ELEC2103 Simulation & Numerical Solutions in Engineering – 2015

Laboratory 5 Laplace Transforms & Circuits 2; random

numbers.
This laboratory covers an introduction to the Laplace transform and continues with circuit
analysis from the last laboratory. There is a little more on symbolic maths and a brief section
on random numbers.
References regarding circuit analysis are to “Introduction to Electric Circuits”, 6th edition,
Wiley, 2004, by Dorf and Svoboda (DS).

Solving Differential Equations

The dsolve function computes solutions to ordinary differential equations. An
example:
>> dsolve('Dy=y')
ans =
C1*exp(t)
Note the use of Dx for the derivative of x. Similarly, D2x represents the second
derivative, etc. Note also the use of t as the default independent variable and of C1 as
an arbitrary constant. We can specify an initial condition and/or change the
independent variable:
>> dsolve('Dy=y', 'y(0)=5', 'x')
ans =
5*exp(x)
If there are several equations in several variables, the solution is returned as a
structure:
>> S = dsolve('Dx+x=2*y', 'Dy+y=0', 'x(0)=1', 'y(1)=2')
S=
x: [1x1 sym]
y: [1x1 sym]
The solutions are the fields of the structure S. Thus
>> S.x
ans =
exp(-t)*(1+4*t/exp(-1))

Laboratory 5 1
ELEC2103 Simulation & Numerical Solutions in Engineering – 2015

The Laplace Transform

The Symbolic Maths Toolbox has functions to handle the Laplace, Fourier and Z-
transforms. In this laboratory, we will consider only the Laplace transform, which is
important in linear systems analysis (including circuit analysis). The Laplace
transform of the function f (t ) is defined by (DS 14.3)


 f (t )e
 st
F (s)  dt
0

Usually, t represents time and s the complex frequency s    j . The derivation of

the Laplace transform is carried out in MATLAB with the laplace function. Some
examples:
>> laplace(sym(1))
ans =
1/s
>> laplace(t^3)
ans =
6/s^4
>> laplace(cos(2*t))
ans =
s/(s^2+4)
MATLAB uses the variable s by default. The inverse Laplace integral (DS 14.5-the
equation is omitted here) is obtained using the function ilaplace. For example:
>> syms s
>> ilaplace(1/s)
ans =
1
Note that, as can be seen from the defining integral, the value of f (t ) for t < 0 has no
effect on F ( s ) ; however, mathematically the formal inverse Laplace transform
produces a function f (t ) which is zero in this range. The function ilaplace returns a
function of t which is correct for positive t but not expressly zero for negative t.
Here are some examples which illustrate some interesting points. Suppose we wish to
find the f (t ) which has the Laplace transform
s
F (s)  .
s  5s  16s  30
3 2

We can use ilaplace:

>> syms s
>> F = s/(s^3+5*s^2+16*s+30);
>> f = ilaplace(F);
>> pretty(f)
/ sin(3 t) 7 \
exp(-t) | cos(3 t) + ---------- | 3
\ 9 / exp(-3 t) 3
----------------------------------- - -----------
13 13

Laboratory 5 2
ELEC2103 Simulation & Numerical Solutions in Engineering – 2015

If we now take the Laplace transform of the result:

>> F = laplace(f);
>> pretty(F)
1 s+1 1
- 3/13 ----- + 3/13 ------------ + 7/13 ------------
s+3 2 2
(s + 1) + 9 (s + 1) + 9
we obtain the original function of F (s ) as (a form of) partial fractions, each term
corresponding to a term of f (t ) . (Note: you may get a slightly different display if you
try this yourself.)
Suppose, however, that
8s 2
F (s)  .
16s 3  30s 2  18s  3
When we use ilaplace to find f (t ) , the result is:
8*symsum((r3^2*exp(r3*t))/(6*(8*r3^2 + 10*r3 + 3)), r3 in RootOf(s3^3
+ (15*s3^2)/8 + (9*s3)/8 + 3/16, s3))
Note that pretty is not used here because it’s harder to read. The answer is a symbolic
sum over the three roots of the cubic equation in s3. The different behaviour occurs
because, in this example, the roots of the denominator polynomial are long
expressions involving cube roots, and the result is complicated. You can check this
with
>> syms s3
>> solve(s3^3 + (15*s3^2)/8 + (9*s3)/8 + 3/16, s3)
(Note that the “=0” is assumed.)
Problems with f in this form are that an error occurs if we attempt to plot it using
ezplot. To get around these problems, we can use variable precision arithmetic with
the vpa function, and replace numbers in f by approximations to a specified number
of decimal digits. Thus
>>f1 = vpa(f, 5);
>>f2 = simple(f1);
>>f3 = vpa(f2,5)
f3 =
0.1136*exp(-0.27229*t) + 0.3864*exp(-0.80135*t)*cos(0.21547*t) -
2.7704*exp(-0.80135*t)*sin(0.21547*t)
and f3 can be plotted with ezplot. Note that simple will give a warning suggesting
you use simplify, but if you do, you may get an answer without the sin and cos
simplification.
Connection between Laplace Transforms and circuit analysis
In the analysis of circuits with sources which vary in an arbitrary manner with time, it
is necessary to solve differential equations, either directly or indirectly. The Laplace
transform (DS Chapter 14) is a powerful tool which allows analysis to be carried out

Laboratory 5 3
ELEC2103 Simulation & Numerical Solutions in Engineering – 2015

in the complex frequency domain (in terms of the complex frequency s) rather than
the time domain (in terms of time t).
Voltages v(t ) and currents i (t ) are replaced by their Laplace transforms V (s ) and
I (s ) , and the impedances of circuit elements are expressed in terms of s as follows:
a resistance of R ohms has an impedance of R ohms;
an inductance of L Henries has an impedance of sL ohms; and
a capacitance of C Farads has an impedance of 1/(sC) ohms.
Rules for combining impedances in series, etc, are as for resistors in dc analysis, and
many of the techniques of dc analysis can be used.
Relationship between Laplace analysis, dc analysis and complex phasor
analysis
Suppose we have a circuit consisting of R, L and C elements and we have obtained
general results in terms of Laplace transforms. Then
Results for steady state dc conditions can be obtained by taking s  0 , and
Results for steady state sinusoidal conditions can be obtained by substituting
s  j where  is the frequency in rad/s.

Exercises
Exercise 1
Use dsolve to find solutions for f (x) and g (x) if
df ( x )
 1.5 f ( x )  g ( x )  0
dx
dg ( x )
 f ( x)  2g( x)  0
dx
with f (0)  0, g (0)  2 . You may find it useful to use vpa and simplify to provide a
compact display on the screen. Note that the current Matlab versions of simple and
simplify may give odd looking results.
Exercise 2
x(t ) 1
2 H
+

t0
+

1
200 F 10 v(t )
e(t ) 1
100 F
- -

In the circuit shown, the 1/100 F capacitor is charged to 5 V with the switch open and
there is no charge on the 1/200 F capacitor. At t = 0, the switch is closed. Write down
three differential equations in terms of the voltages e(t ), v(t ) and the current x(t ) . Use

Laboratory 5 4
ELEC2103 Simulation & Numerical Solutions in Engineering – 2015

MATLAB’s dsolve function to find a solution for e(t ), t  0 and plot the result over
suitable ranges. Use vpa and pretty to display a compact readable solution on the
screen.
Exercise 3
In the circuit shown, component values are in , F and H.
3 1
2 2

+
+

e1 (t ) 4
3
1 e2 (t )
-
-

Use Laplace transform methods to determine the step response of the network shown,
i.e., if e1 (t ) is the unit step function ( e1 (t )  0, t  0 and e1 (t )  1, t  0 ), how does
the output voltage e2 (t ) behave for t  0 ? Such an excitation could be caused by a dc
voltage source switched into the circuit at t = 0. Assume that, initially, the capacitor is
uncharged and the inductors carry no current. Use the method of mesh currents (DS
4.6). Steps in solving the problem are:
(a) Redraw the circuit, showing all elements in the complex impedance form (in
terms of s), and show the voltages as their Laplace transforms.
(b) On the same diagram, show the loop currents (Laplace transform form) as
I 1 ( s ), I 2 ( s ) , say.
(c) Write the equations required to solve for E 2 ( s ), I 1 ( s ), I 2 ( s ) in terms of E1 ( s ) .
(d) Write a script M-file that solves these equations, substitutes the Laplace
transform of E1 ( s ) to find the Laplace transform of E 2 ( s ) , finds e2 (t ) by
taking the inverse transform and plots e2 (t ) for a suitable range of positive t.
Exercise 4
In Exercise 3, the ratio G ( s )  E 2 ( s ) / E1 ( s ) of Laplace transform of output to Laplace
transform of input is the voltage gain. (In the more general setting of linear systems
analysis, such functions are called transfer functions). In the special case when
s  j , G ( j ) is referred to as the frequency response.
Rewrite the M-file of Exercise 3 to calculate and plot the frequency response of the
network, i.e., plot G ( j ) versus . This circuit, in fact, acts as a low pass filter. Do
your results support this description?
Exercise 5
An important class of electronic circuits involves operational amplifiers. Consider the
problem, in the circuit shown, of determining the voltage gain E 2 ( s ) / E1 ( s ) . We will
assume that the op amp has (i) infinite input impedance and (ii) infinite voltage gain.
These are often very reasonable assumptions.

Laboratory 5 5
ELEC2103 Simulation & Numerical Solutions in Engineering – 2015

C2

R3
R1 R2 C1

E2
E1

Because of the first assumption, no current flows into the + input terminal. Because of
the second, since E 2 is finite, the differential input voltage is zero and the + input is
at a “virtual ground” potential. Thus, to determine the voltages and currents, we need
to analyse the following circuit:
C2

R3
R1 R2 C1

V E2
E1

Find the voltage gain. As good a way as any is to use Kirchhoff’s laws to write and
solve two equations in V and E 2 .

Exercise 6: Random Numbers

Computer simulations frequently require the use of random numbers. The idea of
using a deterministic machine to produce random numbers seems to be an oxymoron.
Indeed, modern computers generate pseudo-random numbers. These pseudo-random
numbers pass a series of statistical tests that indicate the numbers “appear” random. In
the early days of computers, John von Neumann suggested the middle-square method
for generating random numbers. This method starts with an initial four-digit integer
called the seed, squares it to get an eight-digit number, and then takes the middle four
digits as the next random number. This will, of course, only produce random numbers
between 0 and 9999. MATLAB’s random number generator, on the other hand, can
theoretically produce 21492 different random numbers before repeating itself (the 1492
has led some to call this random number generator the “Christopher Columbus”
generator). Amazingly, this generator uses shift register and bit manipulations that
require no multiplication or division operations. Even if we were to consume random
numbers at a rate of one million per second, it would take more than 10435 years (about
10425 times the age of the universe) before the sequence would repeat. (Note that this
is only true for MATLAB v.5 and greater.)
Uniform random numbers between 0 and 1 and also matrices of random numbers are
generated in MATLAB using the rand function:
>> rand(10,10).

Laboratory 5 6
ELEC2103 Simulation & Numerical Solutions in Engineering – 2015

The command ‘randn’ produces a Gaussian distribution of random numbers with a

mean of zero and variance of one:
>> randn(10,10).
Unless the initial seed is changed, MATLAB’s random number generator produces
the same random sequence of numbers for every execution. Therefore, the following
command can be used to change the initial seed:
>> rng(100*sum(clock)).
This command uses the software clock maintained by MATLAB to compute an initial
seed for the generator. The variable clock is a six-element vector with the format of
[year month day hour minute second]. Try typing ‘clock’ at the MATLAB prompt.
The command ‘sum’ adds the six elements of the vector and the multiplication by 100
results in a final integer to seed the random number generator.

Calculating pi
One way to calculate  is to randomly and uniformly pick points in the unit square
with vertices (0,0), (1,0), (0,1), (1,1) and calculate the proportion of points which lie
within unit distance of the origin. This number will be proportional to the value of  .
Why? Write a MATLAB function to calculate the value of  using this method. Let
the number of points to pick be an input to your function. This program can be written
with or without a for loop. Note that this method is a very inefficient way to calculate
the value of  . Nonetheless, the procedure outlined here is the “essence” of the
Monte Carlo techniques found in PSPICE that are used in simulating the influence of
the variations of design parameters on integrated circuits.

Laboratory 5 7