6.

003 Homework #13 Solutions

Problems
1. Transformation
Consider the following transformation from x(t) to y(t):

w1 (t) w2 (t) w3 (t)

x(t) × × × + y(t)

p(t) cos(πt)
∞
X
where p(t) = δ(t−k). Determine an expression for y(t) when x(t) = sin(πt/2)/(πt).
k=−∞

X(jω)
1

ω
− π2 π
2

W1 (jω)
1/2

ω
−π π

W2 (jω)
1/2

ω
−π π

W3 (jω)
1/2

ω
−π π

Y (jω)
1/2

ω
1
y(t) = δ(t)
2
6.003 Homework #13 Solutions / Fall 2011 2
2. Multiplied Sampling
The Fourier transform of a signal xa (t) is given below.
Xa (jω)
1

ω
π π
−
2 2
This signal passes through the following system
xb (t) uniform xc [n] xd [n] sample-to- xe (t)
xa (t) × K H(jω) xf (t)
sampler impulse

cos (7πt)

where xc [n] = xb (nT ) and

∞
X
xe (t) = xd [n]δ(t − nT )
n=−∞

and
T if |ω| < Tπ
n
H(jω) =
0 otherwise .

a. Sketch the Fourier transform of xf (t) for the case when K = 1 and T = 1.
Xf (jω)

ω
π π
−
2 2

Use your sketch to determine an expression for Xf (jω) for the following intervals:

0 < ω < π/2: 0

π/2 < ω < π: 2ω/π − 1

6.003 Homework #13 Solutions / Fall 2011 3

π < ω < 3π/2: 0

3π/2 < ω < 2π: 0

b. Is it possible to adjust T and K so that xf (t) = xa (t)?

If yes, specify a value T and the corresponding value of K (there may be multiple
solutions, you need only specify one of them). If no, write none.
2 4 6 8 10 12
T = , , , , , , or2
7 7 7 7 7 7

K=1

c. Is it possible to adjust T and K so that the Fourier transform of xf (t) is equal to the
following, and is zero outside the indicated range?
Xf (jω)
1

ω
3π −π π π π 3π
− −
2 2 2 2
If yes, specify all possible pairs of T and K that work in the table below. If there
are more rows in the table than are needed, leave the remaining entries blank. If no,
enter none.
T K
1
3 2

2
3 2

1
4 2

1
2 2
6.003 Homework #13 Solutions / Fall 2011 4
3. Patterns
The time waveforms for six signals are shown in the left panels below. The right panels
show the magnitudes of the Fourier transforms of x1 (t) to x6 (t), however, the order has
been shuffled. For each panel on the left, find the corresponding panel on the right.
All of the time functions are plotted on the same time scale. Similarly, all of the frequency
functions are plotted on the same frequency scale.
x1 (t) |XA (jω)|

x2 (t) |XB (jω)|

x3 (t) |XC (jω)|

x4 (t) |XD (jω)|

x5 (t) |XE (jω)|

x6 (t) |XF (jω)|

ω
6.003 Homework #13 Solutions / Fall 2011 5
1 C
2 A
3 E
4 B
5 D
6 F
6.003 Homework #13 Solutions / Fall 2011 6
4. Inputs and Outputs
A causal, stable LTI system with frequency response H(jω) has input x(t) and output
y(t). The problem is to determine which of the following inputs can or cannot give rise
to the output y(t) = sin(2π · 100 · t). For each part of the problem, determine if the
statement is True (T) or False (F) and give an explanation.

Part a. x1 (t) is a periodic impulse train of period 0.05 s.

x1 (t)

t
−0.1 −0.05 0 0.05 0.1 0.15
(T or F) x1 (t) can generate the response y(t) = sin(2π · 100 · t).
True. Since x(t) has a period of 0.05 seconds, the fundamental frequency is 20 Hz (=
2π × 20 radians/second). The Fourier transform of x(t) is an impulse train with impulses
at integer multiples of 20 Hz. Thus, the fifth harmonic occurs at 100 Hz. Thus, H(jω)
could be a narrowband filter centered at 100 Hz with a phase shift of −π/2 at 100 Hz.
That would produce the desired output. However, this filter is not causal. An alternative
is to construct a filter with zeros at all the unwanted frequencies.

Part b. x2 (t) is a periodic function of period 0.11 s. Each period consists of five cycles
of a sinewave of the form sin(2π · 100 · t).
x2 (t)

−0.1 −0.05 0 0.05 0.1 0.15

(T or F) x2 (t) can generate the response y(t) = sin(2π · 100 · t).
True. Since the period of x2 (t) is 0.11s, the fundamental frequency is 1/0.11 Hz. Thus,
x2 (t) will contain impulses at the frequencies f = k/0.11 Hz and the 11th harmonic will
appear at the frequency f = 100 Hz. The only remaining issue is whether the area of
this impulse is non-zero. The pulse of sinusoid has a spectrum which is a sinc function
centered on 100 Hz. Hence, its value at 100 Hz is non-zero and so the amplitude of the
impulse at 100 Hz is also non-zero. The filter can be chosen as indicated in part a.

Part c. x3 (t) is a periodic pulse train of period 0.02 s. Each pulse has duration 0.004 s.
x3 (t)

t
−0.04 −0.02 0 0.02 0.04 0.06
6.003 Homework #13 Solutions / Fall 2011 7
(T or F) x3 (t) can generate the response y(t) = sin(2π · 100 · t).
True. x3 (t) can be represented by a uniform impulse train of period 0.02 s convolved with
a rectangular pulse of duration 0.004 s. Thus, the Fourier transform of x3 (t) is a uniform
impulse train, whose period in frequency is 1/0.02 = 50 Hz, multiplied by a sinc function.
Thus, there are clearly impulses at the frequencies ±100, the only issue is whether the sinc
function has a zero at 100 Hz. Since the duration of the rectangular pulse is 0.004 s, the
first zero of the sinc function is at 250 Hz. The filter can be chosen as indicated in part a.

Part d. x4 (t) is a periodic sinc pulse train of period 0.1 s. Each sinc pulse has the
formula
t
sin(π · 0.006 )
t
π · 0.006

x4 (t)

t
−0.2 −0.1 0 0.1 0.2 0.3
(T or F) x4 (t) can generate the response y(t) = sin(2π · 100 · t).
False. x4 (t) is a uniform impulse train of period 0.1 s convolved with a sinc function.
Hence, the Fourier transform of x4 (t) is the product of an impulse train in frequency with
an ideal lowpass filter. Since the period of x4 (t) is 0.1 s, the fundamental frequencey is 10
Hz and there will be an impulse at 100 Hz in the Fourier transform of the impulse train.
But, the total width of the ideal lowpass filter is 1/0.006 = 166.66 Hz so that the passband
is from −83.33 to +83.33 Hz. Thus, there is no component at 100 Hz at the input to the
filter H(jω) and x4 (t) cannot generate the output y(t).

Part e. x5 (t) is a periodic triangular wave of period 0.02 s.

x5 (t)

t
−0.04 −0.02 0.02 0.04 0.06

(T or F) x5 (t) can generate the response y(t) = sin(2π · 100 · t).

False. x5 (t) is a triangular wave which has half-wave symmetry and therefore its even
harmonics are zero. Since the fundamental frequency is 50 Hz, the component at 100 Hz
is the second harmonic and its magnitude must be zero. An alternative approach is to
recognize that the triangular wave is the convolution of a periodic impulse train in time
with a triangular pulse. The triangular pulse has duration 0.02 s and can be generated by
convolving a square pulse of duration 0.01 s with itself. Thus, the Fourier transform is the
squared sinc function. But the sinc function has a zero at 100 Hz. Hence, the component
at 100 Hz at the input to the filter has a magnitude of zero, and x5 (t) cannot give rise to
y(t).
6.003 Homework #13 Solutions / Fall 2011 8
5. DT Radio Demodulation
Commercial AM radio stations broadcast radio frequencies within a limited range:
2π(fc − 5 kHz) < ω < 2π(fc + 5 kHz), where fc = ωc /(2π) = n × 10 kHz and n is an
integer between 54 and 160. The system shown below is intended to decode one of the
AM radio signals using DT signal processing methods. Assume that all of the filters are
ideal.

Part a. Determine the center frequency fc for the AM station that this receiver will
detect.
1
Ω π
ω= = 4 −7
T 10
ω 1
f= = × 107 = 1.25 MHz
2π 8

Part b. Which of the following statement(s) is/are correct?

b1. Increasing the cutoff frequency ωr of LPF1 by a factor of 1.5 will cause aliasing.
b2. Decreasing the cutoff frequency ωr of LPF1 by a factor of 2 will have no effect on
the output yr (t).
b3. Halving the sampling interval T would have no effect on the output yr (t).
b4. Doubling the sampling interval T would have no effect on the output yr (t).
b2

Part c. Which of the following statement(s) is/are correct?

c1. Increasing the cutoff frequency Ωd of LPF2 will change yr (t) by adding signals
from unwanted radio stations.
c2. Increasing the cutoff frequency Ωd of LPF2 will change yr (t) because aliasing will
occur.
6.003 Homework #13 Solutions / Fall 2011 9
c3. Doubling the cutoff frequency Ωd of LPF2 will have no effect on yr (t).
c4. Halving the cutoff frequency Ωd of LPF2 will have no effect on yr (t).
c3