You are on page 1of 26

Practice Exam #1

Challenge 7
Practice Quiz (Lessons 1-7)
Project 1 due on, or before, Friday (class time). Project 1
hardcopy due a week from Friday or earlier. I will be in-class
Friday for Project 1 collection only (no lesson).
Exam #1 TA hours: Sept. 14, NEB501, 1:00- 2:00.

Practice Exam #1

Challenge 07
A signal having a z-transform:
X(z)=(1+0.4 2 z-1)/(1- 0.8 2 z-1 + 0.64 z-2)
is to be transformed into ae discrete-time domain signal having the
form:
x[k] = A( a cos(1)+ b sin(2)); ().
What is x[k] (i.e., inverse z-transform)?
------ This is a computationally intensive problem with a level of
difficulty far greater than a typical examination problem but it provides
an opportunity to show off. -----------------------

Practice Exam #1

Using the slacker friend residuez: (could use manual or residue as well)
-1
-1
-2
>> b=[1, 0.4*sqrt(2)]; a=[1, -0.8*sqrt(2), 0.64]; % X(z)=(1+0.4 2 z )/(1- 0.8 2 z + 0.64 z )
>> [R,p,C]=residuez(b,a)
R=
0.845
0.5000 - 1.0000i
% R1, R2
0.5000 + 1.0000i
% Residues (Heaviside coefficients)

p=
0.5657 + 0.5657i
% p1, p2
0.5657 - 0.5657i
% Pole locations (distinct)
*
C=
[]
0.8-45
>> Mp=abs(p')
% pole magnitude
Mp =
0.8000 0.8000
>> Ap=angle(p')/pi
% phase angles x pi
Ap =
-0.2500 0.2500
% /4 or 45

Practice Exam #1

Therefore, the Heaviside expansion of X(z) is:


R2
R1
(0.5 j1) 1
(0.5 j1) 1
X ( z)

;
j / 4 1
j / 4 1
1 | 0.8 | e
z ) 1 | 0.8 | e
z )
p2

p1

Refer to Table 1 of Lesson 6 (Inverse z-Transform) and Eulers


equation. Noting:
1/(1-z-1) = z/(z- k u[k], an eigenvalue

x[k ] 0.5 j1 | 0.8 |k e jk / 4 u[k ] 0.5 j1 | 0.8 |k e jk / 4 u[k ]

Practice Exam #1

Combine reals, combine imaginaries.

x[k ] 0.5 j1 0.8 e jk / 4 u[k ] 0.5 j1 0.8 e jk / 4 u[k ]


k

1 jk / 4
jk / 4
jk / 4
jk / 4
0.8 e
e
je
e
u[k ]
2

0.8

cos k / 4 2 sin k / 4
Im just sayin.

Practice Exam #1

If you are really lazy, you may have tried to perform a symbolic study as
shown below.
>> syms z
>> f=(z^2 - (0.4*1.414)*z )/(z^2 -(0.8*1.414)*z + 0.64)
f=
(z^2-707/1250*z)/(z^2-707/625*z+16/25)
>> iztrans(f) % inverse z-transform
ans =
sum(1/2*(1/_alpha)^n,_alpha = RootOf(625-707*_Z+400*_Z^2))
??????????????
This answer is basically useless.
For exams, dont rely on MATLAB. I have my ways to frustrate you.

Practice Exam #1

EEL 5525 Exam #1


Fall 2015
Lectures 1-7
Name: ____________________________________
ID: _______________________________________
Instructions:
Complete the exam on the space provided. When time is called, stop
all work and follow the instructions provided. Any work that is not
collected when called for will not be graded. The honor system will be
strictly enforced.
Allowed

Open book
Open notes
Pocket calculators and laptops
Figures and tables
Internet access to Sakai (only)
Disallowed
Unauthorized Internet access
Cell phones
7

Practice Exam #1

Sample
Exam Cover
Sheet

Question 1
Q1: Sampling Theorem and Quantization (Similar to Challenge 04)
You are to analyze the audio recording and playback system shown below.
The input audio frequency range is f[0, 3.5] kHz. The listeners hearing
range is f[0, 8] kHz. The ADC operates at the programmable sample rate
of fs = n8kHz, n an integer.

Record

x(t)
ADC

f[0, 3.5] kHz

fs = n8kHz

Practice Exam #1

Playback
Memory

x[k]

f8 kHz

DAC

Interpolator

y(t)

Question 1
a. What is the lowest sampling frequency fs that will insure that the
original audio signal x(t) can be (theoretically) reconstructed from its
time-series samples x[k] without aliasing?
fs>2B = (2)(3.5kHz) = 7kHz; therefore fs=8kHz, i.e., n=1.
Nyquist sample rate = 7 kSa/a
Nyquist frequency = 3.5 kHz

Practice Exam #1

Question 1
b- The studio is presumed quiet when you begin recording at a sample
rate of fs=8k Sa/s. When played back you hear a 2k Hz buzzing
sound in the captured signal. What is the expected minimum
frequency of the extraneous signal source?
The possible causes of the 2kHz buzz, sampled at fs=8kSa/s, are f=
2kHz, 2k-8k=-6kHz, 2k+8k=10kHz, 2k-16k =-14kHz, 2k+16k=18Hz,
(etc.). You would have heard both the 2kHz and 6kHz tone before
recording but not the others. Therefore the lowest of these remaining
aliased tones is 10kHz > 8kHz.
2k, -6k, 10k, -14k, 18k,
4k, 12k,
0, 8k, 16k,
-2k, 6k, -10k, 14k, -18k,
10

Practice Exam #1

Question 1
c- You sample at fs=8k Sa/s, but now decide to place an ideal audio
analog lowpass anti-aliasing filter in front of the ADC. What should
be the filters passband cut-off frequency?
f0 3.5kHz , also accept fs/2=4kHz.

Audio input range


0

3.5 k 4.0 k
Nyquist frequency

11

Practice Exam #1

Question 1
d- The signed 16V ADC provides an 8-bit output. What is the ADCs
quantization step size in V/bit?

= (2)(16) /28 = 25/28 (Quantization step-size)


= 2-3 V/bit

From other statistics follow.

12

Practice Exam #1

Question 2
z-Transforms
Table: Primitive Signals and their z-Transform
Time-domain
z-transform
[k]
1
u[k]
z/(z1)
aku[k]
z/(za)
kaku[k]
az/(za)2
You are studying a causal signal x[k] that has a z-transform:
X(z)= (z+1)2/((z-1)(z-0.5)2).
The signal has an assumed Heaviside expansion given by:
X(z)= (z+1)2/((z-1)(z-0.5)2 ) = A + Bz/(z-1) + Cz/(z-0.5) + Dz/(z-0.5)2.
What are the Heaviside coefficients?
13

Practice Exam #1

Question 2
Using X(z) = X(z)/z (in-class old school method).
the Heaviside expansion is performed on:
X(z) = X(z)/z = (z+1)2/(z(z-1)(z-0.5)2 )= A/z + B/(z-1) + C/(z-0.5) + D/(z-0.5)2

14

Practice Exam #1

Question 2
X(z) = X(z)/z = (z+1)2/(z(z-1)(z-0.5)2 )= A/z + B/(z-1) + C/(z-0.5) + D/(z-0.5)2
a)

What is A?

b) What is B?

B = (z-1) X(z)/z|z=1 = (z+1)2/(z(z-0.5)2) |z=1


= 22/(1*(0.5)2) = 4/(1/4) = 16

c) What is C?

C = d((z-0.5)2X(z)/z)/dz|z=0.5 = d((z+1)2/(z(z-1)))/dz |z=0.5


= [2(z+1)/(z(z-1)) -(z+1)2(2z-1)/(z(z-1))2)|z=0.5 =
= 2(3/2)/(1/2*-1/2) - (3/2)2(1-1) )/(1/2 -1/2)2 = -12

d) What is D?

15

A = zX(z)/z|z=0 = 1/((-1)(-0.5)2) = -4

Practice Exam #1

D = (z-0.5)2X(z)/z|z=0.5 =(z+1)2/(z(z-1))|z=0.5
= (3/2)2/(1/2 -1/2) = -9

Question 2
Suppose X(z) = X(z)/z = A/z + B/(z-1) + C/(z-0.5) + D/(z-0.5)2 where A = -4, B =
16, C = -12, and D = -9. [Firebreak]
e)

What is x[k]?

Have X(z)=(z+1)2/(z-1)(z-0.5)2= -4 + 16z/(z-1) - 12z/(z-0.5) 9 z/(z-0.5)2 .


Interpret each term in terms of the elements of a z-transform table:
Example: kaku[k] az/(za)2

x[k] = -4 [k] + { 16 - 12 (0.5)k - 18 k (0.5)k)} u[k]

16

Practice Exam #1

Question 2
f) What is x[]?
You can apply the final value theorem if all the poles of (z-1)X(z) are interior to
the unit circle. Since X(z)= (z+1)2/((z-1)(z-0.5)2), the theorem applies.
x[]=lim(z-1)X(z)|z=1= (z-1)(z+1)2/((z-1)(z-0.5)2 )|z=1=(z+1)2/(z-0.5)2 |z=1= (2)2/(0.5)2
= 4/(1/4) = 16
0
0
Check:
0
As k , x[k] = -4 [k] + { 16 -12 (0.5)k - 18 k (0.5)k)} u[k]
Converges to zero even though one of the terms contains a ramp and is k (0.5)k.

17

Practice Exam #1

Question 2
g) What is x[0]?

Apply the initial value theorem:


x[0] = limX(z)|z= = 0.
Check x[k] = -4 [k] + { 16 -12 (0.5)k - 18 k (0.5)k)} u[k],
Therefore:
x[0] = -4 +16 -12 -0 = 0 as expected.

18

Practice Exam #1

Question 3
Sampling and Data Conversion:
A real signal x(t) = sin(2(103)t) + sin(2(6103)t) is presented to the
system shown below.

x(t)

Sampler

fs

19

Practice Exam #1

x[k]

Reconstruction

Shannon Filter

y(t)

Question 3
a- What is the Nyquist sampling rate (Sa/s)?
The Nyquist sampling rate is fN = 2fmax = 2 6kHz=12kHz.
b- If x(t) is sampled at a rate fs=8kHz, what is the reconstructed signal in
the form y(t) = A sin(2fat) + B sin(2fbt)?

The 1kHz signal is not aliased. The 6kHz signal is aliased to -2kHz.
Therefore, y(t)=sin(2103t) + sin(-4103t) or y(t) = sin(2103t) - sin(4103t).

20

Practice Exam #1

Question 3
c- A |x(t)| 8 volt input is presented to a signed n=8-bit ADC. What is
the ADCs fractional statistical resolution in volts/bit?
=(28)/28=24/28=2-4 (independent of |x(t)|)
d- What is the statistical quantization error in bits (i.e., how many
fractional bits are statistically preserved)?
The maximum x(t) is 8=23. Therefore the signed 8-bit ADC produces data
having 1 sign bit, I=3 integer bits, and F=4 fractional bits (i.e.: format [8:4]).
The fractional statistical error becomes
log2(2-F/12) = log2(2-4) 1.8 = -4 -1.8 or -5.8.

21

Practice Exam #1

Question 4
Sampling Theorem and Quantization
The ADC shown is sampled at a fs=12,000 Sa/s rate. Assume the
human vocal input range is limited to 4 kHz.

x(t)

x[k]
ADC

fmax = 4kHz

DAC

Interpolator
y(t)
Playback

fs = 12kHz

22

Practice Exam #1

Question 4
a. To test the system, a hand-held audio signal generator is placed near
the microphone. The signal generator produces the sinusoidal tone
x(t)=sin(2f0t) where f0 = 8kHz. What is the reconstructed signal
y(t)?

The aliased baseband frequency is f = 8000 mod(12000) = - 4000 Hz, or


y(t)=sin(2(-4000)t) or y(t) = -sin(2(4000)t) (shown with phase inversion).

23

Practice Exam #1

Question 4
b. The signal generators frequency is set to f0 = 4 kHz but the gain of

the electronic signal generator used in Part 4.a is set too high, and
producing a saturating square wave x(t) = sign(sin(2f0t)) having a
known Fourier series representation given by:

x(t)
2
x(t )

sin( 2nf 0 t )

n ( odd )

Assume that x(t) can be essentially modeled using only the 1st, 3rd, and
5th harmonics having amplitudes a1= 2/, a3= 2/3, and a5= 2/5
respectively, where f0 = 4 kHz and fs = 12 kSa/s. What is the
reconstructed output signal y(t)?

24

Practice Exam #1

Question 4
f0 = 4000 is reconstructed as a sinusoid at f0 = 4000.
3f0 = 12000 is reconstructed as a sinusoid at f0 = 0 (aliased).
5f0 = 20000 is reconstructed as a sinusoid at f0 = -4000 (aliased).
The reconstructed signal is:
y(t)=(2/)sin(2(4000)t) )+(2/3)sin(2(0)t) +(2/5) sin(2(-4000)t)
= (2/)sin(2(4000)t) + sin(2(0)t) - (2/5)sin(2(4000)t)
(optional) y(t) = (8/5)sin(2(4000)t) -- trig identity

25

Practice Exam #1

Questions?
Sampling theorem
Aliasing
Quantization
z-transforms
Inverse z-transforms
Properties of z-transforms
I will supply figures and limited tables..

26

Practice Exam #1