University of British Columbia

Homework #4

Signals & Systems

ELEC 221

Instructor: Christos Thrampoulidis

First & Last Name:

Student Number:

For instructor use:

Question: 1 2 3 4 Total
Points: 8 12 11 14 45
Score:
Instructions:
• Write your name in the space provided above.

• Please consult the Syllabus for HW grading and late-submission policies.

• The HW is due Tuesday March 28, 11:59pm sharp.

Page 2
1. Problem 1[Warm up]
(a) (2 points) Find and plot the Fourier transform of δ(t − 5).

(b) (2 points) For arbitrary signal x(t) show that if x(t) is real and even, then X(jω) is a real and even
function of ω.

Page 3
(c) (2 points) For arbitrary signal x(t) show that if x(t) is real and odd, then X(jω) is an imaginary
and odd function of ω.

(d) (2 points) Let y(t) be the output of an LTI system with R ∞impulse
R ∞ response h(t) and input x(t).
Using the fact that y(t) = (x ∗ h)(t), show that Y (jω) = −∞ −∞ x(τ )h(t − τ )e−jωt dτ dt. Use this
and an appropriate change of variables to show that Y (jω) = X(jω)H(jω).

Page 4
2. Problem 2[Fourier Transform of periodic signals]
In this problem, we explore the definition of the FT of a periodic signal (this is analogous to what we
did in class for the DTFT a couple of weeks ago).
(a) (1 point) Show that the FT is linear, i.e. if x3 (t) = αx1 (t) + βx2 (t), then X3 (jω) = αX1 (jω) +
βX2 (jω).

(b) (1 point) Verify that Z ∞

1
ejω0 t = 2πδ(ω − ω0 )ejωt dω.
2π −∞

From this observation argue that the FT of ejω0 t is 2πδ(ω − ω0 ).

Page 5
(c) (3 points) Let x(t) be T -periodic and let {Xk }∞
k=−∞ denote the signal’s Fourier coefficients. Write
the Fourier series formula expressing x(t) as a superposition of complex exponentials. By taking
the Fourier transform of both sides of the Fourier series equation and using results of the previous
two parts, show that
∞  
X 2π
X(ω) = 2παk δ ω − k .
T
k=−∞

Page 6
 For the rest of this problem consider the following specific setting. The Fourier series coefficients of
∗ ∗
a periodic signal s, with fundamental frequency ω0 = π/4 are X+1 = X−1 = j, X5 = X−5 = 2 and
the rest are zero. Suppose that s(t) is the input to a band-pass filter with the following magnitude
and phase responses

1 ω ∈ [π, 1.5π]

|H(jω)| = 1 ω ∈ [−1.5π, −π] ∠H(jω) = −ω.

0 otherwise


Let y(t) be the output of the filter.

(d) (1 point) Determine the Fourier Series of s(t).

(e) (2 points) Find and plot the Fourier transform of s(t).

(f) (2 points) Find the filter output y(t).

Page 7
 (g) (2 points) Find and plot the Fourier transform of y(t).

3. Problem 3[Dirac comb].

P∞
Consider the following transformation from x(t) to y(t) where p(t) = k=−∞ δ(t − k) is an impulse train
sin( πt
2 )
(also called “Dirac comb”) and x(t) = πt .

(a) (1 point) Make a sketch of x(t).

Page 8
(b) (1 point) Compute and make a sketch of the FT X(jω) of x(t).

(c) (4 points) Compute and make a sketch of the FT Y (jω) of y(t).

Page 9
(d) (1 point) Compute and make a sketch of y(t).

sin( πt )
(e) (2 points) Consider a system with impulse response h(t) = x(t) = πt . Compute the step
2

response
( of the system, that is the output of the system when the input is the unit step function
1 t>0
u(t) = .
0 t<0

(f) (2 points) Make a sketch (or plot numerically, e.g. using Python) of the step response in the previous
part. If you do this correctly, you should see that the long-term final value of the step response s(t)
exhibits oscillatory behavior, unlike the steady behavior of the input u(t). This behavior is called
ringing and is often not desirable in system design practice.

Page 10
4. Problem [Bode plots].
Let causal LTI system be described by the following first-order linear coefficient differential equation
relating output y(t) to x(t):

dy(t)
τ + y(t) = x(t), (1)
dt
where the coefficient τ > 0 is a positive scalar.
Note: For the plots in this problem you can either directly attach (copy) a figure that you obtain say
using Python or you can sketch here on pen what you see in the numerics.
(a) (2 points) Compute the impulse response and the frequency response of the system.

(b) (1 point) Compute and sketch the step response of the system (see Problem 2f). Confirm that the
value of the step response at time τ is 1 − e−1 . Note that this means the rise time of the step
response becomes sharper as τ decreases. Also, compare the result of that of Problem 2f and check
whether the ringing phenomenon occurs here.

Page 11
(c) (1 point) Make a sketch of the system’s frequency response H(jω). Note that you can view this as
a (pretty poor) approximation to an ideal low-pass filter with cutoff frequency 1/τ .

In graphically displaying CT Fourier transforms and system frequency responses in polar form, it
is often convenient to use a Logarithmic scale for the magnitude of the FT. This allows detail to
be displayed over a wider dynamic range. Typically, the specific logarithmic amplitude scale used
is in units of 20 log10 , referred to as decibels (abbreviated dB). For CT systems, it is also useful to
use logarithmic scale for the frequencies. Plots of 20 log10 |H(jω)| and ∠H(jω) versus log10 (ω) are
referred to as Bode plots.
(d) (1 point) Argue that if the impulse response h(t) of a CT LTI system is a real signal, then plots of
both |H(jω)| and ∠H(jω) for negative ω can be obtained immediately from the plots for positive
ω. How?

Page 12
(e) (1 point) Prove that if we have a graph of the log magnitude (i.e. the logarithm of the magnitude) of
(i) the frequency response of an LTI CT system and of (ii) the Fourier transform of an input signal,
then the log magnitude of the output is obtained simply by adding the former two log magnitude
plots.

(f) (1 point) Show the Bode plots (both magnitude and phase) of the LTI system described by Equation
1. Make the plots for at least two values of τ (e.g. τ = 0.1, 0.5, 1, 2, 10).

Page 13
(g) (1 point) Argue analytically that for frequencies ω  1/τ , the log magnitude 20 log10 |H(jω)| is
approximately zero. On the other hand, when ω  1/τ then the log magnitude is a linear function
of log frequency. Specifically, in the latter case, show that 20 log10 |H(jω)| ≈ −20 log10 ω−20 log10 τ ,
that is the Bode plot has a slop of −20dB per decade (i.e. |H(jω)| decreases by 20dB for every
increase of a factor of 10.

(h) (1 point) Use your findings in the previous part to make a graphical sketch of the Bode plot of log
magnitude for general value of τ . What is the value of the graph at frequency 1/τ ? Confirm that
this explains why the point ω = 1/τ is sometimes called the “3-dB point” or the break frequency.

Page 14
 Consider now a second-order continuous time LTI system described by the following differential
equation for positive scalars ωn , ζ > 0

d2 y(t) dy(t)
2
+ 2ζωn + ωn2 y(t) = ωn2 x(t)
dt dt

(i) (1 point) Show that frequency response of the system above can be expressed as a function of ω/ωn
as follows:
1
H(jω) = 2 .
(jω/ωn ) + 2ζ (jω/ωn ) + 1
Thus, changing ωn is identical to a frequency scaling.

(j) (2 points) Make a Bode plot of the system’s frequency response for ωn = 1 and two values of ζ = 0.4
and ζ = 1.5.

Page 15
(k) (2 points) Similar to part (i) derive approximations for the log magnitude for low frequencies ω 
ωn and high frequencies ω  ωn . What is the slope of the Bode plot for large frequencies? Discuss
how your result compares to that of parts (i) and (j).

——— End of HW #4 ———

Page 16