -1/8sqs(8x)- ... 6. sqs[sqs(nx)]~sqs(nx);
parater, and sqc(x) to denote a square eo- 7. sqs[sqe(nx)]~sqc(nx);
sine wave in the same fashion, a sine wave and this approximates far more rapidly 8. sqe[sqs(nx)]~l;
can be written than the equivalent Fourier series. Now, 9. sqclsqctnxjle L
1Sthere an application in the generation of
sinx~alsqs(x)+a3sqs(3x)+a5sqs(5x), precise time-bases here? Note that a time- These results are very dissirnilar to the
results for sinx and cosx and have some re-
where aj=O when j is even and aj=nfU)l4j markable features: the first point is that
whenj is odd [fU) is a funetion to deterrnine]. 11 multiplication by sqs(2nx) changes the phase
Thus, to synthesize a sine wave with of sqs(nx) by 90°, so that this becomes a
square waves requires the value of lai to -31 way of eonverting between sqs and sqc. To
a11and a top cut fiter as shown in the dia- perform this multiplication logicallyis very
gram, in whieh fis the fundamental fre- -51 easy by using XOR. Let us denote XOR by
queney. In the diagram, the proportions v and the function of a function by o. Then
are adjusted to produce a sine wave since
-71
the 'top hat' is successively rounded by sqs(nx) v sqc(nx)~sqs(2nx);
stages. The final 'top cut' rounds the wave- sqs(nx) v sqe(2nx)~sqe(nx);
form into a sine wave. -111 OUT sqs 0 sqsesqs;
sqc 0 sqcesqc;
sqc 0 sqse L;
No more troublesome sine
sqc 0 sqce l;
generators sqs 0 [sqsfx) u sqe(x)]~sqs(2nx);
Since sqs 0 sqs(x) v sqs 0 sqc(x)=sqs(2nx).

FIGURING IT OUT - The Thevenin equivalent is Fig.32a. What are land U now?
Part 3 (Cont'd) given in Fig. 31b. 2. Replace the 2 A source in 5. Find the Thevenin equiva-
There is another useful equiv- Fig. 32a with an 8 V source lent for the circuit of'Ftg. 32c.
alent circuit. the Nortonequiv- (positive to the top ofthe di- 6. Remove the 3 A source from
f'rom the controlled source. alent, wh ich we shall look at agram) and recalculate 1. Fig. 32a, leaving an open
Solving this equation gives in a future issue. • 3. Use the superposition method circuit. Find the Thevenin
to find land U in Fig. 32b. equivalent for the remein-
I~O.233 A, Wh at is the effeet on land ing circuit.
Test yourself U of replacing the 1 n re-
from which we obtain sistor with a short circuit?
1. U se the superposition method 4. Exchange the voltage and Answers will be gi ven in next
RTb~U/I~1/0.233~4.29 Q. to find the value of I in current sources in Fig. 32b. month's instalment.

ELEKTOR ELECTRONICS MARCH 1993

 FIGURING IT OUT
PART 4 - ANALYSING NETWORKS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Dur aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, ren der some insights into the mathematics involved.

212 but, since Zgflows through the opposite direction tc 12 and that
M any network problems ean
be solved by usingthe meth-
ods described in Parts 1 and 3,
voltage drop, 01' negative pd,
aeross each resistor and a volt-
age rise, or positive pd, aeross
resistor in the opposite
tion toI], this is a negative
diree-
drop,
the polarity of the source tends
to drive the current in the oppo-
particularly by superpositionand thevoltage souree. ApplyingKVL, that is, a voltage rise. Summing si te direction to h.
by deriving Thevenin equiva- the sum ofthe voltage drops across the voltages according to KVL: We now have two unknown
lents. But some networks da not the resistors equaIs the voltage variables, II and Ig and twoequa-
yield to these techniques. Where rise aeross the voltage souree. [Eq. 17] tions. Since these equations rep-
these fail, mesh current ana- The total resistance of mesh resent the voltages found in the
lysis may succeed. One point to 1 is 11 Q, so the voltage drop The same line of argument for circuit when botb sources are ac-
remember: mesh current ana- caused by /) 1SIU). Lookingmore mesh 2 produces the equation: tive, the equations are both true
lysis applies only to networks closely at the circuit, we see that at tbe same time. They are si-
with voltage sourees, not cur- mesh 1 shares branch CE with [Eq.18] multaneous equations. We have
rent sourees. rnesh 2. There is a voltage drop to solve these to find values of Il
The idea behind mesh CUT- across this branch caused by the The negative signs in this equa- and 12 that satisfy both equa-
rent analysis is very simple. 2 Q resistor through which CUf- tion reflect the fact that I1 flows tions.
Unfortunately, as the number rent 12 is f1owing. This drop is through the 2 Q resistor in the There are several methods of
of rneshes increases, the calcu- solving simultaneous equations,
lations soon become tOD complex sorne ofwhich rely on a certain
and tedious to handle, and the Flair for navigating one's way
likelihood of arithmetical mis- 8 C D
through the calculations. Tbe
'0 30
takes increases out of all pro- benefits of personal Flair, if any,
portion. But, because the basic mestl2
are to avoid the awkward frae-
idea is extremely simple, the tions that often arise, and to
computations lend themselves to
a computer algorithm, whereby
arithmetical errors are .elimi- "
:J C
N

~
+
" keep the numbers smalI. All of
these solution methods are Te-
ducible to relatively simple al-
nated. We conc1ude this part with gorithms. Consequently, a sei-
such an algorithm, transcribed E entific formu1a caleulator often
into BASIC, which should run on 930010·11
has such a rcutine as one of its
almost any rnicroeomputer. built-in programs. Using such a
The principle ofmesh curent ealeulator on equations 17 and
analysis is illustrated by refer- Fig. 33. A 2-mesh circuil. 18, we find that the currents
ence to the 2-mesh circuitin Fig. 33. through the resistors are:
Normally, this circuit would be
analysed most easily by super-
position. To begin the analysis,
, 0
we specify two mesb currents, and
11 and h These are hypotheti-
cal currents, flowing in a c1ock- 29V
wise direction around each mesh.
Wecould have them flowing anti- Interpretingthese results as the
elockwise, but this would sim- '0 currents through individual re-
ply alter the signs, making no ststors:
,:)
mesll2

other difference to the ca1cula-

tion. "V
Taking mesb 1 to begin with,
and ignorirtgmesh2, we note that 3 n carries I2=--2.Ja A (opposite
the voltage source is driving the direction to [1);
hypothetical eurrent 11 in the 930010 • 12

direetion indicated by the arrow. 2 Q carries 11-12=1/,-(-%)=1 A

Passing around the mesh in the (same direction as 11,
direction of the arrow, we find a Fig. 34a. Mesh current analysis: circuit.

ELEKTOR ELECTRONICS APRIL 1993

 FIGURING IT OUT· PART 4

for 11 from Fig. 34b.

29 -2
-34 6
-5
-4 29 -2 -5 I : 4 - ~ I =+6x12-(-4)x(-4)
19 -4 12 850 -34 6 -4 =72-16=56.
11 = =5 Note thath is negative because 19 -4 12
8 -2 -5 170
it flows in the opposite direction Here, the co-factors are the sin-
-2 6 -4
to that Indiented by the arrow. Write outeach elementofthe top gle elements 12 and -4, so the
-5 -4 12 Without a programmed calcu- row, multiplied by its co-fac- products can be calculated and
Ia tor, the equations are solved by tor: summed without further ado.
the standard method but, with The rule for evaluating second-
8 29 -5
-2 -34 -4
three (and even more so with
four or more meshes and equa-
tiona), the calculations become
+29XI
-4
6
~I order determinants may be more
conveniently expressed as the
product of the elements on
-5 19 12 -340
12 =-2 a tangle. the leading diagonal minus

•
8 -2 -5
-2 6 -4
-5 -4 12
170 Tosystematiee the solution of
simultaneous equations ofthird
and higherorders, twomain arith-
-2x
1
-34
19 ~I the product of the elements
on the other diagonal. In a
similar way, we find that the
metical algorithms have been de- values of the other two deter-
veloped, which are usually fea- -5x -34 61 minants are:
8 -2 29 tured in textbooks of electrical 1 19 -4
and electronic engineering. One (-34)xI2-(-4)x19 =
-2 6 -34 =-408-(-76) =-332;
ofthese relies on determinants.
-5 -4 12 510 =3 This method, sometimes called
13 = Cramer's Rule, is illustrated in and
8 -2 -5 170 + + +

-2 6 -4 Fig. 34b and we shall discuss it + +

below. For circuits with four or (-34)x(-4)-6x19 =
-5 -4 12 + + +
more meshes, it is possible but = 136-114 = 22.
not desirable to write and solve + +
equations in volving 4th-order or We are now ready to substi-
Fig. 34b. Mesh current analy- higher order determinants. More tute these values into the equa-
sis: solution by determinants. about solution methods in a mo- tion for the value of the original
ment. 930010· 13 third-order determinan:
Another technique for sys-
Knowing these currents and tematizing the solution of si- Fig. 35. Place signs 01 29x56-{-2)x( -332)+(-5)x22 =
the resistances, we can also work multaneous equations 1S based co-factors. =1624-664-110 = 850.
out the voltages at each node. on matrix arithmetic. The most
This is an exercise for the reader commonly used method is the In the same way, we find the
(solution on pA7), 'I'his network augmented matrix in which The co-factor ofan elementis the values ofthe other determinants
can be solved by ueing superpo- the eoefficientsand constan ts are determinant written out again, listed in Fig. 34b and then cal-
sition instead-another exer- manipulated within a matrix but omitting the row and the culate the currents.
eise for the reader). according to a given set ofrules. column in which the element oc- We evaJuated the determi-
Ifthere are tWQ or more volt- While matrixarithmetic is a fas- curs. Thus, the co-factors of ele- nant by taking the elements from
age sources in the rnesh, these cinating branch of mathemat- ments of a third-order determi- the top row together with their
are summed, taking account of ies in its own right and has ap- nant are all second-order deter- co-factors. The same result is
polarity. plications in many field, it has minants. Prefix each product obtained if we take any other
onIy one important application with a plus or minus sign ac- row instead, or any column. This
More meshes in eIectrical and electronic en- cording to the position of the el- fact may be made use of where
gineering, which is the solution ement. Figure 35 shows the place a determinant contains any zeros.
'The network in Fig. 34a has three of simultaneous equations. The sign for the co-factor of each el- By basing the calculations on a
meshes, with three mesh CUT- matrix method is probably slightly ement; in the example, the signs row or eolumn containing one
rents. Itgives rise to these three easier to use than the determi- are +, -, +, so that: or more zeros, some of the ele-
mesh equations: nant method, but determinants ments are zero and the number

mesh 1: SI 1-2h-5Ia= 29;

lend themselves better to pro-
gramming.
+29XI
-4
6
of
~I
products to be summed is cor-
respondingly reduced.

Determinants Higher orders

Later in this article we provide
a computer program in BASIC for
-( - 2)x
1
-34
19 ~I
Evaluating a third-order deter-
rninant is lengthy enough, witb
The value of-34 for mesh 2 shows evaluating determinants up to ample opportunity for arithmetical
that the polarity of the source the fifth order. First we shall -34 errors. Evaluating a fourth-order
+(-5)x
is such as to drive the current look briefly at the usual evalua- 1
19 determinant is even more hor-
in the opposite direction to the tion technique, partly to help rendous. The original determi-
specified 12, the reader understand the pro- Now to evaluate these sec- nant is written out as the foul'
We now have three variables gram, and partly because mesh ond-order determinants. Therule elements of a row 01' column,
and three simu1taneous equa- equations may sometimes in- is the same as for third-order each multiplied by its third-order
tions. Bycalculator, usinga built- clude complex numbers, inte- determinants. Taking the first co-factor,Each ofthesefourthird-
in program for solving three si- grals or trig functions that are determinantasan example, write order co-factors has to be evalu-
multaneous equations, we dis- beyond the scope of the BASIC out each element ofthe top row, ated in the same way as third-
cover that: program. We demonstrate the multiplied by its co-factor and order determinants, giving al-
technique by example, evaluat- prefix each product by the ap- together 12 second-order deter-
ing the numerator determinant propriate place sign: minants to be evaluated. As the

ELEKTOR ELECTRONICS APRIL 1993

 GENERAL INTEREST

number ofmeshes in the circuit The program consists mainly

increases, the computation be- of a set of nested sub-routines.
comes more and more unwieldy. At the heart of the nest, a sec-
This is why the BASTC program ond-order determinant is eval-
10 REM: DETERM INANTS *i(-*
in Fig. 36 is so useful. uated according to the 'leading 20 CLS;n=(l
The pogram is written in LO- diagonal minus other diagonal' 30 WHILE n<2 OR n>5
CQMOTNE BASIC, but we have Te- stated above (sub-routine 200) 40 INPUT"Or-der-";n
stricted it to its bare bones so that The co-factors of a third-orderde- 50 WEND
60 PRINT
it 1Sreadily convertible into other terminant are assernbled by sub-
70 FOR ;=1 TO n
BASIC dialects, such as GW/BASIC. routine 300. Each ofthese is eval- 80 PRINT "Row ";j
We leave it to the reader to re- uated in turn by repeatedly calJ- 90 FOR k=1 TO n
fine the program by addingerror- ing sub-routine 200. In sequence, 100 PRINT" Column ";k
trapping routines to the input these are multiplied by the ap- 110 INPUT d5(j,kl
lines and by formatring the dis- propriate element and sign. The 120 NEXT,NEXT
130 GOSUB 900
play to suit the machine. program takes the elements from 140 ON n GDSUB 10,200,300,500,'700
The program evaluates de- the top row of the determinant 150 PRINT:PRIN""Value~ \: "j z
terminants ofany order up to five. A fourth-order determinant is 160 FOR ;=1 TO 5, PRINT,NEXT
It 1S easy to add further sub- handled bysub-routine 500, which 170 PRINT"Any key for anothar-"
calls sub-routine 300 to evaluate 190 WHILE INKEV$=""IWEND
routines using the same algo-
190 GOTD 20
rithrn to cater for determinante the third-order co-factors, which 200 IF' n=2 THEN FDR j=l TD 2:F'OR k=
of the sixth, seventh, 01' even calls sub-routine 200 for the even- 1 TO 2:d2(j,kl=d5(j,k):NEXT:NEXT:RE
higher orders. It 18also feasible tual evaluation of the second- M SECOND ORDER H<·
to change the input routine and order co-factors. A fifth-order 210 v=d2(1,ll*d2(2,2)-d2(1,2)*d2(2,
1)
extend the program so that it determinant is evaluated by sub-
220 IF n=2 THEN z~v
accepts the coefficients of all the routine 700, which cal ls sub- 230 RETURN
mesh equations at one time, cal- routine 500 and ultirnately sub- 300 IF n=3 THEN FOR J=1 TO 3, FüR k=
culates the required determi- routines 300 and 200. 1 TO 3~d3(j,kl=d5(j,k) :NEXT,NEXT:RE
nants and then direct.ly calcu- Ta asist thereader in expand- M TH I RD OfmER ·IBoi(·
lates the currents. ing or adapting the program, 310 v=0Iw=01~i3=1
320 F'DF~m3=1 TC) 3: b3=1
here is a variable and array list: 330 FOR c3=1 TO 3
340 IF' c3(>m3 THEN d2(1,b3l=d3(2,c3
lld2(2,b3l~d3(3,cJ)=b3=b3+1
b3,b4,b5 The number ofthe column ofthe co-factor cur- 350 NEXT
rently being assembled. 360 GOSUB 200
370 v=v*d3(1,m3l*si3:w=w+v:si3=-si3
c3,c4,c5 Tbe number ofthe column of3rd, 4th or 5th order 380 NEXT
determinants currently being assem bled into 390 IF n=3 THEN z=w
a co-factor (one of these columns is cmitted 400 RETURN
from the co-factor, the column in which the cur- 500 IF n=4 THEN FOR j=l TO 4:FOR k=
rent element occurs), 1 TO 4:d4(j,kl=d5Cj,kl :NEXT:NEXT:RE
j, k Loop counters. M FOURTH ORDER ***
510 x==O:si4=1
m3,m4,m5 Tbe num ber oftbe column of3rd, 4th or 5th order 520 FOR m4=1 TO 4:b4=1
determinants, used to specify for which ele- 5::!.0 FOR c:4=1 TO 4
ment of the top row the co-factor is currently 540 IF c4(>m4 THEN d3C1,b4)=d4(2,c4
being assembled. l:d3C2,b4l=d4C3,c4):d3C3,b4)=d4(4,c
4), b4=b4+1
n Order ofthe determinant being evaluated.
550 NEXT
si3, si4, si5 + 1 or-. 1, thesign for eachelement/co-factor prod- 560 GOSUB 300
uct in 3rd-, 4tb- and 5th-order sub-routincs, 570 w=w*d4Cl,m4l*si4:x=x+w:si4=-&i4
v, w, x,y Output from aub-routines, sums of products of 580 NEXT
each element and ita co-factor; for ex.mple, re- 590 IF n=4 THEN z=x
turning from sub-routine 200, v holds the va1ue 600 RETURN
of the 2nd-order determinant; this ie multi- 700 y=O:si5=1:REM F'IF'TH ORDER ***
710 FOR m5=1 TO 5:b5=1
plied by the element and sign: succeseive values 720 FOR cS= 1 TO 5
of v are accumulated in w (see line 370); sirni- 730 IF c5<>m5 THEN d4(1,b5)=d5(2,c5
larly, in a 4th-order evaluation, successive values ): d4 <:2,b5l =d5 C3,c5l :d4 (3, b5l =d5 (4, c
of w are accumulated in x (see line 570). 5):d4C4,b5)=d5C5,c5) :b5=b5+1
Final value of determinant. 740 NEXT
z '750 GOSLJB 500
760 x=x*d5 C1 ,m5l *$.i5: y=y+)~:si5"-&i5
d20,d3(),d40,d50 Arrays to bold determinants of orders 2 770 NEXT
to 5(these are notDIMmed intheprogram, 780 IF n=5 THEN z=y
as that is unnecessary in this version of 790 RETURN
900 CLS,FDR ;=1 TO n,REM PRINT DET
BASIC; their dimensions are 2x2, 3x3, 4x4
and 5x5 respectively). ***
910 FOR k=l Ta n
920 PRINT d5<J ,k);" " .•
930 NEXT,PRINT,NEXT
Answers to Test Yourself - Part 3 940 PRINT,PRINT
950 RETURN
I. 2.2A
2. 3.5A
3. I=0.6A; U=4 V; no effect
4. 1= 1.125 A; U=-1.75 V
5. Urn= 2 V; Rrn= 3.2 Q
6. Urn= 0.8 V; Rrn= 1.2 Q Fig. 36. BASIC Ii.ting.

ELEKTOR ELECTRONICS APRIL 1993

 FIGURING IT OUT - PART 4
EIl
Test yourself
FOTeach of the circuits in Fig. 37. use the b d
mesh current rnethod to find I to three dec- '0 r--{I'o~-~---(- +
imal places or as fractions. Confirm your
resuJt by using superposition. C
Answers will be given in next month's +
N

instalment.
'0

a 3V -
+

c '0

'0
+ zv
L------1 + - )-------'
sv

930010· 14

Fig. 32. Networks for Test Yourself.

••••••••
SAGE 'SUPER· LINK' PREAMP
T 0 complement our new range of dass
Apower arnphfers we now launch a
-The
mDThe
/

perfect link------
'Controller' Acomplete
ready built stereo control module
NEW PROOUCTS FOR 92/93
Sage Audio have launched many new and
complete dass A preamplifier the SAGE features pure dass A operanon.total exciting produck for 92/93, these indude
'Super-link' system. The preamp is in control over volume, bass, treble, a camplete new range of class Apower
modular form comprising four separate balance, active adjustable gain stage amplifier modules the Supermos200
stereo modules all fully assembled and to Supermos I 000 range with
tested. Assembly is thus straightforward ma The 'Power Supply' The absolutely exemplary perfofmance
consisting of simple mountingand wiring most perfect power supply avaiable, unmatched anywhere in the audio
to produce a top dass matched preamp. virtual zero outpur impedance, noise industry.
All modules can be used independently and total absence of ripple, powers all
or as a whole preamp system, the perfect four modules Specsincludemaximum power output
link to the SAGE Supermos modules rangefrom 50wattsto IOOOW,THO
~he 'Equaliser' An optional lessthan 0,000 I%, slewrateover 700V/
plinth mountable class A phono us, ripple rejection virtual infinite, freq
m!lIIThe 'Selector' A complete, equaliser amplifier for both MC and resp' 0.5Hz - 350Hz and there's much
ready builtstereo signal source selector, MMcartridges, adjustable gain,superior morewecouldfitinthisad.
features, gold phono stereo in/outs performance.
source selection without mechanical or
Forfulldetailsof our fullrangeof dass A audioproducts fullamplifierkitsmainsfllters
electronic switches in the signal path, and price lists please send a large stamped addressed envelope and l2 coins to-
availablewith orwithoutphono equaliser (Overseas please send 6 International reply coupons)
stage, SAGE AUDIO Construction House,Whitley St,
Bingiey, Yorks, BOl6 4JH, England.

ELEKTOR ELECTRONICS APRIL 1993

 m
FIGURING IT OUT
PART 5 - INDUCTOR MATHS
By Owen Bishop t...-

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and otber aspects of the bebaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

nductors are in many ways pacitors is that by which the several ex am pIes in Part 2 to si ne curve tbat lags behind the
I the converse of capaeitere. An
obvious difference 15 that a ca-
farad is defined: calculate charge on a capacitor
in various charging eonditions.
current curve by rr/2 rad (90°).

pacitor has an insulating dielectric C=Q/U, [Eq. 19] Having found Q, we can easily Inductor equations
between its plates, so that no ealeulate U from Eq. 19:
conduction 18 possible. In aetu-
ality, it is an open circuit. By
where C is the capacitance in
farad, Q 1S the charge in coulomb, U = C1 ldt + C·
e f [Eq.21]
The basic equation which de-
fines the henry, the unit of in-
ductance, is
contraet., the coi! of an induc- and U is the pd between the plates
tors is often (though not neces- in volts. (This equation was also An importantpropertyofaca-
pacitor 1S that, if aperiodic cur- L = N<1> (Eq.22]
sarily) of extremely low or neg- given unnumbered in Part 2). I .
1igible resistance. It 15virtually From Eq. 19, we derive an rent is supplied to it (typieally a
a closed circuit. There ia a spe- equation of the form current that varies sinusoidally), Referring to a single coil, its self-
cial term in electronics for this the charge on the capacitor and inductance in henry is the flux
sort of correspondence between Q= fldt+e (Eq.20] the voltage ac ross 'it also vary linkage (the magnetie flux ~ in
two features of properties of a cir- sinusoidally. The charge or vclt- weber multiplied by the num-
cui t or part of a circuit. The mern- where c is the Initial charge, if age curve 18 a negative ccaine ber of turns N, since the flux
bers of such a pair are called any. This equation was used in curve, or we ean think of it as a threads through eaeh turn), when
duals of each other. With re- a unit current flows through it
spect to the ability to conduct a (1 = 1 A). A more useful defini-
current, an open circuit is the 38 tion may be derived from Eq. 22,
dual of a closed circuit, To this 5 I
using Neumann's equation. This
extent, an inductor is the dual I states that:
I
I
of a capacitor. Tbe duality of in-
dlldl (A sec-u
I
U = Nd<P. [Eq.23]
ductors and capacitors extends I
dt
I
to far more than the physical I{A} I

ability to conduct a current. This

I What this equation says 15 that
I
month we contrast the two types I the e.m.f. induced in a coil is
I
of component and the mathe- I proportional t.o tberate of change
gradient '" J
I
matics associated with them. offluxlinkage. The faster the!lux
----~
Although inductors as such changes, the greater the induced
4
are not found in electronic cir- e.m.f. FOT exarnple, if the flux
cui ts as often as they used to be through the coil is inereasing at
(being replaced by opamps in the rate of 2 mWb s+, and the
active fiJters, for example), most coil has 25 turns, then U = 25x2
components havea certain amount 3 = 50 V. The direction ofthee.m.f.
of inductance. The self-induc- dlldl,,3 is such that it opposes the change
tance of the leads of a capacitor in flux according to Lenz's Law.
or resistor becomes ofmajor im- This fact is sometimes indicated
• portance in high-frequency cir- by preceding the right side of
cuits, which is why lead-less sur- theequation with a negative sign
face-mount resistors and capac- but, provided that we state that
itors are sooften used in such ap- U is the back e.m.f., the sign
plications. may be omitted.
We shall take up this discus-
sion again later, but next we
A recap on caps need to examine the equation
We dealt with several aspeets in more detail.
of eapacitor maths in Part 2, but
will brieflyquote the main points Differentials
again so that the duality of the _I{S)
inductor may be better under- The expression d<PIdt in Eq. 23
stood. The basic equation for ca-
9JOO!O·5·" represents what we have termed

ELEKTOR ELECTRONICS MAY 1993

T GENERAL INTEREST

the rate of change offlux. The In other words, the slope or gra- tbevalue4.95wbent ~ 0.4. Iteao and the current is changing at
expression is known as the dif- dient of the curve in Fig. 38 ia 3 also be seen that the slope ofthe maximum rate from positive to
ferential of <c witb respect to t, at all instants oftime. CUTve4sinl.5t when t = 0.4, as negative, dI/dt ~ 6eos7t ~ -6, its
and a whole branch of mathe- Figure 39 shows the curve shown by the tangent to the curve maximum negative value, sbow-
matics is devoted to dealing with foranotherfunction,I = 4sin1.5t. at that point, is 4.95 (taking ae- ing maximum decrease of cur-
expressions ofthis kind. The ex- Referring to the table, a ~ 1.5 count of the different scales on rent.
pression has the form of Olle quan- and b ~ Oin this function. According the two axes). When t = 1t/3, a The practical advantage of
tity(d<P) divided by another quan- tothetable,dIldt ~ 4x1.5eos1.5t point at which the current is at differentials is that they rnake
tity (dt). Differentials are used = 6cos1.5t. This teils us the rate its maximum and is neither in- it possible to calculate rates of
to relate various pairs of quan- of change of I at any instant. For creasing nor decreasing, then change for functions much more
tities: dyldx, for example, relates example,att ~ O.4,dIldt ~ 6eosO.6 dIldt ~ 6cos(1.5x7t/3l ~ 0, corre- complicated than 1= 4sinL5t,
quantities x and y. If a graph 18 ~ 4.95. Thiscan be seen in Fig. 39, sponding to zero change of CUT- and without the need to plot the
plotted with values ofx aJongthe where the curve for 6 cosO.6t has rent. Conversely, when t = 2rr/3 eurves. We sball be looking fur-
horizontal axis and values of y ther into this in future issues.
along the verfical axis, dy/dx dl/dt (A sec-u
represents the steepness of the
I(A)
39 Back to induction
slope, or gradient, ofthe grapb.
If U IS the gate-source voltage Rearranging the terms ofEq. 22
of a FET, with I being the eorre- gives
spondingdrain current, then dIldU no change In current
/,
-:
/:4sin 1.51
speci.fies the amount by which I 1",4.95 I "'-. I <I>~~I [Eq. 24]
N '
changes for given changes in U. ,
I I",2.09(=21t13}
,,
In other words, dl/dU is the ta-ans- 3
I whereLandN areconstant, but
gradient I I
conductance. In the context '" 4.95 I <PandI rnay both vary with time.
I
ofinductors, d<l>/dt is the rate of I maxtrnum Differentiating both si des ofthis
change of flux with time. I eecreese equation with respect to time:
I 01current
Calculating a differential is
the inverse operation to inte-
gration, which we described in I
V _I(S)
d<P
dt
L
N
dI
dt
[Eq. 25]
o .•
Part 2. There is not enough space
2 I
,
,
Writng this equality into Eq. 23,
here to go into the theory of dif- we obtain
~ 2 ,
I

ferentiation, but we can set out dlldl=6cos1.5t I

a table of frequently used ex-

pressions and their differentials.
~.
~3
.> :, I
U~N L
N
dI ~LdI.
dt dt
[Eq. 26]
I
Standard differentials I
~5 I
This is an alternative way of
Functionof I DifferentialdI I dt ,
I
defming the henry. In words, in
~ 6
at" a coil of self'-inductance 1 H, a
a change of current of 1 A s+ pro-
e ' 9lII01D·5· \2
duces a back e.m.f of 1 V.
sin(at+bl a eos(at+b) Given a function which de-
cos(at+bl -asin(at+b) terrnines how a current varies
with time, Eq. 26 alJowsus to cel-
The table uses land t as the U(V)
40a culate the back e.m.f at any in-
variables in place of the cus- stant. Here are some examples.
tomary x and y, so that its con-
tents are more directly applica-
ble to electronics. FOT example,
r A 1=3 (a constant current of
3 Al; dIldt ~ o. There is no
given that I ~ 3t+2, the way I change ofcurrent so the back
varies with time is shown in the e.m.f. is zero.
graph ofFig. 38. Th find tbe rate
of change of 1 at any given time, B I~2t (a Tamp current);
we calculate cU/dt, using the U"'O dl/dt = 2 A S-I. Given a coil
table. The first expression in the withL ~ 0.25H, thebaeke.m.f.
0 5 7 8
function is 3t or, to put it in the --------. t (s) is eonstant at 0.25x2 ~ 0.5 V.
form ofthe first line ofthe table, 93OO!O·~·I30
3t'. Ir n ~ 1, the differential is C I ~ 4sin1.5t (the sinusoidal
31'-'~ 3&. Sinee tO~ 1, the differ- current given in an earl ier •
ential is 3. The second expres- 40b example; dIldl ~ 6eos1.5t.
sion is 2, which can be written I (lJ.A) Given a eoil witb L ~ 0.1 H,
1=3C
as 2&. This makes n ~ 0, and the 300 the back e.m.f at any instant
value ofthe differential is zero. is 0.6 eos 1.5t. The baek e.m.f.
The differential of a constant is r 200

'00 8 follows the differential curve

always zero. Putting these two 0 of Fig. 39, but with ampli-
together, we have -100
tude 0.6 V instead of 6 A.
~ 200

dIldt~3. - 300
D We leave until a later issue
- 400
the interesting case of what
- 500
In this example the differential happens when a voltage is
~ 60'
does not contain t, so the differ- applied to a coil instantly by
ential is independent of time. 'l3OO10·5·1lb closing a switch.

ELEKTOR ELECTRONICS MAY 1993

 FIGURJNG IT OUT - PART 5

duals when applied tocapacitors For the third stage, when Uis de- But, given that the current 18
Duals again and inductors. creasing expcnential ly: sinusoidal, the expression in
Equations 21 and 26 are the volt- Ta complete tbe quartet of brackets must be equal to the
equations, we integrate both I=C. dU peak eurrent, 10:
age equations of a capacitor and dt
an inductor respectively: sides of Eq. 26 with respect to
= 12CHJ.5e2-O.S') 1= 10 sinmt. [Eq. 30]

U = C1 Idt+f c
C'
[Eq. 21]
time:

f U dz e c = LI.
= _600e2-05t j.tA.
A comparison ofEq. 29 and 30
Current flows away from the ca- shows that land U have the
U=L dl [Eq. 26] The integral of dI Idt is I. Re-ar- pacitor at an exponentially de- same frequency and that I lags
dt rangi ng terms gi ves creasingrate. Theflowofcurrent behind U by n/2. Their curves
They tell us the voltage produced is plottecl in Fig. 40b. are plotted in Fig. 42.
across the device by a CUTTent I=±fUdt+f [Eq. 28] Tbe relationship for the in-
duetor contrasts with that of
varying in time according to a Phase relationships the capaeitor as described in
given function. This has the same form as Eq. 21,
Let us see wh at happens if again illustrating the dual na- Part 2. In the inductor, the cur-
we differentiate both sides of ture of capacitance and induc- 41 rent lags the pd by n/2, but in
Eq. 21 with respeet to time: tance, and of pd and current. the capacitor it leads the pd,
I another manifestation of their
We now have a set of foul'
dU = -.!. I equations for solving problems duality.
dt C associated with capacitors and back e.m.f.
= L(dlldt)
Since differentiation and integra-
tion are inverse operations, dif-
inductors. Examples using Eq. 21
and Eq. 26 have already been
".~ Test yourself
ferentiating JI dz undoes the inte- given. Eq. 27 can beused to solve 1. The current in A at time t s
gration, leaving simply I. The dif- problems such as the fo11owing. 930010-5-14
through an inductor,L = 0.1 H,
ferential of the constant cl C ia is 1= 3-0.5t. Calculate the
zero. Rearranging terms gives: A voltage pulse applied to a back e.m.f. and state its di-
100 ~F capacitor has the shape In Fig. 41, an alternating pd is rection.
I =C dU. [Eq. 27] shown in Fig. 40a, and is de- applied to an inductor. Atany in-
dt scribed by this piecewise func- stant, the instantaneous applied 2. The current in A at time t s
'I'his tel1s us the current flow- tion: pd is: through an inductor,L = 0.2 H,
ing tnto or out of a capa ci tor U=O zs o isI= j -a-u.What is tbe back
when the applied voltage varies U=3t 0</,;4 u = Uo cosrut, [Eq. 29] e.m.f. after 2 s?
according to a given function. Ud2e2-O.& 4< t.
It isjust a mathematical way of Calculate the current flowing where U, is the peak value of 3. The voltage in ~V at time t
statingtheobvious: that the CUf- to or from the capacitor and plot the pd. Treating the circuit as a aeross an inductor,L = 100 ~H,
rent flowing into (or out of) a ca- its graph. Joop, by KVL in a clcckwise di- is defined by
pacitor equals the capacitance Solution: for the first stage, rectiou: U=O t,;O
multiplied by the rate ofincrease using Eq. 27: U = 2t 0< t.
dI
(or decrease) ofpd. Uocoscvt -L·- = 0, Calculate the current through
I = C· dU = O. dt the inductor after 3 s.
The interesting point is that
this currentequation forthe C8-
clt dI u,
:. - = _·cosrut.
pacitor has the same form as For the second stage, when U 1S dt L Answers to
the voltage equation (26) ofthe increrasing at a constant rate: Integrating both sides of this Test yourself (Part 4)
inductor. Capacitance and in- equation: a. -0.032 A (-1131)
dU b. 0.259 A (22/85)
ductance are duals. Likewise, pds 1= C·- = 3C = 300j.tA.
dt I = (~~) sinwt. c. 0.250 A (1/4)
and currents (and also rates of
change of pd and current) are This is a constant charging current. d. -0.166 A (-112/676)

42
Author's note
U(V) 1= 'oslnml
The last paragraph of Part 2

, I(A}

U = U,cosml
is misleading, since it implies
that the phase lag of a low-
pass Re filter is always 90°.
r This assumes that the re ac-
tance of the capacitor ts infi-
nite, which is tbe prectical sit-
,12 3rcf2
uation at high frequencies.
Although output lags input at
any frequeney, as impl ied in
that paragraph, the phase angle
is less than 90° at lower fre-
I quencies. I am sorry if I have
I
I mislead the reader. The action
I I
of filters will be discussed in
"','~-----"I:
Q = nf2 more detail in next month's isaue.
!l3OO10-5·15

ELEKTOR ELECTRONICS MAY 1993

 FIGURING IT OUT
PART 6 - FREQUENCIES AND FILTERS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

Capacitive reactance tance of a capacitor, 10 jlF, given

I nPart 5 we looked at what
happens when an alternating
equations above, we derived
equation for I, in which
an

In Part 5 we also showed that,

a signal of'(a) 100 Hz, (b) 1 MHz.
pd is applied to ao inductor. In
Fig. 43, the applied pd is given 1 = (UIOlL)sin wt. if a eapacitor is being charged (a)ül= 628 rad 5-1, as above.
by with a sinusoidally varying pd Xc= 1I(628xIOx1Q-l) = 15911.
U = U,cos wt, [Eq. 31] At the time we said that the ex- (Fig. 44), the current flowing
pression i.n brackets represents into the capacitor is: (b)ül= 6.28x106.
where U, 1S is the peak pd and the peak value, 10,ofa sinusoidally Xc= 1J(6.28x106x10xlO-'=
rois the angular frequency(more varrying cut-reut, immediately 1= C(dUldt). 0.015911.
about this in amoment). We then replacing the expression by 10 to
showed that the current through produce Eq. 32. Now we will take Total impedance
the inductor 1S given by time to look at this bracketed
expression more closely. The ex- The ca.lcu.lations above are for

We compared these equations

[Eq.32]

to demonstrate the facts that I

pression gives the value of the
peak current as

1,= U/OlL. u=uo?_nw_,

1
6 ~_-...lT pu.rely inductive and purely ca-
pacitive circuits, as illustrated
by Fig. 43 and 44. Now, to ex-
amine wh at happens in a cir-
has the same frequency as U, but cuit in which all three elements
lags behind U by rr/2 radi an. This has the same form as the 930010-VI·12 arepresentin series, as in Fig. 45.
well-known equation for resis-
tance: Fig.44
1 R
lelo.sln WI:
1= UIR Differentiating Uosinrot as one
ofthe standard differentiale, we ~R
L

L and demonstrates that OlL 1S the find: ueuo-sfnoit U~

ueuo-cosrot equivalent of resistance. It is 1 = C(ülU.COSwt).
the inductive reactance of the As C, ro andUo are all constants, 9~
inductor and has the symbol XL: we ean replace theirproduct with ~ I
930010-vr-11 a single constant. The equation 930010-Vt·13

[Eq. 33] for current is, in fact, a eosine

Fig.43 equation, where the single con- Fig.45
Like resistance, inductive re- stant is 1a, the peak current. In
First to clarify the quantity ül. actance is one of the elements other words, Two facts are c1ear:
This is related to frequency by ofthe total impedance of a cir- (a)The instantaneous current,
the equation cuit and is expressed in ohm (Q). 1.= ülCU e- i, througheachofL,C,R, and the
Equation 33 shows that induc- source is the same. Ir it were
ül = 2n(, tive reactance is proportional to The argument follows the same not, it would imply that curre.nt
inductaoce, as would beexpected. lines as that above, leading to the was leaking out of, or intö, the
where (t) 1Sin radian per second. Th.eequation also shows that, un- conc1usion that the capacitive circuit at some pcint.
Equation 31 could equally well like resistance, it is proportional reactance is given by (b)KVL applies to tbe instanta-
be written as U=U,cos2n(t. In to frequency: as the frequency in- neous pds. The equation fOTthe
manyinstaoces, particularlywhen creases, so does the inductive Xc= 1I0lC. [Eq.34] loop is:
dealing with audio circuits, fis reactance.
a more convenient quantity to Example: caleulate the reac- Equation 34 shows that Xc has
use. However, it is often eaaier to tanee of a coil, L=O.l roH, given features that are the opposite
deal with the single symbol ro a signal of(a) 100 Hz; (b) 1 MHz. ofthose ofXL.lt is inversely pro- Thereare several ways ofanalysi:ng
then the multiple 2n( when in- portional to both capacitance thechangingpds and currents in
tegrating 01' differentiating, so (a)w=2n(=200n=628 rad s+. and frequency. The high er the thecircuit, Techniques usingcom-
we shalJ use either way of ex- XL =628xO.1x1Q-l= 0.0628 11. frequency, the lower the capaci- plex numbers or differential equa-
pressing frequency, depending tive reactanee. 'I'his is another tions will be considered in a later
on circumstances. Cb)ül=2nx106=6.28x106 rad e-t. example of a pair of duals. issue. Here we adoptan approa.ch
During the discussion of the XL= 6.28xlO6x0.lx1Q-l= 62811. Example: calcuJate the reae- that relies on geometry. The dif-

ELEKTOR ELECTRONICS JUNE 1993

 FIGURJNG IT OUT (6)

Given the values of R, L, and square waves or sawtooth waves,

C, the first step is to calculate for example? As we shall see
XL and XC, using equations 33 later in the scrics, all periodic
and34. Then wecan draw a sketch waves can be treated as if they
ofthe vectors (as in Fig. 48) and were the sum of a number of
find: sine waves ofvarious amplitudes
1. TbetotaIimpedance,Z:apply with related frequencies. For
the Pythagoras theorem to the example, the sawtooth wave of
triangle enclosed by the vectors Fig. 49 is tbe surn of a sequence
R, (XL-XC), and their resultant of terms which goes like this:
Fig.46 Z.
y = 31t-6sint-3sin2t-2sin3l-
ficulty witb analysing this cir- Vx, which are summed by draw- ...-{6In)sinnt
cuit is that the cbanging pds are ing the diagonal of the vector
usually not in phase with each rectangle. Since the vectors rep- 2. The phase angle, ~.
U IV]
other. The graphs ofFig. 46 show resent peak pds, the length of ~ = tan-I[(XL-Xc)IR). [Eq.36)
their relationships, with UL lead- U represents the peak pd of the
ing UR by JTi2 and Uc lagging UR SOUTce.UR is in phase with the Note that since the calculations
by the same amount. current (as always), so it is clear depend solelyon the proper-
The current 1S in phase with frorn Fig. 47 that U is not in tions of the rectangle, not on
UR, since the pd across a resistor phase with 1. This is the effect its absolute dirnensions, the equa-
and thecurrent through it are al- ofthe action ofthe capacitor and tions are independent ofthe mag-
ways in phase. The three vec- inductor. Tbe phase angle, $, nitudes of the pds or the CUT-
tors on the left represent the in this example ts positive (a rent. The circuit behaves the
pds at the instantcorresponding lead) but, ifUc had been greater same, irrespective of the mag-
to the extreme left of the graph. than UL, it wouJd have been neg- nitude of the signal.
Tbe relative lengths of the vec- ative (a lag).
tors depend on the impedances Looking again at Fig. 46, it Example. Given R=100 11, L= Fig.49
found in a particular circuit at can be seen that UR is the dom- 100 mH, and C=22 ~F, connected
a given frequency. Their rela- inant pd in this particular cir- in series, calculate their total The greater the numberofterms
tive directions, however, are al- cuit. UL and Uc are very similar impedance and the phase angle summed, the more accurately the
ways as shown in the figure. in magnitude and, since they when the signal frequency is curve comes to look like a saw-
Comparison with the graph are out ofphase with each other 500 Hz. tooth wave. If we want to know
shows that these vectors repre- (leading and lagging UR by 90°) w=2n(=10001t=3141.6 rad s+; how a circuit will behave with a
sent peak pds. As time passes, they almostcancel each otherout. R = 100 11; sawtooth input, we analyse the
that is, as we move toward the The U,,* curve in Fig. 46 is plot- XL= 3141.6x100xl0~= 314.1611; waveform intoits sinusoidal COID-
right-hand end ofthe graph, we ted by summingthe point values XC= 1I(3141.6x22xlO-')= ponents, and calculate the im-
can imagine this array ofvectors calculated for the other three 1/0.069115=14.46911; pedances and phase angles for
rotating anticlockwise (with an- curves. Its peak value is sligbtly XL-XC= 299.69 11. each component separately, as
gular velccity 00), each vector re- greater than that of UR, and it above. Then we surn the results.
maining the same length and leads UR by a small phase angle, :. Z = ~(1002+299.I2)=315.411; This may involve a lot of calcu-
the angles between them re- both features confirming the re- lations, but it can be done, es-
mainingunchanged. The pds peak sults ofFig. 47. ~ = tan-1299.1I100 = 1.25 rad. pecially with the aid of a com-
in order: UL, then UR,and finally Since a11the quantities are puter.
Ue. This can be seen by examin- represented geometrically, it 1S What about j?
ing the curvea in Fig. 46. Since easy to disco ver tbe equations Simpler circuits
the three vectors retain their tbat relate thern. Figure 48 is the If you are reading this article in
relative magnitude and keep in key diagram for all such calcu- conjunction with an electrical Equations 35 and 36 cater for cir-
phase with each other, we can lations. In this diagram, the vec- or electronics textbook, you may cuits in which all three kinds of
add them for the poeition shown tors represent impedances, not have found that the text quotes impedance are present. If one
on the left of Fig. 46, knowing pds. Because the current is the the reactance of an inductor as source ofimpedance 18absent, or
that rotation does not affect the same for all elements, a diagram jooL and that ofthe capacitor as so small that it can be ignored,
result. of impedance vectors has the -j1wC. The symbolj is used when we use the same two equations
Figure 47 shows the pds added same proportions and angles as expressing complex n umbers and putting the absent impedance
by vector addition. The first step a diagram ofpd vectors. The dif- th is will be explained in a fu- equal to zero. Consider the net-
is to add UL and Ue, which is ference 1S just a matter of scale. ture issue. For the moment, take work ofFig. 50.
easy because they lie in exactly In Fig. 48 the relative lengths +j to indicate that inductive re- Since L equals zero (ass um-
opposite direction (Uc is nega- of vectors differ from those in actance leads the current, and-j ing that the self-inductance of
tive). Call their resultant the Fig. 47, which represent a circuit to indicate that cepacitive reac- the connecting wires and the
reactance voltage, Ux. This with different values of R, L, tance lags it. At present we are leads ofthe capacitor is so small
leaves usjust two vectore, UR and andC. using the geometrical technique
in which lead and lag are clear
from the sketch of the vectors.

Sinusoidal or not?
U=Uo·sin2000 TC t
UX=UL-UC All the discussions so far have f=lkHz

assumed that the source pd or

the current varies in time ac-
9300HI-VI-16
cording to a sine (or cosine) func- ?--"~1-9-3-0010-VI-le
tion. What happens if the wave-
Fig.47 Fig.48 form is notsinusoidal?Whatabout Fig.50

ELEKTOR ELECTRONICS JUNE 1993

that it can be ignored), the equa-
tions simplify to ~M
1.' .- __

u 0.9
:::-------------,
On
-10
0r-----------~~~--~----
0.8 -20

'.7
.. ~_~._ .. _. .. .. --.0_-------_ .. =-_ ... -30
$ = tan-l -XC/R.
0.' -40
-45 ----------_ .••• --_ ••••• -._ .•• --- ----.-- ••••• --- •••••• --
Given the values shown in Fig. 50: 0.5 -50
w=2000n=6283.2 rad s-l:
'.4 -<.
Xc= 1/(6283.2x470x1o-9)=338.6 n
0.' -70
.. Z=;I(3302+338.62)=472.8 n.
0.' -80

$= tarr! (-338.6t.330)=--{).80rod. 0.1 -90

0 -1001"------=_....,.,,,,.. ....,,1
'00 50. 1000 100 500 1000 5000
'iHl] f [Hz] 93ODIII-Y~2tb

XC=338.60:
Fig.53

or Rearranging gives: ber of different frequencies in the

930010-VI-19 UclU = 0.716. range 100 Hz to 10 kHz, to ob-
f= 1/2nRC. serve tbe response, and say what
Fig.51 Redraw Fig. 50 as in Fig. 52; type of filter this is. (Answer
call the total pd the input pd, Example. Find tbe -3 dB pcint next month).
Figura 51 shows tbe vectors. The Uin; and call tbe pd aeross the for a low-pass filter in which
current lags 0.80 rad (about 45') capacitor the output pd, Uout• U'" R=150 n and C=2.2 ~F.
behind the pd. Since the same is about 0.7 cf Ui", The network Test yourself
current flows through all com- attenuated the signal byroughly f= 1/(2nx150x2.2x1Q-<)=482Hz. 1. Calculate the angular veloc-
ponents, the impedance vectors 0.3. Thecalculation assumes that ity 0) for a signal offrequency
are proportional to the pd vec- the current drawn from the riet- 350 Hz.
Other filters
tors. Thus, Z is proportional to work by any load 15 insign ifi- 2. What 1S the reactance of a
the total pd applied to the net- cant. 'I'he filter ofFig. 52 is converted 33 ~H inductor at (a) 50 Hz,
work atany instant. The pd across to a high-pass filter simply by in- (b) 10 MHz?
the capacitor is proportional to terchanging the resistor and ca- 3. Wh at is the reactance of a
Frequency dependence
Xc , and it can be seen that this pacitor. Tbe vector diagram re- 120 ~F capacitor at (a) 2 Hz,
pd lags about 45' behind tbe pd Tbe calculations for F'ig. 50 can mains uncbanged, but Uout is now (b) 10 kHz?
applied to the network. be repeated far lower and high er the pd across the resistar. Thus, 4. Draw the vector diagram for
We can also ca1culate the peak frequencies and graphs plotted output leads input by 45' when the circuit ofFig. 47, given
value of the alternating signal to relate phase lag and signal R andXC areequal and by greater thatR=680n,L=0.3H, C=l~,
across the capacitor. Consider attenuation tofrequency. As these or lesser amounts when the fre- and the signal frequency is
Fig. 50 to be that of a potential graphs show (Fig. 53), phase lag quency is lower or higher. The 250 Hz.
divider circuit. The current approaches zero at low frequency -3 dB point or cut-offfrequency 5. Calculate thecut-offfrequency
through Rand C are equal (of and approaches -90' at high fre- is calculated as for the low-pass- for a lew-pass filter in which
course, no current actually passes quencies. Remember that here filter. R=12 kn and C=2.2 ~F.
through C, but current tlow- we are considering the angle be- A high-pass filter may alsc 6. Given a 10 kO resistor, what
ing to one plate is matched by tween U and Uc (or between Z be constructed by substituting value capacitor is needed to
current flwoing from the other). and Xc), not the ang1e between an inductor for the capacitor in form high-pass filter with its
Thus, the pd developed across U and UR.Attenuation approaches Fig. 52. The -3 dB point of such -3 dB point at 500 kHz?
R and across C is proportional zero at low frequencies and grad- a filter occurs when R=XL, so
to their impedances. The pd aeross ually becomes moreimportantat that at that point f=l/2RL. Answers will be given in next
C is calculated as for a poten- high frequencies. In fact, this net- A resistor, inductor and ca- montb's instalment.
tial divider. Taking pds at their work is a low-pass filter. The pacitor may be combined in a
peak vaJues: frequency chosen forthe worked filter network such as that in
example (1000 Hz) is very close Fig. 54. The output is the pd
Uc= U(XC/Z) = U(338.6I472.8)= to what is known as the -3 dB across the resistor.It is left to the
0.716U., point, This is the frequency reader to ca1culate Uoutforanum- Answersto
(1026 Hz for the specified com- Test yourself (Part 5)
ponent values) at which theout-
put is 0.707 of the input (equiv- 1. U=L(-0.5)=-0.05 V. A COD-
alent to apower attenuation of L C stant back e.rn.f. in the di-
-3 dB or 50%), and tbe phase rection opposing the reduc-
lag is exactiy 45'. It is the fre- ~1,1"~m~R----~ tionof current.
quency at which Xc is equal to
R, giving us the approximately Uin ~ Uout 2. U=0.2x-2e-"=7.33 mA
45' triangle of Fig. 51. We can
find a formula for this frequency 3. l=10000x[t2J30=0.09A. Thecon-
930010-VI-20 by equating R and XC: 930010·VI·22 stant of integration eIL can-
cels outs when we evaluate
Fig.52 Fig.54 the definite integral.

ELEKTOR ELECTRONICS JUNE 1993

 FIGURING IT OUT
PART 7 - MORE DUALS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

T hiS month we exarnine the

converse of several topics dealt
with in previous parts. Although
[Eq. 39]

NodeD:
nothingfundamentally new is in- (Un- UB)I4 = 2 + 3;
volved, since we are looking at
familiar topics upside-down or
if +
ff
[Eq. 40]
back-ta-front, a change of ap-
proach often turns an intractable This provides a set offour si-
B 930010-VII- 11
problern into a solvable one. multaneous equations for cal-
culating the four unknown volt-
Nodal analysis Fig.55 ages. In Part 4, we solved such
equations by usingdeterminants
In Part 4 we saw how to ana- 10 (and a computer program), but
lyse a netwcrk by applying furch- this technique does not work
5A
hoff's Voltage Law and caIcu- +
lati ng the currents circulating in
the meshes oftbe network. This
t here. If you try, you obtain one
or more determinants with zero
value. The potantials are inde-
1Smesh current analysis. Now ac
B terminate under this technique.
C
we look at the dualofthis tech- This makes sense, for we can
nique. We apply Kirchhoff's Cur- 2A 3A
take any one of the potentials
rent Law, ca.lculate the currents <; to be 0 V (er any other volta ge
flowing into and out ofthe riodes ~ ~ that we care to choose) and a11
of the network, and thus find other potentials are relative to
the node voltages. This method, 0 this. For example, adding 100 V
known as node voltage analy- 930010-VII- 12
to all the potentials makes no dif-
sis, is more appropriate when a ference to the currents flowing
circuit is powered mainly by cur- Fig.56 in thecircuit. In particular, if one
reut sources rather than by volt- node 18said to be at 0 V, one col-
age sources. umn of the determinant bas all
We begin with a very simple zero elements, and its value is
circuit (Fig. 55) that has only zero.
two nodes. Apply KCL to node A:

...
U sually, we can evaluate the
potentials by solving the equa-
total current entering A::::: COMPUCATED
tions one at a time, beginniug
NETWORK OF
total current Jeaving A, or with the simplest. Let us take Un
RESISTORS
1/3 + 2/3 = I, so that as the zero level. Eq. 40 gives:
AND SOURCES
t ; lA.

Taking the potential at node B

to be 0 V, the pd across the 2 Q 930010-VII·13 Putting this value into Eq. 37:
resistor is lx2:::::2 V. Thus, the
potential at node A is 2 V. This -20-UA=14, so that
is the one ofthe circuits that we Fig.57
analysed in Part 4, but with the
voltage 80UTcesreplaced by CUT- unknown potentials at nodes A-D Node B:
reut sources. are denoted by UA-Un. Applying (Uc-Un)/5+5 =(UB-UA)/2; Similarly, with Eq. 39:
InFig. 56 (anothercircuit mcd- KCL to each node:
ified from Part 4), there are three -20-UC=-1O, so that
current sources and four nodes. Node A:
Currents are determined either (UrUA)/2=5+2; Node C:
by the current sources or by the (UC-UB)/5-3;
pds across resistors (l = UR). The [Eq. 37]

ELEKTOR ELECTRONICS JULY/AUGUST 1993

 FlGURING IT OUT 7

As acheck, we evaluate Eq. 38: Solvingequations 45 and 46 gives:

~ 11=2.208 A and 12= 1.458 A.
5x-34-7x-20+2x-lO= +
tn

-170+ 140-20=-50. 10V The pd across the 5 n resistor

+

Summing up nodal analysis:

on the circuit diagram. note cur-
a
t is 5x2.208=l1.04 V. Tbus, the
potential atA=10-11.04=-1.04 V.
This teils us that UTh=-1.04 V.
rents provided bycurrent.sources; Note that the 3 n resistor does
evaluate other currents by ref- not enter into the ealeulations
~
erence to known potential dif- + sinee this branch is considered
tn
ferences and resistances. At each '" to be an open circuit and 00 CUT-
node, equate the sums ofthe en- 10V
+ A
MESH 1 MESH 2 rent flows through that resistor,
tering and leaving currents. Salve
the set of equations obtained for
the nodes. It is not always nec-
b
t UTH
B
S) '!) Figure 58e, in which voltage
sources are replaeed by sb ort
circuits, is used for ealculating
essary to work at every node RThby network reduction (see
and solve every equation to ob- Part 1). Resistcrs of6 Q aod 1 Q
tarn the Information required. in parallel result in 0.8571 Q.
so 10 ~....--------,
In series witb 1 Q, this makes
A the rigbthand side of the figure
Norton equivalent equ.ivalent to 1.8571 Q. In par-
c RTH
Tbe Norton equivalent is the B allel with 5 Q, the network is
dual of the 'l'hevenin equiva- equivalent to 1.354 n. In other
lent. In Part 3 of this series we words, RTh=1.354Q.
showed how a 2-terminal net- , We ncw have the quantities
work ofvoltage 01' current sources ,----r----,A required for using equations 43
and resistors could be replaced and 44 to calculate the NOTton
bya sing!e volta ge saurce in d ,-,
,, ,, equivalent:
series with a resistance. This
is the Thevenin equivalent and
,-, IN=-1.04/1.354=-O.768 A
1S a very useful concept for sim- '-----+---4B 930010-VII-14
plifying the analysis ofnetworks. G~1I1.354=0.739 S.
When we drive the Norton equiv-
alent of a network, we obtain a Fig.58 Figure 58d shows the Norton
single current souree in par- equivalent. The 3 Q reststor is
allel with aconductance. Note drawn in dashed 1ines across
BASE CQLLECTOR
AB as a load. Its admittanee is
the three sets of duals: valtage- '. , ,f-+....--~---,
'c 113=0.333S. The current flowing
current, series-parallel, resis-
tance-conduetance. Pigure 57 20mV
through the 3 Q resistor is pro-
+
contrasts the two equivalents.
lf the Norton equivalent cir-
cuit is short-circuited, all of tbe
t portional to its admittance:
1 =-O.768xO.333/(0.739+0.333J=
-0.239 A.
current from the internal current
generator flOW5through the ex- 930010·VII-1S An applicable example
ternal short circuit:
The network of Fig. 59 is in two
[Eq.41] Fig.59 parts but, as they share only one
eommon line, they ean be treated
w here 'sc' stands for short cireuit. eonductanee, divide the short- could be tackled by rnesh analy- separately. The problem is to
Wecan calculate Zgby finding out circuit CUTTentby the open-cir- eis and solving three simulta- ealeulate the value of current I
what eurrent flows when termi- cuit pd. neaus equations, but we will cal- and then pd U across the 750 Q
nals A and Bare connected to- Comparing Eq. 42 with the culate the Nortonequivalent in- resistar.
gether. equations for the Thevenin stead. The 1 kQ and 250 Q resistors
When the NOTtonequivalent equivalent quoted in Part 3, we Figure 58b shows the situa- in parallel are equ.ivalent to 200 Q,
is an open circuit, the current see that: tion: we wish toobtain the equiv- SQ the total current Ilowing from
from the generator flowsthrough alent of the whole network ex- the soureeis 20x1<r1/200=0.1mA.
the Norton conductanee and pro- [Eq. 43] ceptforthe brancb thatcontains This is split between the resistors
duces an open-circuited pd across the 3 Q resistor. Tbe pd across in inverse proportion to their
lt: Also, AB is U~= UTh' To find this, we resistance, so
need to know the current flow-
[Eq.44] ingthrough the 5 Q resistor, and
this is found by mesh analysis.
giving The following example (see Mesh 1 includes the 10 V end Movingon tothe other side ofthe
Fig. 58) shows how the Norton 4 V sourees as weil as the 5 Q and diagram, the eontrolled current
equivalent is used for solving two 1 n resistors, so: souree produces a CUITentof
network problems. Its use may
Substituting from Eq. 37: be contrasted with the way we [Eq.45] 1001=100x80x10~=8x10~A.
used the Thevenin equivalen t
[Eq.42] in Part 3. We are required tc In mesh 2: The three resistore in parallel are
find thecurrentlflowingthrough equivalent to 410.96 Q. Tbe pd
Ta find the NOTton equivalent the 3 n resistor. 'I'his problem [Eq. 46] aeross them is

ELEKTOR ELECTRONICS JULYIAUGUST 1993

-m GENERAL INTEREST

alterthe output from the voltage stead of pds or impedances, le gentsofvery 1arge values and ap-
R
source, U vat-ies in proportion. leads the applied pd and h lags proach 90°. We are back to Fig. 61
The ratio between source voltage it. The technique for analysing again. In practical circuits, the
c and Uis: the currents is the same as for working is not as simple. We can
pds 01' impedances. analyse the circuit graphically
A = 3.288/20xlO~= 164. More often than not, we con- by making scale drawings like
nect a capacitor and an in duc- Fig. 63 and measuring I. This is
The circuit has a voltage gain of tor in parallel, but without a nota precise technique. We need
U=Uo stnrot 164. 'I'his 18 not suprising, for the parallel resistor. Even so, resis- a non-graphic technique for adding
930010-VII-16
circult of Fig. 59 is the equiva- tance plays a part in the circuit, veetors when they are not con-
lent of a common-emitter am- a small resistance being inher- veniently perpendicular to each
Fig.60 plifier. The 20 mV source repre- ent in the construetion of prac- other. 'I'his 1S to be the subject
sents the input to the am- tical capacitors and inductors. of next month's instalment.
plifier, perhaps from a mi- Figure 62 shows the more prac-
crophone. The 1 kQ resistor tical situation. Ifwe think ofthe
'e Testyourself
represents the biasing re- branch that contains the capac-
sistors in parallel. The 250 11 itor as aseries Re circuit with L By using nodal analysis, cal-
resistor represents the input an applied pd ofUosinoot,we cal- culate the potential at each
resistance, hie, at the base culate the phase angle as we did of the nodes in Fig. 64a, tak-
ofthe transistor. Since this last month for pds, except that ingnode C to be at 0V. Calculate
resistance varies with base this now has a lead instead of a thecurrent.alongeach branch.
current, it would be better lag (duals again!). Similarly, the 2. Calculate the Norton equiva-
represented by a controlled induetor current lags the ap- lentofthe networkin Fig. 64b.
IC;IL voltage source. Current Tb plied voltage. omitting the 4 Q resistor on
is the base current, node B For the capacitor: the right. Hence, calculate
is the base terminal, and the current,I, through this re-
node E 1S the emitter ter- sistor.
minal. On the other side, 3. Figure 64c is the equivalent
the controlled current source and for the inductor circuit of a common emitter
represents the amp1ifying amplifierin which bias is pro-
action ofthetransistorwith vided by a feedback resistor,
a small signal guin, hre' of RF• connected between the col-
100. The 10 kl1 resistor rep- The vectors are shown in Fig. 63. lector and base of the n-p-n
930010-VIII·17 resents the output admit- The total current, I, is the vec- transistor. Given the values
Fig.61 tance, hre, the 1 kQ is the col- tor sum of'Jgand Ij. Iftherewere shown in the figure, calcu-
lector resistor, and the 750 Q a third branch in the network, late the voltage gain ofthe cir-
is the load. with pure reaistance, its vector cuit for smaU signals.
c This fairly simple example would lie in the same direction
has been included to al1ay the as U and could be added to the Answers will be given in next
L
suspicions ofthose who may be- other two currents to obtain the month's instalment.
lieve that network analysis is con- total current. If Re or RL is very
cerned with resistances and sources small (ideal components), the
U=Uo sinwt connected in unlikely configu- phase angles are the inverse tan-
93001 O-VII-1 e rations. Figure 59 1S a repre-
sentation ofan entirely practical
circuit and, drawn in this form, 2.
Fig.62 it immediately becomesamenable
to the techniques ofnetwork ana-
lysis descrfbed here and in pre-
a i
vious issues.

LCR in parallel
In the sense that the parallel
J"-
40
+
u connection is the dual of the se-
ries connection, we now examine ,v +
the dual of the LCR circuit de-
scribed in Part 6, Fig. 45. Figure
60 shows the classic LCR cir-
b
i
cuit with a sinusoidal pd applied
to it. At any instant, the same
pd appears across all three
branches. KCL applies to the
instantaneous currents through
10mV
the branches and I is the surn +
Fig.63 of these. To determine the cur-
rents, we use a technique simi-
c i
lar to that employed in Part 6
U=8xl0~x410.96=3.288 V. for theLCR seriescircuit. Figure
61 shows the vector diagram,
An examination of the stages of which ia similar to Fig. 48 ex-
this calculation shows that, if'we cept that, showing currents in- Fig.64

ELEKTOR ELECTRONICS JULY/AUGUST 1993

 FIGURING IT OUT (7)

DSCll1DSCDPE
GOOl.OI)S.ol1OO-l11Gj1~STI1R<\Gi' tne IUHIOIIIlr.lI]·'_Kl'O\AllNfi
HEWLfl P~CVd!!l H4I>l-IIlO!.l11l D\J-'lI;fWi~El D5II WlTl!AllAlOllllESTOI\.OG(

.~~
HEWI.[1pol.W~Dln''''IIXIIoIIIIWAlCl1N;Na TUlMMIX 7~· OOMHr4 CIWINEl
WlTIlMAlOOUl'SroRAGE Ll50 rUTIIDNIX 11100 - llJ(NHI ~ IlIMINEl.
IUTIIONIXl623· I_Kr' 0WiHB.
HlWlU PACIWID 11+1A>HD'1ll1il!Al ClWiNEl

-'~
-~
W\llII,N,l,lOOUfSTllAAGE D5iI WIllIAIiAlOGlJESTOFIolGE
1'Ifll.lWS12\l·l>1111llllW-DWi'ffi.
PllllIJI'Sl211 ~!I!.DU""QWjIffi.
I'IIllll/"Sl12II-T>MKlOOALCllM/jEl
Uom[lSo11
tn!mlnll
~"",0511
IUI!IOm11'lQ.l-5O(U1I1'0IA/iNU
100000.1l~-IIDttt1DUA1. TlIAa
TUI"!lIHI1J4ti5lI-100101H<!J,IALlJW;(
-,=
PIUlllP$Ji.IO·_WAlOWo'NU ,.....inII 1t101l0IlDlolöS·IOO/oIitIDIGIlAl.STlIAAGfIJlAlI1lAO'
"'llUPS32II\·12'1lWt1!~Ctt_El
IUTlIONIX o\S3 -lD!1Wt701lAL GfWINEI-
IEfi1IQ~Il=·I_OU""'O:I.llINEl.
\'"",085

",.,,~
tl5iI
TUTRONiI~
ltllRON111lJ.<
2!D,IIII'CIIA/II~,-l
- COOMIIf' CHII.~!iB.
WlTH.wAI.OGl.I>STIIRAGl'
-""
TEITRtINI'l1.ö-1_DUN..QWj~U """,!'lN IUTIIO.II ...... ·"DIIIl.QlANHEL

O~
=rn

""
""
n"
""

0011 Qf\..1;fIlC'«R.lr.IETlIt
-
n"
nn
""
n"
.lOOB~CVOlTlimA
.1:;E SWIIIIflU ""
."
-
m"
53o!Ylo,IICII(lWA\'fRlfOOEllCYtnIIimI·' 6GH!
1i51~1<S1C15(JWI0!l(»10M!1l n"
8I2OC5WE!POCl\..L!.I0II1~1
II6'IlIISWEEI' OCIL.l-'TDRWlTH eso Pl.UG·IM {S-,UGI\ll ."
S7s.Do\SIIJAAGE 1IOf\MAU$l!lS
~ 5!i1 • 5iJ(Ji] HJDlO mCTRUM
lS82A DUAL IlWINEl 25/(1I1sre;TIIUM
A_ 'YSEiIS
>Mt YEER
'~
~
[1~1II

l5IS.O,~Sl'ECTI!IJ"'A/j"-·SEl!
l~Do\I~Grr'E~TQII
~"
~
3761AEIUIIllIOIJKTQII 0"
tlUl :mw.1'!n"WflllUtnPl.ElMlI\l~
""

MANY MORE MISCEI-LANEOUS MANUFACTURERS TEST EQUIPMENT FOR SALE.

SENO S.A.E. FOR LIST OF EOUIPMENT

ALL EQUIPMENT IS USED, WITH 30 DAYS GUARANTEE

PLEASE CHECK AVAILASULITY BEFORE OROERING. CARRIAGE + V/I,T TO Be /l,OOEO TO ALL GOOOS

TEL lET
Telephone: 0203650702
TELNET UNIT 8, CAVAN$ WAY,
BINLEY INDUSTRIAL
COVENTRY CV3 2SF
ESTATE,

!P'cmlses sllualed erese 10Easlem bypass In Covenlry Wllh easy access 10 MI, M5. M42. M65)

Fax: 0203 650773

Answer to problem set by Fig. 54. 2.5 kHz. 4. See Fig. 66

UoulUm 18low at most frequencies, peaking 5. 6.03 Hz.
sharply at 2 kHz (Pig. 65). This is a band- Answers to 6. 3L8 pF.
pass filter with a narrow pass band. Phase Test yourself (Part 6)
angle is a lag of 80° or more at frequencies L 2199 rad s+
below L5 kHz, zero at 2000 Hz, and Iead- 2. (a) 0.010 Q; n»
2073 Q.
iug by 80° or more at frequencies above 3. (a) 663 Q; (b) 0.133 Q.

~1r-------------------------~~------------'
Uin
0.9

0.8
-3dB
0.7

0.6
I. XL =471.20

0.5
I. R=680n

0.4
III XL -Xc=
-165.411
q>=-13.7~

0.3
III
0.2
III XC=-636.60

0.1

0
100 500 5,000 10,000 930010-VI·2.4
f [Hz] 930010-Vl.23

Fig.65 Fig.66

ELEKTOR ELECTRONICS JULY/AUGUST 1993

 FIGURING IT OUT
PART 8 - COMPLEX NUMBERS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the hehaviour of circuits.
Our aim is to provide more than just a coUection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

W e begin this month's dis-

cussion with a simple and
apparently innocuous equation:
7, ... ; the points between these
represent real numbers of vari-
ous other kinds, including spe-
tha t the term is imaginary.
Note that we can have nega-
tive imaginary numbers as well
imaginarynurnbers. Points within
the four quadrants are defined
by quoting their two coordinates,
cial numbers such as 1L To rep- as positive ones. For example: just as we would on an ordinary
[Eq. 47] resent imaginary numbers, we graph. Each point in the fignre
construct an imaginary num- j = --1-25. is defined by two numbers, one
The first stage of finding the her line which is perpendicu- of which is real and the other of
value of x 18easy: lar to the real number Iine (see The perpendicular lines of which is imaginary. A number so
Fig. 68). The points marked on Fig. 68 take us a stage further, defined is known as a complex
x2 = -1, this represent the integral mul- since they can be taken to be the number. The area ofthe fignre
tiples ofj. Forexample, thepoint axes by which any point in the is the complex numberplane.
but now comes the ostensibly j3 is three times j: whole area of the figure can be Sometimes it is referred to as the
impossible part: defined. Points that are on the Argand plane, and the fignre is
j3 = 3""-1 = -I(9x-1) = -1-9. hori zontal axis (known as the known as an Argand diagram.
real axis) are real numbers.
By convention, we wr ite j3 in Points which are on the vertical Working with complex
It 18 a fact of elementary arith- preference to 3j to make it clear axis (the imaginary axis) are
metic that twa positive numbers numbers
multiplied together give a posi- Consider the pointA in Fig. 68.
tive product. Further, two neg- .-._~_~_-.'-n-i.-_~_~_~~_~_-*n-"--~_~_>-j~~ If this were a point on a graph,
ative numbers when rnultiplied -6 -5 -4 -3 -2 -1 0 2 3 4 5 6 we would say that its coordi-
together also yield a positive 930010-8-11 nates are (2, 3). Since this is an
product. There is no way ofmul- Argand dia gram, we say that
tiplying twa numbers with the Fig.67. this point represen ts the complex
same sign to obtain a negative number(2+j3). Wecannever ac-
product. Yet, here we are ex- tually add 2 toj3, because these
pected to multiply two identical ;6 numbers represent distance in
numbers together and obtain perpendiculardirections. Because
-1. The way out of this impasse ;5 • these distances are both positive,
G
is to recognize that the value of we know that the point is in the
'" j4
~
x does not belong to the set of real
numbers. It is an imaginary
,. first quadrant. Valuesrepresented
by other points in Fig. 68 are:
number. Mathematicians give
~ j3
z •
A
c <l
it the symbol I, but, since this is • « B=2-j4 E=O-j2
;;;1 ·2
likely to be confused with the C=-3 +j2 F=3 -s jü
symbol for electric current, elec- 11 D = -4 - j3 G = 4.6 + j5.2
trical and electronics engineers F
use the symbol j instead. Thus, Each complex number has areal
-6 -5 -4 -3 -2 -1 0 2 3 4 5 6
we define j by the equation: partand an imaginary part. Either
-j1 REAL AXIS
part may sometimes be zero, in
j = -i-I. [Eq.48] which event the number is rep-
-j2 E
resented by a point on one ofthe
Althoughj belongs to the realms axes.
oftheimagination, it helps to un- • -j3
Addition of complex numbers
0
derstand it and to work with it follows the ruJes ofalgebra in that
ifwe represent itin diagrammatte
-j4 •B we are allowed to add a's to a's
form. All real numbers can be -j5 and b's to b '5, but not a's to b's.
represented as points on a con- With complex numbers, we may
tinuous line, the number line -j6 add the real parts and may add
(see Fig. 67). We have marked the imaginary parts, but the two
930010- 8· 12
the positions of the points that parts must be kept separate.
represent integers -5, -2, 0, 4, Fig.68. For example, add (4 +j6) and

ELEKTOR ELECTRONlCS SEPTEMBER 1993

 FlGURING IT OUT 8

(5 + j2): pedance (and with other veetor

4 +j6 quantities besides).
+ 5 +j2 j10

Answer 9 +j8
i' The meaning of j
Another example, add (3 + j2) R=P+Q Before we go on to use complex
;8 •
aod (2 - j4): numbers to analyse practieal
3 +j2 ;7 circuits, let us look agam at that
+ 2 -j4 symbol j. We have given it a
p
Answer 5 -j2 value (Eq. 48), though this is
x'"
;. • not a value like the value ofareal
Subtraction is similar: for ex- "><r- ;5 n umber. There is another way
am ple, subtract (3 + j5) from -c of thinking of j, particularly in
z
(7 + j9): Ci ;4 the context of the Argand dia-
7 +j9 "
;;
;3
gram. We may thiok of j as the
3 + j5 imaginary operator. An op-
Answer 4 +j4 erator symbolizes an operatton.
;2 •
Q For example, the symbol J sym-
Addition and subtraction of ;1
bolizes the operation ofintegra-
complex numbers presents no tion. In logie and in digital elec-
surprises and it might be WOTI- tronics, the operator + symbol-
dered how these operations could -1 2 3 45. 7 8 s 10 izes the OR operation. In the
be relevant to electronics. Figure -j1 REAL AXIS Argand diagram. j symbolizes
69 helps to show the connection. the cperation of a quarter turn
T 930010-8-13 (90°) io the anticlockwise direc-
Fig.69. tion. The convention of writing
Vector addition
j first in an imaginary number
In Fig. 69, there are two points, reinforees the idea that it is an
j10
P and Q, representing the two operator,
complex numbers added in the Given the complex number
first of the three examples in ;' (4 + j3), for example, we inter-
the previous section. The aum ;8 pret this as the following se-
of these tWQ numbers is repre-
.",.
....
.",. I
,," quencecfinstructions, illustrated
sented by point R. In Fig. 70, in Fig. 71:
;7 ,," /
we have the same three points, ,," / 1. Startattheoriginfacingalong
but now these are seen to be the '"x-e ;. "...
/
I the real axis in the positive di-
finishing points of three vec- >- I rection.
<r /
tors, all ofwhich begin at the ori- -c ;5 I
2. Move 4 units forward.
z
gin. Not only can complex num- Ci I 3. Turn 90° antielockwise.
-c
;; ;4
/
bers be drawn as vectors, hut it I 4. Move 3 units forward.
is clear that the vector R is the I This take us to the point in the
;3 I
complex number plane which
surn of the vectors P and Q. I
Complex numbers are a way of I represents (4 + j3).
;2
representing vectors and adding A negative value ofj is inter-
complex numbers is a way of ;1 preted as a 90° clockwise turn.
adding vectors. Now we are re- For example, (5 - j2) means:
ally conneeting with eleetron- 1. As above.
ies. In Parts 6 and 7 we used -1 0 2 3 4 5 • 7 8 s 10 2. Move 5 units forward.
veetors (or phasors) to show -j1 REAL AXIS 3. Turn 90° clockwise.
the relationships between vari- 4. Move 2 units forward.
930010-8-14
ous currents, pds or impedances
in circuits that have sinusoidal Fig.70. A 90° turn anticloekwise fol-
currents or pds applied to them. lowed by another 90° turn anti-
Itwas often necessary to sum two clockwise equals a turn of 180°.
or more vectors. Application of We are facing in the opposite di-
Pythagoras' Theorem makes this 4+ j3
reetion, along tbe real axis, but
relatively easy when the vectors in the negative direction. This
are perpendicular. When they " corresponds to the value of j2:
are not, the calculations may in- j2=j xj = ~-1 x ~-1 =-1.
volve some elaborate trigonorn- Similarly, j3 = -j ( a three quar-
etry. To help us keep the trig ters turn) and j4 = 1 (a complete
under control, we make use ofthe turn).
conventions of imaginary num-
bers and the Argand diagram. Using j
Usingimaginary numbers by no
means implies that thecurrents, Figure 72 shows an LCR par-
pds or impedances themselves allel eircuit to which a sinusoidal
are imaginary. It isjust that the voltage ia applied. This figure is
graphicalrepresentationofimag- o .. REAL AXIS the same as Fig. 60, discussed in
0) start facing
inarynumbers is a useful tool fOT this way 930010·8·15 Part 7. At any instant, equal
dealing with vector quantities voltage is app1ied aeross each
such as current, voltage and im- Fig.71 branch of the circuit. For each

ELEKTOR ELECTRONICS SEPTEMBER 1993

 GENERAL INTEREST

branch of the circuit, the CUT- calculating current (Eq. 49), we becauseofsquaringj. Subtracting nominator:
rent at any instant is calculated are in the position of dividing a this, weobtain theanswer: (6+j17).
by dividing the voltage by the real number by an imaginary (4+ j5)(3 - j2)
resistance er impedance of that number: Division (3 + j2)(3 - j2)
branch. Weconsidereach branch
in turn. Ie = UlXe ~ UwGl-j. [Eq. 52] This is rather more difficult, but = 22 + j7
a simple trick helps. This sub- 13
I, R Aß mentioned earlier, real num- terfuge involves what is known
bers and imaginary numbers do as the conjugate eomplex nUID- = 1.69 + jO.54.
not mix. Ifweare to solveEq. 49, ber. In Fig. 74, there is a point
we must diseuss how to perform A in the first quadrant. Its con-
multiplication and division with jugate is A·, in the fourth quad- Reciprocals
complex numbers. rant. It is clear that if for ex- In circuit calculations, we often
ample, Ais (4 + j2), then A' is arrive at the expressions Vj and
u:uOsinwt (4 - j2). The conjugate is formed V-j. It is useful to know wh at
Multiplication
930010·8·16 by inverting the sign ofthe imag- to substitute for these. Following
Multiplication follows the usual inary part. the rules of division give above,
Fig.72. routines ofalgebra. Forexample, Now eonsider the problem of di- we first note that the j and -j
multiply (4 + j3) by (3 + j2). Set viding (4 + j5) by (3 + j2): are conjugates ofeaeh other, and
this out in the usual format for so:
The current through the re- multiplication: 4 +j5
sistor is 3 +j2 lIj = (lx-j)/(jx-j) = -jll=-j.
4 +j3 [Eq. 53]
IR = UIR. x 3 +j2 The trick is to multiply both nu-
The current is in phase with the 12 +j9 merator and denominator by the Similarly,
voltage, so we represent it in an j8 + j26 eonjugate of the den omina tor.
Argand diagram by a vector lying 12 +jI7-6. This does not affect the result lI-j = (lxj)/(-jxj) = jll = j.
along the real axis in the posi- of division, but gets rid of the [Eq. 54]
tive direction (Fig. 73). The third term has become real imaginary component ofthe de-
The current through the ca-
Continuing the
pacitor is
analysis
Ie = UlXe· [Eq.49] Substituting the result ofEq. 53
into Eq. 52:
Owing to the properties of ca-
I,
pacitors, this current leads the Ic= UwCI-j =j(UwC).
resistor current by 90°. There are
two ways oftakingthis phase dif- The result has no real part and
ference into aecount. One way ~ its imaginary partis positive. The
x
is to state that fact, as we have -c I,
current vector of the capacitor
justdone, and then draw the vec- 0 lies along the imaginary axis
REAL AXIS
tor in the appropriate direction and leads the applied voltage by
as we did in Fig. 61 ofPart 7. The
other way is to say that the re-
,.cc 90°, as shown in Fig. 73. Drawing
veetor diagrams helps us to un-
aetance of a capacitor is "a
z derstand what is happening (it
« is usually worthwhile to draw a
;;;
Xe = -jlwC. [Eq. 50] sketch at least), but the magni-
1 , tude and direetion of the cur-
Eq. 50 is the same as Eq. 34 rent vector has been obtained
Part 6), except that we have in- solely by calculation using com-
corpora ted the phase düference plex numbers,
930010·8·17
into it. Eq. 50 expresses Xc as Similar-ly, we calculate the
an irnaginary quantity, lagging eurrent vector for h:
90 behind the applied voltage.
0
Fig.73.
This hypothetical quality of the h=UljwL = -jUlwL.
impedance is related only to its
6=-4 + j3
representation in an Argand di- This ealeulation by-passes what
agram. The current passing into • j3
happens to the factors wand L
A=4+j2
or out of the capacitor contin- in the denominator. The conju-
j2 •
ues to be subjected to a substantial gate of jwL is -jwL, so wand L
opposition to its ftow! j1 appear temporarily in the nu-
Similarly, Eq. 33 (Part 6) can merator and appear squared in
now be modified to take phase the denominator; these caneel
-5 -4 -3 -2 -1 0 2 3 4 5
into account: out to leave 0) and L in the de-
-tt nomina tor, as at the beginning
XL =jwL. [Eq. 51] ofthe caleulation. The overall ef-
-j2 • feet is that j disappears from
A' =4-j2
Inductive impedance leads the the denominator and -j appears
applied voltage. • -j3
in the numerator, as in Eq. 54.
B' =-4-j3
If we substitute the new ex- 930010·8·18 The outeome of this calcula-
pression for capacitive impedanee tion is that the eurrent veetor
(Eq. 50) into the equation for Fig.74. for the inductor lies along the

ELEKTOR ELECTRON1CS SEPTEMBER 1993

 FIGURING IT OUT 8

imaginary axis and tags the ap- w = 21tf = 628 rad s-l The resultant of le and lt. is lIX = 1I200+lIj157+1I-j72.4
plied voltage by 90°, as shown in Using (J) to evaluate the im- j21.9 mA. Tbe resultant oftbis = 0.005-jO.00637 +jO.0381
Fig.73. pedances at the given frequency, and IR is (14.7+j21.9) mA. The = 0.005+jO.00744.
we arrive at Fig. 76c. magnitude of the resultant is
At time t = 0.001 s, ~(14. 72+21.92) = 26.4 mA. The Then:
Resulting vectors
U = 5sin(628xO.001)= 2.938 V, phase angleis <jl = tao-1(21.9/14. 7)
Ifwe give some numeric values remembering to work in radi- = 56.1". I = UIX = 2.938/(0.005+jO.00744)
toL, CandR, andconsiderone ans. Ifweare notinterested incal- =0.0147+jO.0219.
particular frequency and a par- culating IR, Ie and ti. individu-
ticular instant in time, the ex- IR = 2.938/200 = 14.7 mA; ally, we can simply find the total In milliamps,1 = 14.7+j21.9, as
pressions become simpler. Suppose Ie = 2.938/-j72.4 =j40.7 mA; impedance by applying tbe rules above.
that R = 250 n, C = 1.2 ~F and h = 2.938/j157 = -j18.7 mA. for impedances in parallel: The compleximpedanees can
L = 0.2 H. Also suppose that be used in any ofthe network re-
f = 500 Hz, and that we evalu- duetion techniques described
ate the currents when U = 2 V. in Parts 1, 3 and 4.
The value of co is In the foregoing examples,
the final stages rely on trigonom-
W= 2rr{= 3142 rad sol. etry. Next month we shall show
how to calculate the magnitude
Now calculate the currents: and direction ofthe resultant di-
rectly from the complex num-
IR = 21250 = 0.0080 A; bers. TO BE CONTINUED

Ie =jUmC =jO.0075 A;

Ir = -jU/wL = -j2l628.4 Test yourself

1. Add (3+j2) to (6+j3).
= -jO.0032 A. 2. Add (7+j) to (3-j4).
3. Add (5.2-jO.3) to (-3.6-j2.3).
Express currents in milliamps 4. Subtract (4-j2) from (5+j6).
for convenience: IR = 8.0 mA; o REAL AXIS 5. Multiply (3+j4) by (7+j3).
Ie =j7.5 mA and jj, =-j3.2 mA. 6. Multiply (5-j3) by (3+j).
Finding the resultant of Ie 7. What is the complex conju-
andh is straightforwardas they gate of(4+j5)?
are both complex numbers with 8. Divide (3+j2) by (7+j3).
only imaginary parts: 9. In an LCR circuit (Fig. 74),
R=120 n.z-o.i H; C=47 ~F,
Ie+h = j(7.5-3.2) = j4.3. 930010·8·19
{=50 Hz and U=2sina:t. Cal-
culate the current vectors
The resultant has magnitude and their resultants.
4.3 mAandleadsIRby90°. Adding Fig.75. 10. Recalculate the current vec-
the resultant to IR gives: tors and resultants in the
circuitof question 9when the
I = 8.0+j4.3 [mAl frequency is increased to
150 Hz. What are the mag-
Figure 75 is an Argand dia- nitudesofU andlwhent=2 s?
gram ofthese results. The mag-
nitude of the total current I is
obtained by Pythagoras'Theorem: Answers to
Test yourself (Part 7)
I = ~(8.02+4.32) = 9.1 mA. 1. UA = 3.074 V
UB = -2.272 V
The phase angle is Un = 1.185 V
IAC = 0.741 A
<jl = tan-l(4.3/8.0) = 28.3°. C lAD = 1.259 A
<:'o~ Inc = 0.395 A
Analysing LCRcircuits U = 5"06281 m ~::
~
__
-l

w(221f10)
-s 0 InB = 0.864 A
ICB= 1.136A
{O = 628 .s,
As a final example, we TUn through
the analysis ofa typicalLCR cir- 2. IN = -1.262 A
cuit. Ftgure 76 sets the prob- GN = 0.4309 S
lem, which is to calculate the I = -0.463 A
magnitude and phase angle of
the total currentI when the fre- 3. Perform nodal analysis at B
quencyis 1ooHzattimet= o.ons. c and C; calculate UB and UC;
In Fig. 76b, the impedances
U= 2.94V :
l~ ~: -j72.4Q
UB = 232.1x1Q-9 V
are wri tten inthe form which in- UC=-O.l72V
corporates the infbrmation about voltage gain = -0.17210.01
phase, that is to say, in the form = 17.2
930010·8·20
of complex nurnbers. Given that
f = 100 Hz, we calculate that:
Fig.76.

ELEKTOR ELECTRONICS SEPTEMBER 1993

 FIGURING IT OUT J~~/O-j
PART 9 - REACTIVE NETWORKS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a coUection of rule-of-thurnb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

L ast month we used complex

numbers for circuit analysis, /
Z
reetangular form uses Eq. 56 and
57, the result being written in
but reverted to using Pythagoras / the form ofEq. 55.
and inverse tangents for evalu- / Example: convert z = 50 L36°
ating the final result. What we / into reetangular form
need to da next 15 to formalize /
./ a = 50 cos 36° = 40.45;
this process. We represent a given ,/ b b = 50 sin 36° = 29.39.
complex number z by a point in / Result: z = 40.45 + j29.39.
the complex number plane, quot- /
Most scientific calculators have
ing its coordinates (a, b) in the ./ both conversions as built-in pro-
form / ' grams. Their usefulness depends
a on the fact that adding and sub-
o ~ REAL AXIS
z=a+j6. [Eq.55] tracting complex numbers is eas~
930010 ·IX· n
ier when they are in rectanguJar
These are reetangular coordi- form, but multiplying and dividing
nates, and the complex number Fig.77. is easier when they are in polar
is said to be expressed in reet- form. In the course of an analy-
angular form. Anotber way of sis we may convert from one form
specifying the position of a point e as the argument of the com- gent. These are 1800 apart. Asim- to the other and back again sev-
in a plane is to give its polar co- plex number. ple sketch OT inspection of the eral timcs, according to which
ordinates. The relationship be- signs ofa and b teils us the quad- operation is currently required.
tween the two kinds of coordi- rant in which Z 1Ssituated. Tbe rule for multiplicationis:
Conversions
nate 15 shown in Fig. 77. In this Example: convert z = -4 -j5 multiply the moduli, sum the
figure: The two final examples last month into polar form. First find arguments.
showed how to convert from rect- r = (42 + 52)112 = 6.40. Example: multiply 4 L60° by
a=rcos(J; [Eq.56] angular form to polar form. Given Next, find 2 LlO°. Multiplymoduli: 4 x 2 = 8.
b = r sin (). [Eq.57] z = a + jb, we calculate: tan-I (-51-4) = 51.34°. Sum the arguments:
But both a and barenegative, so 60° + 10° = 70°. Result: 8 L70°.
Substituting in Eq. 55 we find: r = (a2 + 62)112 tbe point must be in the third Conversely, the rule for divi-
and quadrant. This means that sion is: divide the moduli, find
z = r cos e + jr sin e, 6 = tan-I (bla), 6 = 180 + 51.34 = 231.34°. The tbe difference between the
or result: arguments. This short, easily
z = r (cos 8 + j sin 8). [Eq. 58] and write the result in the form remembered form ofthe rule does
ofEq. 59. There is a slight prob- z = 6.40 L231.24°. not specify which modulus to di-
This is the polar form of z. Tbe lern, which is that there are two vide by which, or which argu-
only essential pieces of infcr- possible values ofan inverse tan- Converting from polar form to ment to subtract from which, but
mation in this formula are the commonsense applies, as in this
radius r and the angle 6.We adopt example: divide6 L85°by2L400.
a shorter way ofwritingout the Divide moduli: 612 = 3. Difference
polar form by quoting rand e, A ofarguments: 85 -40 = 45. Result:
but leavingout tbe 'coe' and '[sin', 3 L45°.
which are taken as understccd.
"
all complex numbers having the
U=200rot
Network analysis
same format. This leaves us with:
Figure 78 shows a reactive net-
z = r L8, [Eq.59] work in which we are required
to find the voltages and currents
which rneans the same thing as B for all components. The applied
Eq. 58, but is much quicker to o 930010 - IX - 78 signal has a peak value of200 V
write and much easier to com- and thefrequencyis3183 Hz. We
prehend. In this expression, we begin by rewriting the capaci-
refer to r as the modulus and to Fig.78. tance as a complex impedance:

ELEKTOR ELECTRONICS OCTOBER 1993

 FIGURING lT OUT 9

1 400+ j200
Xc = -jo:.C = -j50.
20-j8
Taking node B to be at 0 V, and 447.214L26.5651'
considering node A at U A V, by 21.5407L-2l.801'
KCL:
930010 -IX -19a ~20. 76L48.37°

Weconvert to polar form toeval-

uate the quotient and state the
UA-200+UA+ UA =0 result in polar form. The peak
40 20 -j50 current in Mesh 1is 20.76A, clock-
stnce I == U / X for each com po- wise, leading the applied voltage
nent. This equation is simplifled: by 48.37".
The complex numbers in the
UA(~+_I_) =5 determinants ofthis example are
40 -j50 few and simple, but they illus-
trate the point that detenninants
may include eomplex numbers
U (-jI50+40) =5 without altering the essential
A -j2000 I,
routine of the mesh analysis.

93OQ10-IX·79b
U -jl0000 Thevenin equivalent
A 40-jI50·
Fig, 79, The complex Thevenin equiva-
Here we have to divide one com- lent is calculated in the same way
plex number by another: convert ably get an error message. This voltage in reetangular form. This as described in Part 3, but using
to polar form and divide: is because you are asking the cal- almplifies to just 50. complex numbers. As an exam-
culator to eval uate tan-I (-50/0). Following the technique de- ple, we calcuJate the Thevenin
U _ 10000 L -90' But we know that the '-j' means scribed in Part 4, we write the voltage, UTH, and impedance, ZTH,
A - 155.24L75.07' 'turn 90° clockwise', so the con- determinant equat.ion for 11: between terminals A and B ofthe
= 64.416L-14.93'. version rnay be written out di- network of Ftg, 81. In this fig-
rectly as 50 L-90°, as in the equa-
15~ -81 ure, Um, defined as the open-eir-
Having found UA. we know tion for Ja. 8+j~ cuit voltage, is the voltage aeross
the pd across the 20 Q resistor Tl the 4 .Q resistor and the induc-
and the capacitor, but need to j in determinants
8-j5
-8
-81·
8+ j4 tor; there is no pd aeross the 2 Q
1
find the pd across the 40 Q re- resiator, because the circuit be-
sistor. Since we are, as usual, tak- Simultaneous equations that in- Tbe value ofthe determinant in tween A and B is open. If I is the
ing the applied voltage as the ref- clude complex numbers can be the numerator is eurrent circulating in the loop:
erence for phase angle, the polar solved by determinants with the
formofthis is 200LO".Calculating method explained in Part 4. 50 (8 + j4) - 0 = 400 + j200. UTH =1(4+ j3)
U'O: Figure 80 is an exam ple of mesh
analysis in which we are required The value ofthe determinant in
10 LO' x (4+ j3)
4-j2+j3
U'O=200LO"- 64.416L-14.93". to find the value of 11. The com- the denominator is
plex impedances are stated be- = 12.127 L22. 8350
Hefe we have to subtract: con- side each component. (8 - j5) (8 + j4) - 64
vert to reetangular form: As usual, we write an equa-" = 84 - j8 - 64 = 20 - j8. The reader may Iike to verifythe
tion for each rnesh, includ.ing the intermediate stages of this eal-
U4D = (200+jO)-(62.241- jI6.596) complex impedances of the ca- Multiplying (8 - j5) by (8 + j4) culation.
=137.759 + jI6.596. pacitor and the inductor: could be done after converting The Thevenin impedance is
them to polar form but, with such calculated with the second method
Convert the result back to polar lj (8 - j5) - 12(8) = 50 + jO; small coefficients, a straightfor- from Part 3, in which we replace
form to express it as a phasor: ward algebraic multiplication Is the voltage source by a short cir-
-lj (8) + 12 (8 +j4) = O. easier. The equation for 11 sim- cuit and then apply the technique
U'O= 138.755 L6.869". plifies to a quotient: of network reduction. The net-
The righthand side of the first work is then equivalent to 4 Q
Nowthat we havethepdacross equation expresses the applied andj3Qinseries, in parallel with
all components, we calculate the -j2 n, and this 1S in series with
currents: 2Q.

Z (4+j3)(-j2) 2
11= (138.755 L6.869")/40 TH 4'3'2+
= 3.4689 L6.869"; +J -J
6-j8
=--+2
T2= (64.416 L-14.93")/20 4+j
= 3.2208 L-14.93"; 6- j8+8+j2
4+j
13= (64.416L-14.93")/50 L-90"
= 1.2883 L75.07". 14- j6
4+j
930010 . IX . 80
If you try to convert a num- ~ 3.694 L -37.235'.
ber such as -j50 into polar form
with a calculator, you will prob- Fig.80

ELEKTOR ELECTRONICS OCTOBER 1993

 GENERA L INTEREST

Omitting intermediate steps, we

Test yourself
arrive at a useful result:
1. Multiply (2 +j3) by (5 - j4) in
Uout!Uin= 11(1+ RIRL + jwRC). reetangular form. Then con-
lEq. 63] vert them both to polar form
and multiply them in tbis form.
IfRLis infinitely large, thisequa- Convert the product back to
tion reduces to Eq. 61, which is the reetangular form to COD-
pa
to be expected. As an example, firm tbat the 'multiplylsum'
B keep other values unaltered and rule 1S true.
L---------+-----------t~ make RL = 10 kQ, with ca = 105, 2. Divide (4 + j7) by (2 - j3) in
930010 -IX - 81
as in the first filter example: reetangular form. Convert to
polar form and divide. Convert
Fig.81. Uoutl Uin = 1/ (1 +0.1 + j) back to reetangular form to
=1/(1.1+j) confirm that the 'divideldif-
~ 1/1.487 L42.27° ference' rule is true.
Filters the filter. Let us see what hap- = 0.672 L-42.27°. 3. Confirm the correctness ofthe
Of all types of circuit, filters are pens if we double tbe frequency analysisofFig.76byconverting
those most concerned with the but keep component vaJues un- The output voltage and phase lag the currents into reetangular
effects of varying frequency. cbanged. Now, wRC ~ 2 and: are botb reduced sJightly by a form and checking that
Consequently, complex numbers 10 kQ load. Il~I2+Ja.
are invaluable for analysing fil- UoutlUin = 1/(j2 + 1) The approach used above may 4. Find the current passing
ter circuits. We begin with the = 1/2.236 L63.43° be applied to tbe analysis of ac- through the resistor in Fig. 78
simplest possible example, a low- ~0.447 L-63.43°. tivefilters, in which oparational (beware inverse tangents).
pass RC passive filter (Fig. 82). amplifiers are employed. Tbe anal- 5. Given a low-pass filter as in
Treating this as a potential di- The output voltage is smaller, ysis is more involved than that Fig. 81, but with R ~ 100 Q,
vider, and assuming that the load showing that this is a low-pass for a passive filter, but the same C = 10 ~F andf= 400Hz, cal-
current is negligible, we arrive fllter, and the phase lag is in- principles apply and, as far as culate Uoutl Uin and the phase
at the ratio creased. Note that when the fre- the topic of complex numbers is angle. What is the effect of
quency becomes high er, the ef- concerned, there is nothing new feeding the output to a resis-
UoutlUin~XCI(R + XC)· [Eq.60] feet ofthe '1' in the denominator to add. ti ve load of 4 kn?
becomes relatively less. Doubling
Writing the capacitor reactance offrequency approximately halves
as a complex number: the output voltage; in other words, R

output falls at the rate of 6 dB

U IU = l/jwC per octave.
out In R+ l/jmC This technique is for calcu-
lating output and phase lag for
Multiplyingthroughout by jmC: any passive Re filter at any fre-
quency.The same principles are
Uout!Uin= 11(jwRC + 1). applicable to a high-pass filter
[Eq.61] (witb Rand C interchanged) or
to an LC filtel'.
FOT example, given that Cl) = 105 The performance of the filter
(f = 15,915 Hz), R = 1 kQ, and is affected if there is an appre-
C = 10 nF, we find that wRC = 1. ciable load on it. In Fig. 83, the Fig.82.
The expression reduces to load resistor is in parallel with
the capacitor. Their parallel im-
UoutlUin ~ 1/(j + 1) pedance 1S Exponential form
= 1/1.414 L45° Answersto
= 0.707 L-45°. ZCL = XCi RLI (Xc + RL)· While on the subject of repre- Test yourself (Part 8)
lEq. 62] senting complex numbers, the 1. 9 + j5.
The output voltage is reduced to exponential form is useful in 2. 1O-j3.
0.707 of the input voltage and Substituting Eq. 62 in Eq. 60: some applications. The relation- 3. 1.6 - j2.6.
there is a phase lag of 45°. The ship between this and the polar 4. 1 + j8.
5. 9 + j37.
frequency on which this exam-
pie is based is the -3 dB point of
form is:
r L8 = reiß. 6. 18-j4. •
7.4-j5.
R
In this equation, e
must be ex- 8. 0.466 + jO.086.
pressed in radians. One advan- 9. m = 314.2; XL = j31.42 Q;
tage of the exponential form is XC=-j67.72Q;IR~0.00833U;
that it makes it possible to take lt.~ -jO.0318U;Ic ~jO.0148U;
logarithms of complex numbers. I ~ 0.0189U, lagging by 63.9°.
Another advantage, more im- Currents in amperes.
UINQ portantfrom the eleetronic point 10.m = 942.5; XL = j94.25 Q;
ofview, is that this form is more Xc = -j22.59 Q;IR ~ 0.00833U;
amenable to the Laplace trans- lt. = -jO.0106U;Ic ~jO.0443U;
formation. This is a topic that we I ~ 0.0337U, leading by 76.1°.
shall deal with in later issue. Wben t = 2 s, U = 88.8 mV,
930010 ·IX - 83
TO BE CONTINUED 1= 2.99 mA.

Fig.83.

ELEKTOR ELECTRONICS OCTOßER 1993

 FIGURING ITOUT
PART 10 - POWER
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

I na11 Dur analyses ofnetworks

and circuits, we have been con-
JQ
'"
This adds up to a total resistor
power of 4.672 W, that is, the re-
sistors are converting electrical
cerned with currents and poten-
tial differences, but have never
considered their joint effect, ,{ f)
"

f)
ii
"
energy into thermal energy at
the rate of 4.672 J s-1 Note that
power. Given a circuit element the direction in which the cur-
with a pd U across it and a CUT- + rent is flowing through the re-
10
rent I flowing through it, the In- sistor makes no difference to the
stantaneous power, Pi, is the "
----.;vI> amount ofenergy being converted.
product of U and 1:

P;= Ul. [Eq.64]

f) lt follows from the Principle
of Conservation of Energy that ~
no energy is created or lost dur- tII
rn
ing this process. The energy is
Instantaneous poweris measured 930010 - X - 84 coming from the power sources
in watts. It 'is the rate of conver- supplyingthis network. The total
SiOD from one form to another at Fig.84. energy supplied must equal the
a given instant of time. In the total energy converted to heat.
case of resistance, electrical en- by the method of mesh analysis Tz = -0.688 A; This is a good check on the cor-
ergy is being converted into heat described in Part 3. First, we write rectness ofthe currentand power
energy. In a capacitor being out the equations for the three 13 = 0.496 A. calculations. We caJculate the en-
charged, energy is being used to meshes: ergy being converted from chem-
build up a charge on the plates Figure 85 shows the currents in ieal form (ifit is a battery) or from
against the opposition ofthe charge 8h - 412 - Ja = 2; each branch ofthe netwcrk, com- 'mechantcal form (if it is a gen-
already existing there. In terms -4h + 612 = -4; bining the currents for two meshes erator) by the 4 V source. Use
of energy, the watt (W) is equiv- -I, +813 = 4. where an element is part oftwo Eq.64: •
alent to the expenditure of en- meshes. We use Eq. 65 to calcu-
ergy at the rate of 1 joule (J) per These are solved as three simul- late the power being dissipated P; = 4x1.184 = 4.736 W.
second: taneous equations, aided if pos- in each resistor:
sible by a scientific calculator or This is more than the energy
1 W = 1 J sec-I. the computer program for de- 3 >2: P = 0.0322x3 = 0.003 W; being dissipated by the resistors!
terminants (Part 4). The results 2 >2: P = 0.6882x2 = 0.947 W; But let us look at the 2 V souree.
With a resistor or reaietive el- are: 4 >2: P = 0.6562x4 = 1.721 W; The energy being converted there
ement, we also have the rela- 1 >2: P = 0.5282x1 = 0.279 W; is:
tionships U = IR and 1 = U I R. 11 = -0.032 A; 7 >2: P = 0.4962x7 = 1.722 W.
Substituting these in Eq. 64, we P; = 2x-0.032 = -0.064 W.
obtain two more equations for in-
stantaneous power: Here the direction ofthe current
does matter. The calculations
P; =[2R, [Eq. 65] showed that that 11 is negative,

and
,{ so that it flows anticlockwise
around rnesh 1 (Fig. 84), that is,
against the polarity of the 2 V
Pi = []2/R. [Eq. 66] + source. The current is supplying
1.184A~ power to the source. Ifthe source
As an exarnple ofthe application <V were a rechargeable cell, the cur-
ofthese equations, let us analyse rent would be charging it. This
a purely resistive network with fact makes the power equations
a steady voltage applied to it. In 0.496A
balance correctly:
such conditions, instantaneous
930010 - x - 85
power is constant. Figure 84 is P"'t=4.672 + 0.064=4.736W
an example of a purely resistive and
network that may be analysed Fig.85. Ps4 = 4.736 W,

ELEKTOR ELECTRONICS NOVEMBER 1993

 UN
'-
"
FIGURING IT OUT 10
m
P ~ 12.5,as indicated by the shaded
V. \
PIW
areas in Fig. 86. Note that the
zs instantaneous power oscillates
, with a frequency which 1Stwice
that ofV and/. Currentand volt-
age are both positive together
ac
from (jj; = 0 to n and also from n
to 21[. Thus,S is always positive.
This is anotherwayofsayingthat
the direction in which the cut-
reut flows through the resistor
makes no difference to the amount
of power dissipated.
Another wayof calculating av-
erage power is
I" 2.5 slnwt

1
üll In this example,

Vrrn, ~ lOd2,
Irrn, ~ 2.5d2, and
-, S ~ (lOx2.5)1d2d2) ~ 25/2
~ 12.5.

-w 930010·K·a~ This is the same result as ob-

tained by the earlier ealculation.

Fig.86.
Capacitive circuit
Ifthe same sinusoidal voltage is
wherePtotis the total power sup- The current through the resistor which gives: appl ied aeross a eapacitor, the
plied to the resistors and the 2 V has the same frequency and is in eurrent is also sinusoidal an d of
source, andPs4 18the power sup- phase with the voltage: sinä 8 ~ (1- cos 28)/2. the same frequency, butleads the
plied by the 4 V source. voltage by 90° (see Parts 2 and 5).
In general, the total power 1 ~ V/R ~ (10 sin 01)/4 Substituting this in Eq. 67: We say that 1 ~ 10 sin(u.t+ 90°).
supphed must equal the total = 2.5 sin Cd, Or we ean eonsider the eurrent
power absorbed. From a practi- Pi ~ 25x(1- cos 20Jt)/2 eurve is a eosine eurve and that
cal point ofview, it is interesting The instantaneous power 18, as ~ 12.5 - 12.5 cos 20Jt. 1 = 10 cos rot, which is trigono-
to note that the power dissipa- before, the product ofthe pd and [Eq. 68] metrically the same thing. The
tion varies widely between re- the current: value 10• that is, the amplitude
sistcrs. Calculations ofthis kind Figure 86 shows the graphs for of the current curve, is deter-
point to the components thatneed . Pi = VI = 10 sin (t.tx2.5 sin ui U, I and Pi plotted on the same mined by the reactanee ofthe ca-
to be higWy power rated. ~ 25 sinä 0Jt. [Eq. 67] scale; U and 1 are in phase, 08- pacitor:
cillating about zero. The graph
AC power We make use ofthe trigonomet- for Pi oscillates about 12.5 W. This 10 ~ Vo/XC ~ VowC.
rie identity: level ia the average or appar-
A sinusoidaJ voltage, V ~ 10 sin 01, ent power, S, since the curve is In this example, Uo = 10 V, and
is applied across a 4 Q resistor. eos 28 == 1 - 2 sin2 S, symmetrical about the line let us assume that C ~ 1000 ~F
and w ~ 400. This makes 10 ~ 4,
and 1= 4 cos ox, The curves fOT
UN
V and Lere plotted in Fig. 87.
11. The instantaneous power is,
PIW
as before, the product of U and
I:
Pi = 10 sin ox x4cos wt
=40 sin ox». cos ox.
u = 1(1 sin"'"
.....--_.1 Onee aga in. we make use of a
trigonometriea:1 identity:
",I
sin 2e = 2 sin 9 cos 9.

The power equation then becomes

Pi ~ (40 sin 201)/2 ~ 20 sin 201.

930010-~-Sl
This equation is also plotted in
-w
Fig. 87: the cxurve has ampli-
tude 20 and frequency twiee that
of V and I. In general, the am-
Fig.87 plitude may be calculated with
ELEKTOR ELECTRONICS NOVEMBER 1993
 GENERAL INTEREST

tg networks that are purely ca-

.4 1",1,857 ~acitive or purely inductive, so
" s: 68.2'
, ~hat the voltage and current are
~Ooout of phase. Ifthe phase angle
,, "5 other than this, a different sit-
,
UR = 3.714 w-rtton arises. Consider the ex-
u = 10 "-- 68.2
I
0
-

am,'le of Fig. 88. The applied

T_" T,n '"" voltage lind the impedance ofthe

capacitor ai e expressed as com-
plex nurnbers t!<:ieeParts 8 and
9). The total imp'edance,
930010· X . 88a ,
Z = 2 - j5 = 5.385 ""_68.2 0
n
in polar form. From this, W<s',"'~l-
Uc =9.285:
"--_ 21.8°;' culate I, which is the same for-,
: 93OD10· x - 5Ub both components:

1= U / Z
Fig.88. = 10 LO° I 5.385 L-68.2°
= 1.857 L68.2° A.
the equation: an jnductor, except that being
the dual (Part 5) of a capacitor, We have been preoccupied with Now look at the power developed
Pi = Urmsxlrmsxsin 2liX same of the praperties are in- sinusoidal signals, but voltage or in the individual impedances.
= [(uo l)/("2x"2)]x sin 2<d verted. Following the same lines current may vary in other ways. For this we need to know the pds
= 1/2 Vo 10 sin 2cd. of argument, ifthe applied pd is The voltage may be a r-amp, for across them:
as befare: example, U = 3t. If such a volt-
One point to notice in Fig. 87 is age is applied to an inductor, the UR = RI = 2 x1.857 L68.2°
that the power curve is sym- U = Uo sin wt. current may be calculated from = 3.714 L68.2° V.
metrical about thex-axis. In other Eq. 28 (Part 5).
words, average power 1Szero. Current lags the valtage: Assuming there is no initial Uc = XCI = -j5I
Instantaneous power 1Szero when current: = 5 L-90° x 1.857 L68.2°
either U or I is zero. Power is pos- 1= -10 cas ca, = 9.285 L-21.8° V.
itive and energy 1Sbeing trans-
ferred to the capacitor when both Instantaneaus power is: The phasor diagram in Fig. 88
U andI are ofthe same sign. Work If L has the value 0.5 H, then shows the relationship between
ie being done against the repul- P. = -Uolo sin 2wt 1= 3t2. At any instant, these quantities. Current in the
sive force due to the charge al- = -1!2 Urms lrms sin M, network is in phase with the pd
ready present: the charge on the Pi = UI = 3t x 3t2 = 9t3. across the resistar. It is 90° out
capacitoris increasing. Power is The instantaneous power as- of phase with the pd across the
negative and energy is heing trans- cillates at twice the frequency, Conversely, we can calculate capacitor.
ferred from the capacitor to the and average power is zero. Since powerwhen we are given the ap- As far as the capacitor is con-
circuit when U and I are of op- the power curve is a negative sine plied current. For instance, given cerned, the voltage-current re-
posite sign. The charge ofthe ca- curve, it is the inverse ofthe power that the current through a 0.1 H lationship is the same as in
pacitor is decreasing. curve afFig. 87. Energy is being inductorisI = 4e2t, we useEq. 26 Fig. 87, in which we saw that
Example. A pd ofU = 3 sio 4OO0t stored when U and I are of op- to calculate that the average power is zero. During
is applied across a 22 ~F capaci- popsite sign (power positive) and a whole number of cycles, the
tor. What 18the instantaneous is returned to the circuit when U U = Ldlldt = 8Le21 = 0.8e2/, capa ci tor neither absorbs nor
power when t = 1 ms? From the andl are ofthe same sign (power and supplies energy.
equation, we identify the values negative). P = UI = 0.8e21 x 4e21 = 3.2e41. For the rcsistor; the power may
of Uo = 3 and OJ = 4000 (the fre- Example. A pd U = 4 sin 5000t be calculated as in Eq. 64.
quency is 637 Hz). Calculate 10: 1S applied across a 15 J..l,H induc- The power after 0.5 s is 3.2e2 In a phasor diagram, all pha-
tor. What is the instantaneous =23.6W sors must have the same angu-
10 = U; I Xc = tt; I oe power when t = 1ms? From the Calculations such as these as- lar frequency. Since power has
= 31 (4000 x 22 x 10-6) equation, we identify the values sume that the network is in a double the angular frequency, we
= 34.09 A. ofUo = 4 and (0= 5000. Calculate steady state, by which we mean cannot show it in Fig. 88, nei-
10: that the function for U or I 1S con- ther can we calculate it by vec-
Thus, the current equation is: 10 = Uo/XL = Uol cd: tinuous, not piecewise. If there tor multiplication. However, since
= 41 (5000 x 15 x 10-6) are abrupt changes in U or 1, a we are concerned with only two
1= 34.09 cos 4000t. = 53.33A. reactive impedance, as might be vectors, I and UR, which have the
expected from its name, reacts same direction, ordinery scalar
Instantaneous power is: Thus, the current equation is: to cushion or even to oppose the multiplication is applicable. We
change. In passing from one steady do not use the moduli ofthe com-
Pi = 1/2x3 x 34.09 x sin(2x4000t) 1= -53.33 cos 5000t. state to another there is a tran- plex numbers directly, since these
= 51.14 sin 8000t. sient period. The analysis of what equal the amplitudes ofthe cur-
Instantaneous power is: happens in such briefperiods is rent and voltage. For a.c. power
Whent= 1ms, a fescinating cne which we shall calculations, we need to use r.m.s.
Pi = 51.14 sin 8 = 50.6 W Pi = -1f2 x 4 x 53.33 x sin(2x5000)t leave until a later issue. values as noted above:
= -106.7 sin 10000t.
U,ms = mod UR d2
Inductive circuit Mixed impedance
When t = 1ms, = 3.7141 "2 = 2.626 V.
Similar behaviour is shown by Pi = -106.7 sin 10 = 58.0 W The examples above apply only

ELEKTOR ELECTRONICS NOVEMBER 1993

 FIGURING IT OUT 10 Im
{nng = modI! ~2 plex impedances for the given the total power dissipated Answersto
=1.857 d2 = 1.313 A. frequency. These impedances have equals the total power sup- Test yourself (Part 9)
already been marked in the di- plied.
From Eq. 64: agram. Next, we calculate the 2. Wh at is the average power 1. Product is (22+j7). Conversion
total impedanee, Z, summing for dissipated by a 5 Q resistor gives: 3.606 L56.31" x 6.403
P = 2.626 x 1.313 = 3.45 W. the two branches and then using connected aeross a source L-38.66" = 23.089 L17.65"
the standard formula for im- U = 20 sin rot? = 22 + j7.
This is the true(or active) power pedanees in parallel. 3. A voltage U = 4 ein 2000t is 2. Quotient is (-1 + j2). Conver-
dissipated in the network. applied aeross a 100 JlF ca- sion gives: 8.062 L60.255° di-
There ia another way of'Iook- (10+ j3)(5- j12) pacitor. What 1S (a) the in- vided by 3.606 L-56.310"
Z = 2.236 L116.565" = -1 + j2.
ing at tbe power calculation. (10+ j3)+(5- j12) stantaneous power when
Knowing U, we can calculate I 1 = 5 ms, and (b) the average 3. /1 = 3.444 - j0.415; /2 = 3.112
86-j105
and the phase angle 'P. Then: power? - jO.830; /3 = 0.332 + j1.245.
15-j9
4. A voltage U = 2.5 sin 30001 is 4. /z= 18.570L21.801". Thecur-
P=Scos <p, 135.7L-50.68° applied across a 100 rnH in- rent through the resistor is
17.49L-30.964° ductor. What 1S the instan- h- 12 clockwise. Its value is
where P is the active power, S taneous powerwhen t = 1.5 ms? -3.448 + j8.6204 or 9.2845
=7.758L-19. 716' [Q] L111.803°. The conversionat
the apparent power (Unnslrms), 5. A40 Q resistor and a 100 roH
and cos <p the power factor. If N ow we calculate the current inductor are connected in se- the laststep maygive theangle
U andI are in phase, as in a purely {=U/Z ries across a voltage source as -68.197", but the negative
resistive network, q> = 0° and U = 5 sin 2000t. CalcuJatethe sign ofthe real part ofthe rec-
= lOLOo /7. 758L-19. 716' active power (i.e., that dissi- tangular form shows the re-
cos <p= 1. The whole ofthe ap-
parent power, S, is dissipated =1.289L-19.716 ° [Al pated in the resistor). sultant to be in the 2nd quad-
in the network. If U and I are 6 Given the circuit of Fig. 89, rant ..
90° outofphase, as in purely ca- The angle between the voltage calculat.e the power factor when 5. Without load: 0.37 L-68.3°.
pacitive 01' inductive circuits, phasor and the current phasor the frequency 1S doubled. With load: 0.356 L-63.6".
= 90° and cos ({J = 0, and none
Cf> is 19.719". The power factor, pr,
ofthe power i5 dissipated in the is the eosine of the angle:
network.
With mixed networks, when p{ = cos 19.716 = 0.941.
U and I are neither whol1y in
phase nor wholly out of phase, Another way of art-iving at the
the power factor teIls us wh at same result is to proceed as fol- To be published in November 1993
proportion of S 18 being dissi- lows. After ealculating Z as be-
pated. We call the dissipated por- fore, convert it to reetangular 305 Circuitsdemonstrates the practical aspects ofelec-
tion the active power, P, and form: tronies through projects that can be built at horne, in a
the non-diesipated portion the small workshop, or in the physies or scienee depart-
reactive power, Q. Z = 7.758 L-19.716" ment of schooIs and colleges. There are projects for every-
The apparent power 1Snot ex- =7.303 - j2.617. [Q] one, covermgAudto &Hi-fi; Computers & Mieroprocessors;
pressed in watts, as there may Music and Electrophontcs: Radio, Television and
be no or little actual conversion The real part of Z is the resis-
Communlcatlons: and Test and Measurement.
of energy, but in volt-amperes tive part ofthe total impedance:
ISBN 0 905705 36 X Price 1:14·95*
(VA). With reactive power, some-
times known as the wattless R = 7.303 Q.
component, there is no con- SMT Projects by Owen Bishop. This book describes
version of energy; it is expressed Having found I as above, we ig- the special features of Surface Mount Technology and
in volt-ampere reactive (VAr). nore the argument (the phase how to work with Surface Mount Devtces on the work-
angle) and take only the modu- beneh. It includes over 20 constructtonal projects, all
lus, since current and pd are in ofwhieh have many useful applications in and around
Power factor
phase for resistive elements. Using the horne and workshop. The projects are se1ected and
We shall now see how to calcu- this, we ealculate the true power destgned to take advantage of the dtsttncttve features
late the power factor of a circuit from Eq. 65: of SMT. They introduce the beginner to SMT in a prac-
as in Fig. 89. If you have a cir- tical 'hands-on' way wtthout requtnng expensive spe-
cuit with impedances marked in P = [2R = 1.2892 x 7.303 cial equipment.
ohms, henrys and farads, the = 12.134 W.
The book is for those who want to know more about
first step 1Sto calculate the corn- surfaee mount devices and to gain practical expert-
The ratio between the true power
and the apparent power gives
ence of handling thern.
the power factor:
ISBN0 905705 35 1 Price 1:9-95*

p{=PlS *Prices for orders received at Dur Dorchester office be-

= 12.134/ (10 x 1.289) fore 15 November 1993 are: 305 Circuits j;12·95: SMT
= 0.941. Projects j;8·25.
U = 10
/"0" TO BE CONTINUED Orders otherthan frorn private custorners, forinstance,
bookshops, schools, colleges, should be sent to:
Gazelle Book Services Ltd, Falcon House, Queen
Test yourself Square, Lancaster LAllRN, England_
1. Calculate the power dissipated
930010· x- 89
in each resistor of Fig. 84 if
Elektor Electronics (Publisbing), PO Box 1414, Dorchester,
the two voltage sources are
Dorset, England, DT2 SYH'II
Fig.89. interchanged. Confirm that

ELEKTOR ELECTRONICS NOVEMBER 1993

 :'

FIGURING IT OUT
PART 11 - QUADRIPOLES
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

Quadripoles, otherwise known

as 4-terminal networks, or 2- .....
I,
+-
I,
usually da, 11 is clockwise and
12 is anticlockwise. This elimi-
nates negatives from the equa-
port networks, are the final ex-
cursion into network analysis tions. Comparingthe coefficients
in the present series. The es- in the mesh equations with those
sential features of a quadripole
are illustrated in Fig. 90. The
.....
I,
Port 2
in Eq. 69 and 70, we see that:

network ia linear, its output being 930010-XI-26

Zu = 5 Q;
directly proportional to its input Z12 = Z2l = 2 12;
and varying continuously with Figure 90 Z22 = 6 12.
it. Current.Jj entersPort 1 byane
terminal and leaves by the other. All z-parameters are expressed
Current.Z; enters and leaves Port2. in ohms.
The behaviour of the network
at any given frequency, and as z-parameter model
seen from the outside, may be
eompletely specified by refer- Based on the z-parameters, a
ence to four impedanees, known Port 1 Port 2 quadripole may be represented
as the open-ctrcuit impedance 930010 - XI-tl by the model of Fig. 92, eu and
parameters, 01' z-parameters. Z22 are represented as imped-
Figure 91
Current 11 produces a pd aeross ances in series with the ports. Ul,
Port 1 as the result of its flow- the pd produced at Port 1 as a
ing through z jj.It also produees I,~
Z,
result of 12, is represented by a
a pd aeross Port 2, for which the controlled voltage source, zlzh
relevant irnpedance is Z21. Simi- accounting for the last term of
larly,I2 produces pds across Ports Eq. 69. Similarly, the last term
1 and 2, as the result of flowing of Eq. 70 is represented by the
through 212 and Z22 respectively.
Summingthesepdsateaeh port: "1Port 1 Port 2
controlled voltage souree z21h.
Figure 93 shows the T-network
afFig. 91 as az-paratneter model.
VI = Znl1 + zJ2l2 [Eq. 69J 930010-XI-12
This model can be used to pre-
V2 = Z21h + z21/2 [Eq. 70J dict the behaviour of the net-
Figure 92 work when connected to exter-
These two equations define the nal circuits. Ifwe know the volt-
z-parameters. With Port 2 open- ages at the ports, we use Eq. 69
circuited, so that 12 = 0, we find: and 70 to calculate the currents.
Conversely, given the currents,
From Eq. 69: Z11 = V11I1 we can calculate the voltages.

From Eq. 70: z21 = V21I1

[Eq.

[Eq.

With Port 1 open-circuited.

71J
72J
so
"1:J Example. In the network of
Fig. 91,I1 = 2A,h=3A. Calculate
VI and V2·
From Eq. 69:
VI = 5 x 2 + 2 x 3 = 16 V.
930010·XI-16
that Zj = 0, we find:
Figure 93 From Eq. 70:
From Eq. 69: Z12 = V1/12 V2 = 2 x 2 + 6 x 3 = 22 V.
[Eq. 73J
From Eq. 70: Z22 = V2112 Fig. 91. Mesh current analysis Mesh 1: VI = (3+2)11 + 2h Determining parameters
[Eq.74J (Part 4) provides equations from Mesh 2: U2 = 211 + 6h
which the z-parameters may be If the circuit of the quadripole
As an illustration, consider the calculated: Note that, instead of defining is known, z-parameters may be
T-section resistance network of a11eurrents as cIockwise, as we ca1cuJated by the usual network

ELEKTOR ELECTRON1CS DECEMIlER 1993

 FIGURING IT OUT - 11

techniques, as we did for Fig. 91. pass filter network.

Theseare open-circuitparameters, "'
ron
which means that the ports can
be considered in turn as open cir-
Practical technique
cuits to simplify the calculations. ; Rl
The a-parameters of a eompli-
As an example, take the 1t-net- eated network may be measured
work of Fig. 94. With Port 2 indirectly. By leaving the ap-
open-circuited, 12ftows through Port 1 Port 2 propriate port open-circuit, we
Rl in parallel with R3 and R2. 930010· XI·15 can make either 11or 12equal to
Their combined resistance is zero. Then, measurements ofthe
4.167 Q. From Eq. 71: Figure 94 othercurrentand thevoltage are
substituted in Eq. 71-74. Onee
these values are obtained, we use
c
them to predict the behaviour
The current flowing through R2 of the quadripole under other
is: input and output eonditions.
IR = h x Rl/(Rl + R2 + R3)
w, ,,~'
radi,
iI Example. At a given frequency,
and with Port 2 open-eireuited
= 5h 1 30 = h 1 6. t "" 1.6 K Hz 0---_>__---- --<> (so thatI2 = 0) and 10 V applied
Port 1 Port 2
to Port 1, measurements made
This generates a pd across R2: on an unknown network are:
930010· XI· 13
Ul = 10 V;
U2 = 5 V;
Figure 95 a
11 = 0.25 A.
Now we can calculate Z21, using With Port 1 open-cireuited (so
4.031..:1.12
Eq.72: that h = 0) and 10 V applied to
Port 2, measurements made are:
Z21 = U21 11 = 2.5 Q. U1 = 0.2 V;
U2 = 10 V;
With Port 1 open-circuited, sim- 12 = 0.02 A.
ilar calculations show that: u·l
w"" 10'
radis
With Port 2 open-cireuited, sub-
stituting in Eq. 69 and solving
Z12 = 2.5 Q; for zll:
Port 1
Z22 = 7:5 Q.
930010· XI· 14 Zn = 1010.25 = 40 Q.
Now suppose that R3ie replaced Figure 95b
bya 10 ~F capacitor (Fig. 96a). Substituting in Eq. 70 and solv-
If co = 104, then Xc = -1 1jwC ing for Z21:
= -j 10 Q. Wecan use this in plaee ao c
ofthe resistance of'Rgin the cal- '21 = 5.0/25 = 20 Q.

"1'J cl"
eulations above. The eornbined
impedanee of R2 and Xc is Similarly, with Port 1 open-cir-
(15 - j10) Q. In parallel with . euited:
5 n, the total resistance is
Z12 = 0.2 10.02 = 10 Q;
5 (15 - jlO) 1 (5 + 15 - jlO) Port 1 Port 2 Z22 = 1010.02 = 500 Q.
=(15-jlO)/(4-j2) 930010· XI· 17
=18.03 L-33.69" /4.47 L-26.57" Figure 96 shows these results
= 4.03 L-7.12". Figure 96
in corpora ted in the z-parame-
Zl1 = 4.03 L-7.12" Q. tel' model. The parameters may
now be used to predict network
Techniques for handl ing corn- behaviour with other values of
plex numbers are described in pd or current at the same fre-
Part 8. queney.
Current flowing through R2
i5:
IR = 11 x 5 1(20 - j10).
Other parameters
Port 1
The z-parameters show how the
Port 2
The pd aeross R2 is: voltages are determined by the
930010- XI· 18
currents and impedances as de-
U2 = IRR2 = 75h 1 (20 - j10) Figure 97 fined in Eq. 69 and 70. We ean
= h x 75 1 22.36 L-26.57" write a corresponding pair of
equations showing how the cur-
Z21 = U2 1 t, rents are determined by the volt-
= 75 122.36 L-26.57" The Port 1 vo1tage source, The output pd is 0.83 times the agesand another setof parameters,
= 3.35 L26.57" Q Z1212. and z22 may be omitted input pd and there is a phase lead the y-parameters. This is an-
from the model(Fig. 95b). Then, of 33.69". other instanee ofduals (seePart5),
From these two parameters we from Eq. 69 and 70: Repeatingthis ealeulation with so it should eome as no surprise
can ealeulate the ga in of this 00 = 2 x 104 gives a gain of to find that they-parametermodel
network,assmingthatitis lightly gain = U21 Ul = Z21 1 Z11 0.95 L18.43°. Gain is increased of a quadri pole has an admit-
loaded. With a high-impedance = 3.35 L26.57" 1 4.03 L-7.12" and phase lead is reduced, as tance in parallel with each port,
load,I2 = O. = 0.83 L33.69". might be expeeted with a high- and a pair ofcontrolled current

ELEKTOR ELECTRONICS DECEMßER 1993

 GENERALINTEREST

sourees (Fig. 97). As might be ex- Wenowhave all the parameters

peeted from the duality, y-par- and can setup themodel (Fig. 99).
ameters are calculated er mea- Use this to make predictions.
sured by short-circuiting Olle For example, evaluate the CUT-
'0 40
or the other of the ports. The rents if U1 = 10 V and U2 = 5 V.
equations for y-parametera are: On the Port 1 side, the current
c through the resistor is the re-
"
1) = YUU1 + Y12U2: '; t sult of the pd across it, which
is:
12 = Y21U1 + Y22U2· I,
10 - h12U2 = 10 - U2/3
Port 1 Port 2
They-parameters are admittances, 't---' = 10 - 5 1 3 = 8.333
93OO'O·)(j·'9
so their unit is the siemens (8).
They are used in calculations 930010· XI ·20
11 = 8.333 1 4.333 = 1.923 A.
similar to those above, if work-
ing with admittances is more Figure 98 On the Port 2 side, the eurrent
convenient. through the6-ohmresistor(equiv-
A third setofparameters con- alent to])6 S admittanee) is 5/6A.
+-1,
sists of the h-parameters. The The current from the controlled
h stands for 'hybrid', because souree is -111 3 = -1.923 1 3 =
these parameters are based on -0.641 A. 12 equals the eurrent
through the 6-ohm resistor plus

"1 'J~
a mix O[UI and Jg, the pd across
Port 1 and the eurrent through the current producedbythe source,
Port 2: which, being negative, flows in
the opposite direction to the arrow:
U1 = hUl1 + h12U2 (Eq. 75] Port 1 P~!.
12 = h2111 + h22U2 (Eq. 76] 9Jool0·XI·21 12 = 5/6 -0.641 = 0.1923 A.
Figure 99
We will find the h-parameters for The current gain of this net-
the eireuitofFig. 91. With Port 2 work is 121 h = 0.1.
short-circuited, as in Fig. 98a, r--------------., This simple example has the
(so that U2 = 0), from Eq. 75: I h" I merit thatits results can be ehecked
easily by mesh analysis ofFig. 91,
with applied vo!tages of 10 V
and 5 V. The reader may care to
But VI / /1 equals the resistance work this as an exercise, and
of the network when Port 2 is confirm that the h-parameter
short-circuited: this is 3 Q in se- technique gives the correct re-
ries with 2 Q and 4 Q in paral- L J
sults.
leL This gives a total resistance Port 1 Port 2
of 4.33 Q. Thus, hu = 4.33 Q. 930010·XI·22 Extending h-parameters
The unit of h11 is ohms.
From Eq. 76, we have: Figure 100 Consider the model afFig. 100,
the circuit details of which are
not relevant for the moment. It
receives input UI from a variable
Figure 98b shows the shorled pd source and its output U2 is fed
network redrawn to make ftow to load resistor R. The h-par-
ameters are marked on the fig-

tu, -
of current easier to visualise. I,
The pd across AB is: b ure. By Ohm's law:
lu,
U = 2(11 + 12) = --412: 12 =-U21 R = -U2/1OOO
(Eq. 77]
11 = -3h
930010· XI· 24

Substituting in the equation for Also, from Eq. 76:

Figure 101 a
h21'
12 = h21I1 + h22U2
h21=-1/3. h. C
=1012 + 0.0012U2·
b

h21 'is a ratio between two CUT- Substituting the value of/2 from

+
rents: it is a pure number with- Eq.77:
out units. R

With Port 1open-circuited, -U2/1000 = lOh + 0.0012U2

as in Fig. 91, so that h = 0, we
have from Eq. 75:
1" .. I1 =-{0.0012U2+ U2/1000)/1O
Port 1 e Port 2 = -U2 (0.0012 + 0001) 110
h12 = U11 U2 = 216 = 1/3.
930010· XI· 23
= -U2 (0.0022) 110
Figure 101 b = -0.00022U2·
Here, the 4-ohm and 2-ohm re-
sistors act as a potential divider, Substituting this value and the
and U2 is divided by 3 to pro- From Eq. 76: This is the admittance of the values of the h-parameters in
vide UI. Again, this is a ratio 4-ohm and 2-ohm resistars in Eq.75:
with no units. series; its unit is the siemens.

ELEKTORELECTRONICSDECEMBER1993
 FIGURING IT OUT - 11 m
Ul = -D.00022U2hll nal is in phase with the input
=-0.00022U2x
+ h12U2
50 + 0.005U2.
, ? signaL Typieally, hib is small,
.. Uv t U2 = -0.011 + 0.005
1" around 20-30 o.
1"
= -D.006.
r
u," hr• ~
t S
h;. ~
r
Uou1 ~
•
A
In the common-collector(emit-
ter-follower) mcdel, hic is large
The pd gain ofthe network is the (1 k.Q or more), and hfc is typi-
reciprocal of this:
I I cally -100, giving large current
gain with no phase shift (one of
U2/ Ul = -1 / 0.006 = -167. 9JOOIO·Xl·25
the main purposes of using a
c-c amplifier). b« is 1, giving
That is, the pd gain is -167. Figure 102 unity voltage gain, as required
The network is an inverting volt- in an emitter folIower, while hoc
age amplifier. It has the fea- has a moderate value.
tures of a common-emitter arn pli- the collector current is virtu- TO BE CONTINUED
fier based on an n-p-n bipolar ally independent of supply volt- Voltage gain = Uout / Uin
junction transistor (Fig. IOla). age. Thecurrent 1S split between =-hreIbR/hie1b Test yourself
In the figure we sbow a pd source hoe and R, which are in paral- = -hre R / hie·
U, but omit biasing resistors, lel. By the rules of current divi- 1. Calculate the z-parameters
which can be considered to be sion: In Fig. 102, the voltage gain is for Fig. 91, given Rj ~ 100,
partofthe pd source. Comparing R2 = 140 and R3 = 6 o.
the two parts of this figure, we 1 ~_h_ 1 x l/hoe (-150 x 103) /103 = -150. 2. Calculate Z11 and Z21 for
L ,", b l/h +R
see that 11 i8 the base current oe Fig. 95a, given that Rj = 100,
Ib, while h Is the collector cur- The output signal is 180 out of 0
R2 = 12 0, C = 1 ~F and
rent, I c-The h- parameters are = -hre1b / (1 + hoeR). phase with the input signaL w= 1()4radis. Whatisthevolt-
given special subscripts to relate age gain ofthis network?
them to the operation ofthetran- Current gain is Other amplifiers 3. In an unknown quadripole
sistor: circuit, Ul = 1 V, U2 = 3 V,
h / IB = -hre / (1 + hoe RL)) The models for common-base 11 = 20 ~ and h = 2.5 mA
h1l becomes hie, the iriput re- [Eq.78] and eommon-collector ampli- Calculate the h-parameters.
sistance; fiers have the same layout as
h21 becomes hfe, the forward Given the values of Fig. 102 thatofthe eommon-emitterampli- Answers to
current ratio, or gain; (hoe = 10-5), the eurrent gain is fier. The subscripts on the par-
h 12becomes hre• the reverse ameters end in 'b' or 'c', but oth- Test yourself (Part 10)
feedback voltage ratio; -150/ (1 + 10-5 X 103) erwise the diagrams are the same. 1. I} = 0.56 A; 12 = 0.04 A;
h22 becomes hoe, the output = -150 / 1.01 = -148.5. The typieal values of the par- 13 = 0.32 A In 3 0: 0.9408 W;
admittanee. ameters are not neeessarily the in 2 0: 0.0032 W; in 4 0:
This is a large current gain, as same as in the common-emit- 1.0816 W; in 1 0: 0.0576 W;
The 'e' in each subscript refers is typical of a common-emitter ter model. For example, the ab- in 7 0: 0.7168 W 4 V source:
to the fact that the transistor amplifier. The negative sign in- solute value of hfbis a little less 2.24 W; 2 V source: 0.56 W
is in the eommon-emitter con- dieates that it is 1800 out of than unity, typieally -0.99. Total for resistors = total for
figuration. Eaeh of these par- phase with the input signal. Tagether with the fact that hob sources = 2.8 W.
ameters is measurable fora given In practiee, it is reasonahle is exceedingly low (a few hun- 2. 40 W
transistor, or ean be obtained . to simplify the model further dred nano siemens, equivalent 3. (a) 0.547 W; (b) 0 W
from a data sheet. hie and hfe without undue loss ofpreeision. to over 1 Mn), this means that 4. 1.627 mW
are measured with the transis- The output admittanee is small current gain is very slightly less 5. 12 mW
tor output short-eireuited; hre (only a few microsiemens), sohoe than 1 in any practieal com- 6. XL = j6; Xc = -j6. Power fac-
end hoe are measured with an can usually beomitted, too.Given mon-base amplifier. Also, sinee tor = 0.944.
open-cireuit input. Given these that hoe is very smalI, the term hfb is negative, the output sig-
parameters, the behaviour ofthe hoeRcanbeomittedfromEq.78,
amplifier ean be predicted, pro- so that
vided that it is operating in the
linear partof'its range, and with current gain ""-hfe = -150.
small signals.
u
The values of hre and hoe for Other characteristies of the o 0
most transistors allow some sirn- amplifier are ealeulated in a j
plifieation of the model. sirnilar way by making suitable
Figure 102 shows a common- assumptions and analysing the
emitter arnplifier with typical network with the use of tech-
values for the h-parameters. niques described in previous 000000
We have omitted the eontrolled parts. Equations may become in-
voltage source, becausea.cis typ- volved, hut the mathematies is
ieally only 0.0001, with the re- simple. For example, to caleu-
sult that the voltage souree ean late voltage gain, we first eon-
usual1y be ignored. R is the load sider the left-hand mesh of
resistor in the collector circuit. Fig.102:
In the model, we are speeifying
that the controlJedcurrent souree
is drawing the collector current
through that resistor; ultimately In the right-hand mesh, ignor-
from the negaive rail through ing hoe:
the low impedanee ofthe power
supply. We can say this because
tMI(
ELEKTOR ELECTRONTCS DECEMBER 1993
-m
FIGURING IT OUT
PART 12 - BUILDING MODELS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a coUection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

[Eq.82]
Fpleigurecircuit.
l03showsa verysim-
The capacitor is
pacitor 1S reduced, causing the
pd across it to be reduced, too.
q = uC = iRC negative sign indicates that the
charge 1S decreasing with time.
previously charged to 9 V by an The effect of this is that the cur- Combining Eq. 81 and 82: Readers will recognize the divi-
external source. The source is re- rentisreduced. Itdecreases from sor RC as the time constant
moved, current i flows through 490 I-lA,fallingeventually to zero q = iRC = dqldtxRC (T) of the circuit.
the resistor, and the pd across (at least, in practical terms: the- Because it contains a deriva-
thecapacitorgradually falls. We oretically, it never reaches zero). dqldt = -<i I RC [Eq.83] tiveor differential, Eq. 83is know
are asked to calculate how lang This circuit is one in which key as a differential equation. The
it takes fOT the current to fall to quantities are changingin time. Equation 83 is a model of the differential terms in such an equa-
100 1-lA. This 1S the reason for uaing the circuit, relating the instanta- tion express rates of change, so
lower-case symbols i and u; they neous current (represented by such equations are widely used
are to indicate quantities that dq I dt) to the instantaneous charge, a models for dynamic systems.

~~1
vary in time. By contrast, quan- with Rand C as constants. The Tbey areidealfor modelling elec-
tities such as C and Rare con-
stant and are represented by
470~ iiiii U [~ capitalletters.
The first step of building a
Derivatives
Given that y is a function of x
Solving
equation
the growthldecay

model of the circuit 1S to write (for ex.ample, y = 4 + 3x), the Equation 83 is a typical ex-
equations to express all the re- symbol dyldx stands for the ample ofthe decay eqaation.
930010· XII· 103
lationships we know. For this first derivative of y with re- It specifies the rate at which
circuit, we know that spect to x. It is the rate of the charge diminishes or de-
Fig.103 change of y with reepect to x. cays. With a positive coeffi-
q = uC [Eq. 79] lfwethinkofagraph in which cient on the right ofthe equa-
y is plotted against x, dyldx is tion, it specifies growth.
One way to ans wer this ques- where q is the charge on the ca- the gradient ofthe graph at any Given a differential equa-
tion is to set up the circuit, in- pacitor at any instant. We also point. Finding the derivative tion, itmay be possibletowrite
c1uding a microam meter to mea- know that (sometimes called the differ- it in the form
sure i, and measure the time ential) of a function is known
with a stop-watch. Timingis not u = iR (Ohm's law)[Eq. 80] as differentiation. There are dyldx = ky.
the difficulty; discharging takes a number of simple rules for
several tensofseconds. 'T'heprob- and finding derivatives, some of Note that the right-hand side
lern is to find a capacitor of ex- which were given in Part 5. consista of only a single term
act1y 470 u.E We are driven to an- i = dq/dt [Eq.81] 'When a functionhas been dif- in y (no x, DO y2, and so on).
other way of answering the ques- ferentiated, the first derivative Also, kmust be a constant, but
tion - writing an equation which The symbol dqldt needs some is a new function, which ean may be positive (growth) or
behaves in exactly the same way explanation. A symbol of ihis also he differentiated. This gives negative (decay), In Eq. 83,
as the theoretical circuit. Such form is know as a derivative, the second derivative, sym- k = -IIRC. The solution of
an equationis a model ofthe cir- more particularly, a first deriva- hol d2yldx2. We can go on and such an equation is:
cuit. tive (see Box 1). In this case, differentiate this again to ob-
dqldt is the rate of change of tain the third derivative, y = Aekx,
Modelling the charge on the capacitor in d3yIctx3,and so on.
coulombs (C) per second (5). By Differential equations may where A is an arbitrary con-
It is easy to find the current at definition, the ampere is a flow contain fugt, second, third er stant. The value of A may be
the instant the external source of charge of 1 coulomb per sec- higher derivatives. The order calculated if a setofbordercon-
is removed. At that instent, we ond (1 Os-tjandEq. 81 expresses of a differential equation is ditions is specified (see main
have a pd of 9 V across a 22 kQ the lass of charge from the ca- that of the highest differen- text).
resistor and pacitor in these terms. tial present. Equation 83 is a
Having jotted down the rele- first-order equation. Next
i = 9/22xl03 = 409 ~A. vant equations, the next step is month we shall look at some
to relate q and dq/dt in a single second-order equations.
But, as soon as thatcurrentstarts equation. Cornbining Eq. 79 and
to flow, the charge on the ca- 80 gives: Box 1 Box 2

ELEKTOR ELECTRONICS JANUARY 1994

 GENERAL INTEREST

= 4.23xlO-3 C 440x10..o = Ae-O.4545

Substitutiog q = 4.23x10-3 C = AxO.6348,
and t = 0 io Eq. 84, the index of
q = A e-tlRC e becomes zero: .. A = (440x10..o)/0.6348

4.23xlO-3 = AeO = A. = 6.931x1Q-4.

We now have a value for A to Having used the border condi-

put io Eq. 84: tions to find the arbitrary con-
stantA, find theinitial charge by
q = 4.23e-O·0967tx10-3 puttiog t = 0 in Eq. 84:
[Eq.86]
This is a particular aolution of q = A = 6.931x10-4.
Eq. 83. Its graph is the curve
drawn as a heavy line in This gives the initial pd:
Fig. 104, and shows how the
charge on the capacitorfalls with u = q/C
time at an ever-decreasing rate.
Knowing the instantaneous = (6.931x10-4)/(220x10-O)
valueofq at any time, we cancal-
culateu and i at thesame instant. = 3.15 V.
From Eq. 79 and 80:
The pd after 15 s is found in a
, ie i = u/R =q/RC. similar way. The index of e is
t [s 1 -------to- -15/RC = -0.6818.
1130010-11U _1001
From Eq. 86:
q = 6.931x1Q-4xe-O·6818
Fig.l04 i = (4.23e-O·0967tx10-3)/RC
= 3.505x10-4.
= (4.23e-0.0967tx10-3)/10.34
tronic circuits, and we shall use lution which relates to thevalues u = (3.505xlO-4)/(220x10-6)
thern to build several models, this and conditions of our circuit. ~ 409.1e-O·0967x1Q-S.
month and next. The key symbol in Eq. 84 is = 1.59 V.
e, the exponential constant. This Wecou1d plota curve for i against
has a negative index, showing t, and this would have a similar First-order equations
Solving the equation that q is decreasing with time. form to the curve for Fig. 86. In
The next step is to solve the dif- The index is proportional to t, but this problern, we want to know A first-crder differential equa-
ferential equation. We need to be has Rand C as divisors to take t when i = 100 1lA. Substituting tion has the form:
able to find the value of q at any into account the effects of sub- i = 100 x 10..0= 10-4:
given time t. Having faund q, stituting different capacitors and dyldx + f(x)y = g(x).
we can easily find what u and i resistors in the circuit. With the 0.2444 = e-0.0967t.
are at that time. There 15 no values given in Fig. 103: Both fex) and g(x) are functions
space to go inta the steps of'solv-: Taking natural logarithrns: of x, but one 01' both may be con-
ing Eq. 83, butapplyingthe rules q = Ae-O·0967t [Eq. 85] stants 01' zero, thus simplifying
quoted in Box 2 gives UB the so- In 0.2444 = -0.0967t the equation.
The quantity A, known as the ar- An equation such as this is
lution:
bitrary constant, is the next -1.4088 = -0.0967t needed for mode Hing the circuit
q = Ae-t/RC [Eq. 84] to be determined. This constant of Fig. 105. Here the capacitor
is there because the technique t = 14.57, is being charged through a re-
For such a simple circuit, this ap- for solving Eq. 83 produces an in- sistor by a variable pd, u. As be-
pears to be rather a complicated finitely large number of general that is, the currentfalls to 100 ~ fore, we writedown relevantequa-
solution but, when looked at an- solutions, depending on the the after 14.57 s. tions:
alytically, it makes sense. We value we allot to A. Figure 104
showsjust 10 ofthe possibilities. UR = Ri = R-dq /dt;
will defer the original task of Falling pd
findingthe current until we have Which one of these represents
examinedin more detail tbe way the circuit in our problem? Consider the same circuit but uc = q/C.
in which q varies with t. At pre- with C = 220 ~F andR ~ 100 kn.
sent, we have what is called the The capacitor is charged, tben al- We can relate these by making
Border conditions lowed to discharge for 10 s, after use of the fact that, at any in-
general solution to Eq. 83. We
need to find the particular so- If the values Rand C were all which time the pd across it is stant, KVL applies:
that were known, we could not 2 V. Fiod the initial pd and the
R find A. Any of the curves in pd after 15 s. U = UR + uC

Fig. 104, or any of many others Tbe differential equation 1S

+-- '" not drawn there, fits the differ- Eq. 83, and the general solution u~R·dq/dt+q/C.
'R
+ ential equation. But we also know is Eq. 84 as above. The border
uc t c
that the capacitor is charged to conditions are that u = 2 V when Rearranging terms and divid-
u
;I 100).1
9 V when t = o. These two values t = 2 s. ing by R to give dq/dt unity as
specify the border conditions. When u = 2, uC = 2x220 x10-6 its coefficient:
930010· XII • 105
Whent =O,C =470IlF andu = 9V, = 440x1Q-6. In Eq. 84, the index
by substitution io Eq. 79: of e is -10/RC = -0.4545. dq/dt + q/RC = u/R.
Substituting io Eq. 84: [Eq. 87]
Fig, 105 q ~ uC = 9x470xlO..o This is the first-order equation

ELEKTOR ELECTRONICS JANUARY 1994

 F[GURING [T OUT - PART [2

that models the circuit. This is the particular solution, In (1/3) = -t2IRC;
The equation may be solved assuming that the source is
°
--
by the routine summarized in switched on at t = and holds a t2 = 1.0986RC.
R
220k
Box 3. This involves calculat- constant value U. Figure 106
ing a quantity known as the In- shows thecurve for thisequation The monostable period is t2-tl UR
+ l
tegration factor, the purpose if U = 6. The charge on the ca- =(1.0986-0.4055)RC = 0.693IRC. U
of which is to produce an equa- pacitor rises rapidly at first, but This is the period quoted in the
tion that is easyto simplify later. at a gradually decreasing rate data sheet for this timer. 100mH

The routine requires integre- until it alm ost, but never quite,
tion: some ofthe prccedures for reaches CU.
this were outlined in Part 5.
Inducfors 930010 XII· 107

We identify the function [(t) The derivative which models the

as the constant lIRC andg(t) as
Charging firnes
action of an inductor is: Fig.107
uiR. Arelated equation is obtained by
dividing Eq. 89 throughout by u = L·dildt.
F(!) = J lIRC dt = t /RC. The C. The pd aeross the eapaeitor tions in amoment. Here is the
Integration factor is eil Re. is q l C, which we will refer to as
The induced e.m.f., ll, is proper- solution at this stage:
uc. Then, dividing: tional to the inductance Land
G(t) = J (uIR) et/RC the rate of change of current, i. i = e-Rt/L J (u/L) eR,1L dt +
uC = U(l-e-'IRC) For the circuit of Fig. 107, the + Ae-R'IL
If we assume that u is a con- equations are:
stant, the integration is simpler. uCIU = 1_e-tIRC [Eq. 90J Note that the index of e is some-
We will now refer to it as U. times + and sometimes -. If U is
Equation 90 ean be used to de- a function of t, for example, u is
G(l) = RC [CU I R) e'l RCJ termine the time to reach any uL = L·dildt. an alternating pd with a function
given pd llC expressed as a frac- such as u. = 2sin rot, we integrate
= CUeURC tion of the applied pd U. In the ByKVL: the flrst expression on the right,
standard 555 timer circuit, the aod obtain a particular solution.
The solution is: monostable period is the time U = uR + uL = Ri + L-di/dt. Here we assume that u. is con-
taken for the pd aeross the ca- stant and call it U. Integrating
q =_e-tIRC·CUe'IRC + Ae-t/RC. pacitor to rise from U/3 to 2U/3, Rearranging terms and divid- the term gives:
where U is the supply voltage. ing throughou t by L gi ves the dif-
In the first term on the right, ferential equation; e-Rt/L CU /L)·eRt/L (L/R)
the indices of e total zero, and llCIU = 113 at tL =IlIR.
eO = 1: dildt + (R /L)i = u /L.
1/3 = 1-e-' iRC Now the solution is simplified to:
q = CU + Ae-tIRC [Eq.88J Applying the rules for solvi ng
-2/3 = -e-' iRC. first-order equations, we have: i = U /R + Ae-R'IL [Eq.91J
This is the general solution. If
the capacitor has zero charge Negatingand taking logarithms: [(t) = R /L; Forthe particular solution, take
when t = 0, we have a border the case in which the pd is 10 V
condition: In (2/3) = -t)IRC F(t) = Rt/L. up to the time t = 0. The eurrent
·is steady (dildt = 0), being de-
0= CU + A, t) = 0.4055RC. Integrating factor is eR'IL termined only by the pd and the
resistor:
.. A= -Cil. g(l) = u/L;
i = 10/(220xl03) = 45.45xlO-6 .
Equation 88 beeomes: J
.'. G(t) = (ul L) eRUL dt.
Substituting in Eq. 91:
q = CU (l-e-'IRC) [Eq. 89J We make simplifying assump-

6 Solving{irst-orrlerequations Integration
Given an equation inthe form Integratinga constant,1IRC:

q
i dyldx + [(x) + g(x). Table I, Part 2, states that
the integral oftn = tn+!f(n+1),
[C X 10-4] 1. Integrate j[(x) dx and call provided that n " -1. Since t
4 it F(x). does not appear intbeexpression
2. The integrating factor is beingintegrated,n =0. Thein-
q=CU(1·e-tIRC) eRx). tegral of /0 is t. Multiplied by
3. Integrate Jgfx) eR%) dx and the constant lJRC, the integral
call it G(x). ofllRCist/RC.
2 4. The solution is: Integrating ellRC : Table 1
y = e-Flx) G(x) + Ae-Ff%). in Part 2 states that the inte-
gral off'Il1is f'Il1/a.Here, a = lIRC.
Note the negative indices at The integral is
stage 4. Ais thearbitrary COD- (etIRC)/(1!RC) = ReetiRe.
staut; find it by using bound-
o 2 3 4 5 ary conditions.
t [sl--.~ 930010· XII 106

Fig.106 Box 3 Box 4

ELEKTOR ELECTRON[CS JANUARY [994

 F1GUR1NG lT OUT - PART 12

Please mention ELEKTOR ELECTRONICS when contacting advertisers

Call us now! We have the widest range 01

components available - At competitive prices!

CABLES
/
CAPACITORS VIOEO HEAOS
I
SPEAKERS CONNECTORS
~
SEMICONOUCTORS

A = 45.45xlQ-6 - U I R. sistance, mathematicaHy, it is easy to in- C = 10 ~F and R = 39 kQ.

We can also investigate what sert new values and determine Rise time is defined as time
Putting this value of A into Eq. 91, occurs ifthe voltage is reduced its behaviour. This is much eas- for the pd across the capaci-
we have the particular solution: not to zero, but to a lower volt- ier than building a circuit, then tor to rise from 10% of U to
age, say, 5 V: replacing resistors, capacitors 90% ofU.
i = U I R + (45.45x10-L 01' inductors. Differential equa- 3.InthecircuitofFig.l07,but
U I R) e-Rt/L [Eq. 92J i = 51R + (45.45x10-L51R) tions are a powerful way of'fig- withR=4.7lillandL=20rnH,
xe-Rl/L. uring 'it out'. a steady voltage of 1 V is ap-
Whent =0, the applied pd is 10V Next month we shall model plied urrti l t = O. Then the
and we have a constant current Or we can model a suddenly in- more complicated circuits with voltage is changed instantly
of 45.45 ~A flowing through the creased voltage to, say, 22 V: second-order equations. to 5 V. Calculate the current
circuit. The voltage SOUTce is 2 us later.
suddenly reduced to 0 V. To find i = 221R + (45.45x10-L22/R) Test yourself
what happens, insert U = 0 in xe-R'IL. Answers to
Eq.92: 1. Writethe particular solution
Figure 109 shows the graphs, for the current in the circuit Test yourself (Part 11)
i = OIR + (45.45xlQ-6- which clearly reflect the action ofFig.l03, butwith C.e 150~, 1. Zl1 = 6.67 Q; Z21= 4 Q;
01R) e-R'IL; ofan inductor in resistingchanges R = 4.7 kQ and an initial pd z12 = 4.67 Q; z22 = 7.47 Q.
in the Ilow of current through of 6 V. Find (a) the pd across
i = 45.45x10-6xe-Rt/ L [Eq. 93] it. We could also modify Eq. 93 the capacitor 1 s after di s- 2. zl1 = 9.83 L-5.57°;
by changing the value of Land charge has begun, and (b) z21 = 1.17 L77.59°;
Figure 108 is the graph of this. seeing what effect this has on the current2 s after discharge gain = 0.12 L83.16°.
It shows the current gradually the extent and timing of CUT- has begun.
dying away at a rate depend- rent changes. 2. Find therise time forthe eir- 3. hl1=50lill;h21=125;h12=1I3;
ing on the inductance and Te- Once a model has been built cuit of Fig. 105, if U = 10 V, h22 = 833 us.

100

.0
80
22 22

ii j =45 .45x 10-6x e-R1/L

i 60
i=R+(45.45X10-6-R )e·R1/L

["Al i
["Al
20
40
.55
'=R+(45.45x10-6-"'R )e·RtIL

20

o 0.2 0.4 0.6 0.8 1.0 o 0.2 0.' 0.6 0.8 1.0
t [~sl--"~ 9300111- XII- 1CH!1 t [~sl--"~ 9JOO10-XII 109

Fig.10B Fig.109

ELEKTOR ELECTRONICS JANUARY 1994

~

FIGURING IT OUT
PART 13 - SECOND-ORDER MODELS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rnle-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

Last month we examined ways by L to make unity the coefficient Substituting in Eq. 97:
of using first-order differential of the first term: [Eq.97J
equations to build circuit mod- i= Ae -1382t + Be--3618L
d2i R di 1. [Eq.94] If D is zero, the roots are both
els. This month we extend the -+-'-+-'l=O
2
methods to include second-order dt L dt LC m and the solution to Eq. 94 is: [Eq. 100J
equations. These allow us to model Equation 94 models the circuit. 'I'his is the general solution. We
circuits of greater complexity, It now remains to solve it and [Eq.98] can apply this to any set of bor-
such as that in Fig. 110. substitute actual values of R, L der conditions. We can imagine

--
and C. If D is negative, we ca1culate the voltage ftuctuating, perhaps
R k = .,j (-Dl2) and the solution is: regularly, perhaps irregularly,
causing a varying current in the
Auxiliary equation i =Ae-{l/2 cos Iü+ Be-ft/2 sin kt circuit. Then, when t = 0, the
",
For an equation of the form of voltage 1S suddenly held con-
Eq. 94, in which R, Land C are [Eq. 99] stant (Eq. 94). The models teils
" constants, there is an auxiliary All three equations are general wh at happens after that. For
equation ofthe form: solutions (see Part 12) and have this example, suppose that the
two arbitrary constants, A and current is 2 rnA wen tirning be-
--.
"c
m2 +fm +s= 0, [Eq. 95] B. Last rnonth we had just one gins, 01' i = 2xl0-3 when t = O. Also
such constant and needed one assume a second border condi-
c tion that the tate ofchange of cur-
in whichfis the coefficientof di/dt border condition in order to find
930010-13-12 and g is the coefficient of i. In theparticularsolution. Thereare rent, dildt, is 0.05 As-t when
Fig.110 Eq. 94, f = R JL and g = 1ILC. two constants, and we need two t = 0. With t equal to zero, e has
Solving Eq. 95 (a straightfor- border conditions for second-order zero index in both terms and so
'I'his includes resistive, inductive ward quadratic equation) for In equations. equals unity. Substituting the
and capacitive elements. By Kirch- is much easierthan solving Eq. 94 values for the first border con-
hoff's Voltaga Law: for i. With values offandg sub- dition in Eq. 100:
Worked example
stituted in Eq. 95:
GivenR=500n,L= 100 mH and 2xlO-3 =A + B
m2 + R m+_1_=0 [Eq.96] C = 2 MF:f = R IL = 500/100xlO-3
As we did last rnonth. we use L LC = 5000; andg = 11LC= 1I100x10-3 .. B = 0.002-A.
lower-case letters for quantities 'I'his is solvable by applying the x2x10-0= 5x1()6.From thesevalues
which are inherent1y variable well-known quadratic formula offandg: Substitute this value for B in
in time and capitals forconstants. which, when applied to Eq. 95, Eq.100:
Substituting equivalent expres- is: D = f2 - 4g = 5000L20x106
i=A(e-1382t _e-3618t)
sions for the voltages, based on = 5x106.
Eq. 26 ofPart 4, Ohm's law, and
m
-f±~ +0.002e-36181
Eq. 19 ofPart 4: 2 This is positive, so the equation
has two real solutions: [Eq.101J
i
Ld-+ R't+-=u
q The value ofthe expression f2 - 4g Ta incorporate the effect of the
dt C (known as the discriminant,D) -f+JD second border condition into the
We will restriet the analysis to determines what kind ofsolution 2 equation, we must first differ-
situations in which u is constant, the equation has: -5000+2236 entiate(seePart5)Eq.101 to ob-
henee duldt = O. Then, differen- lf D is positive, the equation 2 tain an equation for dildt:
tiating both si des of this equa- has two real roots. =-1382
tion: If D is zero, the equation has di = A( -1382e -13821
two equal roots.
dt
2
Ld-+i Rdi i_
-+--O If'D is negative, the equation -f-JD +3618e-36181) _7.236e-36181
dt2 dt C has two irnaginary roots. 2
In differentiatingthe third term, IfD 1Spositive, the two real roots -5000-2236 [Eq. 102]
useq =it, and therefore dq/dt =i. are m1 and m2 and the solution 2 If dildt = 0.05 when t = 0:
The term d2i/dt2 makes this a sec- to Eq. 94 is: =-3618
ond-order equation. Divide through 0.05 = A(-1382 + 3618)-7.236.

ELEKTOR ELECTRONICS FEßRUARV 1994

 FIGURING IT OUT - PART 13

A = (0.05 + 7.236) 12236 the second border condition into

=0.003258. ! (mAl
2 _ j=0.002
this, which involves multiply-
wllent=O ing the indices by 0.001:
Substituting this in Eq. 101:
0.003= A(e-1.382 _e-3·618)
;=0.003258(e-1382t _e-3618/) +0.002e-3.618
+0. 002e -36181
=0.2242A+5.3673x 10-5
Rearranging terms gives: A=0.01314

i=0.003258e-13821
Substituting the new value of A
-0.001258e-36181 into Eq. 101:
i= O. 01314(e-13821 _e-3618t)
[Eq. 103]
This is the particular solution +0.002e-36181
and Fig. 111 shows its graph. It
=0.01314e-1382t
shows that at t = 0 the current
is 2 m.A, as specified. The in- o 2 3 t(ms) _0.01114e-36181
crease of current when t = 0 1Stoo
small to show on this graph. The Yet another solution, with the
tangent to the curve at this point graph shown in Fig.113. Current
would slope up to the right by Fig. 111 increases even further before
only 3.6°, This upward slope i8 damping takes effect.
almost instantly countered by
i (mA)
the relatively strang dampingef-
fects of capacitance and induc-
Critical damping
2
tance. By the end ofthe third mil- The curves we have seen so far
lisecond, the currenthas been al- represent overdamping ofthe cur-
most entirely damped out. rent. Current is reduced to zero
Figure 112 shows what hap- fairly promptly. The model can
pens if we keep the first border I", 0.004131e -'3621_0.002131e -36181 be used also to investigate the
condition unchanged, but have circuit behaviour when it is crit-
the current increasing at 2 A s-t, ical1y damped. We reduce the
instead of at only 0.005 A s-t. capacitance slightly, to l.6 uF,
Substituting dildt in Eq. 102: leaving Rand L as before and,
for comparison with Fig. 113,
A = (2 + 7.236)12236 =0.00413l. keep the same pair of border
conditions.
From Eq. 101: With the new value for C, f
o remains unchanged, but g be-
;=0.004131e-13821 2 3 I(ms)
comes 6.25xl06, and:
_0.002131e-3618t
D = 50002 - 4x6.25x106 = o.
We have another particular so- Fig. 112
lution, matehing the new bor- With a zero discriminant, the
der conditions. Its graph (Fig.112) solution ofthe auxiliary equation
shows current continuing to in- IS
i (mAl
crease, to about 2.12 mA, before m = -f12 = -500012 = -2500.
5
damping takes effect. From the
equations we have established, The general solution takes a dif-
we can calculate the current and ferent form (Eq. 98):
4
the rate of change of current at
i = Ae -2500t + Bte -2500t
any instant from t = 0 onwards.
With this information, we can 3 [Eg. 104]
go on to calculate the volta ge Substitutingthe first bordercon-
across thecomponents, and quan- dition into this:
tities such as the charge on the
2 "'j=O.OO2
capacitor and the magnetic link- wllenl=O 0.002 =A,
age of the inductor at any time.
Eq. 104 becomes:
Specifying loter current
i = 0.002e-25001+ Bte-2500/.
Border conditions need not be re-
stricted to events occurringwhen O~--------'---------~2--------~~ 3 t (ma) Now substitute i = 0.003 and
t = o. We can specify i or di/dt at """10_,..,. t = 0.001 into this:
any other instant after timing
begins. It makes the equation Fig. 113 0.003 = 0.002e-2.5 + 0.00lBe-2.5
s1ightly more complicated, be-
cause there is no simplification der condition, but the second to adefinite value in a specified B = 34.55.
ofthe equations due to the index border condition being that time. Since we have specified i,
ofe beingzero. Let us trythe same i = 0.003 when t = 0.00l. We are not dildt, we go back to Eq. 101 The particular solution is:
circuit, with the same first bor- sayingthat the current increases and proceed from there. Substitute

ELEKTOR ELECTRONICS FEßRUARY 1994

 GENERAL INTEREST

i ee 0.002e-25001+34.55te-25001. of equations ofthe same type as

[Eq 105] 1 (mA) Eq_l03: 'EXP01':"A":?~A:"Ml":
The graph in Fig. 114 shows 6 ?-?C:"B":?-?B:''JM2'':?-?D:GRAPH
current risingto a peak at 5.9 mA y~Ae--CX+ Be--DX:. Tbe program
before beingdamped. The second name is 'EXPOl'. The user is
bord er condition occurs on the requested to key in values for
way down. A, ml, Band m2' As 500n as the
4 final value has been keyed in, the
i:= 0.OO2e -25001+ 34.55le-Z5OOt graph is plotted according to the
Under-damping equation in the program.
This occurs when the discrim- A program can also include
nant has a negative value. For commands to set the ranges of t
example, let UB reduce the ca- and i for the displayed graph,
pacitance drastically, to 10 nF. ----i '"0.002
""he<! 1=0
so that the curve fills the screen
reasonably well. Ifthe rangecom-
The fremains at 5000. butg be-
comes 109 and D becomes mand follows the input of vari-
-3.975xlO9. With a negative dis- ables, one or more ofthese vari-
criminant, we need to calculate ables can be used in the range
o 2 3 I(ms)
k: commands. For example, the range
for i can be set to run from -2 A
k~~3. 975x 109 12~44581 to +2 A. It is also possible to in-
Fig. 114 clude inputs to set the range of
The general solution is given by t direetly from the program. For
Eq.99: 1
those who like to play areund
[mAl with models, adjusting the values
I. O,0Q2e-2liOOI cos 445611 _ O.03626e-25001 sln445611
i c= Ae -2500t cos 44581t

+Be-2500t sin 445811 iae

0

"
to produce the required results,
agraphie calculator is a valuable

[Eq. 106]
Keeping to the same border con-
••
~
too!.

Parallel eireuit
ditions, applying the first con-
dition to Eq. 106, and using the I Figure 116shows a circuit wi th
facts that cos 0 ~ 1 and sin 0 ~ 0: V;V resistance, eapaeitance and in-

0.002 ~ Aeo + 0
us
V V V
' [m~~
~ "
ductance in parallel. Buildingthe
model follows very much the same
sequence as building the model
.. A ~ 0.002. of the series circuit. By KCL:

Substituting th.is in Eq. 106, to-

gether with the values for t and
i under the second border COTI- IDDIIf'_lUl_M
Replacingthe currents by equiv-
dition: alent expressions:

0.003~0.002e-2.5 cos4.4581 du u Ntp .

Fig. 115 C-+-+-~, [Eq.107]
dt R L
+Be-2.5 sin 4.4581
The third term on the right 15
B~-0.03828 obtained by noting thatL ~Ntpl i,
This leads UB to the particular as in Eq. 22, Part 5, where q'l is
solution: the magnetic flux and N 1S the
I, number of turne in the coil.
i~0.002e-25001 cos44581t Differentiating Eq. 107, and as-
- O. 03828e -2500' sin 44581t u c R L
suming that i is eonstant:
d2u 1 du u
The graph of th.is equation has C-+_·_+-~O
dt2 R dt L
an interesting form-Fig. 115.
It shows the current reversin g In obtaining the third term,
many times with gradually de- dq>/dt ~ u IN (see Eq. 23, Part 5),
930010·13·"
creasing magnitude. Oscillations so the term reduces to u I L.
ofth.is type are typical ofan under- Dividing throughout by C gives
damped circuit. The oscillations Fig. 116 the model equation:
take about 1.5 ms to die away.
d2u 1 du u
damped circuits, it is possible produced in this way. Agraphie -+_·_+-~O
dt 2 RC dt LC
to sketch the shape of the graph calculator produees results even
Graphie ealeulator more quickly. We used one when [Eq.108]
roughly after calculating half a
A graph 1S an important aid to dozen points. But the oscilla- we were planning the values to The auxiliary equation is:
visualizing the behaviour of a .tions of an under-damped circuit use in the examples. To make m2+mIRC+1/LC~0
model, and hence the behaviour often need 30 or more points to graph plottingeven quicker, pro-
ofthe circuit 'it is modelling. But produce a reasonably represen- gram the calculator to aecept in which f ~ lJRCand g ~ lJLC.
plotting graphs is a tedious mat- tative curve. variables, then to plot the graph. With given values of R, C and
ter, particularly when equations A computer graphies pack- The following 1S an example L, the discrirninant may be pos-
contain severaJ exponential terms. age may save a lot of time: the of a ahort program used on a itive, zero or negative, yielding
With over-dampled and critically- illustrations to this article were graphic ealculator to plot graphs equations fOT u having the same

ELEKTOR ELECTRONICS FEBRUARY 1994

 FIGURING IT OUT - PART 13

Please meruion ELEKTOR ELECTRONICS when contacting advertisers

8051 TOOLS Use your electronics know-how to

create great sound!
ICE51 i225 Speaker Builder, the world's only loudspeaker
Low eost entry level In-Cireuit-Emulator for the 8031/51. magazine in English, ean show you how to achieve
Operates at IIMhz 28k bytes of program memory, includes excellence in sound from yom stereo system.
many features to aid program development but uses sorne Whether you're modifying end improving your
present speakers or building a new design from
resourees of the target 80S 1.
scratch, Speaker Builder contains the latest in
loudspeaker technolcgy. Within its pages, learn
ICE51+ f495 about Thiele-Small design software, build an in-
expensive measurement mike good enough for the
Advaneed emulator for the 8031/5 I. Operation seleetable professionale, or simply indulge in reading about
up to 16MHz, 64k bytes of program memory, uses none of the experiences of others using their technical
the target (80S I) resourees. skills to improve the sound around them.
Since 1980, Speaker Builder has been exploring
8051 "C" COMPILER f125 both tbe traditional and more unusual formats for loudspeakers. From pipes to
eorrugated boxes to transmission lines and electrostatics, every conceivable
Miero/CS I is an integer 'C' like compiler for the MCSS I medium is discussed. An international forum for speaker enrhusiasts the world
family. Easily custornised for any MCSS I family variant. over, sharing ideas and designs, Speaker Builder has made it possible for
High level language debugger available (f95) thousands to enjoy the bighest quality sound without straining their budgets.
The publishers of Speaker Builder invite you to use your electron.ics kncw-how to
discover the world of better sound. To subscribe, simply return the form below with
8051 BOOK f49.95 your payment or FAX your credit card order and receive 8 issues of Speaker Builder
A very good text on the 805 [ architecture, programming for a low introduetory rate of $45 for a one- year subscription. That's a $5 savings
and applicarion. The book is supplied with free (un- ofTthe regular overseas airmailed rate. Or you may subscribe for two years at $80
supported) IBM/PC based assembler and simulator. (,$10 off).
DOn. Year $45 0 2 Years $80
Other tools avaitabte prices exclude VAT and delivery Remittances in US funds drawn on a US bank onty.

Micro AMPS Limited MCN1SA NlJ;1BER EXP_DATE

66 SMITHBROOK
CRANLEICH,
SURREY CU68.1.1, UK
KILNS,
C!C VISA
ADDRESS

= a ~
Tel: +44(0)483 268999
Fax: +44(0)483 268397
50eaker Builder
PO Box 494, Peterborou9.h, NH 03458-0494 USA
(603)924-9464 or FAX (ö03)924-9467 24 hours

form as Equations 97 to 99. Fig. 117. It begins with a mas- difficult to differentiate and pass which models aseries cir-
Here is an exampleofan under- sive 'kick' of up to -34 A. After beyond the scope of this series, cuit (Fig. 110) in which
damped circuit with R = 1 kD., a few swings of rapidly dirnin- However, there are ways around R = 1 kQ, L = 200 mH and
C = 10 nF andL = 18 mH. ishing amplitude, the current this difficulty, as will be explained C = 100 nF, given the border
RC= 1x10-D,so{= l/RC= 105 is fully darnped out in about next month. conditions that i =-0.01 when
and f2 = 1/(RC)2 = 1xlOlO. 0.15 ms .. t = 0, and i = +0.01 when
LC = 1.8x10-1O, so 4g = 4/LC These examples show what Acknowledgment. The author t = 0.005. Is the circuit over-
= 2.222xlO!O. can be done when we make cer- would like to thank the Casio damped, critically damped 01'
From these we find that tain simplifying assumptions Computer Company Ltd fortheir under-damped? Find the cur-
D = -1.222x1010 and k = 78166. about the model. If these as- valuable assistance. rent when t = 0.0008 s.
The general equation ts: sumptions are not valid, in par-
ticula r, if we can not assurne 2. In another series circuit,
u=Ae-50000t cos 781661 that voltage 01' current is con- Testyourself R = 5 kQ, C = 22 ~F and
Be-SOOOOtsin 78166t starrt, the equations are more 1. Find the particular equation L = 0.5 H. When 1 = 0, i = 0.1
and di/dl = 0.5. Find the par-
Nowforsome border conditions. ticular solution for this cir-
Whenz e O;zz=-1; whent=O.OOOl, i(A) cuit. Find the current when
u = -0.5. Using tbe first eondi- 5 z e ü.ö msand whenz e 100ms.
tion, rememberingthatcos 0 = 1 How long does the current
and sin 0 = 0: °t:-------too~--~~~;0~.1
G-l
__----~.
<, t(ms)
take to fall to 0.01 A? (Plot a
graph 01' solve the equation
whent",O 1=-0.5
-1 =A. -5
when t = 0.0001 ofdi/dl by takingnaturallogs).

For the second condition: -io

Answers to
-0.5 = -e-5 cos 7.8166 -15 Testyourself (Part 12)
+Be-5 sin 7.8166 1. q = ge-1.418tx10-4;
-20
B = -74.22. (a): u = 1.45 V
(h): i = 74.9 ~A.
-25
The particular solution is:
2. 0.857 s.
-30
u=_e-50000t cos78166t
-74. 22e-500001sin 781661 3. 532 ~A.

The graph of this is shown in Fig. 117

ELEKTOR ELECTRONICS FEBRUARY 1994

 FIGURING IT OUT
PART 14 - VARYING INPUTS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

We continue the exploration of equation for a parallel LCR cir- zero: equal:
circuit modelling by extending cuit is similar (see Eq. 108 in
it to deal with applied voltages Part 13). Remember that, al- i 55 OOOC= 400
.g+5000 dd +50 OOOi~O
or currents that vary in time. though we are analysingonly the dt t
C = 400/55 000
'Ib model a circuit with both ca- simplest of series and parallel [Eq. 111] = 7.27 x 10-3.
pacitive and inductive elernents, circuits, more complicated cir- Theauxiliaryequation ofEq.lll
we need a second-order differ- cuits can often be reduced to is The coeffieient of tO is D on the
ential equation. In Part 13 we thesecimplecircuits byusingthe left. There is no term in tO on
showed that, ifthe elements are techniques of network reduc- m2 + 5000m + 50 000 = O. the right, so the coeffieient is
in series (Fig. 118) and the ap- tion that we have discussed on zero:

...
several occasions in previous Solvingthis by usingthe quadratic
R
months. formula (Part 13), we find that D~O.
it has real roots: ml = -10.02
Particular integral and m2 = -4990. The PI is:
_R The CF is:
+ L Solving equations with the form i = t(7.27xl0-3).
~ ~Li ofEq. 109 is da ne in two stages. i = Ae-1O.02t + Be-49901
First we solve the equation as if The complete general solution
~c.. the right-hand side equals zero. where A and Bare arbitrary con- is:
c The method for doingthis, based stants whieh can be determined
on the auxiliary equation, was if we are told two boundary con- i= Ae-lQ02t + Be -4990t
930Q1Q-XIV·118 described in Part 13. The result ditions. +1(7. 27x 10-3)
obtained is known as the com- Thfind the particularintegral,
Fig.118 plementary function (CF). At we assume that the right-hand [Eq. 113]
this stage we can say that: side of the solution has the The first two terms in this gen-
form: eral solution are like those ob-
plied voltage is constant, the eompJemetary function = O. tained in Part 13 for a eonstant
equation has the form: i ~ Ct + D applied voltage. Now we have an
Next, we need to find something additional term to allow for the
d2i R di 1.
-:;-;-,r+_._+- ,~O [Eq.94] else to include in the left-hand where C and D are eonstants. If fact that the voltage is ramp-
dt Ldt LC side ofthe equation, so that the this 1S so, then: ing.
But the right-hand side of the right-hand side equals (lIL)du/dt. In this example we assumed
equation is zero only when we This 'sornething' is known as the di/dt = C that the right-hand side of the
have made the simplifying as- particular integral (PI): and equ.ation has the form Ci + D.
sumption that the applied volt- d2i/dt2 ~O. This was based on the fact that
age is constant. This is an as- CF + PI ~ (lIL)du/dt. on the right-hand side of'Eq. 109
sumption that sei dom can be Substituting these values in the highest power of t is t1 If t
made. Moreoften, thevoltage, u, An example shows how to find Eq.110: is of high er power or there are
varies in a regular manner. Für the particular integral. Suppose
example, it may Tamp up OTdown that in Fig. 118: R = 500 n, 0+5000C+50 OOO(Ct+D) Form of Assumed
at a fixed rate, it rnay vary si- C = 200 ~F and L ~ 100 mH. Let ~400t RH side function
nusoidally, or it may increase u. be a voltage that ramps up at .: 55 OOOCt+50 000D~400t
or decrease exponentially. We a rate of 20t2. Frorn these par- k C
need to be abJe to solve equations ameters, R /L = 5000; lILC [Eq. 112] kt Ct+D
ofthe form ~ 50 000; (lIL)dul dt ~ 400t. The We solve this equation by lock- kt2 Ct2+D +E
modelling equation is: ing at the eoeffieients ofthe pow- k ein t Ccosr-Dsinr
d'i R di 1. 1 du
~+_·_+_·t=_·- ers of t. The highest power of t k cos t
dt L dt LC L dt d2i di . is t1. On the left, its coefficient ekt Cekl
d?+5000 dt +50 0001~400t
[Eq. 109] is 55 OOOC;on the right, its co-
in wh ich u: is a function of [Eq. 110] efficient is 400. For Eq. U2 to
time.Part 13shows howthis equa- To find the CF, put the right- be satisfied by all possible values Box 1. Functions tor particular
tion is obtained. The form ofthe hand side of Eq. UO equal to of t, these coefficients must be integrals (k is a constant).

ELEKTOR ELECTRONICS MARCH 1994

 FIGURING !T OUT - PART 14

other functions oft, different as- of operator D gives:

jfl1A
sumptions are made. Box l1ists
forms suitable for other cases.
I .------------1 D2 L+-'R D.L+-·L=U
1 .

We can now proceed, as usual, 1000 !----- L LC

to find a particular solution (not [Eq. 116]
to be confused with the particu- Now comes the interesting pnint
lar integral). This is the solu- 800 about D. Although we have said
tion obtained when two border it is not a variable, it can be
conditions are satisfied. Let us shown that it ean be man ipu-
assume that, when t = 0, i = O. 600 lated in equations and behave as
l
Also, when t = 0.01, i = 1 mA. if it were a variable. For exam-
Substituting the conditions for ple, we can factot-ize the left-
400 hand side ofEq. 116:
t = 0 in Eq. 113:

(D' R D+-1).·L=ll
+_.
200 L LC
B =-A. There 1S no advantage in using
operator D with such an equa-
The equation becomes: 10 20 30 40 50 60 tion, which 18 easy enough to
93001o.XI~-'19
tt;; solve by the method used above.
i= Ae-1O.02t _Ae-4990t Operator D real1y comes into its
Fig. 119 own when we are modelling cir-
+t(7.27x10-")
cuits in which input voltages
=A(e-1O.021_e-4990t)
plottedon a longertirne seale. We ciable irritial 'kick' ofFig. 119. vary sinusoidal1y (or ccain us-
+t(7.27x1O-") catisee tbat the current increases oidally) or exponentially. For ex-
steadily, keeping pace with the Operator 0 ample, eonsider aseries LCR
[Eq. 114] upwardly-ramping applied volt- Before we model any further cir- circuit as in Fig. 118, but with
SubstitutinginEq.114thevalues age. cuits. we will look ata technique the following vaJues: R = 1.5 n,
when t = 0.01: To sirnulate the conditions in that often si mplifies finding the C = 1 F andL = 0.5 H. The input
which thecurrent begins withno solution ofdifferential equations. voltage, II = -Icos 3t)!6, is sinu-
0.001 = A(e-O.I002 _ e-49.9) charge on the capacitor and no This method 1S based on a dif- soidal. CaJculating coeffieients
+ 7 .27xlO-5 f1ux in the inductor, we make ferent way of symbolizing dif- gives: R /L = 3 and lILC = 2.
= 0.9047A + 7.27x10-5 the first bord er condition i = 0 ferentiation. lnstead of writing Differentiating u and dividing by
and when t = 0 as before. The sec- dildt, we write Di. D is not a L produces sin 3t. The equation
A=(0.001-7.27x10-5J/0.9047 ond eondition is that dildt = 0 quantity or variable, but an op- modelling this circuit is:
= 0.001025. when t = O. In words, the cur- erator, which means 'd__ ifferen-
tiate'. We usually refer to it as d2i di . .
Substi tuting these values in rent is constant at zero when ;u:r+3 dt +2l=sm 3t
Eq. 114, we obtain the particu- t ::::O. This makes B = -A as be- operator D. We are using it to
Iar so1ution: fore and we obtain Eq. 114. representdifferentiation ofi with [Eq. 117]
Differentlating Eq. 114: respeet to t, but it can be used When using operator D, we
i=0.001025(e-1002t _e-4990,) for any other pair of variables, first find the complementary func-
+t(7.27x1O-") di =A(_1O.02e-1U021 for example x. and y. Using this tion in the usual way. The aux-
dt symbol, Dx3 = 3x2. The rules of iliary equation 18:
[Eq. 115] +4990 e-49901)+ 7.27 x 10-3 differentiat.ion apply exaetly as
The graph of Eq. 115 for usual. m2 +3m +2 = O.
0<1<0.01 appears as Fig. 119. Inserting the second bord er con- The idea of D is extended to
It shows the rapidincrease ofcur- dition: differentiating adifferential. We Although we nearly always need
rent; this is necessary in order obtain the second differential: to use the quadratic formula to
to meet the condition thati = 1mA o
=A(-10.02+4990l+ 7.27x10-3 solve this equation, there are
=
when t 0.01 s. Remember that and cases like this one in whieh it.fac-
when we specified thati = 0 when A = (7.27xlO-3)/4979.8 torizes easily:
t ::::0, this only means that i is = 1.46x10-6.
zero at that instant. In actual- Similarly, we have the reverse (m + l)(m + 2) = 0,
ity, it is passing through zero at The pat-ticular equation is: operation, represented by the
that irrstarrt toreach a relatively symbol 11D , which means 'irrte- so that ml = -1 and 1112= -2.
high level soon after. Specifying i = (1.46x10-6)(e-JO·02' _e-49901) grate': Both roots are real and so the
zero current at zero time does not +t(7.27xlQ-3). CF is:
mean that there is no charge on
the capacitor and no magnetic It ean be seen that, because A is i = AE-t + Be-2t.
flux in the inductor. These must so srna ll , the final expression Note that we do not irrtroduce the
have been present as the result dominates the equation. Current constant of'integration when we We make use of operator D for
of a voltage apphed to the cir- increases at the rate of 7.27 mA are working with D. finding the particular funetion.
cuit before t = O. This is why the per second without the appre- Representing Eq. 109 in terms The advantage of D is that
current increases so rapidly. there are three very helpful the-
However, the effeets of this die orems associated with it. Their
away and the current begins to 1. F(D) e'" = eat F(a) proofs are rather lengthy, so we
deerease. After about 10 ms, the 2_ F(D)leat f(x)J = ew F(D + a) fex) do not have spaee he re to ex-
circuit begins to show the ef- plain them. The theorems are
3a. F(D2) sin ax = F(-a2) sin ae
feets of the new voltage ramp- stated in Box 2. They look rather
3b. F(D2) cos ax = F(-a2) cos ax obscure, but they are easy to use
ing up from 0 V at t = O. The ef-
feets of this are better shown on as will be shown by this exam-
the right of Fig. 119, which is Box 2. Operator 0 theorems. pie. First, rewrite Eq. 117 in terms

ELEKTOR ELECTRONICS MARCH 1994

 m GENERAL INTEREST

ofD: Theorem 1 1S the appropriate

i choice since the right-hand side
D2i +3Di + 2i = sin 3t. [mA[ of Eq. 119 is simply an expo-
nential term involving t only in
Factorizing: f
100
the index of e. To apply the the-
orem, we have only to write the
(D2 + 3D + 2)i ~ sin 3t, exponential term as it stands,
but to replace D by a. This makes
whence 0.2
50 the calculation easier than the
previous exarnple. In this ex-
1 . 3t 0.'
L 2 -ain ample, a ~ -0.4545, so we find:
D +3D+2
,,
t
.. 8 -11. 3625 -0. 4545/
[rns]
Notice how we move the terms , ,, • -0.45452 + 250 x-0.4545 + 1000
in D across from Olle side of the -50 .' =- -0. 0128 e-O.4545I
-0.1
equation to the other, just as if
they were variables. Now look This is the PI. Adding this to
at Theorem 3a in Box2. This teIls - 100 the CF, we obtain the cornplete
- 0.2
us that, if we have a function of solution:
D which includes D2, and this is
multiplied by sin ax (in which a i= Ae-41 +Be-246t
is a constant), we can replace Fig. 120 +(-0. 0128 e -".4545' )
D2 by -a2. In this example, we
have t as the variable instead of Given two border conditions, we
x. The constant a is equaJ to 3, canfindAandBin the usual way
so we replace D2 by -32, which and so obtain the particular so-
is -9: lution.
1 This month, we have mod-
--::--..:.:::--::-.ein 3 t elled only series LCR circuits,
-9+3D+2
but parallel circuits can be mod-
1 .
~--·sm 3t elled in a very similar way and
3D-7 solved using operator D.
[To be continued l
Having eliminated the D2, it re-
mairis for us to eliminate the D. Testyourself
We juggle with the fraction on 1. Given the complete solution
the right, multiplying top and ofEq. 118 and the border con-
bottom by 3D + 7: sv
drtions that i ~ 0 and di/dt
3D+7 . ~ 0 when t ~ 0, find the par-
(3D_7X3D+7)·sm 3t ticular solution.
930010·XIV·121

3D+7 .sin 3t 2. Given the complete solution

9D2 -49 ofEq.119 and the border con-
Fig. 121 ditions that i = 0 when t = 0,
Once again, we have D2 in the de- and z e 10mA whenz c 100ms,
nominator, so we use Theorem two border ccnditions. Fig. 120 R /L ~ 250 and lILC ~ 1000. The find the particular solution
3a again and replace D2 with shows the graph ofthe solution, equation is: and calculatet when t = 250 ms.
-9: given i ~ 0 and dildt ~ 0 at t ~ O.
Note how the curve begins with d2· di
3D+7 . 3t -' +250·-+ 1000i 3. Given a circuit as in Fig.118,
-sm zero gradient (di/dt ~ 0).Afterthe dt2 dt with R ~ 100 Q, C ~ 2000 pF,
-81-49
initial upsurge, the current set- ~ -11. 3625e-"·45451 L ~ 10 mH, and u ~ 0.08eO•5',
~-(3D+7)sm ~ t1es down to a sinusoidal form, wri te the modelling equation
130 symmetrical about the z-axis, and [Eq. 119J and the genera1solution. Given
-(3D sm 3t+7 sm 3t) lagging slightly behind the input The auxiliary equation is: that when t ~ 0, i ~ 0 rnA and
130 voltage. di/dt ~ 4 A s-I, find the par-
",2 + 250m + 1000 ~ 0 ticular solution. Calculate the
The ter-m 3D ein 3t means 'three Exponential input current wben t = 100 ms.
times the differential of sin 3f, In Fig. 121 the input to the cp- and its roots are ml = -4 and
which equals 9 CQS 3t and the erational arnplifier rises expo- m2 ~ 246 (approximately). The Answers to
PI becomes: nentially. The opamp is a voltage CFis: Test yourself (Part 13)
follower, so the inpu t to the LCR l. i ~ _0.01e-2500t cos 9354t
-(9 cos 3t +7 sm 3t) circuit follows the same curve. i = Ae-41+ Be-246( - 0.03457e-25001 sin 9354t.
130 The switch is set to charge Cl to Underdamped. At 0.0008 s,
5 V. When t ~ 0, the switch is As before, we use D to find the i ~ -4.85 rnA.
Thus, the general solution (CF set to discharge Cl' It can be PI. Writing Eq. 119 in terms of
+ PI) is: shown by the methods describ- D: 2. i ~ l.0097e-9·lt-D.9097e-9900.9t.
ned in Part 12 that: (D2 + 250D + 1000)i After 0.5 ms the cur-rcnt peaks
Aß-i +Be-2/-{9cos3t+7sin3t)
~ -11.3625e-045451 at 1 A, then falls to 0.406 A
130 u = 5e-0.4545t. at 100 ms. It falls to 0.01 A
[Eq. 118J . -11. 3625e -0. 4545' after 255 ms.
Eq. 118 has two arbitrary con- The coefficients for the equation
,- 2
- D +250D+ 1000
stants so it can be solved, given modelling the LCR circui t are

ELEKTOR ELECTRONICS MARCH 1994

 FIGURING IT OUT
PART 15 - SINE WAVES AND OTHERS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects 01' the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

na11 the discussions in this region. Between t = rr./2 and The effeet of this is simply to Figure 124 shows a signal
I series up to the present, the
only periodie inputs that we have
t = 3rr/2, the waves partly can-
cel out, causing the resultant to
shift the curve up the page. In
electronic terms, the pet-icdic
that has five components:
y = 5 + 3sint -1.5sin 2t
considered have been sinusoidal be f1attened toward the x-axis. signal is superimposed on a con- + 4.5cost + 0.5cos3t
waves. We now have to consider From t= 3rr/2tot = 2R, thewaves stant direct voltage, assuming [Eq. 121J
circuit behaviour when the input reinforce again. We could add a thaty refers to a voltage. In the This might be taken to be the
is periodic but is not sinusoidal. eonstant term to the equation for discussions which follow, this trace of an audio signal Irom a
For example, in what way does Fig. 122, for example: variable might equally weIl rep- musical instrument as seen on
a given circuitrespond to a square resent a current 01' any other the sereen of an oscilloscope. As
wave, a rectified eine wave, 01' a y = 6 + 4sint + 2sin2t quantity which varies per-iodi- we shall later, although a Fourier
sawtooth wave? [Eq. 120J eally. Series has (in theory) an infi-
The situation seerns fraught nite numberofterms, quite often
with complicaticns because there only the first few terms need to
are so many possible forme that be evaluated in practice.
waves may bave. Fortunately, we So far, we have synthesized
are rescued by the work ofBaron Y=4 sint+ 2 sln2t
waveforms by writing out a se-
Jean BaptisteJoseph de Fourier r-ies comprising a constant and
(1768-1830) whose Fourier Series t oneor more sine 01' eosine tenns.
---+
provides a way ofexpressing com- Now we have to find out how to
n
plicated but periodic functions in T werk in the opposite direction.
terms ofsines and eosines. When Given a waveform such as the tri-
a waveform has been analysed angular wave of Fig. 125, we
into a number of sirre and co- !I3OO,.-.v- ,,, analyse it in to its components.
sine waves of different ampli- But, before we leave the simple
tudes, we calculate the behaviour Fig.122 examples above, note these fea-
of the circuit for each compo- tures ofthe series:
nent, considered separately. Then
we m ake use ofthe pr inciple of • a constant (d.c.) term;
@ Y=4slnt
superposition (Part 3) to find ® Y= 2 sin2t
• terms in sin t, sin 2 t (in gen-
the response ofthe circuit to the eral, sin n t, where n is an in-
complete waveforrn. The equa- teger);
tion for the Fourier Series is de- • terms in cost, cos2t ... eosnt;
scribed in detail later, butwe look • the terms have different co-
at some simple examples firat. effieients, which may be pos-
O3OO'.·XV·'>3
itive or negative;
• thefundamental frequeneyis
Summing sines such that one eycle takes 2n
Fig.123
The waveform ofFig.122 isclearly seconds (we will explai n how
not sinusoidal, but it is easy to to deal with other fundamental
see that it is the sum oftwo sine y frequencies later );
wave of different amplitude, one
having a frequency twice that
i. Y= 5 + 3 slnt- 15 sln2t+ 4.5 coat .. 0.5 cos3t
• all frequencies are multiples
of the fundamental.
of the other. These two compo-
nents are shown separately in In practiee, the series may be sim-
Fig. 123. In musical terms, the plified if, for example, the con-
one with the lower frequency is stant term 18zero, 01' the coeffi-
termed the fundamental,.

is the first harmonie. Between

and
the one with souble the frequency

t = 0 and t = n/2, the waves re-

informce each other, so that the
-- t

!»(IO'O-'V-'~'
cients of a11 the aine terms 01'
a11 the eosine terms are zero.
Equation 120 is an example of
aseries in which the eosine terms
a11have a zero coefficient.
resultantFig.122)peaksin that Fig.124

ELEKTOR ELECTRONICS APRJL 1994

 FIGURlNG IT OUT - PART 15

values of y over 1 eycle. This fits

10,
in with the idea ofthe first term
30
beingequivalent to the steady di-
reet voltage on which the signal
voltage is superimposed. In the
case of Fig. 125, it is easy to
20 evaluate the first term simply by
looking at the graph. During
1 eycle, y ramps down linearly
from IOJt to zero. The first term
10 is the mean of IOJt and zero. In
grneral:

Ao~5"
Ao~- 1 12ff
ydt [Eq.123]

-
o rr 2" 0
,- t zn

930010-XV-125
This integral gives the area under
the graph for a period of 1 cycle.
Fig.125 The integral divided by the pe-
riod (2Jt) is the mean value of y.
In some cases, it may be easier

Y = SR + 10 sin! + 5 sin2t + 3.333 sill3t + 2.5 sin4t

y

i Integrating by parts
'Ibis technique 'is used to inte-
grate the product of two ex-
pressions, both of which are func-
tions of the same variable. We
shall eall the two parta u and
10 dv/dt (these symbols do not uec-
essarily ha ve anything to dowith
voltages).
In the example in tbe text,
for finding eosine terms:
,-
n

93001Q-XV-126 u ~ Y ~10,,-5t
du/dt ~ cosnt.
Fig.126
First, differeutiate u:

du/dt ~ -5.

Next, integrate du/dt to ob-

Y=5n:+10sint+5sin2t+ .. + ~Sin19t+~Sln20t
tain u:

v ee (sinnt)/n.
20
The constant ofin tegration is not
required. The integral ofthe prod-
uct 1S given by the formula
10
fu.dv/dt dz e uv-. f v·du/dt dt

~---------,----------,---------.--------+rr
o rr
,- 3rr
,- 930010· XV· 127
-
t
Substitute
evaluate.

example:
2ff
u, v and du/dt, and

Applying this formula to the

Fig.127
1

1
an ~- (lOlI-5t)cosntdt
n 0

~~r[(lO"-5)(sin ntl/n]5ff
Fourier series Although theequation i8 lang and erty ofthe Fourier Series is that, "
The Fourier Series has this form: unwieldy, it breaks down into dis-
tinct parts as listed above. The
with some exceptions, any kind
ofperiodic signal can be analysed
2'
1
- 0-5(sin nt)/ndt)

y =AO + alcost + a2cos2t constant term is AQ• We repre- as an equation of this form.
+a3cos3t + ,., +ancosnt sent the coefficients ofthe eosine Different signal have different The integration is continued in
+b1sint +b2sin2t + b3sin3t terms byal,a2'" " an' and the co- sets of coefficients and our next the text.
+ ... bn sinnt. efficients of the sine terms by task is to calculate these.
[Eq. 122] b1>b2,"" bn. Theirnportant prop- The first term is the rnean Box 1

ELEKTOR ELECTRONICS APRIL 1~94

 GENERALINTEREST

to integrate from -1t to +1t. It

does not really matter where in 11 Y
2rr
. nt
Useful integrals
bn =- SIll dt
the cycle we begin, as lang as rr 0 CalcuJations are often made very much shorter when these in-
we finish 21tseconds later. [Eq.125l tegrals occur in an expression to be evaluated. Integration is
with respect to t, over the limits 0 to 21t or -1t to +7t, and n and
Integrating by parta: mare positive i.ntegers.
Cosine terms
The formula for the coefficients bn ~~1[-(10rr-t)cos ntl~rr Term to be integrated Integral Condition
of the cosine term is: rrn sinnt o
2rr
2rr
r cosnt o
+5 Jo-cos nt dz} sin2nt n
an =-
11 y cosnt dt
rr 0 cos2nt
[Eq. 124l The second term on the right sinnt cosmt o"
equals zero (see Box 2), leaving: cos nt cos mt o n-::t.m.
This is similar to Eq. 123, but or lt n==m
-1
bn ~-IHlO,,-t)cos
2
ntlorr sinnt sinmt o nvm
here we have a product to in-
tagrate. The product isy x cosnt rrn or lt== In
2rr
. and, since y is a function of t,
we must integrate by parts (see
r
+5 Jo-cos nt dti 2rr
r
Example: Josin 3t cos 4t dt~O
Box 1). U sing this technique, we
find that: But cos O.» 1, so
Box 2
an ~~I[(10rr-5t)sin ntl~rr bn ~ lOnlnn ~ 101n.
rrn more terms in the series to obtain easier to type and sums hun-
2rr dreds of terms, if required.
r
+5 Josin nt drl
Thus, the sine terms have the
form:
a good representation. We ex-
tend the series until we obtain
a waveform which we consider Circuit modelling
The second term on the right (lOln) sinnt to be suffieiently elose to the re- AB an illustration ofhow the Fourier
equals zero (see Box 2), leaving: quired waveform: Series may be used in modelling,
for n ::::.
1, 2, 3, . we will see what happens to the
... +2sin5t + 1.667sin6t sawtoothed voltage wave of
an ~~I(lOrr-10rr)sin 2nrr +1.429sin 7t + ... +0.5sin20t. Fig. 125 when it is applied to
rrn How many terms? the circuit ofFig. 129, which is
+(10rr-0)sinOI We are now in a position to as- Figure 127 shows tbe result of a passive high-pass f'ilter.
~~IO+ lOrr sin 01 semble the series from the ini- extending the series up to n == 20, Remember tbat tbe period oftbe
rrn tial term and the coefficients. In the 19th harmcnic. It is clear wave is 21[. Its frequeney is
Eq. 126, we take the series as that, as we take more and more 1/2lt ~ 0.159 Hz. To have any ef-
But, sin 0 = 0, so far as the third harmonic: tenns, the resultingfunction map- feet on the lower harmonics, we
proaehes more and more closely need a eapacitorofrelatively high
y::::. 5rr + 10sint +5sin2t to that of the original triangu- value. With the values shown in
+3.333sin3t + 2.5sin4t lar wave. It is only where there Fig. 129, the -3 dB point is 2 Hz.
As a result of this, there are no [Eq. 125l is a dramatic change that ap- We now need to examine the
eosine terms in the series. This This function is graphed for preciable oseillations are observed. terms of the Fourier Series for y
considerably simplifies matters. 1 eyele in Fig. 126. It shows un- Calculating and summingthe and find what happens to them
dulating sine-like features, but terms is fast on a programmable when fed into the filter. The first
is recognisably the triangular ca1eulator. However, theprogram term, 5n, is a d.c. term and does
Sine terms wave of Fig. 125. For such an memory may not cope with as not pass through. Reealling the
The formula for the coefficients un-sinuscidal wavefonn as a sharp- many as 20 terms. A computer teehnique used in Part 6 to eal-
of the sine terms 15: toothed triangular wave, we need program in BASIC (Fig. 128) is eulate the attenuation and phase

10 FOR t = 0 TO 6.4 STEP 0.05

20 u =PI*5
30 FOR n = 1 TO 20 y
40 u = u + 10/n*SIN(n*t)
50
60
NEXT
PRINT
70 NEXT
ES, t;" ";ROUND(u,3J j"
930010·XV·128 '0

8
Fig.128

6
c
0-11---+-----1
aoollF 4

2
t
01 o ---.
2n

-2 930010·XV·I30
930010· XV -129

Fig.129 Fig.130

RLEKTOR-ELECTRONICS A-PRIL-1994
 FIGURING IT OUT - PART 15 11I
Books from Elektor Electronics (Publishing)
VOUCHER
The following books are currently available: This voucher (original only; no
30 I Circuits f9·95 $14.50 photocopies aecepted) entitles the
302 Circuits f9·95 $14.50 holder to 71/2% discount on the
303 Circuits flO·95 $17.90 purehase price of any Elektor
304 Circuits fl2·95 $23.95 Electronies book listed in the adjacent
305 Circuits (published in April 1994) f14·95 $27.70 box. Two vouchers qualify for 15%
SMT Projects f 9·95 $14.50 discount and three for 30%.
Microprocessor Data Book f10·95 $19.90 More vouchers will be published in
Data Sheet Book 2 f9·95 $18.50 fortheoming issues of
Data Book 3: Peripheral Chips fl095 $19.95 ELektor Electronics
Data Book 4: Peripheral Chips f10·95 $19.95
04/94

How to order: Gazelle Book Services Ltd

Uk änd Overseas privatecustomers (NOT Falcon Hause, Queen Square, Lancaster,
in USA and Canada) from: England LAI I RN
Elektor Electronics (Publishing) Telephone (0524) 68765
PO Box 1414 Dorchester, England DT2 8YH Fax: (0524) 63232
Telephone (0305) 250 995
Fax: (0305) 250996 USA and Canada ONLY from:
Old Colony Sound ~ please see order form
UK and Overseas bookshops, schools, on page 71 for details
colleges, universities and bulk orders
(NOT in USA and Canada) from:

lag of a low-pass filter (Fig, 52), in polar form. The values of R ~3 dB point of the filter. The analysingcireuit behaviour. Next
we calculate the effect of the and C remain the same, but co input amplitude of this cornpo- month, we show how to tackle
high-passfilterofFig, 1280n the becomes 2, 3. 4, ... for succes- nentis 10/12 = 0.8333 and its out- other types of wave and we look
fundamental component, sive harmonics. The outputs are put amplitude is 0.5771. The at- at ways of shortening the cal-
y = lOsint. The total impedance attenuated bydifferent amounts tenuation is 0.5771/0.8333 culations.
of the filter is: and have different phase leads. = 0.69. This is roughly 0.7, the
Z=R~j/oC, These outputs can not be added expected attenuation at the Test yourself
in the usual way, as phasors, ~3 dB level. The phase lead is
where weexpress theirnpedance because they are all of different 0.81 rad., equivalent to 46°, elose A triangular waveform ramps up
of the capacitor as a cornplex frequencies. We therefore ex- to the lead of 45° expected at aecording to the equationy = 2t,
number (see Part 9). Given that press each ofthem as a sine wave the ~3 dB point. repeating every 2n .seeonds.
W= 1,R = 100 and C = 800xl0-6, of given amplitude and period A graph of Uout appears in Express this as a Fourier Series,
with a given phase lead. In this Fig, 129. It shows that filtering as far as the 3rd harmonie. This
Z = 100 ~ j1250, form, the output from the fun- has a dramatic effect on the sa w- signal is applied to a low-pass fil-
damental is: tooth wave, reducing it to a se- ter ha ving the same com ponents
or, in polar form: ries of spikes, where the volt- as Eig, 129, but with Rand C
Uout = 0.7974sin(t + 1.49). age suddenly steps up at the be- interehanged. Caleulate a se-
Z = 1254 L~1.49. ginning of each cycle. The circui t ries for the waveform ofthe out-
Summing the outputs for the filters out the slow change of put signal as far as the 3rd har-
Angles are in radians. Considering fundamental and the first 19 voltage as waveform ramps down. monic.
the filter as a potential divider, harmcnice, we have: This graph has more pronounced
its output is uR, and: undulations than Fig. 127; this Answers to
Uout = 0.7974sin(t + 1.49) is due to the fact that the high-
+0.7900 sin (2t + 1.41) pass filter makes the higher fre- Test yourself (Part 14)
+0.7778sin(3t + 1.34) quencies relatively prominent. 1. i=0.3(e-t~e-2t) + 0.06923e-2t
For the fundamental, the peak + 0.5771sin(12t + 0.81) Inthe example above we have ~(9cos3t + 7sin3t)/130.
values of uin is 10 V (the coeffi- + 0.4241sin(20t + 0.56). illustrated the use ofthe Fourier
cient of the sin t term of the se- Series by modelling a simple 2.i = 0.03316e-4/ ~ 0.02036e-246/
ries). Note that the coefficients de- waveform being passed through ~0.0128e-O.4545t. Current
erease, but much more slowly a cireuit with a simple response. = 774 J.lA.
Uout = 1000/(1254 L~1.49) than in the original funetion. There are easier ways of doing
= 0.7974 L1.49. This reflecte the fact that the this, as we shall see in Part 17. 3. d2i/dt2 + 10000di/dt
higher frequeneies are less at- It is with more elaborate wave- + 5000i = 4eO.5t.
The output signal is attenuated tenuated than the lower fre- forms and more eomplicated cir- i = Ae-5t + Be-9995t
to about 0.8 V peak and leads quencies. Also, the phase lead cuits that the FourierSeries shows + (7.27x10-5) eO.5'.
the input by 1.49 rad. is leass as frequency inereases. its power. With a computer to pro- i = 0.0003277(e-5t ~e-9995t)
We can similarly find Uout for When W = 12, f = w/2n = 1.91, vide help with the calculations, + (7.27x10-5)(eO.5t~ e9995/).
each ofthe harmonics, expressed which is approximately at the this is a useful technique for Current = 0.275 mA.

ELEKTOR ELECTRONICS APRIL 1994

 FIGURING IT OUT
PART 16 - MORE ABOUT FOURIER
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

ast month we described how Mathematieally, we can express stant term; posed on a constant d.c. level
L the Fourier series ia used to
represent any periodie waveform.
this by: • it has eosine terms;
• it has 00 sine terms.
(as, for examp le, is the wave-
formofFig.131b). Ifyou find that
Ta recapituJate, the series COTI- fex) ~ fe-x). An obviousexampleofaneven a wavefortn 1S symmetrical about
sists of: function is the cosine waveform the y-axis, you need not waste
constant (d.c.) term If a waveform is an even func- afFig. 131a, which has a eosine time by trying to evaluate the in-
+ tion, we ean say, without any term and TIO sine terms. Whether tegral ofEq. 125.
eosine terms need for integrations, that: or not it has a constant term de-
+ • it may or may not have a con- pends on wh ether it is superim- Odd functions
sine terms.
The constant term and the coef- These are symmetrieal ab out
ficients of the eosine and sine the origin. If we imagine the
terms are found by integration: eurve spun round through 180 0

around a centre loeated at the

1
Ao ~ 21r.!o ydt
r2rr [Eq.123] crigin, it comes to lie on itself.
Figure 132shows same odd func-
tions. The eharaeteristie features
of an odd function are:
1irry cosnt dt
a,,~- • it has no eonstant terrn;
tt 0 • it has no eosine terms;
[Eq. 124] • 'it has sine terms.
In short, an odd function has
only sine terms. Sinee it is syrn-
1 2rr r
bn =; Joy sin nt dt metrieal about the origin, it ean
never have a eonstant term.
[Eq.125] Between them, even and odd
funetions covermost ofthe wave-
The eonstant term, whieb equals forms we are likely to meet in
the average value ofthe variable electronies, but there are a few
o other reeognizable types of'wave-
(for exarnple , voltage) during 930010_XVI_131

one cycle,ean sometimes be found form that yield series with dis-
by geometrieal means. We are Fig. 131 tinctive features. The next two
often able to simplify the inte- headings deal with these.
grations by using the standard
integralsgiven in Box 2 ofPart 15. Half-wave inversion
A frequent result of simplifica-
tion is that the series for a given In Fig. 133, the waveform be-
waveform has no eosine terms tween t = 0 and t = 1t is repeated,
orno sineterms. Now we will look but inverted, between t = 1t and
at same further ways of sirnpli- t = Zn. This is known as half-
fying the ealculations. wave inversion. The Fourier
series for such a wavefonn has
• no constant term;
Even functions • DO even eosine terms;

An even function is Olle whieh • no even sine terms.

is symmetrieal ahout the y-axis. The effects of half-wave in-
It is as if the y-axis acts as a version and the kinds of sym-
mirror; reflecting the eurve of'the metry are eumu1ative. The wave-
waveform for positive values of form of Fig. 133a has odd sym-
x on to the negative side of the metry and half-wave inversion.
axis. Figure 131 shows some Because it has cdd symmetry
examples of even funetions. Fig.132 (and also because it has half-

ELEKTOR ELECTRONICS MAV 1994

 FIGURING IT OUT - PART 16

wave inversion), it has no eon-

stant term. Beeause it has odd an =.! r3cosntdt+O
symmetry, it has no eosine terms. n: Jo
Beeause it has half-wave sym-
=~[sinntlrr
metry, it has 00 even sine terms. nn:
We are left with aseries that sin
has only odd sine terms. By con- =~[sin nn:l=0.9549 nie
nn: n
trast, the waveform ofFig. 133b
has even symmetry and half- But sin n1t ::= 0 if n is an integer,
wave inversion. It therefore has making c., = O.This resultmeans
no eonstant term and no si ne that there are no coeine terms.
terms. It has only odd eosine We eould have told this at the
terms. start beeause this is essentially
an odd function, raised on a d.c.
yt levelof1.5(comparewithFig.I34).
Half-wave repetition
The constant term 1.5 simply
The waveform of Fig. 134 is re- raises it above the x-axis. If the
peated in the seeond half of the constant term is ignored, sothat,
cycle. For such a waveform, the in effect, it is lowered until it
series has: 0 , passes through the ot-igin,it be-
• possibly a eonstant term; "t comes an odd function and has
• no odd eonsine terms; only sine terms.
• 00 odd sine terms. Finally, we calculate bn:
The example here has a eon- - 930010· XVI· 1

stant term. Removing the eon- bn =.!. r3sin ntdt+O

rr Jo
stant tenn ofthis example would Fig.133
shift the eurve down, giving it -3
=-[cosnt]~
odd symmetry. It therefore has nn:
only even sine terms. This rule -3
=-[cosnn:-eosOJ
does not neeessarily eliminate nn:
the need to ealeulate all three -3
ofAO, an. and bn, but does at least
=-[eosnJr-l1
nJr
provide acheck that your calcu-
lations are eorreet. There are two possibilities to be
eonsidered:
• when n is even, eos n1T. = 1,
Square waves which makes bn ::= 0;
The waveforms wehave analysed o 2, 3n~ • when n is odd, eos nn: = -I,
so far have all been eontinuous 930010· XVI· 1:14 in whieh event bn = 6/n1t
within the period 0 to 21t. Now =1.9099/n.
we will taekle a waveform that Fig.134 The series has only odd eosine
has an abrupt change oflevel dur- terms. This, too, might have been
ing the eycle. Figure 135 is a Y~3 0<1<1( foreseen: when the waveform is
Y~O 1« 1<21t ,- __
square wave with a mark:spaee shifted down to be symmetrical
ratio of 1. We have arbitrarily about thex-axis, ithas half-wave
taken the 'high' level to be 3. inversion, resulting in it having
This is a pieeewise function and, no even terms. Writing out the
to find the scrics, we have to series:
deal with eaeh part of the func- . sin3t
tion separately. First of all , the Y=1.5+ 1.9099(sm t+- -+
--+, 3
eonstant term is the surn of two
sin 5t sin 7t sin 9t
integrals: +--+--+--+ .. )
579
15,' 11
110=- 3dt+-
2n: 0 2n,.
dt'
2
Fig.135
Figure 136 shows the graph of
the series when taken to the 10th
In this example, the seeond in- harmonie (n = 9).
tegral has zero value:
1 Frequency spectrum
110= -[3tlrr +0
2rr Another way of looking at a se-
3 ries is to plot its frequency spee-
=-[rr-Ol=1.5
2n: trum. This has a vertieal line
for eaeh frequeney, the height
This result eouldhave more read- of the line being proportional to
ily been obtained by inspeeting the amplitude of the frequeney.
the graph and noting that the av- In the series we have just eal-
erage value ofy is 1.5. eulated, the amplitude of the
o fundamental is 1.9099. The am-
For the value of an we again
integrate for both parts of the plitude of the 1st harmonie is
waveform: 1.9099/3 = 0.6366. Figure 137
Fig.136 shows the plot ofthe fundamental

ELEKTOR ELECTRONICS MAY 1994

 GENERAL INTEREST

and first 10 harmonics. The even as the fundamental as shown in this pulse wave. This property sound has the tirnbre of a given
Olles are missing and the am- Fig. 140. This 1S plotted as far ofpulse waveforms is made use musical instrument. A1though
plitude ofthe odd ones falls away as the 10th harmonic(there is no ofin synthesizer eireuits. A pulse the amplitude ofthe higher har-
fairly rapidly. Last rnonth we 9th). Compared with the square waveform is a strong source of monies falls offmore significantly,
analysed a sawtooth wave before wave of Fig. 135, as shown in harmonies which can be filtered the limits ofaudibility are reaehed
and after passing it through a Fig. 137, there is rnueh less fall- (formantfiltering) to adjust their around the 5th hannonic. However,
high-pass filter. Figure 138 shows off in the lower harmonies of proportions until the resultant there is an increase of ampli-
the frequency spectra of these tude from the 10th to the 14th,
waveforms; the effect of filter- which has an amplitude almost
ing is obvious. equal to thatofthe 7th harmonie.
This illustrates the faet that un-
A
expectedly strong harmonics may
Pulse waveforms
lnFig. 139 wehave a pulse wave-
i be present in pulse signals.

form, which is another example Other periods

of a piecewise function. This 18
analysed in a similar way to the The formulae described above all
1:1 square wave above. If we refer to a waveform in which
shift it ,,/10 to the left, the pulse one cycle takes 2n seeonds. In
begins at-,,/10 and ends at ,,110. other words, the angular veloc-
It 18 now an even function, so ity is unity. This corresponds to
its Fourier series has no sine a frequeney ofO.159 Hz. This is
terms. By inspection, we can tell rather lower than many of the
that AO = 0.3. frequencies we meet in electronic
For the eosine terms: Fig.137 circuits and, even though the
analysis of different waveforms
11'/10
an=- 3cosntdt is independent of the frequency
J[ -11) 10 " UNFII..TERED of the fundamental, there are
many instances when we want
=.![~sin nll,/IO to know the Fourier coefficients
TC n -Jr/l0
for a frequency that isnot 0.159 Hz.
3 [ai nst
=-Slll--sm-- . -nJr A Adapting the formulae to an-
l
nn 10 10 other frequency is just a matter
6 . n7f of altering the time-scale.
=-Sill-
nx 10 FlLTERED
The equations relating an-
gular velocity, m,frequeney,f, and
= 1. 9099 sin (M/lO)
n 0.'
I I , period (length of cycle), T, are:

Ifn = 1, then sin(nrtl10)= 0.3090;

for n = 2 to 9, we obtain 0.5878, or
08090, and so on to 0.3090. When • , a , s , , • •
" T = 2rt1w.
n = 10, sin(nrtllO) = 0, so there
18Da 9th harmonie. The sequence The phase angle at any given
of values is repeated for n = 11 Fig.138 time t is wt. The Fourier series
to 20) except that they are neg- in these conditions is the same
ative. From n = 21 onwards the Y,,3 G~ ,<~ as before, except that we replace
whole sequenee repeats, but we Y"O ~<f<21[
- t in the eosine and sine terms
are not eoneerned with harrnon- by ox, Conversely, making w= 1
ics of such a high order. restores the terms to their orig-
Written out as far as the 2nd inal form.
harmonie, the series looks like
this:

y=0.3+ 1.9099 (0.3090sin t

+0.5878 sin 21
-,
noo,o-~"'-''19
The formulae forthe eonstant
term is:

Ao =J:. r~(t)dl
TJo
The formulae for the coefficients
2 Fig.139 are:
+0.8090 Si~31 + ..)
an =-
T
21T0
F(I) cos rua dt
Simplifying the coefficients. A
0'

by dividing by n (1, 2, 3, ...):

y=0.3+ 1.9099 (0.3090sin t i 02 b,,=-21'F(t)sinno:tdl
T 0
+0.2939 sin 2t
+0.2700 ein 3t+...) These show the limits of in te-
0 .t gration running from 0 to T, but
The val ues of sin(mt!10) increase theyeanalsorun from-Tl2 toTI2
for the first few terms, so this
partly offsets the effect of divid- , I I I
if'this makes the integration any
easier.
r • • tu
ing them successively by 1, 2, We will derive the Fourier
3, .... In consequenee, the first series for the square wave of
few harrnonics are almest as strong Fig.140 Fig, 135, but wi th aperiod of

ELEKTOR ELECTRONICS MAY 1994

 FIGURING IT OUT - PART 16

Please mention ELEKTOR ELECTRONICS when contacting advertisers

0.01 s., corresponding to a fre- the equation ries has only a single sine term, tegrations are lang, and prone
quency of 100 Hz. At 100 Hz, 2 so this has no effect on the rapid to errar, especially with the fre-
T = 0.001 and w = 2n/0.001 an 2 fall-off of the harmonies. quent changes in sign that re-
Jr(l-n )
= 628.3 rad. At this point we leave the sult from integrating functions
The average value of y is not The den omina tor conteins n2 discussion ofthe Fourier Series, containing sines and eosines. The-
affected by the frequency, so and successive eoefficients (there having shown how to analyse rules we have discussed will
Ao = 1.5 as before. Also, as be- are even terms only) are: 2/31t, many ofthe most commonly oc- help by eliminating the need
fore, we recognize this to be an 2/15n, 2/35n, 2/63n, ... The se- curringwaveforms. Often the in- forsome oftheintegrations. When
odd function, so that an = 0 and you have arrived at the equation
there are no eosine terms. For for aseries, plotting it with a
the sine terms: graphie calculator is a rapidnieans
of cheeking that it is essentially
2 lO.005 correct.
bn=-- 3sinnüJtdt
0.01 °
-3
-;;-::=-(cosnJr-l) Test yourself
0.005nw
1. Tbwhatcategories dothe wave-
If n is even, eos nrt = 1, and forms of Fig. 142 each be-
bn = O.There are no even terms. Fig.141 long. In eaeh ease state what
If n is odd, eos n1t = -1 and elements you would expeet to
b" = 6/0.005nw = 1.9099/n. find in their Fourier series.
The series is exactly the same
as that for the square wave of 2. A sawtooth waveform with

-'
Fig. 135, except that, instead period 2n is defined by these
of sin t, ein 3t, , we now have ~ equations:
sin ox, sin 3ax, The graph of y = 3t O<t<n
the series has the same shape, y = Se-Sn p<t<2n
exeept that the scale on the Derive its Fourier series,
x-axis runs from 0 to 0.001 s.
Thus, all thatwe have said about Answers to
waveforms with aperiod of 2n
app1ies with litt1e roodifieation Test yourself (Port 15)
to periods of other lengths. ao = 2n; an = 0; bn = --4/n.
By contrast with the slow Series is: y = 21t--4sint-2sin2t
Fall-off of harmonics in pulse -L333sin3t-sin4t ...
waveforms, there are other wave- Filtered signal is
forms in wh:ich the rate of fall- U,ut = -3.988sin(t-0.0808)
offis rapid. An exampleis shown -1.975sin(2t-0.1608)
in Fig.l41. This is the waveform -1. 297sin(3t-0 .2358)
of a half-wave rectified alter- 030(1'0- XVI_'" -0.9525sin( 4t-0.3098)
nating voltage. The coeffieients
forthe eosine terms are given by Fig.142

ELEKTOR ELECTRONICS MAY 1994

 FIGURING IT OUT
PART 17 - ALL CHANGE!
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the hehaviour of circuits.
Dur aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, ren der some insights into the mathematics involved.

T herateofchangeofonequan-
tity with respect to changes
in another quantity is known as
as, if not more than, voltage.
Current is a function of both
voltage and resistance:
10 0
i [Al

a differential. As explained in
Part 5, we represent a differen-
tial by a symbol having the form
dyldx. The example given then We might want to know the rate
was d<Udt, the rate of change of at which current varies with re-
magnetic flux cI> through an in- spect to variations in both volt-
ductor, with respect to time t. If age and resistance. Think ofthis
we were to plot a graph oftheflux in terms of a three-dimensional
at different times, then d<PIdt is graph-Fig. 143. The curved
the gradient ofthe graph at any surface represents the way in
given time. which i varies with u and r. We
We are assuming that the de- can consider these var-iations
pendent variable <Pisa function individually. If we cut sections
of the single independent vari- ofthe surface parallel to the i-u
able t. In symbols: plane, we obtain a set ofstraight-
line graphs. The line AB repre-
<P = {(tl. sents changes incurrent with re-
spect to voltage, if resistanee is
Practically a11ofthe circuits and constant at 8 Q. In the other di-
circuit models we have studied rection, parallel to the i-r plane,
exhibit this one-to-one relation- sectioning the surface yields a 10
ship. In many ofthem, the volt- set of hyperbolas. An example
age at a given point is a fune- 1S shown by the line CD, which
tion of time. Given an equation represents changes in current Fig.143
that defines how voltage varies with respect to resistance, ifvolt-
with time(forexample, U = 3 sint), age 1S constant at 4 V.
we differentiate to find the rate We caleulate the slopes ofthese emplified by lineAB. For the set along a curve are doubled. This
of change ofvoltage at any given curves by using partial differ- of lines obtained in this way, the can be seen by comparing curve
instant (in this example, entials. The techrrique corre- gradient is inversely proportional CD with curve CD'.
du.ldt = 3 cost). But there are sponds to cutting a slice through to r. The smaller the value of r,
other cases in which a quantity the surface, parallel to one or the steeper the line, as can be Example
depends on two (or more) other otberofthevertical planes. Cutting seen by eomparing hneAB with
independent quantities. As an a slice parallel to the i-u plane lineA'B' in the figure. Let us put some numbers intc
example, consider the fa miliar means considering r to be con- Similarly, we find the gradi- these equations and see how they
Ohm's law equation for the cur- stant and then differentiating ent of lines parallel to the i-r work out. Take a point P on the
rent through a resistor: Eq. 126 with respect to u: plane. Differentiating Eq. 126 surface, representing u = 4,
while considering u to be con- i = 0.8. The third value neees-
i = u/r. [Eq. 126] 8i/8u = 1/r. stant: sarily follows from the first two.
At this point, the gradient in
As on previous occasions, we are The symbol for the differential is 8i/8r=-u/r2. the direction parallel to the i-u
using lower-case letters for quan- written with a special kindof'd' to plane is
tities thatare varying. Ordinarily indicate that this is a partial The gradient is negatively in-
we would use a capital R for re- differential. We caU the 8 a versely proportional to the square l/r = 1/5 = 0.2.
sistance, assumingthat its value 'curly d'. Because we have de- ofthe resistance, which 18a char-
does not change significantly. cided to make r eonstant, the acteristic of hyperbolas. For a The units of'this gradient are am-
But suppose that the resistor is partial differential (thegradient given resistance, it 18 also di- peres per volt (NY). Tbe gradi-
light-dependent or we are using ofa linedrawn parallel tothei-u reetly proportional to the volt- ent parallel to the i-u plane is
a thermistor. We must alJow for plane) is constant, In otherwords, age; at double the voltage, for ex-
resistance to vary just as much we have a straight line, as ex- ample, gradients at aJI points -ulr2 = -4/25 = -0.16.
ELEKTOR ELECTRONICS JUNE 1994
 FIGURING IT OUT - PART 17

Units are amperes per ohm change of current with respect Percentages 8z18a =u l b:
(AlQ). Now suppose that the volt- to voltage at a given point is
age is increased by 0.1 Vand öi / Su, The actual change in cur- The techniquecan also be applied Sz l Bb = -uo lbt.
the resistance is decreased by rent produced by the small change when the changes are expressed
0.5 Q. The change in current re- in voltage is as percentages. Using the re- We will use this example to il-
sulting from a change in volt- sistor of the previous example, lustrate the method of estimat-
age is 0.2 A V-I. A change of (8ildu)xf1u. we might specify that the resis- ing errors and werk in percent-
0.1 V produces a current increase tance decreases by 0.5% and the ages. The rum is to calculate the
of 0.2xO.1 = 0.02 A. Similarly, Similarly, the change in current current decreases by 2%. maximum positive errorinz that
the change in current result- resulting from a small change Calculating the changes: my be brought about by resistor
ing from a change in resistance in resistance is: tolerances and the precision of
is-0.16AQ-l Achange of-0.5Q f1r = -0.0051' the input voltage. Factars that
pro duces a current change of (Bi l drvx ör, and will result in an increase in z
--D.16x--D.5= 0.08A Both changes f1i = -0.02i. are:
are currentincreases and their Provided that both f1u and f1r are • Increase in u; suppose this 18
effects are cumulative. The total smalI, these two may be added Total change in power is: precise to ±2%; L1u = 0.02u.
change in current is 0.02+0.08 (taking account of sign) to find • Maximal value of a; suppose
= O.IA. Thenewcurrentis 2+0.1 the overall change in current. i2Ll!' + 2irf1i = -0.005i2r-0.04i2r tolerance is 1%; L1a = O.Ola.
= 2.1 A = i2rx-0.045 = px-0.045. • Minimal value of 6; suppose
This result is only approxi- tolerance is 1%; L1b = -0.016.
Another example
mate, because we have consid- The percentage change in power
ered the effects ofvoltage and re- Wewillworkthis examplestraight is -4.5%. Using an equation of the same
sistance separately. Figure 144 through, without comment, as form as Eq. 127:
1S an enlarged view ofthe surface a practical summaryofthemethod. Three or more
ofFig.143, in the region ofpoint A 33 Q resistor has a current of f1z = (al b+1) x 0.02u
P. Increasing voltage by 0.1 V 2.5 mA passing through it. The The technique rnay be extended +(ulb)xO.Ola
is equivalent to moving from P power dissipated in the resistor to include three or even more + (-u l al b2) x -0.01b,
to Q. Decreasing resistance by is p = i2r. We are asked to find variables. Partial differentia-
0.5 Q is equivalent to moving the change in power when the re- tion is no longer j ust a matter so that
from P to R. Both produce an sistor increases by 1 Q and the of cutting slices through a three-
increase of current (an uphill current decreases by 0.1 mA. Ta dimensional surface, for the f1z = 0.02u (2alb + 1)
climb on the sloping surface of begin with, p = 0.00252x33 'gt-aph' is a surface in space of [Eq. 129J
the graph). We have simply added = 206.25 uw The change in power four 01" more dimensions. But
these increases. But really, we resulting from the change in re- the principle isjust the same. If Equation 129 is used to calculate
want to knowthe change thatoc- sistance is: z is a function of c, b, C, ... , then the change in z for any combi-
curs in going from P to S. Going nation of values of u, a, and 6.
&- &- &-
from P to Q and then adding on (op I 8r)xi1r = i2xI =6.25x10-6. 8z=-xi"ia+-xi"ib+-xi"ic+ .. Forexample, ifu = 5 V,a= 100Q
Iia i5b l5c and b = 200 Q, then
the current increase that was cal-
culated on the basis of being The change in power resulting [Eq. 127]
at P is not quite the same thing from the change in current is: f1z = 0.02x5(200/200+ 1) = 0.2.
as going directly from P to S. AB am example, considerthe non-
The change in level between Q (öp I 8i)xf1i = 2irx-0.001 inverting amplifier circuit of The output voltage may be up
and S is not quite the same as = 0.165x-0.0001 = -16.5x1~. Fig. 145. The output voltage, z, to 0.2 V greater than its calcu-
that between P and R. Having is related to the input voltage, lated value. With the exact values
said this, we still find this tech- The total change in current is u, by the equation: given, z = 7.5 V, but errors may
nique a useful one, provided that (6.25-16.5)x10-6 = -10.25x10-6. take it up to a maximum of7.7 V.
we keep to smaU changes in u The new power level is z = u(alb+1) [Eq. 128]
and r. If PQ is smalI, the in- (206.25-10.25)xl0-6 = 196 Jlw. Related rates
crease in going from Q to S 1S The values of the resistors are
not very different to the increase We can easily check this result represented by variables a and In the examples above, the two
in going from P to R, so simply by war king out the power when b. Here we have three indepen- (or more) independent variables
adding the increases 1S precise i = 2.4 mA and r = 34 Q. For this dent variables, and can wt-ite have been independent of each
enough. pair of values, p = 0.00242x34 three partial differentials by dif- other. But it may happen that
= 195.84 pw. This, of course, is ferentiating Eq. 128 appropri- they are related to each other be-
a precise value and it might be ately: cause they are both functions of
Small increments
wondered why we need to go into same other quantity. Very often
Let us summarise this in math- the complications of differenti- I5z18u~alb+1; we find that two variables are
ematical terms. The small change ation to abtain only an approxi- functions of time. The input to
in voltage ts L1u. The rate of mate value. The reason 1S that a circuit may be a ramping volt-
the method of partial differen- ageor a sine wave; in either case,
tiations ia applicable to equations the voltageis a function oftime.In
much more complex than the the same circuit, there may be
simple one used here to if lus- photoresistor responding to a
trate the method. It can be ap- periodically changing light level;
plied to any combination of re- the resistance of this ia also a
sistances and currents, it being function of time. Ir a current in
necessary only to insert the ap- the circuit is affected both by
propria te values of L1i and L1r the input voltage and the reais-
into the equations once the ini- tance of the photoresistor, we
tial differentiationhas beendone. have a circuit which can be ana-
lysed by partial differentials.
Fig.144 Fig.145 For example, suppose that

ELEKTOR ELECTRONICS JUNE 1994

 GENERAL INTEREST

the input is given by u = 2sin4t. di/dt = 72.77 mA s-t. partial differentials 8z I 8x terms. The series is:
The resistance of the photore- and 8z I 8y when y = 3n/2 ~ 6 [(sin 2t/2)
sistor is given by r = 100+40t2 If instead we need to know the (a) z = 2x+3y + (sin4t14) + (sin6t/6)
(a light source is approaching the
rate of change of current when (h) z = 3x2/4y + ... ]
sensor and the inverse square the voltage has a particularvaJue,
law applies), The current is we calculate the corresponding 2. A 22 ).lF capacitor has a p.d,
i = u l r, We want to know the values of rand t, then substi- of 9 V across it. The charge q
rate of change of i. tute these in Eq. 131. is given by q = cu. Ifthe volt-
The standard equation CEq. 127) In this example, the time-re- age increases by 0.2 V and
tells us that: lated variables are synchronised the capacitance decreases by
with each other. This is because 0.5).lF, use the method of'par-
8i 8i
Si= 8u xl!.u+ &- xl!.r the differentials still incJude t, tial differentials to calculate
This need not be the case. If, for by how much the charge
This equation does not require example, we state that the volt- changes.
the change to occur in any spec- age is ramping upward at3 V s+,
ified length oftime. Now we will the equation is: u = 3t. Differ- 3. In a voltage-regulatingcircuit
introduce the time element by entiating this gives du/dt = 3. (Fig. 146), a constant volt-
saying that the changes must The rate of change is constant; age U = 10 V is applied. The
take place in a short length of it does not depend on how lang band-gap reference produces
time, M. Dividing both sides by has elapsed since we began tim- a reduced voltage with atempco
Llt: ing. We might also specify that of 40 ppmrc. The resistor
the resistance is increasing at the has a tempco of 100 ppml°C.
Si 8i Su: 8i l!.r rate of 411 s-l, so that dr/dt = 4. When the temperature is 25 -c,
-=-x-+-x-
l!.t 8u l!.t &- l!.t The variables are still both time- u isexactly 2.5 V andr=7.5 kl1.
dependent, since they are rates The current through the re-
If Lit is made very small, ap- ofchange, butthey are not syn- sistor is i = (U-u)/r. What is
proaching zero, the expressions chronised in the time-scale. the tempco of the current?
containing it become differen- Substituting these differentiale
tials: in Eq. 130 gives:
di 8i du 8i dr
-=-x-+-x- di/dt = 3/r-4ulr2.
dt 8u dt &- dt +
[Eq. 130]In this case, it is possibJeto spec- .!:!.=10V
ify any pair of values for u and
The expression on the left is the rand calculate the correspond-
rate of change of current with ing rate of change of i. For ex- 2V5

time. We call it the total dif- ample, when u = 2 V and r = 411, 930010·XVJI·146
ferential of z with respect to t. then:
Note that Eq. 130 does not con-
tain any finite quantities such as di/dt = 3/4-8/16 = 0.25 A s-t. Fig.146
M, and soitgives an exact result,
not an approximation. Let us As wen as rates of change with
apply this equation to our ex- respect to time, we often encounter
ample. Calculating the partial rates of change with respect to Answers to
differentials from the equation temperature. Thevaluesofmany Test yourselt (Part 16)
given (the same as in the first ex- components, inc1udingresistors,
ample this month): capacitors and regulators are 1. (a) Odd function; no con-
all functions oftemperature. Tbe stant term, no eosine terms;
8il8u = 1Ir; temperature coefficient, often expect to find only sine
known as the tempco of a re- terms.
sistor, may be ofthe order of200 (b) Half-wave repetition,
parts per million per degree Celsius neither even Dar odd; expect
Calculating the other differen- (er per kelvin which, in this con- to find a constant term,
tials in the usuaJ way from the text, amounts to the same thing). even cosine terms and even
equations given: As a differential, and now using sine terms.
the variable t to represent tem- (c) Even function; expect to
du/dt = 8cos4t; perature insteadoftime, we have: find a constant term and co-
sine terms.
dr/dt = 80t. duldt = 2rxl0-4
2, 'I'his has half-wave repeti-
Substituting in Eq. 130: Given the tempcos of resistors tion; with the constant term
and other components in a cir- ignored, it is an odd func-
di/dt = (8 cos 4t)/r-SOut I r2. cuit, the method described above tion; expect to find a con-
[Eq. 131] can be used to calculate the rate stant term and even sine
of change of current or other terms.
We can use Eq. 131 to find the quantitu with respect to tem- Ao = 3n/2
rate of change of current when perature changes. an = 0
one of the quantities has a spec- bn = ~3/n(cosnn+1)
ified vaJue. For example, when When n is even, cos n1t = 1,
Test yourself
t = 0.1 s, u = 0.7788 V and and bn = ~6/n.
r = 100.4 11. Then, substituting 1. Given that z is a function of When n is odd, cosna = -1
in Eq. 131 gives: both x and y, calculate the and bn = O. Only even

ELEKTOR ELECTRONICS JUNE 1994

I.. ·
FIGURING IT OUT
PART 18 - THE LAPLACE TRANSFORM
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voItage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

ystems of many kinds, from set. We can also perform the in- on a time function. By this we
S electronic circuits to the mo-
tion of planets and the spend-
verse transforrn, using the eX
function on the calculator, and
mean that a quantity (such as
voltage)is specified by a funetion
ing patterns of supermarket cus- recover the original set. fit), in whieh time is the inde-
tomers, may be model1ed with At one time, logs were used pendent variable. For example,
differential equations. Setting extensively to assist in calcula- u = 3 sin rot. We say that u is in
up the equations is usually fairly tions, either in the form of log the time domain. We sometimes
easy. Solving the equations may tables, or on the slide rule. This write u(t) and i(t)focthe function
be unexpectedly dfficult. Ob- is because they have the prop- instead of ((t), the lower ca se
viously, a solution may be more erty that adding logs is the equiv- letter indicating the quantity
readily reached if we assume alent of multiplying ordinary involved. The Laplace transform
thatcertain values are zero or re- numbers. For example: of a time function (Ct) is F(s),
main constant, but this may make where:
!.
the model unrealistic, or at least
limit its applicabi1ity. Various
In 2 + In 3.5 = 0.6931 + 1.2528
= 1.9459 = In 7. F(s)= s: ((t) e-~ dt
930010-XVIIl-147

techniques have been devised

to assist the solution of differ- Taking logs, adding them, and [Eq. 132] Fig.147
ential equations. One of these takingthe antilogofthe sum, has The main condition attached to
is the operator D, described in given us the product. Adding is this transform 1S that the inte-
Part 14.Anotbertechnique, known easier than multiplying, espe- gral must be convergent, which ,,(tl = 1 0+ < t < +=.
as Euler's method, is easy to use cial1y when there are many dec- means that the integral must ap- Note that we specify u up to the
but, since it reHes heavily on ap- imal places. proach adefinite limiting value instant before the switch is elosed
proximations, is not entirely re- Finding a fractional power of as t becomes large, and not be- (t = 0-) and from tbe instant after
liable. A third and powerful tech- a number is a troublesame op- eorne infinitely large itself. Most the switch is closed (t = 0+). lt is
nique is the Laplace transform. eration. For example, find the ofthe time, functions met in elec- undefined for the instant of clos-
value of 3.52.4. The log transfor- tronics conform to this require- ing. The second ofthe two func-
mation comes to OUT aid because ment. It 1S also specified tbat tions above is the one we trans-
Transforms multiplying a log by a number (Ct) = 0 when t = O. If it is not, form. Substituting in Eq. 127:
Given a set of values, we trans- is the equivalent of finding the we insert starting conditions into
power. In this example: the equatians. U(s)= fe-<tdt
form them by performing a spec-
ified mathematical operation on
each ofthem. A well-known ex- 2.4 x In 3.5 = 2.4 x 1.2528 Unit step function - -1[ e -"]- 0 -li
_ - s
s
ample is the logarithm.ic trans- = In 20.22
form. Take this set of numbers: The Laplace transform is not
that is, unduly complicated when the As t approaches infinity, e-st ap-
0.5,2,3.5,7,20.22. time function is a simple one. proaches zero; when t is zero,
3.52.4 = 20.22. As an example ofthe transform, e-st = eO = 1. The integral re-
The logarithmic transform of we look at the unit step fune- duces to 1/s. Similarly, ifthe step
these is obtained by taking the Multiplyingthe log, then finding tion, which, in effect, 18equiva- 15other than a urrit step, for ex-
natural logs, or logs to any other the antilog, has given us the lent tojust turning on the power arnple u = a, the transform is
base, if preferred. Natural logs power. Multiplyingis easierthan switch-see Fig. 147a. Up to als.
can be found with a pocket cal- taking powers, especially when the instant ofthe starting time,
culator and we obtain the trans- they are fractional. Summing the switcb 1.5 open. The voltage Other transforms
formed set (shown here to 4 dec- up, we use the log transform be- across the resistor is u = O. The
imal po'ints): cause it makes certain maths switch is closed when t = 0 and In a similar way, we can find the
operations easier to do. We use the voltage across theresistorin- transforms ofother common func-
-0.6931,0.6931, 1.2528, the Laplaee transform for the stantly rises to u = 1. We have a tions, As might be expected, the
1.9459,3.0067. same reason. unit step-see Fig. 147b. The integrations required are more
function describing this is piece- complicated than that for the
The members ofthe original set wise: simple step function above.
Laplace transform However, there is no need to worry
map in a one-to-one manner with
the members ofthe transformed The Laplaee trans form operates u(t) = 0 about this as the transforme of

ELEKTOR ELECTRONICS JULViAUGUST 1994

 FIGURING 11"OUT - PART 18

all the most frequently met func- tion, we need to know how to sistsofseveral terms, usually in- that timing begins. Again, we
tions are available in a table transform these. Box 2 shows the duding a constant, a first fiffer- multiplythe transform by the co-
(see Box 1). The transforme in transforms ofthe terms ofa dif- ential (dy/dt), a second differen- efficient of the original terrn, so
Box 1 are mostly needed when ferential equation and the way tial (d2y/dt2), and possibly dif- the transform of 5000[di(l)ldt]
we are performing the reverse to use these and the traneforme ferential termsofthird or higher is
transform, as we shall see later. in Box 1is explained in the ex- orders. An example taken from 500081(s) - 5000i(0+).
Since our calculations usually ample below. Part 13 is a model of aseries
begin with a differential equa- A differential equation con- LCR circuit in which R = 500 Q, In Part 13, we stipulated that
C = 2 pF, and L = 100 mH. The the current is 2 mA when tim-
Function differential equaticn, based on ing begins, so i(O+) = 0.002 and
Transform Conditions
Eq. 94, is: the transform ts 500081(8)-10.
1 (unit step function} 118 9\(8) > 0 Note that instead ofhaving to dif-
a (step function) als 9\(s) > 0 d2i (t)! dt2 +5000 (di(t)/dt) ferentiate i(t), we have simply
t (ramp function) 1/82 + (5x106)i(t) = 0 multiplied its transform by s to
[Eq. 133] give sl(8). Multiplying by 8 is
e'" (growth function) 1I(8-a) 9\(8) > a Note that we have replaced i in much easier than differentrat-
e-at (decay function) 1/(8+a) 9\(8) >-a Eq. 94 with i(t) toemphasisethat ing, which is thereason forusing
1--€'" alS(8-a) 9\(8) > a i varies in the time domein ~ it the transform.
1S a funetion of t - but it repre- • The transform ofthe extreme
1-e-'" a/sfs-r.a) 9'(8) >-a sents exactly the same quantity. left-hand term, d2i(tlldt2, is also
sin fit (0/(82+(02) 8>0 Dur aim is to discover the way i made according to Eq. 132 and,
cos ox s/(82+w2) 8>0 varies with time. In other words, assuming that no current is fiow-
to obtain the function i(l). To ing at the instant the power is
solve this equation, we will first switched on, it produces
9\(s) is the real part of s in those cases where s is a complex find its transform. Itcan beshown
number. that, if we find the transform of 82[(8)-si(O+)-di(O+)Idt.
a and ro are constants. each term individual1y and sum
the transforms for each side of The second differential is ob-
the equation, this gives the trans- tained by multiplying 8[(S) by s;
Box 1 - Laplace Iransforms of functions. form of the whole equation. as before, it is much easier to mul-
Working from right to left: tiply by s than it is to differen-
Term Transformed term • The transform ofOis 0 (a unit tiate. In this expression, i(O+)
Constant a als step in which a = 0 - see Box 1) = 0.002, as above. di(O+)Idtis the
Funetion r(tl F(s) • The transform of (5x106)i(t) rate of change of current when
presents a problem because, until timing begins. In Part 13, we
First derivativef"(t) 8F(8)-f(O+) we have solved the equation, we said that this is 0.05 A s-l, so
Second derivative {""(t) 82 F(s)-sf(O+)- r '(0+) do notknow the form ofthe func- the tranform under these start-
tion i. Represent it by a symbol, ing eonditions becomes
lntegral f:r(t) dt F(8)/8+{CO+ )/8 using the corresponding capital
letter, and say that its trans- 82[(8)-0.0028 - 0.05.
form is1(s). Note that the trans-
Box 2 - Laplace transforms of differential equations. form 1S in the 5 domain, which Summing the transforms of
can be shown to be the frequency the terms on each side of
domain. Since any constant mul- Eq. 133, we obtain a new equa-
1. Faetorise the denominator ofthe original fraction, ifpossible.
tiplier also multiplies the trans- tion in the frequency domain:
2. Match the factors against one or more ofthe formats shown in
form, the transform of(5x106)i(t)
Box 4. 82[(8)-0.002s-0.05
is (5x106)1(8).
3. Box 4 sets out the form ofthe partial fraetions; write them out +500081(8)-10
• The transform of5000(di(t)ldt)
as an identity.
can be obtained by integration +(5x106)1(8)=0
4. Clear fractions bymultiplying both sides ofthe identity by the
according to Eq. 132. Skipping
original denominator.
over the intermediate stages, The next step is to simplify this
5. Equate coefficients of each power of x, obtaining equations for
which are a conventional inte- in order to find l(s), the Laplace
constants A, B, e!c. gration by parts, wefind that the transform of Eq.l33. Examination
6. Solve these equations to find A, B, e!c. transform of di(t)/dt is sl(8 )-i(O+). shows thatI(s) is a faetor in three
As above, we have bad to state of tbe terms. The equation be-
Example: Express (3x+5)1(xLx-12) as a partial fraction. the transform in terms ofl(s), be- comes:
Step 1: (3x+5)1(xLx-12) = (3x+5)1(x+3J (.<-4). cause we do not yet know the
Step 2: There are two factors, both have the form (z+a), So, form ofi(t). The term i(O+)is the 182 +5000s+(5x106 )11(8)
there are two partial fractions, both with the form current flowing at the instant =0.0028+10.05
N(x+a).
Step 3; (3x+5)/(x+3) (x-4) es A!(x+3)+B/(x-4).
Factors in denominator Partial fractions
Step 4: 3x+5 = A(x-4)+Blx+3)
= Ax-4A+Bx+3B. X+a N(x+a)
Step 5: Equating ccefflcients ofx: A+B = 3. x2+ax+b (AHB)/(x2+a.nb)
Equating constants: -4A+3B = 5. (x+a)2 N(x+a) and B/(x+a)2
Step 6: A = 417 and B = 1717;partial fractions are:
(x+a)3 N(x+a), B/(x+a)2 and C/(x+a)3
(3X+5)I (x2-X-12) = 417 (x+3)+17/7 (x-4).
(x2+ax+b)2 (Ax+B)!(x2+ax+b) and
Further worked examples appear in the text. (Cx+D)/(x2+a.nb)2

Box 3 - Partial fractions. Box 4 - Formals for partial fraclions.

ELEKTOR ELECTRONICS JULY/AUGUST 1994

 GENERAL INTEREST

so that: usingBox 1, toobtainanequa- calculate initial and final cur-

tion for i(t). rents and voltages.
0.002s+10.05
[(s)
S2 +5000s+(5x106) The important point to no- Q Test yourself
-';:;0.5 sH
tice is that 'at no stage is there
The expression below the line can any need to differentiate or in- + 1. Use the table ofBox lto find
be factorised, usingthe quadratic tegrate. The tables oftransforms the transforms of (a) 2e-6',
30mV and (b) 3cos4t.
equation formula (available on cover almost every case, and ta- 930010-XVIII-149
many scientific calculators) to bles more extensive than Boxes
give: 1 and 2 are available for the in- 2. Use the table ofBox 1to find
frequently used functions. The Fig.149 the inverse transform of (a)
[(s)~~2s+10.05 3/s(s--3), and (b) 28/(s2+49).
onIy maths required is simple al-
(s+ 1382)(s+3618)
gebraic manipulation, mostly
[Eq. 134J the finding of partial fractions. of i(O+)~ 6/100 ~ 0.06 A. Sub- 3. Write the transform of this
We are working toward getting This does not seem to appear in stituting in Eq. 136: equation and solve it, using
the termsoftheequation into the GeSE maths syllabuses, so a the starting conditions given:
same form as Olle or more ofthe simple routine for this is out- 120[(s)+0.5s[(s)-0.03 ~0 d2i(t)1dt2+di(t)/dt-6 ~ 0, given
transformed expressions in Box 1. lined in Boxes 3 and 4. or that nü+) ~ 0 and di(O+)/dt~ 5.
Many of these have a single fac- 120[(s)+0.5s[(s) ~ 0.03
tor, such as (s-.a), beneath the Transformed circuits [Eq. 137J 4. In the circuit ofFig.151, the
line. The equation above could Equation 137 is the equation input voltage is 10 V at t = 0,
be turned into this form by ex- which woulld be obtained by ap- and ramps down according
A
pressing the fraction as partial plying the network rules to the to the equation u ~ 10-200t.
fractions. Wben this is done, we circuit of Fig.149. In other words, (a) Express the current in this
find that: Fig.149is thecircuitofFig.148 circuit as a differential equa-
~ 500mH transformed into the frequency tion; (b) find the transform
[(s) 0.003258 0.001258 domain. Note how the induc- of the equation and simplify
s+1382 s+3618 tance depends on s, which has it; (c) find the inverse trans-
(Toconfirm this working, add the 930010·XVIII-148 the dimensions offrequency. The form to obtain an equation
two partial fractions and verify circuit includes a voltage source, which shows how the current
that you get back to the fraction Fig.148 representing the voltage due to varies in time.
ofEq.134). A search through the theinitial current. FromEq.137:
table of Box 1 shows that ex-
pressions of this form are trans- Figure 148 shows a circuit in [es) ~ 0.03/(0.5s+120)
forms of the decay function, which the switch is in position U(t) = 10-20ot +
i(t) = e-at. We are ready to per- A lang enough to reach a steady Before we can reverse the trans-
form the reverse transform, read- state. The switch is changed to form, we must multiply the nu-
ing from right to left in Box 1 position B when t = O. At any merator and denominator ofthe 2H5

and substituting appropriate instant the voltage aeross the two fraction by 2 to obtain 5 in the
values for a and 5: resistors is 120i(t) and, because denominator instead of 0.055: 9300tO-XVIII-151

it depends on the rate of change

i( t)~0.003258e-1382 of current, the voltage across i(s) ~ 0.06/(s+240) Fig. 151
-0. 001258e -3618 the inductoris 0.5di(t)/dt. Applying
the rules for networks that we Box 1 shows that this is the trans-
This is precisely the same re- first met in Part 4, we can state form ofthe decay function. In the Answers to
sult as was obtained in Part 13 that by KVL, and with the switch transform. a = 240 and: Test yourself (Part 17)
using the straightforward tech- in position B:
niques for solving differential i(t) ~ 0.06e-240t
equations. It serves to confirm 120i(t)+0.5di(t)/dt ~ 0 1a lizl&. ~ 2+3y;
that the Laplace transform does [Eq.135J The graph ofthis function is given l5e I ~ ~ 2>:+3.
give the correct result, in spite This differential equation de- in Fig. 150. Itshowsthatthecur-
of the apparently roundabout scribes the behaviour of the cir- rent starts at 0_06 A and decays 1b 15e1&.~ 3x12y;
route from start to finish. The ad- cuit but, sinee it does not take exponentially, beingvery close to 15e/~ ~ _3x2/4y2.
vantage of the Laplace method into account the initial state of zero after 20 ms.
is that it can be used with dif- the circuit, we do not yet have 2. Charge decreases by 0.2 ]lC,
ferential equations that do not an equation for i(t). Because the from 198.0]lC to 197.8 jrC.
.t.
yield readily to the ordinary tech- voltage source is now switehed
niques. out of the circuit, it is hard to !~l 1\'
__~~ii~~r.~~~1 3. dr/dt ~ 0.75;
Ithas taken several paragraphs see how we can allow for it, This duld! ~ 100x10-6;
to work through this example, is where the transform helps i5ilou ~ =Lr;
but the steps in the calculation out. Transforming Eq. 135: i5ilor ~ (u-1O)lr2;
are few: dildt ~ 86.7xlO-9 when
120[(sl+0.5[s[(s)-i(O+)} ~ 0 i~ 1mA, equivalent toa tempco
1. Transform the differential equa- [Eq. 136J of 86.7 ppm °C-I.
tion, using Boxes 1 and 2. Now we can take the initial CUT- [93001O-XVIIIJ
2. Simplify the equation to ob- rent, i(O+),into account. We find ,
I""'j
tain an expression for 1(5). its value by noting that in the _,..."..,iO

3. Insert irritial values and re- steady state with the switch at
cast the expression so that it A, assuming that the resistance Fig.150
consists ofterms ofthe same of the inductor is negligible, a
typeas thetransforms in Box 1. volta ge of 6 V across a resis- Next month we look at how
4. Find the inverse transforms, tance of 100 Q causes a current to model capacitors and how to

ELEKTOR ELECTRONWS JYLY/AUGUST 1994

 FIGURING IT OUT
PART 19 - BEGINNINGS AND ENDINGS
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

T hiS month we take the

Laplace transform a step
further by considering how to
equation for a capacitor (Eq.
27, Part 5). Using Kirchhoff's
current law, we ean write:
Urs) = 50[1Is - 11(0.1+ s)].
[Eq. 139]
The inverse transform 1S:
A 51

"
B
use it to find the initial and u(t)!50 + 0.2du/dt = l. u(t) = 50(1 - e-O.1).
+
final states of a circuit. But, The differential equation de- [Eq. 140] 10V
befare we da this, we will look scribes how voltage varies in Figure 153 shows the graph 2mF

at circuits containing capaci- the circuit from the moment SI of Eq. 140. The voltage across
tive elements. is closed. Transforming this the resistor and capacitor

51
equation, using the tables of
transfarms given in boxes 1
°
starts at V, increaeing at an
exponentially decreasing rate 930010 XIX ·154

and 2 oflast month's article: until it reaches 50 V.

U(s)!50 + 0.2[(sU(s) - u(O+)] Fig.154
= 1/s.
As always, the use of the Capacitor current [Eq.141]
200mF
transform means that instead To model the currents in a ca- The initial charge is now rep-
of differentiating a function pacitive circuit, we need to resented by q(O+). To werk with
of u, we simply have to multi- begin with a voltage equation. equations which contain inte-
ply Urs) by s, which is a much In Part 5 we gave the voltage gral expressions, we use the
930010 XIX· 152
easier operation. Assuming equation of a capacitor in the transform given in Box 2 of
that the capacitor is uncharged form: Part 18. The transform of
Fig.152 to begin with, so that u(O+) = 0,
the equation simplifies to: f;i(t)dt is I(s)/8 +i(O+)ls.
A simple example 1Sthe ca- U(s)(1150 + 0.2s) = 1Is
pacitor of Fig. 152, which ini- Urs) = 501s(1 + LOs). [Eq.21] Note that to integrate an ex-
tial1y has no charge and is Converting the right-hand side In this equation, u is the volt- pression, we divide its trans-
charged when 81 15 c1osed. The into partial fractions (see last age across the capacitor at any form by s, This is easier than
current being supplied to the month, Box 3): given time, C is the capaci- integrating.
circuit rises instantly from Uis)» 50ls - 500/(1 + l.Os). tance, I is the current entering As a practieal example, con-
zero to 1 A. Last rnonth, we de- [Eq. 138] or leaving the capacitor at any sider the circuit of Fig. 154, in
scribed this as a unit step The second expression on the given time, and c is the initial which a capacitor is eharged to
function. The currents arriv- right will obviously transform charge (if any) on the capaci- 10 V with the switch in posi-
ing at and leaving node X at back into the decay function, tor. Rewriting the equation tion A. When t == 0, the switch
any instant are shown in the but we must first reduce the with symbols to remind us that is moved to poaition B. If the
figure. The current through coefficient of s to unity. Divide voltage and cur-rent are both cut-rent in the circu.it at any
the resistor is obtained by ap- top and bottom of this expres- functions of time, we obtain: given time is i(t), the voltage
plying Ohm's law. The current sion by 10: across the resistor is
that is charging the capacitor 50/(0.1 + s). u(t)~ 1ft i(t)dt+q(O+ )/C Ri(t) = 1000i(t). The voltage
is obtained from the current The transform becomes C 0
aeross the capacitor is given by

, , , • , 10
~ - - - - - -"..(:'--- ~~-- - - - - ---

~i"
11--,
(';Al·-,
1._,.
2
t (8)----'
00

ac .e
e ... ''}

10 .,
·10
o 10 ~ se 1I3OO10·XlX·155

" "t
(5) ----.
93001D· XIX· 153

Fig.153. Fig.155

ELEKTOR ELECTRONICS SEPTEMBER 1994

 FIGUIUNG IT OUT - PART 19

Eq. 141 aso This is 1/C8 = 1000/8.

Element Voltage equation Transformed voltage
The voltage equation of an
Jcf';(t)dt+q(O+ )/C inductor, given in Part 5, is:
C 0
u(t) = u(O+) U(8) = u(O+) /
Step voltage, u(O+) 8
uL=Ldi(t)
Substituting C = 0.002 Fand
Resistance, R u(t) = i(t) r U(8) = 1(8) r dt
q(O+)/C = u(O+) = 10 V, the ini-
tial voltage across the capaci- Capacitance, C u(t) = (1IC)!i(t)dt U(8) = [(8) I C8 which transforms to Ls I (8).
tor, into the equation for Inductance, L u(t) = Ldi(t) / dt U(8) = L8 I (8) The impedance is Ls = 0.25 s.
Kirchhoff's voltage law: These transforms are sum-
Table 1. Circuit transforms (voltages). marised in Table 1. This is the
1000i(t)+500 f:i(t)dt+10=0 table to use when working
i(O) = lim-0.018/(8+0.5). u(O) with voltages according to
As usual, the equation con- Substituting 'infinity' directly KVL, in order to calculate cur-
=Iim 508[1/8-1/(0.004+8)]
tains an unknown funetion, in into this equation leads to a rents. In this example, we surn
this ease i(t). We want to find meaningless indeterminate re- =Iim [50-508/(0.004+8)] the transformed voltages
it and we da not want to have sult. We overcome this prob- around the circuit, according
to integrate it. So we trans- lem by looking at the equation = lim [50-50]=0 10 KVL:
form the equation as in Boxes to see wbat is the highest [(8)1000/8+0.2581(8) = 68
1 and 2, Part 18: power of S. The highest power The init.ial voltage is zero. [(8)[100018+0.258] = 618
10001(8) is 51, which is s. Divide top and Next, for the final value: l(s)[(1000+0.2582)18] = 6/8
+ 500[1(8)18+[(0+)18] bottom by 8: [(8) = 6/(1000+0.2582).
u(~)
+ 10/s = O. [Eq. 144]
i(O)=lim -0.018/8 =l.im 508[lfS-1!(0.004+8)] This 15 the first time that we
The initital current, 1(0+) = O. HO
Substituting this in the equa- 8/8+0.5/8 have wanted to reverse-trans-
e lim [50-508/(0.004+8)]
tion: . 0.01 HO form an equation with s2 in the
= Irm
10001(8)+500[(8)/8+1018 = 0 1+0.5/8 e Iim [50-0]=50 denominator. Looking at
1(8)(1000+500/8)+10/8 = 0 HO Box 1, Part 18, the obvious
1(8)(10008+500)/8 = -108 As s becomes very large, 0.5s candiate for the function is
1(8) = -10/(10008+500). becomes very small. Eventu- The final volta ge is 50 V as sin tat, which transforms to
Dividing top and bottom of Ihe ally, it is virtually zero and: originally deduced. tot (s2+w2). First, we multiply
fraction by 1000: i(O) = -0.01/1 = -0.01. the fraclion of Eq. 144, top
[(8) = -0.011(8+0.5). This is the same result as ob- LCcircuit and bottom, by 4 to obtain s2 in
[Eq. 142] tained from our knowledge of the denominator:
The reverse transform is electric circuits. The initial Now to look at something a lit- 1(8) = 24/4000+82.
5
i(t) = -0.01e-O· '. value theorem has its uses tle more complicated: a circuit This makes m2 = 4000, so that
[Eq. 143] when the behaviour of the cir- that contains both capacitive w = 63.246. The numerator is
Figure 155 shows the current cuit is not so obvious, as is and inductive elements 24, which is m multiplied by
from the moment the switch 18 shown by a later exarnple. (Fig. 156a). The capacilor is 24/63.246 = 0.3795. Wilh the
changed. The current is already charged to 6 V when calculated value for 00, the
-0.01 A (-10 mAl to begin the switch is closed. It helps equation becomes:
wi th , as might be expected Final value the analysis to draw a tran- 1(8) = 0.3795m/ (m2+82).
when the capacitor has 10 V The final value theorem tells formed circuit, as in Fig. 156b. The reverse-transforrned equa-
across it and is discharging us the final value of the cur- The voltage across the capaci- tion is:
through a 1 kQ resistor. The rent, voltage or other quantity. tor is represented by the volt- i(t) = 0.3795 sin 63.246t.
negative sign indicates the re- 'I'hie theorem states that: age source. Closing the switch [Eq. 145]
verse directicn of the current. is equivalent to introducing a The current oscillates in the
((~)=Iim 8F(8) step voltage into the circuit; circuit with an amplitude of
The current decays exponen- HO
tially to zero. the transform of this is 6/s. nearly 4 V and with a fre-
Using this theorem to find the The voltage equation for the quency ( = w/27r = 10.07 Hz.
final value of the current in capacitor 1S given inEq. 141. The equation states that the
Initial value Since we have already allowed oscillations continue indefi-
the previous example:
The initial value of a current i(~) = lim-D.0181(8+0.5). for the initial charge on the ca- nitely, but resistive and other
or volta ge may be easy to cal- Substituting 8 = 0 in the above: pacilor, Ihe lerm q(O+)/C is los ses in the circuit will
culate, as in the example i(~) = O. zero. Thus, the transformed eventually damp them out.
above. Putting t = 0 in As the capacitor discharges, voltage across the capacitor is The circuit is a tuned circuit
Eq. 142, we find: the current eventually falls to [(8)/C8. In Fig. 156b we have with aresonant frequency
i(O) = -O.OleO = -0.01 A. zero. written the transformed im- given by the equation
'I'here is another way to arrive If we apply these theorems pedance, which is the trans- i= 1I2",J(LC) = 10.07 Hz.
at this result, using the La- 10 Ihe Iransform of Eq. 139: formed vollage divided by 1(8). As in the other examples, we
pi ace transform. We apply Ihe have been able 10 check Ihe
initial value theorem, which Laplace transform method by
states that: independent means.

II'j!J
(0)= tim 8F(8)
H- Current transforms
We perform the Laplace trans- If you are working with cur-
form, mulliply it by 8, and find 0.25$ rents and wish to calculate
the limit of the product as s vollages by applying KCL,
tends to infinity. In the exam- Table 2 supplies the required
930010· XIX· 156
ple above, from Eq. 141, and transforms. This table could
taking limits as s approaches have been used with the exam-
infirrity: Fig.156 ple of Fig. 152. Note thai Ihe

ELEKTOR ELECTRONICS SEPTEMBER 1994

 GENERAL lNTEREST

capacitive and inductive trans- much maths, there are several

forms are the duals (see Element Current equation Transformed current
circuit modelling 01' circuit
Part 5) of those in Table 1. simulation programs for per-
Step current, i(O+) i(t) = i(O+) I(s) = i(O+) 1s sonal computers. Next month,
LCRcircuit Resistance, R i(t) = u(t) 1r we conclude this series by
I(s) = Urs) 1 r
looking at the subject of cir-
The behaviour of the circuit in Capacitance, C i(t) = Cdu(t) 1dt I(s) = CsU(s) cuit simulations and using
Fig. 157a 1S not readily pre- Inductance, L i(t) = WL) Ju(t) dt I(s) = U(s) 1Ls one of these programs to in-
dictable. Figure 157b is its vestigate the behaviour of cir-
transformed equivalent. Ap- Table 2. Circuit transforms (currents). cuits of varying degrees of
plyingKVL: complexity.
5001(s )+0. 18I(s)+ 108I(s )/s transform will help us find the = O.
= 2/s degree of damping.
l(s) (500+0.1s+108/s) = 2s.
Final value is lim 8I(s) Testyourself
,-t0
Multiplying both sides by s: 1. Draw the transform of a cir-
Initial and final states e Iim 20s/(82 +50005+109)
I(s) (5008+0.1s2+108 = 2 ,~o cuit like Fig. 152, except
l(s) = 2/(0.182+500+108). If we are interested onJy in the that the reeistance is 220 Q
=(20xO)/(02 +5000xO+I09)
Multiply top and bottom by 10 initial and final currents or and the capacitance is
to get 82 in the denominatcr: voltages, there is DO need for =0/(0+0+ 109 )=0. 180 J.lF.Find equations for
1(8) = 20/(82+5000s+109) the reverse transform. The ini- Urs) and ur!), and calculate
[Eq. 146J tial and final value theorems Tbe current is zero to begin the initial and final volt-
If we try to factorise the de- require only the transformed with and finally dies away to ages.
nominator, wi th the aim of equation. In the previous ex- zero.
finding a pair of fractions, we ample: 2. Given the circuit of
find that the denominator has l(s) = 20/(82+5000s+109). Fig. 157a, but with the ca-
imaginary roots. We must try
Help at hand
pacitor replaced by 10 pF,
another approach. One of the Initial value is tim 5[(S)
We have tt-ied to illustrate the ca1culate the equation for
standard functions not in- various modelling techniques current against time. Is
c1uded in last month's Box 1 is = lim2081(s2 +50008+ 109).
,~- with examples in which the this circuit underdamped,
e-at sin cit, which has the mathematics is not too in- critically damped 01' over-
transform: The highest power of S 1S s2, so volved. With circuits that are damped? Hint: the trans-
to ! [(s+a)2+w2J. divide throughout by s2: more complicated than those form used previously with
Our plan is to remould the ex- we have analysed here, the this circuit is not used now.
Initial value is
pression in Eq. 146 into this principles of analysis remain
form. The expansion of (8+a)2 =lim20s/(s2 +5000s+109) the same, but the calculations
is (s2+2as+a2). So the 5000 in soon begin to spread over
Answers to
Eq. 146 corresponds tc 2a in = lim (201s )/(1 +50001s+ 109 1s' ). many pages. There is little Testyourself (Port 18)
the expansion. In other words, new in such calculations; one 1a 2/(8+6)
a = 2500. Now a2 = 625x104, so As s becomes very large, 201s, just has to keep a clear head
the 109 in Eq. 146 is made up 500018 and 109182 approach and a void a rithmetical errors. 1b 3s/(s2+16)
of 625x104+w2. This makes zero: irritial value is 0/(1+0+0) For those who do not relish too
oJ2 = lOL625x104 = 99375x104. 2a I_e3t
From this we find that
w = 31524. (0) st
A
® 2b 4sin 7t
The 20 in the numerator 5000

consists of (tJ multiphed by a B

3 I(5) = 1!(s-2l-1/(8+3)
factor. 'I'his factor is 20/(J) = + + i(t) = e2t-e-3t
;(I)~ I(')~
20/31524 = 6.34xlQ4. We can av 'I,
IlOW rewrite Eq. 146 in the 4a 2.5di(t)/dt+20i(t) = 1O-200t
100mH 0.15
form of a standard transform: 2
[(s)= 10s-200+1.25s
4b
I(8)=6.34x104 x 10nF 8' (2.5s+20)
lO'/S
31524
x 930010 XIX- 157

(8+2500)2 +315242 4c i(t) = 1.75-10t-1.25e-8'.

Fig.157
The reverse transform yields: [93001O-XIXJ
i(t) = e-250Otsin 31524t.
[Eq. 147]
Tbe graph of this equation
(Fig. 158) is a sine wave with
exponentially decaying ampli-
'i"..
'A' ;=,,'-5;0315241

tude. In cornparison with the •.•

circuit of Fig. 156, the pres-
ence of the resistor means that
"'
1.0 1.2
the osci11ations are damped in
·0.2
a few milliseconds. A similar t(rns)-+-

circuit was analysed in Part 13 ·0.4

(Fig. 115). This is an under- ·0.6

damped circuit; with different

·0.8
component values, the circuit !!3OO'O·XIX·'511

might be overdamped or ci-iti-

cally damped. The Laplace Fig.158

ELEKTOR ELECTRONICS SEPTEMBER 1994

 FIGURING IT OUT
PART 20 - USING A COMPUTER
By Owen Bishop

This series is intended to help you with the quantitative aspects of electronic design:
predicting currents, voltage, waveforms, and other aspects of the behaviour of circuits.
Our aim is to provide more than just a collection of rule-of-thumb formulas.
We will explain the underlying electronic theory and, whenever
appropriate, render some insights into the mathematics involved.

he circuits that we have

T analysed in this series
have all been simple ones, usu-
V voIts-source
I amps-source
-output gnd
-output: gnd
+outputpos
+OUtput:output
V=O.OOOOOO
v=O.OOOOOO
Ex=none
Ex=none
0I=t2.000000
01=5.000000
ally consisting of no more than A Al pt:pos p2:OUtput v--3.000000
four components. The reason A R2 pt:output p2:gnd v=2.000000
for this is to illustrate the
principles and mathematical Fig.159
techniques as clearly as posi-
ble. Analysing a circuit of the [27.00C]
complexity of for exarnple, a Componentname amps Componentname amps Componentname amps
radio receiver is just a matter. _
of breaking its circuit down V volts-souree 399.9992m I amps-souree 5.000000 R Rl 399.9992m
inta self-contained seetians R R2 5.3999992
and steadily working out the
characteristics of each part. No
new principles or techniques Fig.160
are required, but there is DO
[27.00C]
denying that an analysis of
even a fairly basic circuit can Component name amps Component name amps Component name amps
soon fill several sheets of
paper. As the calculations get V U2 433.4301m V U3 -306.211m V U5 649.4064
longer, the risk of arithmetical R R1 38.46074m R R2A 267.7502m R R2B 433.4301
errors increases - not to men- R R3 -165.680m R R4 -649.406m R R5 -127.219m
tion the boredom that may set R R7 343.1954m
in at a very early stage!
During the 20 months that
this series has been rtm ni ng, Fig. 161
the pages of this magazine
have shown a marked increase through its paees on something code for specifying electrical or Age© for WindowsT'\or SAW,
in the number of advertise- more demanding. We shall also electronie eircuits. SPIeE has as we shall call it) is based on
ments for simulation software. look at an -alternative ap- been in existenee for a number an updated variant of SPIeE,
No longer is it necessary to proach, using the mathematt- of years and is used by several though it will also aecept speci-
cover sheet upon sheet with cal paekage Mathematicaw, simulation programs. Spiee- fications written in SPIeE.
close-packed numbers. No and aome of its associated elec-
longer is it necessary to wield tronie software.
a soldering iron, or switch on
an oscilloscope, to investigate a Spiee
circui t that defies analysis. A ----
circuit can be assembled on the
screen of your computer and
then subjected to a wide range
SPIeE is a set of rules or a
( 10mH
bloeklng

I~~~~tor

of tests to determine its char- 3V

aeteristics. One such simulator 100mH
program is SpiceAge© for +
c
Windowsn ... In this final part
R2 uc:slnw t
12V N

of the series we look at what

this software ean da. First of
all, to illustrate the prineiples ~ 22~F ~
involved, we shall try it out on
930010·XX·164
~ 93oo1O-XX-159
some of the circuits that we
have analysed in previous
months. Then we shall put it Fig.162 Fig.163 Fig.164

ELEKTOR ELECTRONICS NOVEMBER 1994

 FIGURlNG IT OUT - PART 20

Writing out a netlist may be

V source -outout 0 -output.t v=1.oooooo Ex=sine Fr=5oo.00oo· unfamiliar at first but the con-
R res pl:2 p2:3 v=loo.oooo ventions are easy to learn. An
Lind pt :3 p2:4 v=loo.0ooom alternative method for input-
Ccap pl:4 p2:0 v=22.oooou ting a circuit specification is to
Lblock pl:l p2:2 v=1O.ooooM Blockinginductor10 preventlow create a schematic diagram of
impedance of source swamping results the circuit on the screen, using
a sc.hematic capture progr-am
Fig.165 that can automatically produce
a compatible netlist. SAW has
[27.00C] a aister program. GESECA,
J which produces SpiceAge©
11 netlists from schematics.
'Pn
275 1 80

250
17 150
Analysis
J 1 20 Once the netlist has been en-
225
90 tered, it is saved to disc and
200

175
'"" /
60
30
then subjected to a variety of
analyses. We wanted to check
150 "- I": V "
o
30
on the calculation done in
Part 3, which was to find the
125 -<0 current flowing throug.h R2. It
<, V
is possible to 'attach' up to four

"" ...
100 90
./ 120 probes to the circuit. Often
75 these are used to measure the
.-- >< .....
--
150
voltages between various pairs
50

25
p r- k 180
of nodes. They an also be used
for measuring the current
33 39 47 56 68 82 100 120 150 180 220 270 330 390 470 through any specified compo-
930010-XX-166 f (Hz] nent. We set a probe to mea-
sure the current through R2,
Fig.166 which automatical1y causes
the program to ca1culate cur-
rents through an components.
~r---------, cuit of Fig. 162, which is de- k for kilohms, M for megohms, The resulting table of the qui-
rived from Fig_ 21 in Part 3. and so on. No other parame- escent state of the circuit is
We have given names to the ters are imperative, but we shown in Fig. 160. It shows
riodes of the circuit, though cou1d include tolerance, tem- that the current through R2 is
SAW would be equaUy content perature and temperature co- 5.3999992 A. Our result in
with numbers. efficient, if these were to be Part 3 was 5.4 A, using fewer
Components are specified in taken into account in the ana- significant figures. Note the
6V
any order. In the netlist lysis. R1 is specified in the line much smaller current (approx-
(Fig. 159), R2 is on the bottom above as a 3 Q resistor con- imately 400 mAl through Rj,
line of the list. The letter 'R' nected between 'pos' and 'gnd'. also listed as flowing through
specifies the component type - The voltage source is to de- the voltage source with which
930010-XX-167 resistor. Next comes its desig- liver a constant 12 V. The line it is in series. The analysis is
nation, 'R2', to distinguish it begins with V to specify a volt- done in a fraction of a second,
from other resistors. The code age source. Its negative output and no mistakes are made; al-
Fig.167 'pl:output' teils SAW that terminal, -out, is connected to ready we are beginning to ap-
pin 1 of the resistor is con- 'gnd' and its positive terminal, preciate the benefits of the
nected to the node named 'out- eout, goes to 'pos'. 'Ex' stands computer. Above the cable,
The procedure is straightfor- put'. Code 'p2:gnd' shows that for 'excitation'. This specifies '27.00C' indicates that the ana-
ward. The circuit is specified pin 2 is connected to the node the waveform, whether it is a lysis is performed with an am-
by a netlist, which consists of named 'ground". With a non- sine wave, a step, a ramp or bient temperature of 27 °C. If
a list af the components, their polarised device such as a re- some other recognisable func- tempcos had been specified for
parameters, and the circuit sistor, it does not matter which tion, or random noise. Here we the resistors, we could investi-
no des to which they are con- pin is called p l., and which is give it 00 excitation, but gate the circuit performance at
nected. Then we are able to p2. The final entry on this line simply make its offset 'Of" any other temperature.
run a wide range of tests on it is 'v=2.000000'. This is the equal to 12 V. Similarly, the
and to display the results as value, which in this case is current generator, specified by
tables or graphs. As an ele- 2 Q. There is no need to specify
Dodging determinants
'1', has no excitation, but has
mentary example, take the cir- the unit, but we use the suffix an offset of 5 A. Figure 163 shows a resiator
network with fcur meshes, re-
quiring four simultaneous
B Vcc negative:gnd positive:vcc v-accccco equations and a fourth-order
R load pl:vcc p2:c V=1.oo0000k %=10.00 determinant to solve it, as ex-
V source -oulput:gnd -outputb v-tooeeon Ex=sine Fr=1.oo000k Of=610.ooom plained in Part 4. The netlist
>TRI npn.lib base:b colleelor:c emitter:gnd has specifications similar to
those of Fig. 159, except that
it lists three voltage sources
Fig.168 and seven resistors. The quies-

ELEKTOR ELECTRONICS NOVEMBER 1994

m--CENERAL [NTEREST

cent analyis Fig. 161 shows [27.00Cl

the currente through a11 corn-
ponents. The circuit was one of
the 'Test Yourself" questions of
Part 4. Tbe task was to find
4.75
the current through the 3 n re-
sistor. FOT convenience, we
4.5
have named the resistors by
the value of their resistances.
4.25
As can be seen in the printout
of the analysis, Fig. 161, the
current through R3 15
-165.680 mA. The negative 3.75
sign shows that it flows from
pin 2 (node 5) to pin 1 (node 4). 3.5
This agrees with the result
found in Part 4. 3.25
If we run an analysis of
power dissipation, we obtain a
table similar tc Fig. 161, but 0.5 1.5 2 2.5 3 3.5 4 4.5 5
t[ms]
quoting power instead of cur- 93001Q.-XX·169

rent. The analysis shows that

the highest dissipation is Fig.169
1.69 W, in the 4 n resistor.
[OdB=445.6ml [27.00C)
Reactance dB UM
0
The software really comes into <P['I
its own when the circuit con- -2.5 180
tains reactive components and 150
is excited by an alternating -5
120
signal. In Fig. 164 we see a 90
-7.5
basic RLC circuit, which first 60
appeared as Fig. 45 in Part 6. -10 30
We were asked to find tbe total
impedance of the circuit and -12.5
~--------------------------------~~-to
30
the phase angle. Figure 165 60
shows the netlist; note that it -15
90
includes a massive 10 mega- 120
-17.5
henries blocking inductor (not 150
present in the original circuit., -20J-----------------------------------------------~,80
see later) in series with the
voltage source. The voltage -22.5
source has sine-wave excita- 100 1k 10k 100k 1M 10M 100M

tion, amplitude 1 V. For calcu- 930010-){X·170 I [Hzi

lating inductance, the software
replaces the source by a sm all Fig.170
resistance equivalent to its in-
ternal resistance and applies
an alternating signal of small lated in Part 6. The reason for
amplitude to the no des on adding the blocking inductor to
either aide of each component, the circuit is that the voltage
in order to calculate the im- source is virtually a short-cir-
pedance and phase angle. cuit during the analysis. With- Cl + 81
For this circuit, we set up a out the large inductor, the in-
'spot impedance' test, which ductance between 0 aod 2 is ", ::>
o
~
- :
finds the impedances at a se- almost zero; in effect, the soft-
3V
lected frequency, in this case, ware looks into the low-imped-
500 kHz. We si ted the probes anee source. The blocking in- 1mV
to measure the impedances be- duetor makes it look the other
tween the pairs of nodes to way: into the RLC in aeri es. 'V
which the reaistor, capacitor Figure 166 shows another 1kHz
----+------+-----+--~~~~
and the inductor were con- analysis performed on this cir- 930010-XX-171

nected. The readou t gave the cuit. Here, the frequency is

following results: nodes 2-3 swept over the range 30 Hz to Fig.171
(resistor): 100 .0, <l> = 0°; nodes 500 Hz and the graph shows
3--4 (inductor): 314.159 n, <I> = the response of the cireuit at
90'; nodes 4-0 (capacitor): all frequencies in this range. of the inductor increases (the right shows phase to be a
14.4686 n, <I> = -90'; nodes 0-2 We selected this range to illus- curve rising to the right), as steady _90 for the capaeitor
0

(total of RLC): 329.691 n, <I> = trate how the impedance of the frequency inereases. The and +90 for the inductor. 'I'his
0

72.34°. These results are in ac- capacitor decreases (the CUTve graphs of phase angJe are hori- analysis took just under 10 s
cord with the results calcu- falling from the left) and that zontal lines. Tbe scale on the with a 486 running at 25 MHz;

ELEKTOR ELECTRONICS NOVEMBER 1994

I GENERAL

it could be even faster if a

INTEREST

maths coprocessor was in- B batt negative:gnd positive:vcc v=3.000000

stalled. CC1 p1:in p2:1 v=4.700oou %=20.00
CC2 p1:4 p2:out v=4.70000u %=20.00
RR1 p1:vce p2:3 v=10.0000k %=2.00
Transistor amplifiers %::2.00
RR2 p1:4 p2:2 v=100.000k
A very basic amplifier circuit RR3 p1:4 p2:gnd v=1.00000k %:::2.00
is shown in Fig. 167, derived RR4A' p1:1 p2:2 v=10.0oo0k
from the circuit of Fig. 101a RR4B p1:2 p2:gnd v=90.0000k
of Part 11. Figure 168 is its RRload p1:out p2:gnd v=100.000k %=2.00
net-Hst. For the moment, we Vsource -output:gnd -outputln v=1.00000m Ex:::sine Fr=100000k
will ignore the entries l%=10' >TR1 BCW32 collector:3 base:2 emitter:gnd
and Of=610,OOOm',which were >TR2 BCW32 collector:vcc base:3 emitter:4
not there when the first simu-
lations were run. A transient
analysis with the probes on Fig.172
b-gnd and c-gnd (correspond-
ing to the input and output of [27.00Cl
this circuit) showed the input UM
as a sine wave of amplitude
'"
10 m V, and frequency 1 kHz.
The output was an almost
...
+7

-s
square wave, falling rapidly to +4
78 mV as the input rose above -a
approximately 0.7 V and j-is- +2
+1 4 6 8 10 12 14 16 18 20
ing to 6 V when the input fell
below 0.7 V. The transistor o
-1
was being driven alternately
-2
between non-conduction and -3
saturation, aeting as a switch. -4
Entering 'Of=0.610V' on the -5
netlist gives the source an off- -6
set of 0.61 V which biases the -7

transistor into its ampifying -6+---~~~~--~~---r--~--~~--~

o 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
range. Figure 169 shows the 930010-XX-I73 t[ms]
result of the transient ana-
lysis, with the input (lower
curve) plotted on a xl.O scale. Fig.173
The output is sinusoidal, with
[27.00Cl
a phase angle of 180°; in other
words, this is an inverting
amplifier. Here we have in fact
run ten analyses simultane-
ously with ten slightly differ-
ent values for the load re-
sistor. Thi s is achieved by
specifying the tolerance of the
resistor by the statement
'%=10'. We then instrueted
SAWto select ten raudem val-
ues fOTthe resistor within the
range 1000 n ±100 n and pro-
duee a transient for eaeh
value. This result shows how
the performance of the ampli-
fiel' would vary subject to tol-
erance variations in the value
of the load resistar. It shows
that although the amplifi-
eation varies, the output re- 250 500 750 1000 1250 1500 1750 2000 2250 2500 2750 t[jlS]
93001o-XX-174
mains sinusoidal. We are also
able to give the resistor a tem-
perature eoeffieient and inves- Fig.174
tigate how the amplifier be-
haves when swept over a
specificed range of tempera- from 100 Hz to 100 MHz. The begins to fall from 180° to 90°. characteristics of production
tures. In Fig. 170, the fre- two horizontal lines are the The transistor model used types. SAW includes.a library
quency response of the ampli- input amplitude and phase. in this simulation is a 'general of ZETEX transistors and it is
fier is plotted on a decibel The steeply sloping line shows purpose' n-p-n transistor. For also posnible for the user to
scale (we could alternatively how output amplitude falls off greater precision, we can write a netlist of transistors of
call for a linear scale) for the above ab out 3 MHz. At the make use of a library of more other typea, based on data
range of input frequencies same point, the phase angle complex netlists based on the sheets provided by the manu-
 FIGURING IT OUT - PART 20

[27.000Cl
facturers. Using these more
I~]r---~------~~--~--~--~--,"--~--~~~ complex models gives maxi-
+250 mUID precision to the simula-
tion, but remember that the
+200 '-I=,r===;i;;:::=';:!'!!!'!!~l!"! more complex the netlist, the
langer it takes to perform an
analysis.
Figure 171 shows an
+100
amplifier taken from SMT
Projectst), page 93. This is de-
scribed in the book as a low-
voltage a.f amplifier with a
voltage gain of 30. Keying in
the netlist (Ffg. 172) Irom the
schematic in the book took Da
-100
more than five minutes. Note
-150
the 20% tolerance of the elec-
trolytic capacitors and the 2%
-200 tclerance of the resistors. The
100 kQ potentiometer, VRj,
-250 appears in the netlist as two
resistors in series. R4A and
0 200 400 600 800 1000 1200 1400 1600 1800
R4B, with vaJues set to 10 kQ
I ",sI and 90 kQ. We can edit this
930010-XX-175

pair of values when necessary

Fig.175 to sirnulate adjuatmerrt of
VR1. The source amplitude is
[27.000Cl
10 mV, which is within the
UM range quoted for this ampli-
30 fier. The irrteresfing point 1S
that the transistors are mod-
els of the BCW32, which is a
25
surface-mount type. One of
the problems of designing in
20 SMT is that it is not possible
to use a conventional bread-
board for bench testing: the
15 components are just too small
and fiddly. The only course is
to use ccnventional compo-
10
nents and trust that the de-
sign will work when realised
5 in SMDs. With simulators
such as SAW,the design takes
place on the computer and the
2 3
• 5 6 7 8 9 10 11 12 13
930010-XX-176
14
I[s]
miniscule size of the compo-
nents is of no consequence.
We decided to check up on
Fig.176 the performance of this cir-
cuit. The transient analysis-
UM see Fig. 173-shows its out-
15 put response to one cycle of
<jJ[1
the source and, as expected,
"
13
+180
this is 180 out of phase. The
0

+150
12
thickening of the line of the
+120
output is the result of sweep-
11 +90 ing the components through
10 +60 their tolerance ranges. We
9 +30 stipulated 20 runs, with the
8 o values being selected at ran-
7 -30 dom. The graph shows that
6 -60 variations due to tolerance
5 -90 make virtually no difference
•
3
-120 to the output. The sm all nu-
merals beside the curves .indi-
-150
2 -180 cate wh ich of the 20 runs cor-
1 res ponds to each curve, but
O~~~~~~~~~~~~ __;-__-; __ the curves are so elose to-
o 2 3 4 5 6 f[Hz] gether (fortunately) that it is
93OO10-XX-1n not possible to say which 1S
which. Instead of random se-
Fig.l77 lection, values can be equally
I GENERAL INTEREST

spaced over a given range. fn{27]:=

For both methods of selection, NSolve[{6i[1]-3i[2]-i[3]==0, -3i[1]+ 10i[2]-5i[3]==2, -i[1]-51[2]+ 13i[3]-7i[4]==-3, -7i[3]+ 11i[4]==5), {i[1], i[2], i[3], i[4])]
the scale may be linear, loga- Out{27]=
rithmic or antilogarithmic. In [[i[4]->0.649408, i[1]->0.267751, i[2]->0.433432, i[3]->0.306213]]
a second analysis, we edited
the netlist to increase the re-
sistor tolerances to 10%; the Fig.178
amplifier continued to perform
perfectly, so it would be rea- [n[43]:=
sonable to specify only 10% euit when the input voltage DSolve [[y'[t]==-y[t] / rc, y(0]==O.004231, yeti, t]
tolerance for the resistors. ehanges rapidly from one
Out[43]=
The gain of the amplifier as steady value to another. An
example of this type of calcu- -[-y(t]->0.00423/EU"J)
read from Fig. 173 is approxi-
mately x8, which is rather low lation appeared in Part 13,
Fig. 110, which was an RLC [n[44]:=
when compared with a gain of
cireuit similar to Fig. 164. We Plot [O.00423/EA(0.0967*tl, [t, O.10J]
x30 quoted in the book. How-
entered the netlist (omitting Out[44]=
ever, this is partly because of
using only medium-gain tran- the blocking inductor) with
sistors. The netlist specifies R = 100 Q, C = 1.6 p F and 0.0040
BCW32 transistors which L = 100 mH. The source was
0.0035
have an hfe of 400. The BCF33 1 V, switched to zero at t = O.
transistors in the original de- Figure 174 shows the current 0.0030

sign have a gain of 800. But, through the resistor. As in the 0.0025
by making R4A equal to 1 kQ worked example in Part 13,
and R4B equal to 99 kQ, we Fig. 113, there is a surge of 6 10
were able to increase the current which dies away in 930010-XX·179
amplifier gain to x50. By set- about 2.75 ms, with what is
ting the value of the battery to elose to critical damping.
various levels, we faund that Changing C to 10 nF and re-
gain depends on V cc- With two peating the analysis, we find Fig.179
NiCd cells giving 2.6 V, the that the circuit ie under-
gain falls to x4. With a single damped. The eurrent alter-
[n[3]:=
alkaline cell at 1.5 V, the gain nates with decreasing ampl i-
Plot3D[u/r], Ir, 0.01,101, tu, 0,101, ViewPoint ->[15,15,101,
is only x1.6, so it qualifies as a tude, eventually beeoming
AxesLabel->l"u", "r", "i"! , BoxRatios ->11,1, I}]
low-voltage amplifler, but with zero (Fig. 175).
low gain. With V cc reduced to
ouue):
1.3 V (a single NiCd cell), re- Fourier
sponse begins to depart from
the sine eurve and there is no There are many other facili-
gain. ties in SAW for whieh we da
The original description not have space to describe
quated the power eonsumption here. Its ability to analyse
as 0.6 m.A, whieh is an aver- logic circuits and mixed
age value determined by mea- logie/analogue cireuits will be
surements taken on the prcto- a topic for our new series 'The
type. We set up a probe to Digital Solution', which begins
monitor the eurrent through next month. But we can not o
o
the battery and ran a fre- finish this account without
queney analysis. 'I'his showed looking at Fourier analyses. u
eurrent rising with frequeney Again we take an example 10 93OO'1).](x·180

over the audio range; it was that we have previously dealt

0.6 mA at 7 kHz, and 1.4 mA with. In Part 15, Fig. 127, we
at 10 kHz. We also measured found the Fourier series for a Fig.180
output over a range of fre- triangular wave of period 21t
quencies and found that out- seconds, amplitude 51t volts,
put was level up to 50 kHz, displaced 5" volts, so that its we have plotted the output, Fig. 177. We can read off the
above which it began to fall. minimum value is 0 V. We after filtering. This has the amplitudes of the fundamen-
The -3 dB frequency was then analysed what happens same form as that calculated tal and harmonics: 10 V, 5 V,
607 kHz. These 'measure- when it is passed through a in Fig. 130, Part 15, being a 3.33 V, 2.5 V ... This is the
mente', each obtained in a high-pass filter. For the SAW sharp spike at the beginning same as Fig. 138 of Part 16,
few seconds, show how useful analysis, we netlisted an Re of each cycle. The flrst spike is but here we have values for
this kind of software is to the high-pass filter with not as high as later spikes many more of the higher har-
designer and experimenter. C = 800 )lF and R = 100 Q, since we are simulating areal monies. Note that there are
which has a -3 dB point at circuitInot using a pure math- also same intermediate har-
2 Hz. The input to the circuit ematieal funetion as III monics of relatively low ampli-
Transients was the triangular wave. A Part 13) and it begins in an tude, resulting again from the
In previous examples we have transient analysis of this unexeited state. Having run fact that we are analysing a
used the tr-ansient routine to shows its triangular shape; the transient analyses, we can simulated circuit, not a math-
analyse a circuit driven by an the duty factor is set to 0, so perform a Fourier analysis on ematieal equation. The graph
alternating source. The rou- that it begins with a rapid each. The analysis of the tri- also displays information on
tine is also valuable for find- rise, followed by a steady fall angular wave is plotted as a the phase angle of each har-
ing out what happens in a cir- (Fig. 176). On the same graph frequency spectrurn in morric, indicated by the small
 squares. anyone who has maths to do. Graphics
FIGURING IT OUT - PART 20

tention on those aspects of

I
Routines of special interest to Mathematica relevant tc
Alternative approach the electrical engineer include The graphie output of Math- electronics. Unfortunately,
.its ability to differentiate and ematica includes 3D plots, this article was almest com-
With SpiceAge©and compara- integrate. More than that, it with colour shading to en- plete when we received our
ble simulation software we can solve differential equa- hanee the effeet. Figure 180 copy of Nodal so a more de-
build a circuit by netlisting it tions. Our first modelfing ex- is a plot of the equation tailed examination of this
(instead of physically assem- ample (Part 12) produeed a i = u / r, which was illustrated promising software will have
bling it from real compo- differential equation (Eq. 83) by Fig. 143 in Part 17. The to be deferred until another
nents). Then we use the rou- for a charging capacitor: va1ue of u ranges frotn 0 to 10, occasion.
tines of the simulator to test dqldt = -qIRC. the value of r runs from 0.001
it (instead of probing it with a Figure 179 shows this en- to 10, so as to avoid a 'division
multimeter, an oscilloscope 01' tered as a command, using the by zero' error. When gen er-
The choice
a frequency analyser). The ap- variable y in pl ace of q, and ated, the default viewpoint In this final part of 'Figuring
proach is essentially a practi- quoting the bord er condition looks over the high part of the it Out', we have demonstrated
cal one even though it is car- y(O) = 0.00423. This value is surface, so we have specified two complementary ap-
ried out on a computer. The al- the value of A that was used another viewpoint which cor- proaches to computer-assisted
ternative approach is theoreti- in Eq. 85. The program then responds to that used in circuit design. We believe that
cal, in which we work with calculates and displays the so- Fig.143. the choice between a practical
laws and equations. lution of the equation in sym- approach and a more analyti-
To highlight the differenees bolic form; interpreting the cal one is very much a matter
in the two approaches, let us conventions for printing out:
Electronic applications
of personal preference. Seme
return to the network in A purely mathematical ap- will like one, sorne will like
Fig. 163. Following the tech- y = 0.004231e'/RC proach is excellent for those the other. We like both.
niques described in Part 4, we who are well up on their the- [930010-XXj
would reduce this to a set of This is the equivalent of ory, but many users would
simultaneous equations, one Eq. 86, for prefer to be provided with rou-
for each mesh of the network, tines adapted to eleetrical and Acknowledgments
based on Kirehoff"s Voltage 0.00423/e,/RC" 0.00423e-l/RC electronic applications. These We wish to thank Those
Law. Writing out the equa- are available as a library of Engineers Ltd for their valu-
tions is easily done; solving This funetion is plotted out by files (known as Notebooks) able advice and assistance
them is a lengthy matter. The the computer for t = 0 to published as an Electrical with SpiceAge© for
recognised technique is to t = 10, and corresponds to the Engineering Pack. The windows": Wolfram
evaluate the fourth-order de- thick-lined eurve of Fig. 104. notebooks cornprise text to ex- Research Europe Ltd for their
terminant based on the coeffi- Another differential equa- plain the applications and help with Mathematica® and
eients of the equations. With tion (Eq. 94) used in this se- ready-made routines for use their Electrical
Mathematica we need only ries for modelling an RLC cir- by engineers. These make di- Engineering Pack; and
to key in the equations and in- cuit: rect use of the computationaJ Goth, Goth and Chandleri for
struct the software to solve routines of Mathematica, assistance with Nodal.
them. Figure 178 shows the d2ildt2+(RIL)dildt+911LC)i = O. aeting as an interface between
screen for this computation, the practical engineer and the
which was the 27th one in the This equation 1S purely sym- mathematics. To help expedite
Reference
current session. The command bolic and it is one of the the calculations, the pack in- SMT Projects, Owen Bishop,
'NSolve' means 'solve the fol- strengths of Mathematica cludes a number of routines published 1993 by Elektor
lowing equation and give nu- that it can work exclusively in not already present in Math- Eleetronies (Puhlishing).
merical results'. The first set symbols. Its solution of the ematica, which are specially
of curly brackets shows the equation 15: applicable to electronics.
four equations; note that 'i[l]' These include converting be-
is the standard way.to key in i = C1e-(ICR-'HCL+C2R'jltl/12CLj+ Answers to
tween a linear magnitude
a suffixed symbol such as il, +C2e-(ICR+'I-4CL+C1R2)],J/12CLj. scale and the decibel scale, TestYourself Part 19)
and also that the double and converting from polar co- 1. See Fig. 159.
equals sign '==' is Math- Cl and C2 correspond to the ordinates to complex (rectan- U(8 )=220[118-11(25.22+8).
ematica's equivalent of '='. constants A and B, deter- gular) coordinates. Another u(t)=220(1...,-25.22'). Initial
The second curly bracket lists minable by knowing the start- useful routine seleets the voltage = 0 V. Final voltage
the variables, i I to i4 for which ing conditions. Here, the equa- nearest standard resistor = 220 V.
values are required. The out- tion is in a form that can be value to a specified value.
put appears on the screen applied to any combination of These routines are simple 2. 1(s)=20/(82+5000+105). Tbe
within a few seconds. This values for R, L, and C. If we enough in themselves, but den omina tor has real roots,
gives '1 0.267751 and insert the values used in the their real benefit is that they 20.08 and 4980, and we ean
i2 = 0.433432. We were asked example in Part 13, we obtain can be written into other more obtain partial fractions ..
to find the current, i, through the particular solution given complicated routines. l(s)=4.032[1I(s+20.08)-
the 3 .Q resistor: in Eq. 100. We eould, of Nodal is a circuit analysis 1/(s+4980)1.
course, have put these values package written to interface i(t )=4.03 2( e- 20.80'...,-4980') .
i = i,-i2 = 0.267751-0.433432 into the original calculation, with Mathematica. It accepts The current rises steeply to
= -0.165681 A. along with starting conditions, circuits specified by netlists a peak of 3.93 A at about
when we would have obtained and analyses circuit behaviour 1 rns, then dies away
This is the same result as ob- the numerical solution di- by me ans of symbolic equa- smootWy, and much more
tained from simulation. rectly. For those who are fa- tions, numeri ca 1 results and slowly, without oscil1ating,
Mathematica is essen- miliar with the Laplace graphics. Like Mathematica, falling to 7.7 nA in 1 s. It is
tially a mathematica1 package Transform (Parts 18 and 19) it favours the analytical ap- overdamped.
and it provides a wealth of the program can perform proach to circuit design, but
mathematical routines for transforms in both direetions. concentrates the engineer's at-