EE1-01

IMPERIAL COLLEGE LONDON

DEPARTMENT OF ELECTRICAL AND ELECTRONIC ENGINEERING

EXAMINATIONS 2014

ExamHeader:
EEE/EIE PART I: MEng, Beng and ACGI

ANALYSIS OF CIRCUITS

Friday, 6 June 10:00 am

Time allowed: 2:00 hours

There are THREE questions on this paper.

Answer ALL questions.

Q1 carries 40% of the marks. Questions 2 and 3 carry equal marks (30%
each).

Any special instructions for invigilators and information for

candidates are on page 1.

Examiners responsible First Marker(s) : D.M. Brookes

Second Marker(s) : P. Georgiou

© Imperial College London

A NALYSIS OF C IRCUITS

Information for Candidates:

The following notation is used in this paper:

1. The voltage waveform at node X in a circuit is denoted by x(t), the phasor voltage by X and the
root-mean-square (or RMS) phasor voltage by Xe = √X2 . The complex conjugate of X is X ∗ .

2. Component and source values in a circuit are normally given in Ohms, Farads, Henrys, Volts or
Amps with the unit symbol omitted. Where an imaginary number is specified, it represents the
complex impedance or phasor value.

3. Times are given in seconds unless otherwise stated.

4. Unless otherwise indicated, frequency response graphs should use a linear axis for phase and log-
arithmic axes for frequency and magnitude.

Analysis of Circuits c Imperial College London 1/6

1. a) Using nodal analysis, calculate the voltages at nodes X and Y of Figure 1.1.
[5]

Figure 1.1 Figure 1.2

b) Use the principle of superposition to find the current I in Figure 1.2.

[5]
c) Draw the Thévenin equivalent circuit of the network in Figure 1.3 and find the
value of its components. [5]

Figure 1.3 Figure 1.4

d) Assuming the opamp in the circuit of Figure 1.4 is ideal, give an expression for
Y in terms of U and V . [5]

Analysis of Circuits c Imperial College London 2/6

 e) The graph of Figure 1.5 plots the output voltage, Y , against the input voltage,
X, for the circuit shown in Figure 1.6. The graph consists of two straight lines
that intersect at the point (10, 10) and that pass through the origin and the point
(20, 12) respectively. Assuming that the forward voltage drop of the diode is
0.7 V, determine the values of the resistor, R, and the voltage source, V . [ 5 ]

12
10
Y

0
0 5 10 15 20
X

Figure 1.5 Figure 1.6

f) Determine the gain, YX , for the block diagram shown in Figure 1.7. The rectan-
gular blocks are drawn with inputs at the left and outputs at the right and have
gains of F and G respectively. The open circles represent adder/subtractors;
their inputs have the signs indicated on the diagram and their outputs are W and
Y respectively. [5]

Figure 1.7 Figure 1.8

g) In the circuit of Figure 1.8, the RMS phasor Xe = 230 and the component values
shown indicate complex impedances. Determine the value of the RMS current
phasor J˜ and of the complex power, Ve × I˜∗ , absorbed by each of the four com-
ponents.
[5]

Analysis of Circuits c Imperial College London 3/6

 h) Figure 1.10 shows a transmission line of length 100 m that is terminated in a re-
sistive load, R, with reflection coefficient ρ = +0.6. The line has a propagation
velocity of u = 2 × 108 m/s. At time t = 0, a forward-travelling (i.e. left-to-
right) pulse arrives at X with amplitude 4 V and duration 1.5 µs, as shown in
Figure 1.9.
Draw a dimensioned sketch of the waveform at Y , a point 60 m from the end of
the line, for 0 ≤ t ≤ 3 µs. Assume that no reflections occur at point X. [5]

Figure 1.9 Figure 1.10

Analysis of Circuits c Imperial College London 4/6

2. a) Show that the transfer function of the circuit of Figure 2.1 can be written in the
form

Y ( jω) 1
H( jω) = =  2
X ( jω) jω
ω0 + 2ζ ωjω0 + 1

and express the values of ω0 and ζ in terms of the component values L, C and
R. [5]
b) Give expressions for the low and high frequency asymptotes of H( jω) and the
angular frequency at which they have the same magnitude. [3]
c) Determine the magnitude and phase of H( jω) at ω = ω0 . [2]
d) Show that |H( jω)|−2 may be written as a polynomial with real coefficients in
 2
x where x = ωω0 . By differentiating this polynomial, or otherwise, show that
p
the maximum value of |H( jω)| occurs at ω = ω0 1 − 2ζ 2 . [6]
e) Determine values of C and R so that ω0 = 5000 rad/s and ζ = 0.1 given that
L = 100 mH. [2]
i) Sketch a dimensioned graph of |H( jω)| in decibels using a logarith-
mic frequency axis. Your graph should include both the high and low
frequency asymptotes in addition to a sketch of the true magnitude
response. [3]
ii) If x(t) = 3 cos ω0t, determine the average power dissipation of the
circuit and the peak value of the energy, 12 Cy2 (t), stored in the ca-
pacitor.
[3]
iii) Determine the values of ω for which ∠H( jω) = −45◦ and −135◦ .
Hence sketch a dimensioned graph of ∠H( jω) using a straight-line
approximation with three segments. Your graph should use a loga-
rithmic frequency axis and a linear phase axis. [6]

Figure 2.1

Analysis of Circuits c Imperial College London 5/6

3. In the circuit of Fig. 3.1, the input, X, is driven by a voltage source as shown.
Y ( jω)
a) Derive an expression for the transfer function, X( jω) and determine the corner
frequencies in its magnitude response. [4]
b) With the capacitor temporarily removed from the circuit, determine the Thévenin
equivalent voltage and resistance of the remainder of the circuit at the terminals
of the capacitor. [4]
c) Derive the time constant of the circuit, τ, in two ways: (i) from the Thévenin
resistance found in part b) and (ii) from the denominator corner frequency found
in part a). [2]
d) If the input voltage, x(t), is given by
(
−2 for t < 0
x(t) = ,
+3 for t ≥ 0

determine an expression for the output waveform, y(t). Sketch its waveform
over approximately the range −τ ≤ t ≤ 4τ. [7]
e) Assuming that the opamp in Fig. 3.2 is ideal, determine the transfer function,
V ( jω)
U( jω) . [4]
f) By considering the voltage across the capacitor, explain why an input voltage
discontinuity of ∆ u will result in an output voltage discontinuity of the same
amplitude. [2]
g) If R = 20 kΩ, C = 20 nF and the input voltage, u(t), is given by
(
sin 1000t for t < 0
u(t) = ,
2 cos 2000t for t ≥ 0

determine expressions for the output v(t) for both positive and negative t. [ 7 ]

Figure 3.1 Figure 3.2

Analysis of Circuits c Imperial College London 6/6

A NALYSIS OF C IRCUITS

**** Solutions 2014 ****

Information for Candidates:

The following notation is used in this paper:

1. The voltage waveform at node X in a circuit is denoted by x(t), the phasor voltage by X and the
root-mean-square (or RMS) phasor voltage by Xe = √X2 . The complex conjugate of X is X ∗ .

2. Component and source values in a circuit are normally given in Ohms, Farads, Henrys, Volts or
Amps with the unit symbol omitted. Where an imaginary number is specified, it represents the
complex impedance or phasor value.

3. Times are given in seconds unless otherwise stated.

4. Unless otherwise indicated, frequency response graphs should use a linear axis for phase and log-
arithmic axes for frequency and magnitude.

Analysis of Circuits c Imperial College London 1/19

1. a) Using nodal analysis, calculate the voltages at nodes X and Y of Figure 1.1.
[5]

We can immediately label the voltages on the bottom left and top left nodes as
6 and 6 + 9 = 15 respectively. We now write down KCL equation at node X to
obtain
X − 15 X −Y
+X −6+ = 0
2 3
⇒ 3X − 45 + 6X − 36 + 2X − 2Y = 0
⇒ 11X − 2Y = 81

KCL at Y gives
Y −X Y
+ +4 = 0
3 2
⇒ 2Y − 2X + 3Y + 24 = 0
⇒ −2X + 5Y = −24

357
Combining these gives 55X − 4X = 405 − 48 ⇒ X= 51 =7
−10
from which 5Y = −24 + 14 = −10 ⇒ Y= 5 = −2
Several people wrote X−93 + · · · instead of
X−15
3 + · · · for KCL at X and a few
Y −(−6) Y
wrote · · · + 2 + · · · instead of · · · + 2 + · · · for KCL at Y . I advise labeling
the nodes explicitly with their voltages on the diagram; a voltage source fixes
the difference between two node voltages rather than the voltage at a particular
node. Thus, the node with a ground symbol has, by definition, a voltage of 0,
the node at the + side of the 6 V source has a voltage of 6 and the node at the
+ end of the 9 V source, therefore has a voltage of 6 + 9 = 15.

Figure 1.1 Figure 1.2

b) Use the principle of superposition to find the current I in Figure 1.2.

[5]

If we short circuit the 12 V voltage source, the 7 Ω and 1 Ω are shorted out by
8
the central link that carries I and so we have a current of IA = 3+2 = +1.6 A.
If we short circuit the 8 V voltage source, the 3 Ω and 2 Ω are shorted out and
−12
so we have a current of IB = 7+1 = −1.5 A.
By superposition, the total current is therefore 1.6 − 1.5 = 0.1 A.

Analysis of Circuits c Imperial College London 2/19

 Some people treated the two ends of the wire carrying I as separate nodes. Each
of the sub-circuits is simple enough to write the current down directly. A few
people however used nodal analysis to solve the sub-circuits and often made
mistakes. The most common mistake by far was not to define which node was
ground (and therefore at 0V ); unless you do this, nodal analysis is hopeless
since all the node voltages are undefined.

c) Draw the Thévenin equivalent circuit of the network in Figure 1.3 and find the
value of its components. [5]

We can find the open circuit voltage by ignoring the 3 k resistor (since there is
no current flowing through it). The 8 mA will therefore flow upwards through
the 7 k resistor resulting in an open-circuit voltage of VAB = −8 × 7 = −56 V.
To find the Thévenin resistance, we treat the current source as an open circuit.
The 4 k resistor now plays no part and the Thévenin resistance is therefore
7 + 3 = 10 k.
So the complete Thévenin equivalent is:

A more complicated approach is to do a full nodal analysis. If we define ground

to be terminal B and label the top and bottom of the current source as X and Y
respectively, then we want to find the voltage at A when a current I is flowing
into it. KCL at A, X and Y gives
A − X A −Y
+ − I = 0 ⇒ 11A − 7X − 4Y = 28I
4 7
X −A
+ 8 = 0 ⇒ A − X = 32
4
Y −A Y
+ − 8 = 0 ⇒ 3A − 10Y = −168
7 3

from which the solution is

   −1      
A 11 −7 −4 28I −56 10
 X  =  1 −1 0   32  =  −88  +  10  I
Y 3 1 −10 −168 0 3

easily found using the simultaneous equation solver on the calculator by solving
first with I = 0 and then with I = 1.
From the top row, A = −56 + 10I, which gives the Thévenin component values
directly.
Several people thought the 7 k and 3 k resistors were in parallel rather than in
series (presumably because they shorted A and B together). When calculating

Analysis of Circuits c Imperial College London 3/19

 the open-circuit voltage, some people thought that the 8 mA current would di-
vide between the 7 k and 3 k resistors; in fact no current flows through the 3 k
resistor when A and B are open-circuit. Some people calculated the component
values but lost marks because they did not draw the diagram as the question
asked. Many people got the sign of the voltage source wrong; often they cal-
culated the correct answer of −56 V but then drew the wrong polarity on the
diagram.

Figure 1.3 Figure 1.4

d) Assuming the opamp in the circuit of Figure 1.4 is ideal, give an expression for
Y in terms of U and V . [5]

This is an non-inverting op-amp circuit and so we can write down Y = W 1 + 25

5 =
6W .
W −U
Applying KCL at node W gives 12 + W −V
4 = 0 from which W − U + 3W −
3V = 0 and hence W = U+3V
4 .
Putting these together gives Y = 6W = 6 × U+3V
4 = 1.5U + 4.5V .
Although the two inputs of an ideal opamp with negative feedback are at the
same voltage, they are two distinct nodes nevertheless and hence give rise to
two separate KCL equations (because current does not flow into one input and
out of the other). A few people incorrectly wrote down a single KCL equation
that included both nodes.A surprising number of people had sign errors in their
algebra (in other questions as well) and wrote Y = −1.5U + 4.5V ; if an input
is connected only to the + terminal of an opamp, then its gain will be positive
and if it is connected only to the − terminal, its gain will be negative.
In this question, as in others, quite a few people gave answers that were equiv-
alent to the correct answer, but that did not quite answer the question, e.g.
2Y = 3U + 9V ; if the question asks for an expression for Y , then the last line of
your answer should be “Y = . . .”.

Analysis of Circuits c Imperial College London 4/19

 e) The graph of Figure 1.5 plots the output voltage, Y , against the input voltage,
X, for the circuit shown in Figure 1.6. The graph consists of two straight lines
that intersect at the point (10, 10) and that pass through the origin and the point
(20, 12) respectively. Assuming that the forward voltage drop of the diode is
0.7 V, determine the values of the resistor, R, and the voltage source, V . [ 5 ]

The diode turns on when Y = V + 0.7. For this to be when Y = 10 V we must

have V = 9.3.
To determine R, we can apply KCL at node Y when X = 20 and Y = 12:
12 − 20 12 − 10
+ =0
20 R
from which −8R + 40 = 0 which gives R = 5 k.
Alternatively, we can say that, since the slope of the second part of the charac-
2
teristic is 10 = 0.2, this must be the gain of the potential divider formed by 20 k
and R. To get a gain of 15 you need 20 k = 4R from which R = 5 k as before.
Yet another way to solve the problem is to write the KCL equation when the
diode is on:
Y − X Y − 0.7 −V
+ = 0
20 R
⇒ (Y − X)R − 20V = 14 − 20Y

Now we can substitute the two points (10, 10) and (20, 12) (both of which lie
on the “diode on” part of the graph) to get a pair of simultaneous equations

0R − 20V = 14 − 200 = −186

−8R − 20V = 14 − 240 = −226

which we can solve for R and V .

Several wrote down incorrect equations for either the “diode on” or “diode
off” situations. The diode acts either as an open circuit (“off”) or else a 0.7 V
voltage source (“on”). When the diode is off, there is no current through R, so
−X
KCL at Y gives Y 20 = 0. Several people assumed the “diode on” equation,
Y −X Y −0.7−V
20 + R = 0, was valid all the time including at X = Y = 0.
Some people treated the circuit as a potential divider; this is possible but, since
neither resistor is connected to ground, it is easy to make errors and most of
those that tried this method got it wrong. The commonest error was to write
R
Y = X 20+R which implicitly assumes the top of the diode is at 0 V. The correct
potential divider expressions involve voltage differences across resistors (i.e.
the voltage at one end minus the voltage at the other) giving Y − (V + 0.7) =
R 20
(X − (V + 0.7)) 20+R or alternatively X − Y = (X − (V + 0.7)) 20+R . Using
nodal analysis is much less error-prone for circuits that are even a little bit
complicated. Several people took the voltage across the diode to be −0.7 V
instead of +0.7 V; when the diode is on, the voltage and current polarities are
the same as for a resistor (since they both absorb power).

Analysis of Circuits c Imperial College London 5/19

 12
10

Y 5

0
0 5 10 15 20
X

Figure 1.5 Figure 1.6

f) Determine the gain, YX , for the block diagram shown in Figure 1.7. The rectan-
gular blocks are drawn with inputs at the left and outputs at the right and have
gains of F and G respectively. The open circles represent adder/subtractors;
their inputs have the signs indicated on the diagram and their outputs are W and
Y respectively. [5]

For node W , we can write the following equation: W = X − GFW from which
1
we get W = 1+FG X.
F
− 1 X = F−FG−1


For node Y,we can write Y = FW − X = 1+FG 1+FG X so the gain
Y F−FG−1
is X = 1+FG .
Some wrote down the correct equations but were not sure what to do next; if you
want to get Y in terms of X, you need to eliminate W . Some people wrote down
dimensionally inconsistent equations such as W = X − GF; when writing down
the equations for a block diagram, gains (such as F or G) must always multiply
signals (such as X, W or Y ). Although the question asked for Yx , many people
FX
instead gave an expression for Y in terms of X (e.g. Y = 1+FG − X). This is, of
course, equivalent but it is always a good idea to answer the question that was
asked rather than a different question of your own creation.

Figure 1.7 Figure 1.8

g) In the circuit of Figure 1.8, the RMS phasor Xe = 230 and the component values
shown indicate complex impedances. Determine the value of the RMS current
phasor J˜ and of the complex power, Ve × I˜∗ , absorbed by each of the four com-
ponents.
[5]

Analysis of Circuits c Imperial College London 6/19

 Xe
The RMS current through he capacitor is −23 j = 10 j.
Xe 2760−1150 j
The RMS current through the inductor is 12+5 j = 169 = 16.3 − 6.8 j.
2760−540 j
The total RMS current is therefore J˜ = 16.3 − 6.8 j + 10 j = 169 = 16.3 +
3.2 j.
Power absorbed by source is −Xe × J˜∗ = −230 (16.3 − 3.2 j) = −3.76+0.73 j kVA.
Power absorbed by capacitor is Xe × (10 j)∗ = −2.3 j kVA.
2
Power absorbed by resistor is I˜ R = |16.3 − 6.8 j|2 ×12 = 313×12 = 3.76 kW.
2
Power absorbed by inductor is I˜ ZL = |16.3 − 6.8 j|2 ×5 j = 313×5 j = 1.57 j kVA.
As expected the powers sum to zero and the complex power absorbed by a pas-
sive component (C, R or L) has the same phase as the component’s impedance.
Many people, presumably following their training from school, gave “exact”
answers like J˜ = 2760−540
169
j
. Answers in this form are useful in engineering only
in very rare cases since physical quantities such as resistance and current never
have exact rational values except when they are 0; in questions based on com-
ponent values, you should always give answers in decimal form. Component
values are typically only accurate to 2 or at most 3 significant digits.
Many people made errors doing complex arithmetic manually; note that the
calculators supplied in the exam are able to do complex arithmetic but practis-
ing beforehand is advisable. Also, if you use the calculator, it is still advisable
to write down intermediate values so that you will get some marks even if the
230
final answer is wrong. A common error was to say −23 j = −10 j instead of
+10 j. Some took S = |I|Z˜ instead of S = |I| ˜ Z. Others stated the correct ex-
2

pression but actually used S = I˜2 Z instead. In order to determine J, some peo-
ple began by calculating the total impedance of the three passive components
(= 13.6 − 2.65 j Ω ); although this is entirely correct, it is easier to calculate
the capacitor current separately and add it to the current through the induc-
tor/resistor combination (as is done above).
Ṽ
Calculating the current through the capacitor is very easy as I˜C = −23 j . Quite
12+5 j
a few people used the current divider formula to calculate it as IC = 12+5 j−23 j J˜
˜
which is correct but much more effort and usually gave an answer that was not
purely imaginary.
Quite a few people gave the power supplied by, rather than absorbed by the
source; indeed several gave this as their only answer which was not what the
question asked for. One or two were troubled by the idea of a voltage source
“absorbing” power at all; if there is only a single source, the complex power
it absorbs will always have a negative real part (meaning that it is actually
supplying average power to the rest of the circuit) but the imaginary part can
have either sign.
The correct formulae for the power absorbed by a component with complex
2
˜ 2 Z = |Ve∗| . Several people omitted the squares,
impedance Z are S = Ve × I˜∗ = |I| Z
the conjugation and/or the modulus signs from the expressions even though the
conjugation
√ was given in the question); a few sprinkled their equations with
2 factors or else multiplied the expressions by 21 which is not necessary if
you are using RMS quantities like Ve and I. ˜ From the formula S = |I| ˜ 2 Z it
is clear that S and Z must have the same phase (i.e. complex argument) so the
powers absorbed by R, C or L must respectively be real, negative-imaginary and

Analysis of Circuits c Imperial College London 7/19

 positive-imaginary; many people gave answers that violated this constraint.
Several people used the correct formula S = |I| ˜ 2 Z but used the total current
J˜ when calculating the power absorbed by the passive components. In fact,
J˜ splits and flows partly thought the capacitor and partly through the induc-
tion+resistor;in the formula, I˜ needs to be the current that actually flows through
the component in question.
1
Several people talked about jωL or jωC and some said the question was im-
1
possible without knowing ω; however, in this case jωL = 5 j and jωC = −23 j
so you do not need to know ω explicitly.

Analysis of Circuits c Imperial College London 8/19

 h) Figure 1.10 shows a transmission line of length 100 m that is terminated in a re-
sistive load, R, with reflection coefficient ρ = +0.6. The line has a propagation
velocity of u = 2 × 108 m/s. At time t = 0, a forward-travelling (i.e. left-to-
right) pulse arrives at X with amplitude 4 V and duration 1.5 µs, as shown in
Figure 1.9.
Draw a dimensioned sketch of the waveform at Y , a point 60 m from the end of
the line, for 0 ≤ t ≤ 3 µs. Assume that no reflections occur at point X. [5]

Figure 1.9 Figure 1.10

The velocity, u, is 200 m per µs. The forward wave takes 0.2 µs to reach Y and
a further 0.6 µs to reflect from the end and return to Y . Therefore the waveform
at Y is the sum of two overlapping waves: (i) a pulse of amplitude 4 V beginning
at t = 0.2 µs (ending at t = 1.7 µs) and a pulse of 4 × ρ = 2.4 V beginning at
t = 0.8 µs (ending at t = 2.3 µs). Where the pulses overlap, their combined
voltage is 4 + 2.4 = 6.4 V.

0
0 1 2 3
Time (µs)

Most people who tried this got it right but quite a few did not attempt this
question at all. A few people used a value of ρ = −0.6; in general, ρ can have
either sign but in this question you are told that it is positive. Several assumed
a value of ρ0 = +1 at X even though the question explicitly said there were no
reflections at X; this made the question somewhat harder. On a “dimensioned
sketch” you should mark the values on the X and Y axes where interesting things
happen: in this case, this means marking “0”, “2.4”, “4” and “6.4” on the
vertical axis and “0.2”, “0.8”, “1.7” and “2.3” on the horizontal axis (unlike
on the graph above).

Analysis of Circuits c Imperial College London 9/19

2. a) Show that the transfer function of the circuit of Figure 2.1 can be written in the
form

Y ( jω) 1
H( jω) = =  2
X ( jω) jω
ω0 + 2ζ ωjω0 + 1

and express the values of ω0 and ζ in terms of the component values L, C and
R. [5]

Viewing the circuit as a potential divider, the transfer function is

1
jωC
H( jω) = 1
jωL + R + jωC
1
= 2
( jω) LC + jωRC + 1
1
=  2
jω
ω0 + 2ζ ωjω0 + 1

q
1 2ζ 1
where, by identifying coefficients, ω02
= LC and ω0 = RC from which ω0 = LC
q
ω0 RC RC
and ζ = 2 = 2 LC = 2 CL .
√ R

One person said they could not do this question because they had not revised the
topic of “resonance”. Actually, the question is pretty much self-contained, so it
should be possible to do it even so (revision is still a good idea though). Many
people find ζ hard to write (or at least write clearly). Rather than matching the
coefficients between the transfer function and the formula given in theqquestion,
1
pc
some just remembered the formulae from the lectures: ω0 = a = LC and
b
ζ = √4ac = √RC
4LC
; this method works but relies on a good memory and also on
the notation in the question exactly matching that used in the notes (luckily this
was true). Many people did not precisely answer the question that was asked:
if the question asks for an expression for ζ , then your answer should end with
a line of the form ζ = . . . rather than with some vaguely equivalent equation
such as 2ζ = √RC or even 4ζ 2 = R2 C .
LC L

b) Give expressions for the low and high frequency asymptotes of H( jω) and the
angular frequency at which they have the same magnitude. [3]

 −2
LF asymptote: HLF ( jω) = 1. HF asymptote: H( jω) = 1
LC ( jω)−2 = jω
ω0 .
The asymptotes have the same magnitude at ω = ω0 .
The asymptotes are complex-valued functions of ω that specify both the mag-
nitude and the phase at low or high frequencies respectively. Some people just
gave the magnitude of the asymptotes; this is incorrect.
Some people confused “HF asymptote” with “value at ω = ∞”. In this exam-
 −2
ple, the HF asymptote is ωjω0 but the value at ω = ∞ is H( j∞) = 0. Notice
that, because the asymptote is a function of ω, it tells you how the gain varies as
ω approaches infinity not just when it actually equals infinity. Another way of

Analysis of Circuits c Imperial College London 10/19

 looking at the distinction is that the asymptote is an entire line whereas H( j∞)
is just a single point on the line.
Note that at ω = ω0 the values of the asymptotes are 1 and −1 respectively;
these have the same magnitude (as required by the question) but not the same
phase. Some people tried to find a value of ω at which the two asymptotes had
exactly the same value (i.e. both magnitude and phase); this is not possible for a
real-valued ω. Many people did not distinguish clearly between an asymptote
 −2
and its absolute value e.g. writing false equations like H( jω) = ωjω0 =
 −2
ω
ω0 .

c) Determine the magnitude and phase of H( jω) at ω = ω0 . [2]

 2
At ω = ω0 , jω
ω0 = −1 so H( jω0 ) = 1
−1+2ζ j+1
=−2ζ
j
. This has a magnitude
of |H( jω0 )| = 2ζ1 and a phase ∠H( jω0 ) = − 2 . Note that ζ = ω02RC is always
π

positive.
Several people said that H( jω0 ) = 1 since the high and low frequency asymp-
totes both have magnitude 1 at ω = ω0 (albeit with different phases). However,
at a resonance, the true gain is often not well appriximated by the asymptotes;
that is the main point of this question.
 2
Some ignored the j2 factor in the first term and set ωjω0 = +1 instead of −1.
Ignoring j in a complex number is like ignoring the difference between “North”
and “East” when navigating using a map. Pretty much the entire behaviour of
a quadratic resonance arises from the fact that at ω0 the impedances of the
inductor and capacitor cancel out.
−j
Several people had surprising difficulty giving the magnitude of 2ζ
; some said
−1 j
2ζ
(which is negative) and others said (which is complex). The magnitude
2ζ
of a fraction is the magnitude of the numerator divided by the magnitude of the
denominator and is always real and positive. Note that ζ is always real and,
in our circuit, is bound to be positive (although in other circuits it might be
negative).
Others had difficulty with the phase of − 2ζ
j
; many tried to work it out using
arctan( ) which is valid but definitely overkill. If in doubt about the phase of a
simple complex number, plot it on an Argand diagram.
Although the question asked for “the magnitude and phase” of H( jω0 ), many
people just wrote down its value as a complex number (and therefore lost
marks).

d) Show that |H( jω)|−2 may be written as a polynomial with real coefficients in
 2
x where x = ωω0 . By differentiating this polynomial, or otherwise, show that
p
the maximum value of |H( jω)| occurs at ω = ω0 1 − 2ζ 2 . [6]

To find the magnitude squared, we take the sum of the squares of the real and
imaginary parts; the middle term in the denominator of H( jω) is imaginary
while the other two terms are real (since j2 = −1):

Analysis of Circuits c Imperial College London 11/19

  2 2
−2 −1 2 jω jω
|H( jω)| = H( jω) = + 2ζ +1
ω0 ω0
 2 ! 2
ω 2ζ ω
= 1− +j
ω0 ω0
 2 2 !
2ζ ω 2

ω
= 1− +
ω0 ω0
√ 2
(1 − x)2 + 2ζ x


=
= (1 − x)2 + 4ζ 2 x
= x2 + 4ζ 2 − 2 x + 1

Setting the derivative of this polynomial to zero to find its minimum (it must be a
minimum rather than a maximum because the coefficient of x2 is positive) gives

4ζ 2 − 2
xp = − = 1 − 2ζ 2
2

Hence
 2
ωp
= 1 − 2ζ 2
ω0
p
⇒ ωp = ω0 1 − 2ζ 2

This is the minimum of |H( jω)|−2 and so must be the maximum of |H( jω)|.
Many people found this quite hard because they were not completely familiar
with the facts that if z = a + jb is a complex number, then |z|2 = zz∗ = a2 + b2
and also zk = |z|k for any integer k (we use this above with k = −1). Many
people either ignored the j or ignored the modulus signs; both of these mistakes
make the algebra much harder as well as giving the wrong answer. Actually
|z|2 is much easier to work with than |z| which involves square root signs. Some
multiplied the numerator and denominator of H( jω) =  jω 2 1 jω by its
ω0 +2ζ ω +1
0
complex conjugate; doing this is almost always a bad idea in algebra because
it converts quadratic expressions into quartic expressions.
1
Quite frequently, people assumed that if H( jω) =  jω 2 then |H( jω)| =
ω0 +2ζ ωjω +1
0
1
 2 formed by deleting all the j factors; this is not a valid way
ω
ω0 +2ζ ωω +1
0
of calculating the magnitude of a complex number or expression. Alterna-
tively, several assumed that |H( jω)|−2 = H( jω)−2 which is true for real num-
bers but not for complex numbers. An approach that used even more alge-
bra was to calculate H( jω)2 , i.e. to square it before taking the magnitude.
Others just squared all the terms to get the (incorrect) squared magnitude
 4  2
−2
|H( jω)| = ω0 + 2ζ ω0 + 12 ; you need to add together all the real
ω ω

and all the imaginary terms and then square the two resultant sums.

Analysis of Circuits c Imperial College London 12/19

 If z = a + jb, then |z|2 = z × z∗ = (a + jb) (a − jb) = a2 + b2 , i.e. |z|2 equals
the sum of the squares of its real and imaginary parts (this can also be seen
by applying Pythagoras’ theorem to the Argand diagram). Some people used
the first expression rather than the second which results in much worse algebra:
 2 2  
2
  
2

−2 jω jω jω jω jω jω
|H( jω)| = ω0 + 2ζ ω0 + 1 = ω0 + 2ζ ω0 + 1 ω0 − 2ζ ω0 + 1 =
. . .. A few ignored the j or else did not take the complex conjugate; either error
results in the wrong answer entirely.
Despite the instructions in the question, some substituted x = ωω0 or even x = jω
 2
instead of x = ωω0 . Others didn’t make any substitution at all and worked
entirely in ω which is fine but messier.
Even though the questions told you to use |H( jω)|−2 , a few brave people used
|H( jω)|2 instead. It is, of course, true that the maximum of |H( jω)|2 is at the
same value of ω as the minimum of |H( jω)|−2 but differentiating |H( jω)|2 is
much more effort.
Many people who could not get the right answer “adjusted”
p their algebra so
that the last line was a triumphant ⇒ ω p = ω0 1 − 2ζ 2 but the algebra
adjustments counted as additional errors and so lost them additional marks.
Better to confess that the answer is wrong but you cannot find your mistake.

e) Determine values of C and R so that ω0 = 5000 rad/s and ζ = 0.1 given that
L = 100 mH. [2]

q
1 1 1
From ω0 = LC , C = Lω02
= 0.1×25×106
= 0.4 µF = 400 nF.
ω0 RC 2ζ 0.2
From ζ = 2 , R = Cω 0
= 0.4×5×103−6
= 0.1 kΩ = 100 Ω .
Most people got this right. However it sometimes involved a great deal of alge-
bra. On the whole it is easiest to manipulate symbolic algebra into its simplest
form first and only then substitute numerical values. This avoids wasted work
such as taking square roots and then later on squaring the result (as many peo-
ple did in this question).

i) Sketch a dimensioned graph of |H( jω)| in decibels using a logarith-

mic frequency axis. Your graph should include both the high and low
frequency asymptotes in addition to a sketch of the true magnitude
response. [3]

|H(jω0)| = 14.0dB

0
|Y/X| (dB)

-20
ωp=4950 ω0=5000

-40
1 2 5 10 20 50
Frequency (krad/s)
p √
From part d), the peak is at ω p = ω0 1 − 2ζ 2 = 5000 × 0.98 =
5000 × 0.9899 = 4950 rad/s.

Analysis of Circuits c Imperial College London 13/19

 1
From part c), the gain at ω0 = 5000 is 2ζ
= 5 = 14 dB.
The gradient of the HF asymptote is −2 or, equivalently −40 dB per
decade, meaning that at ω = 10ω0 the gain has fallen to approxi-
mately −40 dB.
Several people derived the correct asymptote expressions but drew
them with the wrong gradient. If the asymptote is Aω k then it has
a gradient of k on the log-log axes that we use for magnitude re-
sponses. This the LF asymptote has a gradient of 0, the HF asymp-
tote has a gradient of −2 and they cross at ω0 = 5000 rad/s. Several
people drew the LF asymptote with a gradient of +1 (presumably
remembering an example from the notes).
Several people showed the gradient of the HF asymptote as only
−20 dB per decade.

ii) If x(t) = 3 cos ω0t, determine the average power dissipation of the
circuit and the peak value of the energy, 21 Cy2 (t), stored in the ca-
pacitor.
[3]

At resonance, the total impedance is jω0 L + R + jω10C = R since the

impedances of the L and C cancel out (we can see this either by
substituting numerical values for L and C or algebraically because
ω 2 LC
ω02 LC = 1 implies jω10C = jω0 0C = − jω0 L). So the average power
hx2 (t)i
dissipation is R = 12 32 × R1 = 45 mW.
Using phasors and the result from part c), Y = YX ( jω) × X = 2ζ1 j X =
1
0.2 j ×3 = −15 j.Hence the peak capacitor voltage is 15 and the peak
1 2
energy stored is 2 C × 15
= 45 µJ.
2
Several people said the average power dissipation was XR (which
hx2 (t)i 1 2
X̂
applies to DC levels) rather than R = 2 R which applies to a
sine wave with peak voltage X̂. Many said the peak energy was
2
measured in Watts rather than Joules. The formula P = VR requires
that V is the voltage across the resistor R. Several thought the power
2
dissipation in R was (x−y)
R even thought he voltage x − y is not the
voltage across the resistor but includes the inductor as well.

iii) Determine the values of ω for which ∠H( jω) = −45◦ and −135◦ .
Hence sketch a dimensioned graph of ∠H( jω) using a straight-line
approximation with three segments. Your graph should use a loga-
rithmic frequency axis and a linear phase axis. [6]

!
1 2ζ ωω
H( jω) =  jω 2 so that ∠H( jω) = − arctan  0 2 . Since
ω0 +2ζ ωjω +1 1− ωω
0 0
tan(−45◦ ) = −1 and tan(−135◦ ) = tan(+45◦ ) = +1, we need the
argument of arctan( ) in the expression for ∠H( jω) to equal ±1. An
equivalent geometrical approach from the Argand diagram is that

Analysis of Circuits c Imperial College London 14/19

 the real and imaginary parts of a complex number with an argu-
 2
ment of ±45◦ must have equal magnitude. So we need 1 − ωω0 =
 2
±2ζ ωω0 or, equivalently, ωω0 ± 2ζ ωω0 − 1 = 0. The roots of this
p
equation are ωω0 = ∓ζ ± ζ 2 + 1.
√
For ζ = 0.1, this gives ωω0 = ∓0.1 ± 1.01 = ±0.905, ±1.105. Tak-
ing the positive frequencies, ∠H( jω) = −45◦ at ω = 4525 and ∠ YX ( jω) =
−135◦ at ω = 5525.

Phase (rad)
4525 rad/s
-1

-2
5525 rad/s
-3
1 2 5 10 20 50
Frequency (krad/s)

The central segment of the straight-line that passes through (4525, −π4 )
−3π
and (5525, 4 ) passes through (ωa , 0) and (ωb , −π) where ωa =
4525 2

2
= 4095 and ωb = 5000× 5525


5000× 5000 5000 = 6105. The squared
frequency ratios arise because we wish to double the phase shift rel-
ative to that at ω0 . Alternatively, for those with a good memory, the
formula given in the lecture notes gives slightly different values of
ωa = 10−ζ ω0 = 3972 and ωb = 10+ζ ω0 = 6295; this is the straight
line approximation plotted above as a dashed red line. The solid
blue curve shows the true phase (not requested).
Most people found this difficult. It is a lot easier if you recognize
straight off that if a complex number has an argument of ±45◦ then
its real and imaginary parts have equal magnitude.
Quite a few people assumed that the corners in the phase response
were at 0.1ω0 and 10ω0 which is true for linear factors. For quadratic
factors however, the corresponding formulae are 10−|ζ | ω0 and 10+|ζ | ω0
which comes to the same thing if |ζ | = 1 (its maximum value) but not
otherwise.

Figure 2.1

Analysis of Circuits c Imperial College London 15/19

3. In the circuit of Fig. 3.1, the input, X, is driven by a voltage source as shown.
Y ( jω)
a) Derive an expression for the transfer function, X( jω) and determine the corner
frequencies in its magnitude response. [4]

1

−1
The circuit is a potential divider, and the impedance of 3R||C is 3R + jωC =
3R
1+3 jωRC so the transfer function is

Y ( jω) R 1 + 3 jωRC
= 3R
= .
X ( jω) R + 1+3 jωRC 4 + 3 jωRC

1
The numerator corner frequency is ωn = 3RC and the denominator corner fre-
4
quency is ωd = 3RC .
Most people did this correctly but quite a few people made algebraic errors. It
is helpful to write gains in terms of dimensionless terms (like ωRC or ω 2 LC)
because it is them easy to spot dimensional incompatibilities resulting for alge-
bra errors. Expressions like R + 3 or R + jωC should never occur because they
are not dimensionally consistent.
Several people implicitly took the impedance of a capacitor to be jωC instead
1
of jωC .
A few people wrote down KCL equations at Y (correct) and/or at X (incor-
rect). You cannot apply KCL at X because you do not know what current flows
through the voltage source.
A good final check is to verify that the transfer function gives the correct gain
at ω = 0 (i.e. with the capacitor an open circuit) and ω = ∞ (i.e. with the
capacitor a short circuit). This simple test will detect most algebra errors.

b) With the capacitor temporarily removed from the circuit, determine the Thévenin
equivalent voltage and resistance of the remainder of the circuit at the terminals
of the capacitor. [4]

To determine the Thévenin equivalent voltage, we assume no current flows

through the capacitor and determine the voltage across it as Vth = 0.75X.
To determine the Thévenin equivalent resistance, we connect the two grounds
together, short-circuit the voltage source and measure the resistance of the re-
sultant network, which consists of R and 3R in parallel. The resistance is there-
fore Rth = 0.75R.
Most got this right although quite a few gave the Thévenin resistance as 4R.A
few gave a dimensionally incorrect value such as 14 which cannot be right since
all the resistors in the circuit are proportional to R.

c) Derive the time constant of the circuit, τ, in two ways: (i) from the Thévenin
resistance found in part b) and (ii) from the denominator corner frequency found
in part a). [2]

1
The time constant is (i) RthC = 0.75RC or alternatively (ii) ωd = 0.75RC.

Analysis of Circuits c Imperial College London 16/19

 4C
Mostly correct. Again several answers were dimensionally wrong, e.g. τ = 3R
does not have dimensions of “time”.

d) If the input voltage, x(t), is given by

(
−2 for t < 0
x(t) = ,
+3 for t ≥ 0

determine an expression for the output waveform, y(t). Sketch its waveform
over approximately the range −τ ≤ t ≤ 4τ. [7]

Y
From part a), the DC gain of the circuit is X (0) = 0.25. For t < 0, y(t) =
−2 × 0.25 = −0.5.
For t ≥ 0, the steady state solution is ySS (t) = +3 × 0.25 = 0.75.
We can calculate y(0+) in two ways:
(i) by ensuring the capacitor voltage, y(t) − x(t), does not change instantly and
(ii) by noting that y(0+) = y(0−) + YX (∞) × (x(0+) − x(0−)).
Using method (i), at t = 0− the capacitor voltage is y(0−) − x(0−) = −0.5 −
(−2) = 1.5. At t = 0+, we therefore still have 1.5 = y(0+) − x(0+) = y(0+) −
3. From this we get y(0+) = 1.5 + 3 = 4.5.
Alternatively, using method (ii) we have y(0+) = y(0−)+ YX (∞)×(x(0+) − x(0−)) =
−0.5 + 1 × (3 − (−2)) = 4.5.
t
For t ≥ 0, the output is therefore given by y(t) = ySS (t)+(y(0+) − ySS (0+)) e− τ =
t
0.75 + 3.75e− τ where τ = 0.75RC from part c). The dashed line in the plot be-
low shows the asymptote y(+∞) = 0.75.

4
3
y(t)

2
1
0

-1 0 1 2 3 4
Time (t/τ)

Many people assumed that the output was continuous (i.e. had no discontinu-
ity at t = 0). The most common reason given for this was that the capacitor
voltage could not change instantly; although this true, it would only force Y to
be continuous if the capacitor were connected between Y and ground which it
isn’t.
A few people gave complex values for the steady-state value of y(t) and/or the
transient amplitude; these quantities are always real-valued.
Several people gave the formula for y(t) for t ≥ 0 but did not say explicitly what
it was for t < 0.

e) Assuming that the opamp in Fig. 3.2 is ideal, determine the transfer function,
V ( jω)
U( jω) . [4]

Analysis of Circuits c Imperial College London 17/19

 From the standard gain of a non-inverting amplifier, the gain is UV = 1 + RZ =
R+Z
R where Z is the impedance of the 3R||C combination. Note that this is just
the reciprocal of the gain YX . Making use of the previous result, we therefore
V ( jω) 4+3 jωRC
have U( jω) = 1+3 jωRC . Alternatively, we can regard the circuit as a potential
divider with V as the input and U as the output (since the inverting input of the
opamp is constrained by negative feedback to equal U). Its transfer function is
thus the inverse of Fig. 3.1. Yet another way is to use nodal analysis by doing
KCL at the junction between the R and 3Rresistors (noting that the voltage
at this node is equal to U) to get: UR + U−V 3R + (U − V ) jωC = 0 from which
3U + U − V + (U − V ) j3ωRC = 0 ⇒ U(4 + j3ωRC) = V (1 + j3ωRC) and
the result follows.
Some gave the transfer function as UV ( jω) = 1 + 1+3 3jωRC which is correct (and
got full marks) but it is more conventional to write transfer functions as a ra-
tional polynomial in jω (i.e. one polynomial divided by another).
Many people included an extra factor of R in both numerator and denominator:
V ( jω) 4R+3 jωR2C
U( jω) = R+3 jωR2C . This is, of course, still correct but is a bit lazy and makes
for a lot more calculation in part g).

f) By considering the voltage across the capacitor, explain why an input voltage
discontinuity of ∆ u will result in an output voltage discontinuity of the same
amplitude. [2]

If u(t) suddenly changes by ∆ u, then negative feedback will ensure that the
inverting input of the opamp changes by the same amount and, since the voltage
across the capacitor cannot change instantly, V , must jump by the same amount.
Many people said (correctly) that this followed from the gain at ω = ∞: UV ( j∞) =
1. However the question asked you to consider the voltage across the capacitor.
Relatively few people got the chain of cause and effect right: the input discon-
tinuity causes a change at V+ and the negative feedback then adjusts V so that
V− becomes equal to V+ .
In many cases, the answer given was rather vague and did not form a logical
explanation; it is not sufficient to say “voltage across capacitor cannot change
instantly” without relating the voltage across the capacitor explicitly to the
input and output voltages.

g) If R = 20 kΩ, C = 20 nF and the input voltage, u(t), is given by

(
sin 1000t for t < 0
u(t) = ,
2 cos 2000t for t ≥ 0

determine expressions for the output v(t) for both positive and negative t. [ 7 ]

V ( jω) 4+3 jωRC

From part e), the transfer function is U( jω) = 1+3 jωRC with RC = 4 × 10−4 .

At ω1 = 1000, ω1 RC = 0.4 and so UV ( jω1 ) = 4+1.2 j 544 360

1+1.2 j = 244 − 244 j = 2.23 −
1.475 j. Using phasors, for t < 0, U1 = − j and so V1 = − j × (2.23 − 1.475 j) =
−1.475−2.23 j. Hence, for t < 0, we have v(t) = −1.475 cos 1000t +2.23 sin 1000t.

Analysis of Circuits c Imperial College London 18/19

 At ω2 = 2000, ω2 RC = 0.8 and so UV ( jω2 ) = 4+2.4 j 976 720
1+2.4 j = 676 − 676 j = 1.44 −
1.065 j. Using phasors, for t ≥ 0, U2 = 2 and so V2 = 2 × (1.44 − 1.065 j) =
2.89 − 2.13 j. Hence v2,SS (t) = 2.89 cos 2000t + 2.13 sin 2000t.
To determine the transient amplitude, we note that v(0−) = −1.475, vSS (0+) =
2.89 and that ∆ v = 1 × ∆ u = 2. Thus v(0+) = v(0−) + ∆ v = −1.475 + 2 =
0.525 and the transient amplitude is 0.525 − 2.89 = −2.365.
t
Thus, for t ≥ 0, we have v2,SS (t) = 2.89 cos 2000t + 2.13 sin 2000t − 2.365e− τ
where τ = 3RC = 1.2 ms.
Several people got the wrong answer when evaluating 4+1.2 j
1+1.2 j ; it is well worth
learning to use the complex arithmetic capabilities of the provided calculator
to save time and reduce errors.
A few people wrote things like v(t) = 4+1.2 j
1+1.2 j sin 1000t which does not make
sense: never mix j and t in the same expression. Phasors are complex but
do not involve t while waveforms are real and do not involve j.
Note that u(0+) is the value of the waveform u(t) at the specific time t = 0+
(where the + means just a tiny bit after t = 0). Its value must therefore be a
specific voltage and must be a real number that does not depend on t. Likewise,
the transient amplitude is given by A = v(0+) − vSS (0+) is also a real number
that does not depend on t.
Rather than using phasors, some people took the gain to be 4 (i.e. the DC
gain) even thought he input signal was a sine wave. This may arise from con-
fusion between “steady state” and “DC”; for t > 0 the “steady state” input is
a continuous cosine wave: u(t) = 2 cos 2000t.
Giving an “exact” answer like v(t) = − 360 544
244 cos 1000t + 244 sin 1000t makes no
sense at all in an engineering problem where the component values and fre-
quencies are only ever approximate. Much better to give the answer in decimal
to, say, 3 significant figures.

Figure 3.1 Figure 3.2

Analysis of Circuits c Imperial College London 19/19