Circuit Analysis Ii: (AC Circuits)

Will Moore
MT 12
CIRCUIT ANALYSIS II
(AC Circuits)
Syllabus
Complex impedance, power factor, frequency response of AC networks
including Bode diagrams, second-order and resonant circuits, damping
and Q factors. Laplace transform methods for transient circuit analysis
with zero initial conditions. Impulse and step responses of second-order
networks and resonant circuits. Phasors, mutual inductance and ideal
transformers.
Learning Outcomes
At the end of this course students should:
1. Appreciate the significance and utility of Kirchhoff’s laws.
2. Be familiar with current/voltage relationships for resistors, capacitors
and inductors.
3. Appreciate the significance of phasor methods in the analysis of AC
circuits.
4. Be familiar with use of phasors in node-voltage and loop analysis of
circuits.
5. Be familiar with the use of phasors in deriving Thévenin and Norton
equivalent circuits
6. Be familiar with power dissipation and energy storage in circuit
elements.
7. Be familiar with methods of describing the frequency response of
AC circuits and in particular
8. Be familiar with the Argand diagram and Bode diagram methods
9. Be familiar with resonance phenomena in electrical circuits
10. Appreciate the significance of the Q factor and damping factor.
11. Appreciate the significance of the Q factor in terms of energy
storage and energy dissipation.
12. Appreciate the significance of magnetic coupling and mutual
inductance.
13. Appreciate the transformer as a means to transform voltage, current
and impedance.
14. Appreciate the importance of transient response of electrical
circuits.
15. Be familiar with first order systems
16. Be familiar with the use of Laplace transforms in the analysis of the
transient response of electrical networks.
17. Appreciate the similarity between the use of Laplace transform and
phasor techniques in circuit analysis.
2
Circuit Analysis II WRM MT12
AC Circuits
1. Basic Ideas
Our development of the principles of circuit analysis in Circuit Analysis I

was in terms of DC circuits in which the currents and voltages were
constant and so did not vary with time. Here in Circuit Analysis II we
extend our analysis to consider time varying currents and voltages and
thereby we are able to extend our analysis to include capacitors and
inductors. In our initial discussions we will limit ourselves to sinusoidal
functions. We choose this special case because, as you have now
learnt in P1, it allows us to make use of some very powerful and helpful
mathematical techniques. It is also a common waveform in nature and it
is easy to generate in the lab. However as you have also learnt in P1,
any waveform can be expressed as a weighted superposition of
sinusoids of different frequencies and hence if we analyse a linear circuit
for sinusoidal functions we can, by appropriate superposition, handle
any function of time.
Let's begin by considering a sinusoidal variation in voltage
v Vm cost
3
in which  is the angular frequency and is measured in radians/second.
Since the angle t must change by 2 radians in the course of one
period, T, it follows that
T  2
However the time period T  f1 where f is the frequency measured in

Hertz. Thus
2
T  2f
This is a simple and very important relationship. We naturally measure

frequency in Hz – the mains frequency in the UK is 50Hz – and it is
easy
to measure the time period, T  1 f  from an oscilloscope screen.
However as we will soon see, it is mathematically more convenient to
work in terms of the angular frequency . Mistakes may be easily made
because in practice the word frequency is commonly used to refer to
4
both  and f. It is important in calculations to make sure that if 

appears, then the correct value for f = 50 Hz, say, is  = 100 rads/sec.
A simple point to labour I admit, but if I had a pound for every time
someone forgets and substitutes  = 50 . . . . . . . . !!
In our example above, v Vm cost , it was convenient that
v Vm at t  0 . In general this will not be the case and the waveform will
have an arbitrary relationship to the origin t = 0 or, equivalently the origin
may have been chosen arbitrarily and the voltage, say, may be written
in terms of a phase angle, , as
v Vm cost  
Alternatively, in terms of a different phase angle, , the same waveform

can be written
5
v Vm sint  
where
  2
The phase difference between two sinusoids is almost always measured

in angle rather than time and of course one cycle (i.e. one period)
corresponds to 2 or 360. Thus we might say that the waveform above
is out of phase with the earlier sinusoid by . When     2 we say
that the two sinusoids are said to be in quadrature. When    the

sinusoids are in opposite phase or in antiphase.
2. RMS Values
We refer to the maximum value of the sinusoid, Vm, as the “peak” value.
On the other hand, if we are looking at the waveform on an oscilloscope,
it is usually easier to measure the “peak-to-peak” value 2Vm, i.e. from
the bottom to the top. However, you will notice that most meters are
calibrated to measure the root-mean-square or rms value. This is found,
as the name suggests, for a particular function, f, by squaring the
function, averaging over a period and taking the (positive) square root of
the average. Thus the rms value of any function f(x), over the interval x
to x+X, where X denotes the period is
6
1 xX 2
frm   f y dy
s X x
For our sinusoidal function v Vm cost
The average of the square is given by
1T 2
V cos2t dt
 m
T 0
where the time period T  2 . At this point it's probably easiest to
change variables to t and to write cos2  

1
2
1 cos 2. Thus the
mean square value becomes
1 V m2 2 Vm2
  1 cos  d 
2 2 0 2 2
The root mean square value, which is simply the positive square root of
this, may be written as
Vrms = Vm /√2 ≈ 0.7 Vm.
7
Since we nearly always use rms values in our AC analysis, we assume
rms quantities unless told otherwise so by convention we just call it V as
in:
V = Vm /√2.
So, for example, when we say that the UK mains voltage is 230V what
we are really saying is that the rms value 230V. Its peak or maximum
value is actually 230√2 ≈ 325 V.
To see the real importance of the rms value let's calculate the power
dissipated in a resistor.
Here the current is given by i v R
i  Rm cos t  m cos
V
t
I
8
where Im Vm R and the power, p vi , is given by
2 2
V
p  Rcos t
m
If we want to calculate the average power dissipated over a cycle we

must integrate from t  0 to t  T  2. If we again introduce t ,
the average power dissipated, P, is given by
1 V 2 2 2
P . m  cos d
2 R 0
1 V 2m 1 2
P .  1 cos d
2 R 20 2
P Vm2 2R
If we now introduce the rms value of the voltage V Vm 2 then the
average power dissipated may be written as
P V 2 R
Indeed if the rms value of the current I I m 2 is also introduced then
P V 2 R I 2R
9
which is exactly the same form of expression we derived for the DC
case.
Therefore if we use rms values we can use the same formula for the
average power dissipation irrespective of whether the signals are AC or
DC.
3. Circuit analysis with sinusoids
Let us begin by considering the following circuit and try to find

an expression for the current, i, after the switch is closed.
10
The Kirchhoff voltage law permits us to write
di
L dt  Ri Vm cost
This is a linear differential equation, which you know how to solve.

We begin by finding the complementary function, from the homogeneous
equation:
di
L dt  Ri  0
which yields the solution:
i  A exp Rt L
We now need to find the particular integral which, for the sinusoidal
"forcing function" Vm cost , will take the form B cost C sint .
Thus the full solution is given by
i t  A exp Rt L B cost C sint
11
We see that the current consists of a "transient" term, A exp Rt L,
which eventually decays and becomes negligible in comparison with the
"steady state" response. The transient response arises because of the
sudden opening or closing of a switch but we will concentrate here on
the final sinusoidal steady state response. How long do we have to wait
for the steady state? If for example R 100 and L  25mH then
R L  4103 sec1 and so after only 1ms exp Rt L exp  40.018
and so any measurements we are likely to make on this circuit will be

truly 'steady state' measurements. Thus our solution of interest reduces
to
i  B cost C sint
In order to find B and C we need to substitute this expression back

into the governing differential equation to give
LC cos t  B sin t  RB cos t  C sin t Vm cos t
12
It is now a simple matter to compare coefficients of cost and sint

to obtain expressions for B and C which lead, after a little algebra, to
Vm
i R cos t  L
R 2  L 2
sint
If we now introduce the inductive reactance X L  L we can write this

equation as

Vm R
XL
i sint 
2 2 cost  R  XL
2 2

R X L  R 2  X 2L

The expression in curly brackets is of the form
coscost  sinsint cost  
and hence
Vm
i
R 2  X L2 cost 
where

X 
1
L
  tan 
 R
Thus we see that the effect of the inductor has been to introduce a
phase lag  between the current flowing in the circuit and the voltage
source. Similarly the ratio of the maximum voltage to the maximum
13
current is given by R 2  XL 2 which since it is a combination of
resistance and reactance is given the new name of impedance.
It is apparent that we could solve all networks containing combinations

of resistors, inductors and capacitors in this way. We would end up with
a series of simultaneous equations to solve – just as we did when
analysing DC circuits – the problem is that they would be simultaneous
differential equations which, given the effort we went through to solve
one equation in the simple example above, would be very tedious and
therefore rather error-prone. Fortunately there is an easier way.
We are saved because the differential equations we have to solve are

linear and hence the principle of superposition applies. This tells us
that if a forcing function v1(t) produces current i1(t) and a forcing function
v 2 t  provides current i 2 t  then v 1t  v 2 t  produces i1 t  i 2 t .
The
trick then is to choose a more general forcing function v t v 1 t v 2 t
 in
which, say, v 1 t  corresponds to Vm cos t and which made the
differential equation easy to solve. We achieve this with complex
algebra.
You should know that

exp j t cost  j sint .
where j = (-1), [electrical engineers like to use i for current]
so let’s solve the differential equation with the general forcing function
14
v t Vm exp j t Vm cost  j Vm

where
sint
v1 t Rev t ReVm exp j t Vm

cost .
The solution will be of the form

i (t ) I exp j t
where will, in general, be a complex number. Then in order to find that
part of the full solution corresponding to the real part of the forcing
function, Vm exp j t we merely need to find the real part of
i t . Thus
i1 t ReI exp jt ReI exp jt  
 I cost   
Let's illustrate this by returning to our previous example where we tried

to solve: di
L dt  Ri Vm cost
Now, instead, we solve the more general case:
di
L dt  Ri Vm exp j
t
15
and take the real part of the solution. As suggested above an
appropriate particular integral is i I exp j t which leads to
j LI exp j t  RI exp j t Vm exp j t
The factor exp j t is common and hence
R  j LI Vm
in which R  j L may be regarded as a complex impedance. The

complex current I is now given by
Vm Vm
I  exp  j
Rj R2  
L L 2
with  tan1 L R and hence
Vm
i t ReI exp j cost 
R2 
t 
L 2
which, thankfully, is the same solution as before but arrived at with
considerably greater ease.
Let us be clear about the approach. We have
16
(i)introduced a complex forcing function Vm exp j t knowing that in
reality the voltage source must be real i.e. ReVm exp j t .
(ii)We solved the equations working with complex voltages and

complex currents, V exp j t and I exp j t (or rather V and I since the
time dependence exp jt cancelled out).
(iii)Since the actual voltage is given by ReVm exp j t  the actual
current is given by ReI exp j tReI exp j exp j t I cost

  .
(iv) Since the differential of is simply
and since we always take out as a common

factor, you may see now that our differential equations turn into
polynomial equations in jω (and you knew how to solve these at GCSE!)
This is a very powerful approach that will permit us to solve AC circuit

problems very easily.
17
4. AC Circuit theory -- Example
Let's now do an example to show, formally, how we can solve AC

problems. Let's imagine we want to find the steady state current, i2,
flowing through the capacitor in the following example
The two KVL loop equations may be written
Ldi
Ri  dt 1 R imi E
1 1 2
and
cost 2 1 1
R i  i  C  i dt
2
0
18
Replacing Em cos(t + ) by Em exp j (t +)=E1 exp j t where
E1 =E m exp j and further introducing I1 and I2 via
i1 =I 1 exp j t and i 2 =I 2 exp j t we obtain
2R  j L I1  R I 2  E1
R  1 I 2 RI 1 0
 j C 
and, after a little algebra
E1 E m exp j 
I 2 L 2 

R  j  L   Mj
CR  N
C 
where M R  L CR and N L  2 C . We note that this may

be
written, introducing tan N M
Em
I 2 exp j 
M2N2

and hence the actual current i 2 ReI 2 exp j t  may be written

as
Em
i 2t cost 
M2N
 2 
19
5. Phasors
We have just introduced a very powerful method of circuit analysis. In

essence we have introduced the use of complex quantities to represent
sinusoidal functions of time. The complex number A exp j (often written
A) when used in this context to represent A cost   is called a

phasor. Since the phase angle  must be measured relative to some
reference we may call the phasor A0 the reference phasor.
Since the phasor, A exp j , is complex it may be represented in

Cartesian form x  jy just like any other complex quantity and thus
x = A cos 
y = A sin 
A=
x2+y2
Further since the phasor is a complex quantity it is very easy to display

it on an Argand diagram (also in this context called a phasor diagram).
Thus the phasor A exp j  is drawn as a line of length A at an angle  to
the real axis.
20
We emphasise that this is a graphical representation of an actual
sinusoid A cos(t +). The rules for addition, subtraction and

multiplication of phasors are identical to those for complex numbers.
Thus addition:
For multiplication it is easiest to multiply the magnitudes and add the

phases. Consider the effect of multiplying a phasor by j
j Aexp j =exp j  2 Aexp j = Aexp j (+  2)
which causes the phasor to be rotated by 90o.
21
Similarly, dividing by j leads to
A exp j   exp  j  2 A exp j   A exp j  

1
j
 2
i.e. a rotation of –90o.
We finally note that it is usual to use rms values for the magnitude of
phasors.
22
6. Phasor relations in passive elements
Consider now a voltage Vm cost applied to a capacitor. As we have
indicated we elect to use the complex form Vm exp j t and so omit the
"real part" as we calculate the current via
dv d
i =C dt =C dt (Vm exp j t )= j CVm exp j
t
If we now drop the exp j t notation and write the voltage phasor Vm as
V and the current phasor as I we have
1
I=j or V I
j C
CV =
In terms of a phasor diagram, taking the voltage V as the reference
from which we confirm two things we already knew
23
1
(i) the ratio of the voltage to the current - the reactance.
C
is
(ii) the current leads the voltage by 90. The pre-multiplying factor j
describes this.
For the inductance an analogous procedure leads to
V = j LI
Where the reactance is now jωL and, if we now take, say, the current as
the reference phasor we have
and here the current lags the voltage by 90.
24
[It is important to get these relationships the right way around and as a
check we may use the memory aid “CIVIL” – in a capacitor, the
current leads the voltage CIVIL and in an inductor, the current lags
the voltage CIVIL.]
Finally for a resistor we know that the current and voltage are in phase
and hence, in phasor terms
V =I R
25
7. Phasors in circuit analysis
We are now in a position to summarise the method of analysis of AC

circuits.
(i)We include all reactances as imaginary quantities j L (= j X L ) for
an inductor and 1 j C   j X c  for a capacitor.
(ii)All voltages and currents are represented by phasors, which usually

have rms magnitude, and one is chosen as a reference with zero phase
angle.
(iii) All calculations are carried out in complex notation.
(iv) The magnitude and phase of, say, the current is obtained as
I exp j  . This can, if necessary, be converted back into a time varying
expression 2 I cos(t +).
26
Suppose we wish to find the current flowing through the inductor in the
circuit below
The reactances have been calculated and marked on the diagram. The
left hand voltage source has been chosen as reference and provides
10V rms. The right hand source produces 5V rms but at a phase angle
of 37 with respect to the 10V source. If we introduce phasor loop
currents I1 and I2 as shown then we may write KVL loop equations as
10  5 I1  j 10 I1  I 2 
5 exp j 37o  4  j 3   I 2  I1  j 10   j 5I 2
where we have noted that 5 exp j 37o =5cos37 o + j 5 sin37o = 4 + j3 . It is
routine to solve these simultaneous equations to give
I 1
7  j  and I 2= 6.5+ j 8
12.5 12.5
and hence the current I  I1  I2 becomes
-
27
0.5 j9
I 12.5  0.72   86.8  0.72 exp  j
o
86.8o
Since rms values are involved, if we want to convert this into a function
of time we must multiply by 2 to obtain the peak value. Thus

i t  1.02cos t  86.8o 
In our example we do not know the value of  but it was accounted for in
the value of the reactances. Since everything is linear and the sources
are independent it would be a good exercise for you to check this result
by using the principle of superposition.
We have used mesh or loop analysis in our examples so far. It is, of

course equally appropriate to use node-voltage analysis if that looks like
an easier way to solve the problem.
As an example let's suppose we would like to find the voltage V in the

circuit below where the reactances have been calculated corresponding
to the frequency, , of the source
28
It's probably as easy as anything to introduce two phasor node voltages

V1 and V. The two node voltage relationships may be written as
V1 10 V1 V V1  0
  
10 j5  j5
0
and
V  V1 V  0 V  0
  0
j5
5  j10
5  j10
We note that in writing these equations no thought was given to whether

currents flowing into or out of the nodes were being considered. As in
the DC case it is merely necessary to be consistent. It is now
straightforward to solve these two simultaneous equations to yield
8. Combining impedances
As we have seen before the ratio of the voltage to the current phasors is
in general a complex quantity, Z, which generalises Ohms law, in terms
of phasors, to
V =ZI
where Z in general takes the form
Z =R e + j X e
where the overall effect is equivalent to a resistance, Re, in series with a

reactance Xe. If Xe is positive the effective reactance is inductive
whereas negative values suggest that the effective reactance is
capacitative.
Consider the circuit below
30
V  R  j L  1 
  I j
C
Thus the combined impedance Z is given by
Z  R  j L    R  j
1
C
X
This may be visualised on an Argand or phasor diagram
We note that the reactance may be positive or negative according to the
relative values of L and 1 C . Indeed at a frequency = LC we see
that X =0 and that the impedance is purely resistive. We will return to

this point later.
31
It is straightforward to show, and hopefully intuitive, that all the DC rules
for combining resistances in series and parallel carry over to
impedances. Thus if we have n elements in series, Z1, Z 2 , Z 3 Z n
Where
n
Zeff  Z1  Z2  Z3 Z n  ∑Zi
i 1
and similarly for parallel elements
1 1 1 1 N 1
Z eq = Z1+ Z + += ∑i
2 Z 3 Z i
=1
We note that the inverse of impedance, Z, is known as admittance, Y.

Thus, as in the DC case it is sometimes more convenient to write
n
Yeq = ∑Yi
i =1
32
and finally since Y is also a complex number it may be written
Y =G + j B
where G is a conductance and B is known as the susceptance.
33
Example
Find the equivalent impedance of the circuit below
Using the usual combination rules gives
1
Z 1Z (R + j j C Z
2
= Z1 + Z 2 = 1
R
L)+ j L
j C
+
Which we can simplify to
R  j L
Z
1 2LC  j C
R
34
9. Operations on phasors
We have just introduced a method of analysing AC circuits in terms of

complex currents and voltages. This method inevitably involves the
manipulation of complex phasor quantities and so we list below the
results for manipulating these quantities, which are, of course, simply the
standard rules for complex numbers. Sometimes it is easier to use the
a+jb notation and sometimes the r exp j r notation is easiest. We
summarise below the important relationships.
Addition and Subtraction
If
I1 =a+ jb and I 2 =c + jd
then
I1 ±I 2 = a ±c + j(b ±d)
where the real and imaginary parts add/subtract
35
Multiplication
Here it is easiest by far to use the r  rotation.
If I1  r1 1  r1 exp j1 and I 2  r2 exp j2
then I1 I 2  r1 r2 exp j 1   2  r1 r2 1 

2
when we see the amplitudes multiply and the arguments add
For division we have
2  1  1 
I1 r1 exp j 1 r1 r
1
r 2 j 
2
I2  r2 exp j 2 exp 2
r
the amplitudes divide and the arguments subtract.

36
Complex conjugates also appear.
If
I  a  j b  r exp j  r 
then the complex conjugate I*, is given by

I *  a  j b  r exp  j  r 

from which we see
I  I *  2ReI ; I  I *  2 j Im I
where Re{ } denotes the real part the Im { } denotes the imaginary part.
37
Rationalising
We are often confronted with expressions of the form
a+j
b c+
jd
And sometimes we wish to rationalise them. We do this in one of two

ways. The first is to multiply top and bottom by c  j d . This gives
ajb aj
 c  j d  ac2  bd   bc  ad
cb  j d c  j c  j   c  d 2   j  c 2  2 
d d d
Alternatively, we can write a + jb as r1 exp j 1 with r1 = a 2 +b 2 and
tan1 =b a . Similarly c + jd may be written as r2 exp j  2 and hence
a  jb r1 exp j1  r1 exp j r

 1  2  1 1  2
c  jd r2 exp j2  r2 r2
where the amplitudes divide and the arguments subtract.
There is no golden rule as to which approach to take – it is determined

by the problem at hand. However, if you do not have a pressing need to
rationalise the expression, we will see some quite good reasons why
we may often prefer to stick with the factorised form and not rationalise
at all.
38
10. Phasor diagrams
Although direct calculation is easily carried out using phasors it is

sometimes useful to use a phasor (Argand) diagram to show the
relationships between say a voltage and current phasor graphically. In
this way it is easy to see their relative amplitudes and phases and hence
gain quick insight into how the circuit operates. As a simple example
consider the circuit below
Since the current, I, flows through both elements it is sensible to choose

this as the reference phasor. Having made this choice the voltage drop
across the resistor, VR =IR , whereas that across the capacitor,
Vc  I jC   j CI . The sum of these voltages must equal V. The
phasor diagram is easily drawn as
39
from which it is clear that the voltage lags the current by an angle . The
angle  may be obtained from the diagram as =tan 1

(1 CR).
Let's consider another example
We could obtain the relationship between I and V by using the

equivalent impedance derived earlier. However we will use a phasor
diagram to show the various currents and voltages that appear across
the various components. Since the voltage, V, is the same across each
arm it is sensible to choose this as the reference phasor.
The relationships are
I 2  jCV V  I1 R  jLI1 and I  I1 

I2
The two phasor diagrams are
40
or, combining onto a single diagram
where  denotes the phase angle between V and I. In the diagram

above I leads V by .
41
11. Thévenin and Norton equivalent circuits
The Thévenin and Norton theorems apply equally well in the AC case.
Here we replace any arbitrarily complicated circuit containing resistors,
capacitors, and inductors by a circuit whose behaviour, as far as the
outside world is concerned, is entirely equivalent.
The two choices are the Thévenin equivalent
and the Norton equivalent
The methods for determining V, I and Z are identical to those used in the
DC case. In general
(i) Calculate the open circuit voltage, Voc
(ii) Calculate the short circuit current, Isc
42
From which
V =Voc , I =Isc Voc

Z=
and I sc
We note, of course that in the absence of dependent sources it is often

easier to "set the sources to zero" and simply calculate the terminal
impedance Zab. We emphasise again that when a voltage source is "set
to zero" it is replaced by a short circuit whereas when a current source
is "set to zero", no current flows, and hence the source is replaced by
an open circuit.
Finally we note that the equivalent impedance Z is also frequently

referred to as either
(i) internal
impedance or
(ii) output impedance.
43
Example
Find the Norton and Thévenin equivalents of
Noting that there are no dependent sources it is easiest to calculate Zab

directly with the voltage source replaced by a short circuit. This is easy
since the circuit then reduces to an inductor in parallel with a resistor.
Thus
j 40.20
Z =Zab =20+ j =16+ j 8
40
We now need to calculate the open circuit voltage between a and b.

This is easy since the circuit is essentially a voltage divider
20
Voc  5 exp j
50
20  j 10
40
50 exp j
  22.3 exp  j
10
5

exp j 63.4

53.4
 22.3   53.4
V
Thus the Thévenin equivalent circuit takes the form
44
In order to find the Norton equivalent we need to find the current flowing
between the terminals a and b when they are shorted together. In this
case the circuit becomes
and
50 exp j 10 50 exp j 10

Isc   1.25 exp  j 80 1.25  
80 j 40 40 exp j
90 
and hence the Norton equivalent becomes
45
We have elected to find the Norton equivalent directly. However it is
equally possible to transform between Thévenin and Norton equivalents
directly as we did in the DC case. It is left as an exercise to confirm
that
Hence we could have worked out the Norton current source in our
example directly from the Thévenin equivalent as
Voc 22.36 exp  j 53.04

I 
Z 16  j 8
22.36 exp  j 53.4

 17.89 exp j 1.25 exp  j 80 1.25  
8026.6
which, of course, is the value we previously calculated.
46
12. Power in AC circuits
Let us consider the power flowing into a network as shown
We are usually interested in the average power P rather than the

instantaneous power p, especially for constant and sinusoidal currents
and voltages.
In general we can write the current and voltage, for the sinusoidal case,
as
i  I m cost
v V m cost  
where voltage is out of phase with the current by an angle . The phase
angle  is, of course, determined by the actual network. The
instantaneous power p is given by
p =iv =Im Vm cost cos(t +)
47
which may be expanded noting that
cosA  Bcos A  B 2cos A
cosB
to give I m Vm
p= {cos+cos(2t
+)} 2
or, introducing the rms values of current I =I m 2 and voltage
V =Vm 2
p =IV {cos+cos(2t +)}
The first tem is constant whereas the second is periodic. Thus if we

average over a cycle the average power entering the network is given by
P =IV cos
The factor cos  - the cosine of the phase angle between the current and
the voltage - is known as the power factor. For the networks we are
interested in, the power factor is determined by the elements in the
network.
The figure below shows the instantaneous power flow into the network
for three different values of .
(a) Shows the case where the network is purely resistive and there is
no phase angle between the current and the voltage and hence the
power factor, cos=1 (“unity power factor”). The power flow
fluctuates sinusoidally but is always into the network and is
48
therefore all dissipated by the resistive element(s) with an average

value of P  IV .
(c) Shows the other extreme where the voltage and current are in
quadrature ( = /2 or 90). In this case the energy flows into the
network for half the time but flows out again for the other half. The
power factor cos is zero, P 0and there is no power dissipation.

The power flowing in must therefore be being stored in the reactive
component(s), e.g. an inductor, for half of the cycle before being
recovered again. As you found out at the end of P4 Electricity and
Magnetism we are sometimes interested in this reactive power or
reactive volt-amps (VAR), which is measured by its peak value
VAR  IV in this case.
(b) Shows the case for 0 <  < /2. There is an average power
dissipation of P =IV cos in the resistive element(s) but also
energy flowing back and forth to the reactive element(s)
and
VAR IV sin .
The combination of resistors, capacitors and inductors in the network

can be thought of as equivalent to impedance R + j X .
49
In phasor notation we now have
V =I R + j I X
=I (R + j X )=I Z exp j

and
See here that V cos  IR so that P  IV cos  I 2R confirming that the

power is only dissipated in the resistor.
Also observe that V sin  IX so that the reactive volt-amps is
VAR  IV sin  I 2 X .
50
We can illustrate this easily in a simple case: consider an inductor in

series with a resistor
We again take the actual current i(t )= 2 I cost where I represents

the rms value. The energy stored in the inductor is given by
1 2 2 2
W= Li =I L cos t
2
and hence the power flow due to the stored energy is given by
∂W
∂   I L 2 sint
2
cost
t   I 2 L sin 2 t
and we recognise, for this circuit, that X =L and hence
∂W
∂ I X
2
sint
t   IV
sinsint
51
where we have used the phasor diagram to observe that I X =V sin .
This quantity, which has a peak value of I 2 X or, equivalently, IV sin , is

evidently the reactive volt-amps, VAR. The expression is equivalent to
those developed previously.
An alternative way to write these expressions is in terms of the complex
phasors I and V (= V exp j ) themselves. It is straightforward to write

average power, P as
P = I V cos
=Re {V I * }
[If I is at an angle  and V is at an angle ( + ), say, then I* is at an

angle (– ) and VI* is at an angle {( + ) + (– )} =  and the projection
onto the real axis is VI cos .]
Similarly the reactive volt-amps, VAR, denoted by Q, may be written
Q = I V sin
=Im {V I * }
Evidently these two quantities, P and Q, may be regarded as the real

and imaginary parts respectively of a complex quantity V I* whose
magnitude is IV and phase is . Thus
52
V I * =S =Re {V I * }+Im {V I * }
= I V cos+ j I V sin
=P + j Q
The quantity S is simply called the volt-amps or VA and
S = VI =
P 2 +Q 2
53
13. Maximum Power Transfer
The internal or output impedance of a source restricts the amount of

power that we can extract from it. Let us imagine that the source is
represented by a Thévenin equivalent whose impedance may be written
Rs + j X s . We further assume that the source is connected to a load
ZL =RL + j X L .
Let us now assume that RL and XL may be chosen independently and

ask what their values should be in order to maximise the power
delivered to the load. We note that power can only be delivered to the
resistive part of RL since XL cannot consume power. Thus the average
power delivered to the load, PL, may be written
PL =I 2 RL
and the current I is given by
Vo Vo
I
)
= Rs + j X s +RL + j X L =( Rs +RL + j X s+ X L
)
(
54
hence
2 Vo 2 RL
PL = I RL =
{(R s + RL +) j X(s + X L
)}2
Since X L and RL are to be chosen we could simply differentiate with
respect to X L and RL to maximise P. However it is clear that P will be
maximised by setting XL = - Xs and hence
Vo 2 RL
PL =
(Rs +R 2
)L
dPL
If we now solve
dRL =0 to find the maximum we find R =R
L .s
Thus for maximum power transfer we require
*
ZL = RS - jXS = ZS
where the asterisk denotes complex conjugate.
Under these conditions the power delivered to the load PL is given by
V2
PL=I 2 R L = 4Ros
and the power dissipated in the source is given by
55
V 2o
Ps=I 2 Rs= 4 Rs
and hence the efficiency  is given by
Powerdelivered to theload PL =50%

= =
Totalpowerdissipated
PL +Ps
Thus half the power is dissipated in the source! This is clearly an
impractical proposition in the case of many generators and, indeed, is
undesirable in the case of the electricity supply companies!! Indeed it
would be spectacular to watch, albeit briefly, the effects of connecting a
matched load to a 500 MW power station. In this case, efficiency is
more important and hence Ps must be minimised in comparison to PL. In
the purely resistive case, for example, this requires that RL >>RS . On
the other hand in small signal electronic circuits handling small amounts
of power, matched operation is desirable because the maximum
available power in these applications is often rather small to begin with.
We emphasise that these results apply only if RL and XL can be chosen

separately. If, say, the load is restricted to being wholly resistive then it's
easy to show -- try it -- for maximum power transfer, we should choose
RL such that
RL  Z s
56
Equally if the load consists of a fixed reactive element, XL, but an

arbitrary resistive element, RL, then for maximum power transfer we
should choose
RL = Rs2 +(X s+ X L
2
57
14. Power Factor correction
We have seen that AC power can be expressed as P IV cos where
cos  denotes the power factor. The power factor has a significant
economic impact on the distribution and consumption of power since the

electricity utilities try to supply customers with a specified constant rms
voltage (240V in the UK). If the load has a small power factor then the
power company needs to provide a larger current to deliver the same
power to the load than would be the case if the power factor were high.
This is undesirable since the large current is supplied through
conductors of finite resistance, which leads to unnecessary losses. It is
particularly undesirable for the power companies since they have to
generate more power to supply the same average power to a customer
with low power factor equipment than would be required in the power
factor were high. The power companies have a very limited sense of
humour about this and tend to impose penalty charges to 'low power
factor' customers. So, what should we do if we assume that our load
has been designed to have as large a power factor as possible? For
example, most common AC motors have a lagging power factor of 0.9 at
best and often much worse.
Let us assume a load with a lagging power factor, cos  , P I L V cos
58
Since the voltage, V, is fixed by the supply company we must find a way
to reduce the current supplied to I such that the supply "sees" a load
with an overall power factor of unity, P IV , without compromising the
power
delivered to the load. One way to achieve this is to connect a reactive
(no power dissipation) element across the load such that the phasor sum
of the current passing through this element and IL add to I (which is in
phase with V). Since the load has a lagging power factor it is clear that a
capacitative element is a suitable candidate. Thus
and it is clear that
I c  I L sin 
from which a suitable value of capacitor may be found.
59
Example
An electric motor consumes 2kW at a lagging power factor of 0.8 when

supplied at 240V and 50 Hz. Find the value of the capacitor needed to
correct the power factor to unity.
Since P I L V cos it follows that I L P V cos and hence
I
I c P sin whence, since V  c we find
 V cos  c
Ic P
C  tan 
 2fV 2V 2
f 83F
60
15. Frequency Response
In our previous analysis we have considered networks of resistors,

capacitors and inductors and calculated voltages and currents in the
steady state due to sinusoidal forcing functions. In many practical cases
the frequency of this forcing or driving function may be variable or,
indeed, there may be several driving functions at different frequencies.
Since the reactance of an inductance, L , and a capacitor, 1 C , vary with
frequency as shown below then, unless the network is purely resistive,

its response will inevitably depend on frequency.
The sensitivity of a network to frequency is sometimes undesirable, as in

an amplifier intended to amplify equally over a range of frequencies. On
61
the other hand it is sometimes desirable, as in an electrical filter
designed to separate out certain frequencies in a communication
system. In either case it is important to examine the response of a
network as a function of frequency.
As a simple example to illustrate frequency dependence consider the

circuit below in which a constant amplitude voltage source, say 1V,
of variable frequency is connected to a resistor and inductor in
series.
Our previous analysis with e  2 cost gives
2
i t cost
 R 2  L 2 
with  tan1 L R
62
At frequencies low enough that R  L , we see that   0 and hence
i 2
R cost . Thus the current is of reasonable amplitude and in phase
with the voltage.
At high frequencies on the other hand, L  R and hence   90
2
giving i  cos 
L t 90 . In this case the current is relatively small,
due to the large impedance at high frequencies, and 90 out of phase
with the driving voltage. This behaviour, together with the phasor
diagrams is shown below
63
 2
 
1 2
We note that the magnitude R  L 2  and the phase
 tan1 L R

vary smoothly with frequency as shown below.

1/R
64
16. Frequency Response Function
In the last section we assumed a specific input voltage 10 . It's more
convenient, however, to deal with the ratio of the phasor output to the
phasor input.
Further, the input phasor may represent a current or voltage and the
output phasor may also represent a voltage or current at some other part
of the circuit. This leads to the idea of a frequency response function.
As a trivial example, consider the voltage divider circuit below
If we assume that the output is not connected to anything (open circuit)

then
R2
Vout
G  R R
1 2
Vin
65
In this simple case the frequency response function, G, is constant. In
general, however, the frequency response function, G, will be a function
of frequency and hence, for an arbitrary network we can write
Gj
Vout
Vin
where, since we use the complex " j" notation, we have described the
frequency response function as Gj.
There are two common ways of illustrating frequency response functions

graphically and these are described in the next two sections:
66
17. Polar Diagrams
For a given frequency response function the substitution of a particular
value of frequency, , will produce a particular complex number, Gj,
which can be represented as a point in the complex plane (Argand

diagram). A different value of  will lead to a different complex value for
Gj and is represented by a different point in the complex plane. Thus
we may represent the behaviour of the frequency response function
Gj as the frequency varies as the locus of Gj x j y with

as a parameter.
We'll now illustrate this by way of several examples. The most trivial is
the resistive potential divider for which we found
R2
Gj
 R1  R2
67
which is frequency independent and hence represented by a single point
Consider now
and suppose we now define the frequency response function of interest

to be
Gj  R  jL
V
I
68
i.e. the complex impedance. The polar diagram now takes the form
A further example might be
but now the frequency response function of interest is
1
Gj 
I
V R
jL
69
i.e. the complex admittance. It is not quite so straightforward to draw the
polar diagram in this case. It's probably easiest to re-write this as
1 1
Gj
1
 1
R 1 jL R  R  1
jT
Where we have introduced T L R which, since T must be dimensionless,
clearly T has the units of time. It is often called a time constant and
is something we will return to later. If we introduce the x
and y co-ordinates of Gj via
1 1 jT
Gj x  jy  
R 1 T
2
Hence
1  T
1 1
x  R  1 T ; y R  1 T
2 2
from which it is straightforward to eliminate T to give
2 2
 1   1 
 x  y 2 
2R   4R 

which is the equation of a circle. In order to find which part of the circle
is involved we note that ArgGj tan1 T and hence the polar
diagram takes the form shown below where we also show the locus of
the phasor diagram as  varies
70
As a final example we consider
 1 jC  R Vi
V0V V
1 2  C 
R 1 jC R 1
j
71
and hence the frequency response function Gj, defined as V0 V1 is
given by
Gj
1 jT
1
jT
where we have again introduced a time constant T  RC . In this case it

is clear that
G 1 ; ArgG  2 tan1 T
and hence the polar diagram is again semi-circular and takes the form
Since all frequencies are passed by this network with equal magnitude
this is an example of an "all pass" network.
There are several problems with this approach. This first is that these
diagrams are difficult to sketch without resort to computation for all but
the simplest frequency response function. The second is that the
72
frequency  appears as a parameter and is lost in the representation

unless it is specifically marked on the diagram. The representation is
compact but an alternative approach (below) in which G j and
ArgGj are drawn as a function of frequency, , on two
separate diagrams is more common.
73
18. Bode diagrams
An alternative to the polar (Argand) diagram method of representing the
frequency response of a function Gj is to plot its magnitude, G , and
phase, ArgG, as two separate functions of frequency, . As we will

see later, many frequency response functions can be factorised into
combinations of standard forms. Thus, in general, we might end up with
a frequency response function of the form
G1  jG2  j 
G  j
G1 G2   j 
 G3  jG4 ArgG1  G2 G3 ArgG4
G3 G 4 
Arg Arg 
 G j G j
where Gi j may be any of the forms we have come across before, i.e.
1 1
jT 1 jT
jT 1 jT
or, as we will see later, they may also be quadratic in form
1 2 jT  jT  .
2
It would, of course, be perfectly possible to plot

Gj  G1 G2  G3 G4  directly as a function of frequency but it
would almost certainly require resorting to a calculator/computer in all

but the simplest cases. Fortunately this difficulty can be removed by
74
electing to plot the logarithm of G j rather than G j itself. This
approach turns the multiplication and division of the individual
magnitudes G1 , G2 etc. into their addition and subtraction – a much
easier proposition. We also decide to use a logarithmic scale for

frequency so as to be able to represent a vastly wider range of
frequencies than would be possible with a linear scale. The ordinate of
the phase, ArgG, frequency response is drawn linearly but with the
same logarithmic frequency scale. We elect to use logarithms to base
10 throughout and to measure the magnitude or amplitude in (somewhat
eccentric) logarithmic units called decibels (dB).
As an historical aside, this unit was first used to indicate the loudness of
sound. Thus a sound of intensity W2 was said to be louder than a
sound of intensity W1 by log10 W 2 W1  bels or 10log10 W 2 W1
decibels (dB).
This was subsequently adopted by electrical engineers for the ratio of

electrical powers, W2 and W1. When the powers happen to be
developed in the same or equivalent resistance then
W2 V 2
 V 
10log10
10log 10 V 20log10 

   
2 2
W1 1
2
 V1 
10log10  I22    I 2 
2
10 
 I1  20log  I1 
Nowadays electrical engineers routinely use this interpretation of

decibels, dB, for voltage or current ratios without reference to the power.
i.e.
75
GdB  20log10 Gj
This approach to representing the frequency response of Gj is due to

Hendrik Bode and yields two graphs
(i)The magnitude is plotted as 20log10 G dB against frequency on a
logarithmic scale
(ii)The phase or argument, ArgG is plotted against the same

logarithmic frequency scale.
We note that the usefulness of this approach is that if

G  G1 G 2 G3 G4 then on a Bode plot we merely plot 20log10
G as,
GdB, where
GdB  20log10 G1  20lg10 G2  20log10 G3  20log10 G4
G1dB G 2dB G3dB G 4dB
Therefore we can build up the frequency response of a complicated

function merely by adding and subtracting the frequency responses of its
component parts.
The most complicated Gi j that we will meet in this course will be the
ratio of two polynomials and since polynomials can always be factorized
there are very few basic elements that we need to look at. We will
consider these in detail before sketching more complicated looking
function.
Bode diagrams for standard forms
(i) GjK
In this case Gj K and ArgGj0 . If K=18, say,
GdB  20log10 K  20log10 18  25.1dB
We note that K>1 corresponds to “gain” whereas K<1 describes

“attenuation”. If K 0.25 , say, then GdB  20log10 0.25 12dB . The
negative sign arises because K<1.
(ii) Gj jT
We see that
GdB  20log10 T  ArgG

and 90
77
and note that the magnitude Bode plot is a straight line with a slope of
20 dB/decade (i.e. a change of 20 dB for each ten-fold change in
frequency – note log10 10  1, log10 100  2 etc.).
It is easy to generalise these results to Gj jt n to give
GdB = 20.n.log10(ωT) and Arg{G} = n.90°
(iii) Gj1 jT
We start by considering the magnitude plot
G
dB
 20log 1 T 2 10log
10
1 T  
2
10
and look at the value of this expression for low frequencies T 2 
1
and high frequencies T 2  1. For low frequencies
GdB 10log10 10dB

i.e. a low frequency asymptote which lies on the 0 dB axis.
At high frequencies, on the other hand,
GdB 10log10 T   20log

10 T
2

which gives a high frequency asymptote, of slope 20 dB/decade.
We note that these two asymptotes meet when T 1 or 1 T . This

frequency is sometimes known as a “corner frequency” and the point
at which the asymptotes meet is knows as a “break point”.
The asymptotes give a general idea as to what the frequency response

looks like but they are only asymptotes and so to find the actual value of
the function we have no alternative but to work out a few values. The
maximum difference between the asymptotic value and the actual values
occurs at the break point T 1. Here the actual value is

GdB 10log10 1T 2 10log10
 23dB
or, strictly, 3.01 dB. Thus the frequency response takes the form below
where the actual response has been sketched alongside the asymptotes
79
The phase-frequency response is given by
Arg G Arg 1 jT  tan1 T 
which is a smooth curve varying between 0 and when T is very small and 90
when T  1.
It is often convenient to use the straight-line approximation shown below
in which tan1 T  is approximated by a slope of 45/decade in the

region 0.1T 10. (See HLT p167.)
We note that the actual curve and the approximation cross when T 1
and ArgG 45 in both cases and that the worst error is less than 6°.
80
Straight line approximation
Example
Gj 251 j0.01
The general approach is to regard this equation as containing two terms

25 and 1 j0.01. We plot the Bode magnitude and phase frequency
responses for each constituent part separately and then merely
add them up to obtain the overall frequency response.
The magnitudes of each term in dBs are drawn initially in the left hand
diagram below. The K  25 factor gives a constant gain of
20log10 25 28dB . The 1 j0.01 term exhibits a break point at T
=
0.01 sec or a frequency of 100 rad/sec. The diagram on the left below
shows the actual response where the 3 dB correction has been made at
the breakpoint.
81
We now turn to the phase plot and note that the first factor (25)
imparts zero phase shift and hence the phase response is due entirely
to
1 j0.01 and is shown below
We summarise below the asymptotes for magnitude and phase for the
four (two really) common factors.
82
Examples
We begin by considering the circuit below
Suppose we are interested in G jV0 Vi we have
83
jT
V0
 
1 1 jT
R R
 jC
Vi
where T  RC . We now look at the solution in terms of the two factors
(i) The jT term results in a linear gain of 20 dB/decade which passes
through 0dB when 1 T .
(ii) The factor 1 jT 1 has a break point at  1 T .
The overall response is shown below as the sum of the two components.
The phase response consists of two factors
(i) jT leads to a phase (Argument) of + 90.
84
(ii) The term 1 jT 1 gives a phase lag increasing from zero at
low frequencies to 90 at high frequencies over the range 1 10T to
10T .
The asymptotic and total phase shifts are shown
We note that this is an elementary example of a high pass filter. We

see from the Bode diagrams that frequencies lower than 1 T are
attenuated whereas those higher are passed essentially unaltered –
note for these frequencies that GdB 0dB i.e. G 1 and Arg
G0 .
We now move on to consider the voltage gain of an amplifier that, in a

particular case, might be given as
 0.5
Gj

1 0.01jj 110 5
j
We look again at the solution in terms of the four individual factors
85
(1) The gain associated with the constant factor,  0.5, is given by
20log10  0.5  20log10 0.5  6dB .
(2) The j term results in a linear gain of 20 dB/decade which passes

through 0 dB at =1 rad/sec.
(3) The factor 1 0.1 j1 has a breakpoint at  10 rad/sec.

(4) The factor 110 5 j  1
has a breakpoint at 105 rad/sec.
The overall magnitude response is shown below as the sum of the

individual responses. The individual asymptotes are shown in the top
diagram together with the full response below which shows a 'mid-band'
 34 
gain of 34dB, corresponding to a numerical gain of 50 10 20 
 
86
When dealing with the phase plot we have three basic factors:
(1) - 0.5 j leads to a phase (Argument) of -90.
(2) The term 1 0.01j1 gives a phase lag increasing from zero
(or thereabouts) to -90 (or thereabouts) over the range 10
rad/sec to 1000 rad/sec as shown by the dotted line below.
87
(3) The term 1105 j behaves in an analogous fashion over
the frequency range 104 to 106 rad/sec.
The asymptotic and total phase shifts are shown below.
Finally, for completeness, we include the quadratic factor that we will

meet later when we discuss resonance.
G  j 12 j T  j T 
2
This is an extremely common factor that occurs in many fields of applied

science. The parameter  is called the damping factor, 0<<1, and 1 T
corresponds, as we will see later, to the resonant frequency. However,
for the time being, we can just regard it as a function whose frequency
response is to be sketched.
88
We begin with the magnitude response
G  20log 1 T   2 2

 2
T 2
and, as before, look

dB for10the high and low frequency asymptotes.
At low frequencies, T  1, and
GdB  20log10 10dB
which gives the low frequency asymptote as a straight line on the 0dB
axis.
At high frequencies, T  1, and only the highest powers of T are
retained. This gives
GdB  40log10 T 
The resulting high frequency asymptote is therefore a straight line with a

gradient of 40 dB/decade.
These two asymptotes meet in a breakpoint at 1 T and, at this point

the actual value of the function is given by
GdB  20log10 2
89
Since 0<<1 this value is negative for 0<<0.5 and positive 0.5<<1.
The asymptotic and actual magnitude frequency responses are plotted
below.
We finally note that when there is a negative peak, 0<<0.5, that it

occurs at a frequency a little less than 1 T .
The phase response of this function is given by
 
ArgG  tan 1  2 T 2 
1 T 


At low frequencies Arg G~0 and at high frequencies
Arg G~ tan 1  2 T 180 whereas when T 1; Arg  90 . This
suggests a straight-line asymptote as shown below. The actual form of

the response depends, of course, on the particular value of  as
indicated on the right-hand side diagram.
90
Check out HLT page 168 to see accurate plots of the inverted form of
this second order function.
91
19. Resonance
As we have seen previously, the reactance of an inductance increases

with frequency whereas the reactance of a capacitor decreases with
frequency. This suggests that in certain circuits it may be possible to
find a frequency at which the two reactances are equal and opposite.
When this occurs the impedance is purely resistive and a condition of
resonance occurs. Under these circumstances, depending on the
circuit, a large current may flow or a large voltage may develop across
part of the circuit.
Let’s illustrate this by considering the series RLC circuit below.
The impedance is given by
1 
Z  R  j L  
C 
92
Clearly Z is a minimum when L = 1/C when it has the real value Z =

R. The frequency at which this occurs is called the “resonant”
frequency, , which is given by
1
0 
LC
The voltages and current are related by
VS VR VL VC

j
 I
 R  jL  
C 
At resonance, L 1/ C , and the phasor diagram takes the

form
And we see that, at resonance,
VC VL
93
and that the current, I¸ and the voltage, VS, are in phase and are related
by
VS IR
As the frequency changes from its values at resonance the current, I,

falls from the maximum value of VS/R. At resonance the impedance
(Z=R) is a minimum and the admittance (Y=1/Z) is a maximum. We plot
below the variation of current (and Y) with frequency. The graph is an
example of a resonance curve.
If R is small the current at resonance, VS/R, is very high and the

resonance curve is sharp. On the other hand if R is large then the
current is small and the resonance curve is much flatter.
At resonance, as we have seen, the magnitude of the voltages across

the capacitor and the inductor are equal, VL  VC , and may be larger
than the source voltage itself! In order to see this we note that
94
VL  0 LI , VC I / 0C and VS IR . Thus
VL VC 0L
VS  VS  R  0CR
1
VL VC
which can be made very high by making R small. The ratio, or
VS VS
represents the ‘voltage magnification’ and is known as the “quality

factor”, Q, of the circuit. Thus
magnitude of voltage across L or C at resonance

Q
magnitude of voltage across whole circuit
0L 1
 
R 0CR
A high Q also results in a sharp response curve. This suggests that the
circuit acts as a narrow band filter since it only passes significant current
near the resonant frequency,  0 .
[In the lab, when you build a tuned circuit for a radio receiver, the
inductor will always come with some resistance and this will limit the Q
L
that is achievable. We sometimes therefore refer to Q  R0 as the “Q
of the coil”.]
95
In order to have a quantitative measure of the sharpness of this peak we
could consider the frequency range where the magnitude has fallen to
some fraction of the maximum value. It is conventional to choose points
at which the magnitude has fallen by 3dB to 1/ 2 of the maximum.
These points are called ‘3 dB or half-power’ points. The later term arises
from the fact that -3 dB corresponds to a factor 1 2 in current
or voltage and hence to a factor of 1/2 in power.
If we write the maximum current at resonance as I 0  VS / R then the
current at any frequency is given by
I R

 1 
2
I0 R 2   L 
 C 
We now wish to find the spread of frequencies between the 3dB points
where the magnitude has fallen by 1/ 2 , i.e. 2  1  ,
We formally need to solve
96
R 1

 2
2
1 
R2  L 
C 
or, equivalently,
  1
L  C  R
We take the negative sign to correspond to the lower frequency, 1.

Solving the resulting quadratic gives
R 2
R 1
1      
2L 2L LC
where the positive square root has been taken in order to give a positive
1.
Similarly the higher frequency, 2, is obtained by taking the +R sign. In

this case we obtain
R 2
R 1
2     
2L 2L LC
The frequency difference between the 3dB points is therefore
97
R
  2  1  L
0L
which together with our previous definition of Q  R permits us to
write
0 L 0
Q  
R

and so the higher the Q the narrower the bandwidth, , and the
sharper the resonance peak.
0
magnification’ we could just as easily used Q 
as an alternative
We note that although we first defined Q in termsof the ‘voltage
definition.
We also note that the resonant frequency  0 is not positioned midway
between the two 3dB frequencies  2 and  1 . It is, in fact, the
geometrical mean of the two as can be seen by multiplying the

equations for  2 and  1 together
0  12
and hence, on a Bode plot where a logarithmic frequency scale is used,

 2 and  1 will be symmetrical about 0 .
98
Let us finally return to the circuit and consider the voltage across the
capacitor, VC. It is easy to write
I VS
VC  
jC 
 1 
jCR  j L  
C  
and hence
VC 1

1 jCR  j2
VS LC
Now, if we recall that 0 = 1/(LC) and Q   0 L / R  1/ 0CR then we
may write
VC 1

  2
V
S
1   
1 j   j 
Q 0   0 
We see that the denominator here is of the standard form we discussed

earlier where the Q factor is evidently related to the damping factor, ,
via
1
Q  2
The Bode diagram corresponding to Gj  VC /VS are shown below

where it is again clear that a high Q – low damping -- results in a
sharp resonance peak.
99
[i.e. Exactly this frequency response function is plotted in HLT page
168.]
100
Example
In a particular series RLC circuit, R = 10 , L = 1 mH and C = 0.1 F

and the source VS = 2.0 V rms.
The resonant frequency, 0, is given by
1
0   105 rad/sec
LC
which may be expressed as f0  0 / 2  15.9 kHz. The Q of the
circuit is
 L
Q  0R  10
and hence the bandwidth,   0 / Q , or f0  f0 / Q  1.59 kHz.
The current at resonance is
I 0  VS  0.2 A
R
and the magnitude of the voltage across either the capacitor or the
inductor at resonance is
V  QVS  20V
101
General remark on Q
We have defined Q in a way that is convenient for our ‘electrical’

purposes. However, since all resonant systems – mechanical and
electrical – have a common basis in energy, it is also possible to show
that
energy stored
Q  2
energy dissipated per cycle
Parallel resonance
Another common circuit displaying resonance is the RLC parallel circuit

shown below.
The admittance of the circuit (Y = 1/Z) is given by
102
1 
Y  R j  C
1
  L 
This is analogous to the equation for Z of the series resonant circuit but
with impedance/resistance/reactance replaced by
admittance/conductance/susceptance. The current will be a maximum at
the resonant frequency
1
0
 LC
and will have the same properties as the voltage has in a series
resonant circuit.
By analogy, Q for the parallel circuit will become
R
Q   0CR
0L
but will still equate to


Q 0 .
103
A practical resonant circuit
Because inductors always have some resistance, a more practical

parallel resonant circuit is given below.
Unfortunately the analysis is more complicated but with a little maths we

can show that this practical circuit is equivalent to
with
2  L  2  L 
2 2
R* R and L*  R
R 2 L
104
For high enough Q, we can use the simple approximations in HLT (page
167).
105
20. Mutual Inductance
The action of the inductor relies on the presence of a varying current to

give rise to a varying magnetic field, which then induces a voltage in the
coil that produced it. Let’s wind a conductor around a high permeability
core and pass a current i through it:
You learnt in P4 that when there is a closed magnetic path (i.e. the
complete path is not shown in this figure) and no flux leakage,
 = (A/l) Ni and v = N d/dt
so that
v = L di/dt , where L = N2 (A/l) = N/i
We call this self inductance
106
Now let’s take two windings following around core and once more
assume no flux leakage. For good measure, suppose the first wire has
N1 turns and the second wire has N2 turns.
A flux is produced in the core from both currents and totals

 = (A/l) N1i1 + (A/l) N2i2
Equally, if this flux changes, it induces a voltage in both windings
V1 = N1 d/dt and V2 = N2 d/dt
And this is what we call mutual inductance.
Combining these equations we can write
di di2
v1L 1 dt
1
M dt
and
di di2
v 2 M dt1 L 2
dt
where
2 2
L1 = N1 (A/l) and L2 = N2 (A/l)
are the self inductances of the two coils and
M = N1N2 (A/l)
107
is the mutual inductance between the two coils.
Perfect Coupling
Note that in this ideal world of perfect flux coupling, M = (L1L2).
Magnetic Coupling Coefficient
In practice we always get some flux leakage, i.e. not all the flux stays in
the core and therefore not all the flux links all the coils. We should
therefore expect that
L1 < N 2 (A/l) and L < N 2 (A/l)

1 2 2
and
M < (L1L2)
We can then write
M = k . (L1L2)
where k (0 < k < 1)

is known as the
“coupling
coefficient”.
The dot notation
We assumed above that both coils were wound in the same direction
around the core. Of course if we reverse the direction of one winding, its
flux and its induced voltage will change sign so it is important that we
know the relative directions of our two windings. The dot notation is
used to specify the ends of the coils that have the same polarity.
In order to see how to use the dot notation, let's consider the coupled
circuit below.
109
When, as here, we define i1 and i2 as both “into” the dotted end and also
v1 and v2 as both “to” the dotted ends, the terms in our mutual coupling
equations will all have the positive signs that we saw above, i.e.
di di2
v1L 1 dt
1
M
and dt
di1 di2
v 2 M L
dt 2
dt .
All your analysis can be performed using these standard equations to

represent mutual coupling, but some people like to redraw the circuit
with the mutual coupling replaced by dependent sources, thus:
v1 v2
Strictly speaking the dots are now superfluous but have been retained to
help avoid errors when marking the polarity of the induced e.m.f.. We
emphasise that we have taken account of the coupling by introducing
110
the voltage sources. We may now analyse each circuit independently,

giving of course:
di di2
v1L 1 dt
1
M
and dt
di di2
v 2 M dt1 L 2
dt
Should you choose to define a current or a voltage in the opposite

direction, you will have to change the sign of that variable in these
equations. Therefore I strongly recommend committing these two
equations to memory along with notation that the voltages are measured
to the dotted ends and the currents go into the dotted ends so that all
the terms appear positive. Otherwise it is very easy to finish up with the
wrong sign. If you decide that you actually want to work with the current
coming “out of” the right hand side say, it is a simple matter to define it
as a new variable, e.g. i3 = –i2.
Finally we note that if you are working in the frequency domain

with phasors, the equations become
V 1  j  L1I 1  j MI 2
V 2  j MI 1  j L2I 2
111
21. Mutually coupled circuits in series
Using the definitions of v1, v2, i1 & i2 above and our two mutual coupling
equations, the series connections here can be expressed by the
following three further equations:
v  v 1 v 2
i  i1  i2 .
From these five equations, we can eliminate v1, v2, i1 & i2 to obtain:
v  L di  M di M di
1
dt L
di
v  Ldt
1
 L 2 2M  dt dt
di
2
dt
112
and so we see that effective inductance of the circuit is L1  L2  2M . In
this case the coils have been connected so that the mutual effect is to
increase the effective inductance. If the coils were re-arranged and
connected so that the dots appear as
the connections can now be expressed by

v  v 1 v 2
i  i 1  i 2
and hence
di di di di
vL M M L
1
dt dt
2
dt dt
and in this case the effective inductance is L1  L2  2M .
We note that when the coupling is perfect, M  L1L2 , the effective
inductance in the first configuration (series aiding) may be written as


Leff  1L  L2
2

whereas in the second configuration (series opposing),
113
Leff   L  L2
1
2

For the special case of L1  L2  L gives Leff  4L in the first case and
zero in the second case! (Both answers have simple practical

consequences – can you think what these might be?)
It is left as an exercise to confirm that you can use the dot notation
confidently to show that the circuit below
has an effective inductance of 12mH. It's easy but take care!!
114
22. The transformer
In essence a transformer is nothing more than a pair of coupled coils

with the expectation that we will supply power to one side (which we call
the “primary” winding) with N1 turns and take power out of the other side
(the “secondary” winding) with N2 turns. We call N1 / N2 the “turns ratio”
N.
I2 ’
Our standard equations are
V 1  j  L1I 1  j MI 2
V 2  j MI 1  j L2I 2
and, because we like to se a current coming out of the secondary, we

need the supplementary equation
I2’ = – I2
115
In an “ideal transformer” with no flux leakage, L2 L1  N 22N 2 
1 N and
2
M L1L2 so we can write the equations for V1 and V2 in terms of L1
alone as:
V1 = j L1 I1 + j NL1 I2
V2 = j NL1 I1 + j N2L1 I2
From which we deduce that

N
V 2  2 V1  1
NV N 1
Thus we can use a transformer to “transform” power from one voltage to

another by selecting an appropriate turns ratio. For example, to "step
down" the mains 230V & 50 Hz to something more useful for electronic
circuits, say 12V. Alternatively a transformer can be used to "step up"
voltages from a generator for transmission over large distances.
Since our transformer does not dissipate any power, it is tempting to

think that V.I is the same at the primary and secondary and that when
the voltage is 'transformed' by a factor N the current will be 'transformed'
by the inverse factor 1/N. However, this would to be to ignore the phase
shift in the output current. By eliminating I2 we obtain
V1 V1 '

I1 
j  L1 
M
1
I 2' 
j  L1 NI
2
116
from which we see that the ideal current relationship will be approached
by making L1 large. That’s why we like lots of windings!
We conclude this section by working out the input impedance of an

ideal transformer when it is connected to a load ZL. When the load is in
'
place we have V2  I 2 Z L and the equations become
V1  jL1 I 1  jM I 2'
'
I 2 'Z  1  jL2I2
L jMI
The input impedance is defined as Zin V1 I1 which, after a little algebra
gives
V 
j  L Z  2 M 2  L L 
Zin  I1 1 L 1 2
j  L2  Z L
1
For our ideal ideal transformer with M 2  L1L2 when the inductances are
large enough we approximate
L
Z in L 1 ZL
2
or
117
Z in  ZL
2
N
Therefore, as well as being useful for transforming currents and voltages

transformers are also used to transform impedances. In this case the
source "sees" an effective load of ZL N 2 . This can be useful when

coupling signals to very low impedance loads. Indeed if you elect to use
earphones in the radio you will build in the DBT exercise you may have
occasion to use a transformer in this way.
118
23. Transient response
In our previous analysis of circuits we were concerned with the 'steady

state' solutions for current and voltage, which occur long after the source
or sources have been connected to the circuit. Since the current/voltage
relationships for capacitors and inductors involve time derivatives they
clearly do not respond instantly to abrupt changes in current and
voltage. It takes time for the effects of an abrupt change to die away
and for the final steady state conditions to be reached.
When we began discussing AC circuit theory we set up the governing

differential equation and proceeded to solve it using standard
mathematical techniques. The full solution consisted of two terms. The
first, the complementary function, corresponded to the transient
response which eventually decayed away whereas the second, the
particular integral, led to the final steady state response. In the RL
circuit you built in the laboratory the steady state was achieved in a
matter of milliseconds and so the transient component was not evident
in the measurements you took. However, a millisecond, or even a
microsecond, can be a long time in electrical engineering and hence
transients deserve our attention.
We will now discuss in more detail the effects of abruptly connecting or

disconnecting sources to electrical circuits. Previously we have only
dealt with sinusoidally varying sources but we will now remove this
restriction. In order to introduce the topic logically we will begin by
considering first order systems; so called because they contain only
resistive and one reactive element and lead to first order differential
equations before moving on to second order systems containing
119
resistors and two reactive elements. At first we will solve the differential
equations using standard 'classical' mathematics before introducing a
powerful (and simple) technique based on the Laplace transform.
First order transients
We begin by asking how the voltage across the capacitor increases with
time as it is charged by the battery after the switch is closed. We
assume the capacitor is initially uncharged.
When the switch is closed
E  i R v
The current/voltage relationship for the capacitor gives
dv
i C dt
and so
120
dv
E  RC dt 
v
or
dv v
 
E dt RC
RC
where, evidently, as we have seen before, the product RC must have the
dimensions of time. Thus we set RC=T and we formally now need to
solve
dv v E
 
dt
  0 . A suitable candidate is A expmt  which leads to an auxiliary

dv v
dt T T
equation
T
1
m 0 ; m
We begin by considering the1
complementary function as theTsolution
of
T
t
and hence the complementary function is A exp  T . The particular
integral is determined by the specific forcing function applied. In this

case it is constant (a battery) and hence a suitable particular integral
is
v  E . The full solution is provided by
121
v = Complementary function + Particular integral
t
v = A exp T  E
The unknown constant, A, is determined by the initial conditions, i.e. the

value of v just after the switch is closed. Since the capacitor was initially
uncharged
vt 0  0
where the notation, 0+, is used to indicate time just after closure of the
switch. This condition leads to A  E and hence the full solution is
given by
v  E1exp  
t
T

.
122
The factor T  RC is known as the time constant and indicates how

quickly the capacitor charges.
We can now make some general remarks about the form of this solution
(i) The complementary function, Ae  R T , determines the transient

response which decays away at a rate determined by the time
constant, T. The unknown constant, A is determined by the initial
conditions.
(ii) The particular integral leads to the final steady state solution. As
such it is determined by the specific forcing function – constant in
this case, (sinusoidal in our previous discussion) – and not by
the initial conditions. It is, of course, independent of the initial
conditions.
Let us now look at what happens after the capacitor has charged up to
a voltage E and is discharged through a resistor in the manner shown
below.
123
Again we want to determine the variation of v with time. The governing
differential equation is as before but with zero forcing function
dv v
 0
dt T
which gives a complementary function
t t
v  B exp   E exp 
T
T
where the unknown constant, B has been determined from the initial
condition that the capacitor was charged to in voltage E
v 0   E
Formally we note that the particular integral is zero indicating, as it must,

that no voltage appears across a discharged capacitor.
124
Let us now consider the following RL series circuit where we assume no

current to be flowing in the inductor before the switch is closed.
Kirchhoff leads to
di
L dt  i R 
E
or
di R E
 i
dt L
L
As before we note that L R T must have

the units of time and hence
di i E
 
dt125 T
L
which is mathematically, formally, equivalent to the differential equation
we previously encountered. The complementary function is again given
by
A exp  t T
with time constant T L R and unknown constant A. The particular

integral is
ET E

L
R
and hence the full solution

E
i  A exp 
t T R
The initial condition, i 00 leads to A  

E
and hence
R
E t 
i 1 exp  
R T
which again, has the standard form with a time constant, T L R .
It is left as an exercise to show that if we discharge the inductor through

E t
the resistor that the current decays as i  exp  .
R
T
126
General remark
In the case of both the RC and RL circuits we have essentially solved a

first order differential equation with a time constant T. This led to a
complementary function for, say voltage, v, but it could equally be
current – it depends on the specific case – of the form
v  A exp 
t T
together with a particular integral giving the final steady state value, VSS
v VSS
The unknown constant, A, is determined by the initial condition v v 0  .

Thus
v 0   A VSS
or
v = [v(0+) – vSS].exp(-t/T) + VSS
or
v = [initial value – steady state value].exp(-t/T) + steady state value
or, in words,
127
the variable rises (falls) from its initial value to its final steady
state value exponentially with a time constant T
128
This observation permits us to sketch the transient response of many

simple circuits by inspection.
Initial Conditions
In order to sketch the transient response directly we need to be able to

find the voltage and current in a resistor, inductor and capacitor
immediately after a switch has been open or closed.
Resistor
Since the current and voltage are related by V IR it is clear that any
instantaneous change in, say, voltage will be accompanied by an equally
instantaneous change in current.
Inductor
Since the current and voltage are related by
di
v  Ldt
a sudden change in current would result in an infinite voltage. Since this

is implausible we can conclude that closing a switch to connect an
inductor to a source will not cause current to flow at the initial instant,
t  0  , i.e.
i 0  0
i.e. it will act as an open circuit.
129
If, of course, a current I0 was flowing before the switch was closed it will
continue to flow and
i 0   I 0
These observations permit us to introduce the idea of an equivalent

circuit at t  0 
Capacitor
In this case
dv
i C dt
and arguments analogous to those above require that voltage does not
change instantly since the current cannot be infinite. If the capacitor is
initially uncharged then
v 0   0
130
i.e. the capacitor acts as a short circuit. If the capacitor is initially

changed to V0
v 0  V0
24. Example
If the capacitor is initially uncharged it acts like a short circuit at t  0 
and hence
131
i 0   E R
The final steady state current is zero and hence the transient response
may be written as
i t  exp  t
E
R
RC
We note that if the resistance were small (zero) that i t  would be very
large for a very short time. Although the R  0 case could not happen in
practice it is often a useful "special case" and the discontinuity in current
is described by an impulse function, which has infinite magnitude for a
vanishingly short time.
As a further example we investigate the current flowing in the circuit

below after the switch is closed.
132
When the switch has been open long enough for the inductor to act as a
short circuit to DC a current i  E R1  R 2  and a time constant T L
R1
permits us to sketch
133
Second order transient
Consider
Kirchhoff gives
di
E  i R  Ldt c
v
dv c
which together with i C gives
dt a second order differential equation
for the voltage across the capacitor as
d 2v
LC dv c RC c vc E
dt 2 dt
We now solve the equation by our standard method of first obtaining the
complementary function by setting E  0 on the right hand side and
looking for solutions of the form A exp mt . The auxiliary equation takes
the form
134
LC m 2  RC m 1
0
which has solutions
R 2 1
m  R 2
2L 4L LC
In order to make some general comments we now recast this into non
circuit-specific notation following our discussion of resonance. In that
context we introduced
1 L
0 an Q R
0
LC d
We also noted that Q was related to the damping factor

1 R C
 via    . This permits us to write
2Q 2 L
m  0      2 

1

or
m  
with   0  and   0 2 1 .
135
We note that the particular integral, - corresponding to the steady value
of vc – is simply v c  E . The full solution is therefore
v c  E  A exp  t  B exp   texp t
where the unknown constants are determined by the initial conditions,

v c  0 and i  0 . We note that the latter condition is formally equivalent
dvc
to requiring  0 . The full solution now takes the form
dt
vc  E 
2
E
  exp t    exp 
texp  t
It is clear that according to the value of  the factor  may be real or

imaginary. We consider these cases in turn
Case 1:  R 2 L C
1;
In this case  is real and >. It is convenient to write the solution

as
vc  E 
2
E
 exp  t  exp 
t
where we see that the transient solution of the response is the difference
between two exponentials and is negative. The voltage v c follows the
136
usual gradual rise to its steady state value typical over overdamped
systems.
We note that the current is given by
dv
i C dtc
or
2EC
i  02 
exp  t  exp 
t
Thus the current is proportional to the difference in two exponentials of

equal amplitude but different decay rates and hence rises from zero to a
maximum before finally falling to its steady state value of zero
137
Case 2:  1 ; R 2 L C
For this special case of critical damping the general solution is no longer
valid since the auxiliary equation has two equal roots. The
complementary function in this case must take the form
C  Dt exp  0 t
and hence the full solution becomes
v c E  C  Dt exp  0 t
which with initial conditions v c  i  0 gives
v c  E 1 1 0 t exp  0 t 
which gives a response graphically similar to the previous case.
138
Case 3:   1 ; R 2 L C
In this case  becomes imaginary and so we write
 0 2
1 j 0 1 2
 j n
with
n  0 1  2
It is now straightforward to recast the general solution of page 136 by

replacing  with j n as
    
v c  E1 
 cos n t   sin n t exp t
  n  

In this case the solution is very different from before. The transient
solution is oscillatory, at a frequency n , but its amplitude decays at a
rate determined by    0  and we speak of underdamped

oscillation.
The frequency n  0 1  2 is the natural frequency. The same term is
sometimes applied to 0 but the two are strictly only equal when  1
(critical damping).
139
(See HLT p169)
Case 4:  0; R
0
In this case there is no damping since R  0 means there is no element

in which energy can be dissipated. The voltage across the capacitor in
this undamped case is given by
v c  E 1 cos 0 t 
140
We observe that for small damping, representing minimal loss, that the
frequency of the oscillatory transient n  0 1 2  0 , the resonance
frequency in the steady state. This is not surprising since this is the
frequency at which energy naturally oscillates between the
reactive elements and this is actually what is happening during
transient oscillation. Naturally with increased damping the effect is
reduced.
We therefore see that the damping factor  (or Q) has an important role
in describing the transient behaviour:
If the damping is less than critical (<1) the transient

behaviour is oscillatory otherwise it is not.
141
25. The Laplace Transform
The classical approach we have just used to find the transient response
and indeed the final steady state response for simple circuits may be
extended to ever more involved circuits but at the expense of ridiculously
increased complexity. An alternative approach is required. Since we are
concerned with turning forcing functions, often constants, on and off the
mathematical technique of choice to solve the differential equations is
based on the use of the Laplace transform. This is particularly useful for
our purposes because
(i) it gives rise to the ability to write circuit equations in a very

general way just as the "j" method does for sinusoidal forcing
functions.
(ii) the initial conditions are dealt with automatically. The solution does
not contain any arbitrary constants.
(iii) Transients are dealt with automatically.
(iv) and, very importantly, the answers can be looked up easily in

H.L.T.!!
We recall that the Laplace transform of a function of time f t 

is defined as
Lf t  F s  f t exp 

st dt
0
142
where negative values of t are excluded. In mathematics since only

positive values of the variable t are permitted this is often called 'one-
sided' transform.
Arguably one of the most useful properties, from our point of view, of the
Laplace transform, is that for a given f t  there is one and only one
F s and vice versa F s and f t  are called transform pairs.
Fsf t 
Hence if we know the value of F s and we want to know f t , or

vice versa, we simply look it up in the tables!! What could be easier?
Although you will have worked out Laplace transforms in maths we'll
introduce a few 'electrically useful' transform pairs below.
(i) the unit step, ut 
This function, which is useful to describe the closing of a switch say is a

function that switches from 0 abruptly to 1 at t  0 . Thus the terminal
voltage in the circuit below would be written 10ut 
143
The Laplace transform Us is given by
 
1
U s   ut exp  st dt 
 1exp  st dt
 0  0 s
Hence
ut 
1
s
are Laplace transform pairs. Thus given Us the corresponding

1
s
function of time is the unit step, ut
(ii)
. exponential decay
f t exp t ; t 
0

1
F s exp s  t dt
 0 s

144
(iii) damped sinusoids
 cost
f t exp t ;t
0 sint
Here it's probably easiest to consider the composite function
g t exp t cost  j sintexp    jt
since this is formally equivalent to the exponential function we have just

discussed we can write the transform down directly as
1 s 
Gs   j
 s    s   2  2 s   2 
j 
and hence by equating real and imaginary parts we find
145
s 
Lexp t cos ; Lexp t
t
s   2  sint
s   2 
2 2
(iv) Laplace transform of a derivative
If the Laplace transform of f t  is F s

then
L df  s F s  f 0
dt 
Where f 0   represents the initial value of f t . The

corresponding expressions for the second derivative is
d 2f  2
L 2   s F s  s f 0   f 0
 dt 

where f    denotes the first derivative.
Fortunately for you in the P2 course (though not in the P1 course), we

will only use Laplace in the case of zero initial conditions when these
expressions simplify to
 
L df  s F s  L d 2f  s 2 F s 
2
dt  dt 
and 
(v) Laplace transform of an integral
 F
L  f t dt 
t
 0 s s
146
Having listed a number of Laplace transforms let's now illustrate their

use formally by returning to the RC circuit we considered previously
but redrawn below
Since the switch is assumed to be closed at t  0 the governing

differential equation may be written very generally as
u t 
dv v E
 
dt T T
Where, as before, T  RC and the unit step ut 0 t  0 and 1, t 

0.
We now take Laplace transforms of all terms in the equation to give
V s E 1
sV sv 0
  .
T Ts

If we assume, as before, that the capacitor is initially uncharged

v 0   0 , gives
147
1
V s .
E
T ss 1 T

The form of V s is almost familiar, but not quite. In order to be able to
find the inverse transform we need to rearrange it as
1
V s E 1 
s s 1 T


which are functions we have met before and permits us to transform

back to the time domain to give
vt  E ut  exp  t T
which, if we restrict ourselves to t  0 , may be written in a more familiar

way as
vt  E1 exp  t T
which is, of course, the solution we obtained previously but now with
considerably greater ease since we had no arbitrary constants to find.
We emphasise that the procedure involves
(i) transform the differential equation and include the initial

conditions
148
(ii) manipulate the equation to obtain V s is an easily 'invertible' form.

This, sadly, may involve the use of partial fractions in more
involved cases.
(iii) transform back to the time domain to obtain v t .
It would be a useful exercise to return now to the second order RLC

circuit, transform the governing differential equation to obtain the
Laplace transform Vc s of v c t  with zero initial conditions as,
 
 s2  R L 
Vc s E1  R  1 R
2
s
  s 
  
 2L  LC  1L 

It is now routine – try it – to transform back to the time domain to
reproduce the damped and undamped cases we discussed previously.
We end this section by listing a further few useful properties of the

Laplace transform.
149
The Shifting theorem
We show below a function f t  along with a "shifted" version. The
function ut T  is a unit step that is switched on at t T .
If F sdenotes the Laplace transform of f t  then
Lf t T u t T  exp sT Fs
i.e. when the time function is shifted along by T the Laplace transform is
multiplied by exp  sT .
This property is very useful when building up the transforms of

complicated functions. Suppose we want to consider a pulse that might
correspond to the opening and subsequent closing of a switch. The time
function, gt , might look like
150
that is a unit step up at t  0 together with a unit step down at t T .

Thus
Lgt  Lut  ut T 

1 exp  sT 1 exp 
  
sT s s s
In reverse if we are required to find the inverse

transform of, say,
s 
exp 
sT. s   2  2
s 
we recognise this as the inverse transform of , i.e.
s   
2
 2
exp t cost but "shifted" by T. Thus the inverse transform is given by
1 
 
s
L exp  sT   exp t T cost
s     
2
 2
T 

151
Thus we look up the transform in tables as usual and then introduce the
shifting function.
The tables which are available to you in the examination are to be found
in H.L.T. and are reproduced below for convenience.
Unit step at t  0,ut 1s
t n 1 n  1!
 1 sn
e at 1 s  a
ss 
1 a1 e at
1
 a s s 2
coshat
cos at
 a2  s
sinat s  a 2 2
sinhat a s  a 2 2
1 a 1  cosat 
2
 a s  2
1 a at  sinat 
3
a  2
te at
1 s s  2
a 2
e at
1 at 
1 s s 2 2
 sinat  at
1 2a 3
 a  2
cosat 
1 s 
t 2asinat
a2 s s 
t cosat
e at
cos bt
 a2 
e at
sin bt

1 s2 
a 
2
152 2
s s 2

2
26. The Laplace transform in circuit analysis.
If we wanted to find how the voltage across a certain component varied

with time due to some arbitrary forcing function it would be perfectly
possible, using appropriate combinations of node-voltage and loop
equations, to set up the required (simultaneous) differential equations.
We could then take Laplace transforms of both sides of all the equations
or, to put it more technically, we would transform from the time domain to
the s-domain. The resulting simultaneous algebraic equations could
then be solved to find an expression for the Laplace transform of the
desired ‘output’ in terms of the Laplace transform of the ‘input’ source.
Once this frequency response function has been obtained it becomes a
routine matter to introduce the actual transform of the source and then
transform back to the time domain by looking up the inverse transforms
in tables. However it is possible to miss out the initial steps of writing the
differential equations and transforming them by working directly with
transformed voltages and currents V(s) and I(s) together with the
appropriate s-domain version of ‘Ohm’s Law’ for resistors, capacitors
and inductors.
Representation of circuit elements in the s-domain.
Resistor
The v-i equation for a resistor in the time domain is
vRi
153
The Laplace transform of this equation is
V(s)  R I(s)
The transfer function V(s) / I(s)  Z(s)  R is the s-domain

impedance
of the element.
Inductor
In this case the time domain v-i relationship is
di
v  L dt
which, when transformed into the s-domain assuming zero initial

conditions becomes
V(s) = sL I(s)
154
Capacitor
Finally we consider the v-i characteristics of a capacitor as
dv
i  C dt
which becomes in the s-domain, assuming zero initial conditions,
I(s) = sC V(s)
or
V(s) = (1/sC) I(s)
I hope that, by now, although this approach is new, it is beginning to feel

a little familiar. When we were interested in sinusoidal sources (forcing
functions) we found that the mathematical technique which enabled us
to solve the differential equations most easily required us to write ‘Ohm’s
Law’ for the resistor, inductor and capacitor, in phasor notation as
155
1
VRI V  jL and V I
I jC
respectively. In the more general s-domain case the relationships are
1
V RI V  sL I and V  I
sC
which are seen to be identical if we merely replace j by s. Thus all the

methods we developed in the AC theory ‘j’ case carry over to the s-
domain by simple substitution.
Example
Let’s revisit the RC circuit we have already analysed twice before

already i.e.
Where v 1 (t ) Eu(t ) since the switch is closed at t=0. Since the
capacitor is assumed to be uncharged the circuit may be redrawn s-

domain notation below. We have also re-arranged the circuit to make it
clearly of the form of a voltage divider.
156
We can immediately write
V2 s 
1
sC R  sC
V1 s  1 V
1
1 sCR s
which, introducing T = RC and V1(s)  E / s

gives
E
V2 (s) 
s1 sT

which is the same function we obtained previously but we have found it

much more easily and directly here.
As a further example consider the circuit below together with its s-

domain representation.
157
In this case
R V
V2(s)  1
1 s
R
sC
which with V1(s)  E / s and T = RC gives
E
V2(s) 
1
s
T
The inverse is readily looked up in tables to give
v 2 (t )  E exp  t /T
We know, from physical reasoning, that when the switch opens,

v 2 (0 )  E , whereas when it has been open for a long time the
capacitor
acts as an open circuit and hence v 2 ()  0 . These observations are
confirmed, of course, by the expression we have just obtained.

However
it can also be checked from the form of V(s) by using the initial and final
value theorems. We know
158
 
 E
initial value  Lims  sV2 (s)    E 
s   s   Lim s   E
Lim 1  1  sT 
s  T 
whereas the final value is given by
 
 E
final value  Lims 0 sV2 (s)  Lim 
s 0  s  0
1
s  T 
We note the procedure is
(i) Introduce ‘generalised’ s-domain impedances, sL and 1/sC.
(ii) Find the appropriate transfer function. In more complicated cases

this may/will involve writing node-voltage and/or loop equations.
(iii) Substitute the Laplace transform of the input signal (forcing

function)
(iv) Convert back to the time domain by looking up the

inverse transforms in tables.
As a final example let’s consider the following circuit where the input
voltage is a pulse of duration T and we are required to find the voltage
v 2 (t ).
159
The generalised impedance of the RC parallel combination is given by
R /1  sCR and hence, after a little trivial algebra
V2 s 1 1

V1 s  2T1 s  1/T1
where T1 = CR/2.
We have already shown how the shift theorem may be used to obtain
the Laplace transform of a pulse by regarding it as the superposition of
two time-delayed step functions. The details were given previously
which permit us to write
V1s  1  exp  sT 
E
s
and hence
160
E 1
V2 s 
2T1 ss  1/ sT1

1  exp  sT 
1 

E 1   1  exp  sT 
2 s s  1/T1 
The inverse of the expression in curly brackets is readily obtained as

u(t )  exp  t /T1 . The second bracket merely indicates, as we have
seen before that the full solution consists of the difference between two
identical, but time shifted, expressions. Thus
v 2 t   u(t )  exp t /T1  u(t T )  exp (t T ) /T

E
1
2

which, depending on the relationship between T and T1, might look
something like
161
27. Disclaimer!!
The notes are reasonably self-contained but they certainly don’t pretend
to tell the whole story so please do consult one or more of the huge
number of textbooks on the subject.
162

Circuit Analysis Ii: (AC Circuits)

Uploaded by

Copyright:

Available Formats

Circuit Analysis Ii: (AC Circuits)

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Circuit Analysis Ii: (AC Circuits)

Uploaded by

Copyright:

Available Formats

Will Moore

Our development of the principles of circuit analysis in Circuit Analysis I

Let's begin by considering a sinusoidal variation in voltage

However the time period T  f1 where f is the frequency measured in

This is a simple and very important relationship. We naturally measure

both  and f. It is important in calculations to make sure that if 

In our example above, v Vm cost , it was convenient that

Alternatively, in terms of a different phase angle, , the same waveform

The phase difference between two sinusoids is almost always measured

that the two sinusoids are said to be in quadrature. When    the

For our sinusoidal function v Vm cost

The average of the square is given by

where the time period T  2 . At this point it's probably easiest to

change variables to t and to write cos2  

Here the current is given by i v R

where Im Vm R and the power, p vi , is given by

If we want to calculate the average power dissipated over a cycle we

Indeed if the rms value of the current I I m 2 is also introduced then

3. Circuit analysis with sinusoids

Let us begin by considering the following circuit and try to find

The Kirchhoff voltage law permits us to write

This is a linear differential equation, which you know how to solve.

which yields the solution:

i t  A exp Rt L B cost C sint

and so any measurements we are likely to make on this circuit will be

In order to find B and C we need to substitute this expression back

LC cos t  B sin t  RB cos t  C sin t Vm cos t

It is now a simple matter to compare coefficients of cost and sint

If we now introduce the inductive reactance X L  L we can write this

The expression in curly brackets is of the form

coscost  sinsint cost  

resistance and reactance is given the new name of impedance.

It is apparent that we could solve all networks containing combinations

We are saved because the differential equations we have to solve are

You should know that

where j = (-1), [electrical engineers like to use i for current]

v t Vm exp j t Vm cost  j Vm

v1 t Rev t ReVm exp j t Vm

The solution will be of the form

i1 t ReI exp jt ReI exp jt  

Let's illustrate this by returning to our previous example where we tried

Now, instead, we solve the more general case:

j LI exp j t  RI exp j t Vm exp j t

The factor exp j t is common and hence

in which R  j L may be regarded as a complex impedance. The

with  tan1 L R and hence

Let us be clear about the approach. We have

(i)introduced a complex forcing function Vm exp j t knowing that in

reality the voltage source must be real i.e. ReVm exp j t .

(ii)We solved the equations working with complex voltages and

time dependence exp jt cancelled out).

(iii)Since the actual voltage is given by ReVm exp j t  the actual

current is given by ReI exp j tReI exp j exp j t I cost

(iv) Since the differential of is simply

and since we always take out as a common

This is a very powerful approach that will permit us to solve AC circuit

Let's now do an example to show, formally, how we can solve AC

The two KVL loop equations may be written

Replacing Em cos(t + ) by Em exp j (t +)=E1 exp j t where