EEL3552C Lab Manual

Department of Electrical and

Computer Engineering

University of Central Florida

Signal Analysis & Analog

Communications

Revised Orlando, August 2018

 Table of Contents

0. Safety Rules and Operating Procedures page 3

1. Experiment 1 (Spectrum Analysis) page 5 (1 week)
2. Experiment 2 (MATLAB) page 12 (1 week)
3. Experiment 3 (Periodic Signal Spectra) page 22 (1 week)
4. Experiment 4 (Low Pass Filter) page 39 (1 or 2 weeks)
5. Experiment 5 (Amplitude Modulation) page 60 (2 weeks)
6. Experiment 6 (Frequency Modulation) page 93 (2 weeks)
7. Lab report - Cover Pages Sample page 114
8. Lab report Outline & Grading key page 115
EEL 3552C Lab Manual

Safety Rules and Operating Procedures

1. Note the location of the Emergency Disconnect (red button near the door) to shut off power
in an emergency. Note the location of the nearest telephone (map on bulletin board).
2. Students are allowed in the laboratory only when the instructor is present.
3. Open drinks and food are not allowed near the lab benches.
4. Report any broken equipment or defective parts to the lab instructor.
Do not open, remove the cover, or attempt to repair any equipment.
5. When the lab exercise is over, all instruments, except computers, must be turned off.
Return substitution boxes to the designated location. Your lab grade will be affected if your
laboratory station is not tidy when you leave.
6. University property must not be taken from the laboratory.
7. Do not move instruments from one lab station to another lab station.
8. Do not tamper with or remove security straps, locks, or other security devices.
Do not disable or attempt to defeat the security camera.
9. ANYONE VIOLATING ANY RULES OR REGULATIONS MAY BE DENIED ACCESS TO THESE FA-
CILITIES.

I have read and understand these rules and procedures. I agree to abide by these rules and pro-
cedures at all times while using these facilities. I understand that failure to follow these rules and
procedures will result in my immediate dismissal from the laboratory and additional disciplinary
action may be taken.
———————————————

Page 3
EEL 3552C Lab Manual

Laboratory Safety Information

Introduction
The danger of injury or death from electrical shock, fire, or explosion is present while conducting experi-
ments in this laboratory. To work safely, it is important that you understand the prudent practices necessary
to minimize the risks and what to do if there is an accident.
Electrical Shock
Avoid contact with conductors in energized electrical circuits. Electrocution has been reported at dc voltages
as low as 42 volts. 100ma of current passing through the chest is usually fatal. Muscle contractions can
prevent the person from moving away while being electrocuted.
Do not touch someone who is being shocked while still in contact with the electrical conductor or you may
also be electrocuted. Instead, press the Emergency Disconnect (red button located near the door to the
laboratory). This shuts off all power, except the lights.
Make sure your hands are dry. The resistance of dry, unbroken skin is relatively high and thus reduces the
risk of shock. Skin that is broken, wet, or damp with sweat has a low resistance.
When working with an energized circuit, work with only your right hand, keeping your left hand away from
all conductive material. This reduces the likelihood of an accident that results in current passing through
your heart.
Be cautious of rings, watches, and necklaces. Skin beneath a ring or watch is damp, lowering the skin
resistance. Shoes covering the feet are much safer than sandals.
If the victim isn’t breathing, find someone certified in CPR. Be quick! Some of the staff in the Department
Office are certified in CPR. If the victim is unconscious or needs an ambulance, contact the Department
Office for help or call 911. If able, the victim should go to the Student Health Services for examination and
treatment.
Fire
Transistors and other components can become extremely hot and cause severe burns if touched. If resistors
or other components on your proto-board catch fire, turn off the power supply and notify the instructor. If
electronic instruments catch fire, press the Emergency Disconnect (red button). These small electrical fires
extinguish quickly after the power is shut off. Avoid using fire extinguishers on electronic instruments.
Explosions
When using electrolytic capacitors, be careful to observe proper polarity and do not exceed the voltage
rating. Electrolytic capacitors can explode and cause injury. A first aid kit is located on the wall near the
door. Proceed to Student Health Services, if needed.

Page 4
EEL 3552C Lab Manual 1. EXPERIMENT: SPECTRUM ANALYSIS

1 EXPERIMENT: SPECTRUM ANALYSIS

1.1 Objective:
Analyze the spectral content of a simple signal.

1.2 Equipment:
The equipment used in this experiment are:

Oscilloscope: Rohde & Schwarz RTM 3004

Function Generator: Tektronix AFG 3022B
Bring a USB Flash Drive to store your waveforms.

1.3 Implementation:
A waveform representing amplitude, as a function of time, is called a time domain display. It is also possible
for a waveform to represent amplitude as a function of frequency. This is called a frequency domain display.
A Spectrum Analyzer is an instrument, which can display the frequency domain of a signal. However, the
RTM3004 Oscilloscope has the capability of producing both time domain and frequency domain displays.
A sine wave is the simplest signal for spectral analysis. The amplitude of the sine wave can be determined
on the vertical scale and the frequency can be determined on the horizontal scale.
The units of amplitude used in this experiment will be dBV, which is dB relative to 1 VRMS (0 dBV = 1VRMS),
according to the formula:

Vsignal
dBV = 20log( ) (1.1)
Vref

where Vsignal is the RMS voltage of the signal and Vref = 1 volt RMS.
Another common unit of amplitude used for spectrum analysis is dBm, which is dB relative to 1 milliwatt,
according to the formula:

Psignal
dBm = 10log( ) (1.2)
Pref

where Psignal is the power of the signal in milliwatts and Pref = 1 mW.

Page 5
EEL 3552C Lab Manual 1. EXPERIMENT: SPECTRUM ANALYSIS

1.4 Pre-lab Questions:

a. Calculate the amplitude in dBV of a 4 KHz, 2 volt peak-to-peak, sine wave.
b. Calculate the peak-to-peak voltage of a −10 dBV, 4 KHz, sine wave.

1.5 Procedure:
1.5.1 Time Domain
a. Set the Function Generator for a sine wave output with a frequency of 4 KHz. Set the amplitude to 2 volt
peak-to-peak with zero DC offset.

Oscilloscope Reset:
b. Connect the Channel 1 output of the Function Generator to Channel 1 input of the Oscilloscope using a
BNC to BNC cable. Turn on the Channel 1 output of the Function Generator.

c. Set the oscilloscope to the default state (PRESET).

d. Before beginning the exercise, configure the oscilloscope to use the maximum sample rate.

Switch the oscilloscope to MAX. SA. RATE mode. You can achieve that by clicking on the second upper
label near C1 and change the GSa/s Acquire Mode to High Resolution after a menu appears on the
right of the screen.

View the time domain waveform on the Oscilloscope:

e. Press Auto Set.

This sets the vertical, horizontal, and trigger for a stable display of the waveform.

f. Press Channel 1 button to display the menu on the right side of the screen and select an Input Impedance
of 50 Ohms under the Termination label.

For this experiment, the output of the Function Generator must be terminated into a 50 Ohm load.

The time domain waveform should now be displayed.

1.5.2 Frequency Domain

View the frequency domain waveform of the input signal on the Oscilloscope:
a. Change the Horizontal Scale to display 10 cycles. You can achieve that either by rotating the second
upper knob until 10 cycles appear on the screen. Or by double clicking on the top pane where the the value
of seconds per division appears (the label to the left of Run). This is called the Timebase. A scroll menu
containing specified durations per division appear.
Since, there are a total of 12 horizontal divisions and a total of 10 vertical divisions. For a 4KHz sine wave
where the center of the screen is at 0s time. You can find the time duration per division :

Page 6
EEL 3552C Lab Manual 1. EXPERIMENT: SPECTRUM ANALYSIS

Fig. 1.1

1
Each cycle of the sine wave is 4000 s = 0.00025s = 0.25ms = 250µs. Then 10 cycles will correspond to
10 × 250µs = 2500µs

But the total number of horizontal divisions is 12. Then, for 10 cycles to appear we need the time duration
per division (i.e., Timebase) to be 2500/12 = 208.33µs/div. But the available durations per division are
specified. So pick the value appearing in the scroll menu just above the calculated valued, that is 500µs. Or
click on the menu button and adjust the value to the calculated one 208.33µs/div.

b. Press on the FFT icon touch button on the right top pane.

The screen is split into two panes; see Fig. 1.1. The upper pane (2) displays the previously defined voltage-
time trace for the measured signal and the settings for the horizontal system. That is, the waveform in the
time domain. The zoom and position information is displayed in between the two panes (3). Only the signal
segment that is displayed on the monitor and bound by the two vertical white lines is used for the FFT.
The lower pane (4) shows the result of the FFT analysis, that is, it will display the spectrum of the signal.

Page 7
EEL 3552C Lab Manual 1. EXPERIMENT: SPECTRUM ANALYSIS

The FFT displays the frequency domain by expressing the input signal as a combination of simple sig-
nals and produces a waveform to display the amplitude verses frequency of the component frequencies
of the input signal.

c. Now, either
i) Double press the FFT button under the Analysis grouping and a menu will appear on the right of the
screen (Fig. 1.2),
ii) or by clicking on the FFT label on the bottom pane, a menu will appear (Fig. 1.3), then Press on Menu.
The right menu will appear (Fig. 1.2).

Fig. 1.2 Fig. 1.3

On this right pane menu (Fig. 1.2), these should be selected, if not already set:

FFT Source: C1 (Channel 1)

FFT Window: Hanning
Vertical Scale: dBV RMS
Automatic RBW : off

d. First, adjust the time duration per division (timebase) on the top pane from 833.33µs/div to 200ms/div
and then select a maximum signal segment for the FFT.

Page 8
EEL 3552C Lab Manual 1. EXPERIMENT: SPECTRUM ANALYSIS

Fig. 1.4

Span (4 in Fig. 1.4) shows the size of the currently displayed frequency range. The span, time-
base and selected segment are directly linked to one another. As the selection for the timebase
and the range in the time domain displayed between the two vertical lines increases, so too
decreases the frequency band that can be set. Center defines the frequency at the center of
the segment displayed on the monitor. The minimum step size ∆f can be set indirectly via the
number of points.

e. Set now the value of the RBW (5 in Fig. 1.4) to the maximum (i.e. by clicking on RBW on the bar between
the two screen and adjusting the value to the maximum allowed).
f. On the bar between the two panes that display time and frequency domains respectively, set Start (1 in
Fig. 1.4) value to 0, you can achieve that by clicking on the Start label where calculator form menu appears
and you can set the value. This will be the value of the minimum frequency displayed.
g. Similarly, adjust the Stop (2 in Fig. 1.4) value to a value above 4KHz, let’s say 8KHz. N.B. Usually
you need this value of 8KHz to fall between the minimum value and maximum value allowed for the Stop
value. Therefore, you need to readjust the time duration per division (timebase) so that is satisfied. Adjust
the center frequency to 4KHz you can set this value through clicking on Center in the middle bar. See the
blue frame comments.
Using the Cursors:
First, for a better view of the frequency domain screen. Move the double arrow bar of the lower screen
upwards so that the frequency domain is the only display on the oscilloscope. There are a variety of ways
to measure the amplitude. One method is to use the Cursors.
a. Press the Cursors button twice on the front panel to activate the cursors. At the first click a bar will appear
showing the values at the used cursors. After the second click a right menu will appear. On the Type, set to
Vertical & Horizontal, so that both types of cursors appear on the screen. Set the Source to FFT: Spectrum,
so that the frequency domain cursors appear.
Cursors 1 and 2 are the horizontal cursors.
Cursors 3 and 4 are the vertical cursors.
Click on number 1 on the cursor line and move it to the desired point on the spectrum. Its peak value in
our case. The amplitude measured by the Cursor 1 is now displayed on the bar near label L1 and the
corresponding frequency value near label f1.
b. Does the amplitude agree with the value that you calculated in the Preparation?

Page 9
EEL 3552C Lab Manual 1. EXPERIMENT: SPECTRUM ANALYSIS

c. Save the Screen Image to your USB Flash Drive to include in your lab report by clicking on the camera
icon in the Action section of the oscilloscope buttons.
d. Change the frequency of the Function Generator to 2 KHz.
Does the spectrum display accurately represent the input signal?
Does it appear on the frequency domain display?
Can you adjust any of the labels in the middle bar so that the peak value appears?
Are you able to adjust accurately the setup so that the center frequency is at 2KHz.
e. Change the frequency of the Function Generator to 1 KHz.
Does the frequency domain waveform confirm that the signal is 1 KHz?
Are you able to answer the questions for part d.
Taking a Measurement:
a. Decrease the amplitude of the Function Generator output to −10 dBV as measured in the frequency
domain display. Now, switch to the time domain display. Press on the measure button. A menu will appear
on the right of the screen. Set the Measure Place to 1. Turn the Measure 1 bar to on. Choose the Type to
the type of measurement (peak to peak). Read the peak-to-peak voltage measurement that will appear on
the bottom left of the screen above the channels bar. Does this measurement agree with the value that you
calculated in the Preparation?
Using the Zoom Feature:
a. Turn the time domain on again by pressing the CH1 button. Turn the Zoom on by pressing the Zoom
button. Press on the part of the signal you want to zoom in. By using the two fingers motion that you use on
the i-pad for zooming, double-tap with two fingers to zoom in and back out so that you can adjust the signal
part you want to inspect to the right resolution.

1.6 Report:
In your report, describe what you have learned in this experiment. Compare your experimental measure-
ments with the theoretical calculations. Remember to insert the picture that you saved as part of your report.
Write all conclusions. Follow the instructions listed in the Appendix regarding outline and required
analysis.

Page 10
EEL 3552C Lab Manual 1. EXPERIMENT: SPECTRUM ANALYSIS

Fig. 1.5 : The connection of Section 1.5.1

1.7 Hints & Comments

1.7.1 Procedure
1.5.1 Time Domain: Students should not mistakenly connect the input from the function generator directly
to the oscilloscope. Instead, they should construct a circuit with a 50 Ω resistor and then connect the output
of that resistor to the oscilloscope. A figure can help resolve this misunderstanding (see Fig. 1.5 above)
Students must not forget to adjust the input impedance of the oscilloscope to 50 Ω instead of 1 M Ω.

Page 11
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

2 EXPERIMENT: MATLAB
2.1 Objective:
The purpose of this experiment is to be able to implement a signal analysis problem through simulation by
either using MATLAB.

2.2 Equipment:
The equipment used in this experiment are:

Desktop or laptop computer

MATLAB software

2.3 Simulation
MATLAB
MATLAB is a programming languange for technical computing. MATLAB is case sensitive. A variable in
MATLAB is an array. An array has dimensions m × n, m is the number of rows and n is the number of
columns. If m = n = 1, the variable is a scalar. If m = 1 and n > 1, the variable is a row vector. If n = 1 and
m > 1, the variable is a column vector. If both m and n are greater than one, the variable is a matrix, and if
m = n, then the variable is a square matrix. The coefficients of an array can be real or complex numbers.
Please refer to Slide 2.1 for some examples on data types and data accessing.

Slide 2.1

A previously typed command can be recalled with the up-arrow key. When the command is displayed at the
command prompt, it can be modified if needed and executed.

Page 12
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Slide 2.2

See Slides 2.2 and 2.3 for defining some special matrices (all-zero, all-one, and identity matrices) and some
basic commands. See Slide 2.4 for some helpful commands.
For predefined functions and constants see Table 2.1 and Table 2.2.
For plotting functions see Slide 2.5.

Compact representation of a vector: for example instead of A = [1 3 5 9 11 13] we can say A = 1 : 2 : 13.
π
Another examples is B = 0 : pi/100 : 2 ∗ pi is a vestor that starts at 0 takes steps or increments of 100
stops when 2π is reached. If you omit the increment MATLAB automatically increments by 1.

For element-wise operations on array see Slide 2.6. and Slide 2.7 for some examples.

For U = [u1 , u2 , ..., un ] and V = [v1 , v2 , ..., vn ] we have:

U. ∗ V = [u1 v1 , u2 v2 , ..., un vn ]
U./V = [u1 /v1 , u2 /v2 , ..., un /vn ]
U.∧V = [uv11 , uv22 , ..., uvnn ]

For inequalities and equations see Slide 2.8. And for logical operators see Slide 2.9.
For plotting use Slide 2.10, 2.11 and Fig. 2.1.
For multiple plots check Slide 2.12.

Page 13
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Slide 2.3

Slide 2.4

Page 14
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Slide 2.5

Page 15
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Slide 2.6

Slide 2.7

Slide 2.8

Page 16
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Slide 2.9

Slide 2.10: Attributes on the plot

Slide 2.11

Page 17
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Fig. 2.1: Plot of sinc function

Slide 2.12

Page 18
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Fig. 2.2: stem of sin(n)

Slide 2.13

For plotting discrete time signals use stem (see Fig. 2.2)
For using complex numbers check Slide 2.13.
For using loops and conditional statements check Slide 2.14 and example in Slide 2.15.
In MATLAB you can save programs in a file and execute them from the command line. The file is called an
m-file, and can be assigned a name. You can execute the programs in the file by entering the name of the
file on the command line. You can save all our scripts on an m-file. It is a good idea to always start an m-file
with clear all. You can write a function and call it inside your m-file
function [outputs] = function_name(inputs)
Save the example in Slide 2.16 as myfactorial.m

Page 19
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

Slide 2.14

Slide 2.15

Slide 2.16

Page 20
EEL 3552C Lab Manual 2. EXPERIMENT: MATLAB

2.4 MATLAB Implementation

Problem 1
Let

x1 (t) = t2 − 2t + 3 − 10 ≤ t ≤ 10
π
x2 (t) = 4cost(2πt − ) + 3sin2πt − 2 ≤ t ≤ 2
8
sin(t)
x3 (t) = sinc(t) =
t
x4 (t) = e−at

a. Plot x1 (t) and x2 (t) .

b. Plot x3 (t) for −2π ≤ t ≤ 2π
c. Plot e−at for a = 0.1, 1, 3 and 0 ≤ t ≤ 10 on one plot. Use different colors or markers for each plot.

Problem 2
a. For the functions in Problem 1, plot their discrete versions using stem where n varies on the integers.

x1 (n) = n2 − 2n + 3 − 10 ≤ n ≤ 10
π
x2 (n) = 4cost(2πn − ) + 3sin2πtn − 2 ≤ n ≤ 2
8
sin(n)
x3 (n) = sinc(n) = −5≤n≤5
n
x4 (n) = e−an 0 ≤ n ≤ 10 f or a = 0.1, 1, 3

For each of the above functions compare the graphs of Problem 1 and their corresponding samples in
Problem 2 by plotting both on the same plot.
Problem 3
For functions x1 (t), x2 (t), x3 (t), write a program that plots a graph of a function y(t) for −4 < t < 4 such that


x1 (t)
 − 4 < t < −2
y(t) = x2 (t) −2<t<2

x3 (t) 2<t<4


a. Write the above program using a for loop and conditional statements.
b. Write the above program using vector evaluations. i.e., plot(t1 , x1 , t2 , x2 , t3 , x3 )

Page 21
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

3 EXPERIMENT: Periodic Signal Spectra

3.1 Objective:
To understand the relationships between time waveforms and frequency spectra.

3.2 Equipment:
The equipment used in this experiment are:

Oscilloscope: Rohde & Schwarz RTM 3004

Function Generator: Tektronix AFG 3022B
Bring a USB Flash Drive to store your waveforms.

3.3 Theory:
3.3.1 Motivation
Consider for a moment the special class of systems that are called linear time-invariant systems. When a
signal is applied to this system, the output can be expressed in terms of linear differential equations with
constant coefficients. An example of such a system is shown in Figure 3.1. The differential equation that
describes the input/output relationship in such a system is given below

d R R
vo + vo = v i (3.1)
dt L L

For a finite input vi (t) we expect to solve this differential equation to find the corresponding output vo (t).
The difficulty involved in this task is that the input has to be a very simple signal. One way to deal with this
complication is to express the input as a linear combination of simpler inputs. The choice of simpler inputs
will affect the difficulty that we may encounter in solving the problem.
There is an intriguing possibility regarding the simpler input signals. Let us, for a moment, assume that we
choose as an input a function that repeats itself under the operation of differentiation. We can show that
such a function at the input will yield the same function as an output multiplied by some algebraic polynomial
involving the parameters of the function and the parameters of the differential equation.

Fig. 3.1: A Simple linear Time-invariant System

Page 22
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

A function that demonstrates this behavior, that is to repeat itself under the operation of differentiation, is the
function est , where s = σ + jω is a complex number. Since some of the signals that we are interested in are
energy-type signals that may exist for positive or negative time it is wise to choose σ = 0. This discussion
indicates that if the input signal vi (t) in equation (3.1) is of the form ejωt , then we can readily find the output
vo (t) of the system. Furthermore, if we can express the input signal vi (t) as a linear combination of complex
exponential signals (of the form ejωt ), we can still find, without difficulty, the output vo (t) of the system (the
reason being that the system in (3.1) is linear, and as a result if the input vi (t) is a linear combination of
complex exponentials, then the output vo (t) can be expressed as a linear combination of the outputs of the
complex exponentials involved in the input). One version of the Fourier series expresses an arbitrary signal
(like vi (t)) as a linear combination of complex exponentials. This is the first reason that the Fourier series
representation of a signal is so useful.
A close relative of a complex exponential ejωt are the signals cosωt and sinωt (remember Euler’s identity:
ejωt = cosωt + jsinωt). Consider now the following set of familiar signals.

f1 (t) = asinω0 t
f2 (t) = asin2ω0 t
f3 (t) = asin3ω0 t
. .
(3.2)

. .
. .
fn (t) = asinnω0 t

In Figure 3.2 we show a plot of f1 (t) and f2 (t). More generally, we can say that as n increases, the rate
of variation of fn (t) increases as well. One way to quantify the above qualitative statements is to take the
derivative of fn (t) with respect to time. In particular,

d
f (t) = anω0 t (3.3)
dt

From the above equation we observe that as n increases the maximum rate of variation of fn (t) with respect
t time increases. We can represent each one of the fn (t) functions in a little bit different way than the one
used in Fig. 3.2. That is, we can represent these functions by plotting the maximum amplitude of fn (t) (i.e.,
a) versus the angular frequency of the sinusoid (i.e., nω0 ). The product of these two values gives you the
maximum rate of variation for the signal under consideration. In Fig. 3.3 we are representing the functions
f1 (t),f2 (t), and f3 (t) following the aforementioned convention. Obviously, larger values for the location of
the plotted amplitudes imply faster varying time signals; also the larger the amplitudes plotted at a particular
location the faster the corresponding signals vary with respect to time.
Now consider a simple example of an amplitude-modulated signal.

f (t) = A(1 + cosωm t)cosω0 t ωm << ω0 (3.4)

In Fig. 3.4 we plot the aforementioned signal f (t). As we can see from the figure, the amplitude varies
slowly between 0 and 2A. Its rate of variation is given by ωm , the modulating frequency; ω0 is referred to
as the carrier frequency. The amplitude variations form the envelope of the complete signal, and represent
any information being transmitted. Note that

A
A(1 + cosωm t)cosω0 t = Acosω0 t + [cos(ω0 − ωm )t = cos(ω0 − ωm )t] (3.5)
2

Page 23
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Fig. 3.2: Plots of f1 (t) and f2 (t)

Fig. 3.3: Representation of f1 (t), f2 (t) and f3 (t)

Fig. 3.4: Plot of an amplitude-modulated signal f (t)

Page 24
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Fig. 3.5: Representation of the amplitude-modulated signal f (t)

The representation of the aforementioned signal, using the new conventions, is illustrated in Fig. 3.5. As we
observe from Fig. 3.5, larger carrier frequency corresponds to a faster varying signal; also larger modulating
frequency results in a faster varying signal.
The previously discussed signals vary in some sinusoidal fashion. They do not carry any information but we
have used them to focus on an alternative representation of the signal, that conveys to us the information of
how fast the signal changes with respect to time. Another variation of the Fourier series representation of a
signal represents the signal as a linear combination of sines and cosines of appropriate frequencies. As a
result, the Fourier series representation of the signal, which is an extension of the alternative representation
of the simple signals that we discussed above, gives us an idea of how fast the signal changes with respect
to time. This is the second reason that the Fourier series representation of a signal is so useful.

3.3.2 Fourier Series

Let our space S of interest be the set of piece-wise continuous real time signals, defined over an interval
(0, T ). Let us also consider the set of signals
1
φ10 (t) = √ (3.6a)
T
2
φ1n (t) = √ cosnω0 t n = 1, 2, ... (3.6b)
T
2
φ2n (t) = √ sinnω0 t n = 1, 2, ... (3.6c)
T

where ω0 = 2π
T . The aforementioned signals (see equations (3.6)) are orthonormal. They also constitute
a complete set of functions. These two properties allow us to state that an arbitrary signal in S can be
expressed as a linear combination of the signals in equations (3.6), such that
∞
X ∞
X
f (t) = C10 φ10 + C1n φ1n + C2n φ2n (3.7)
n=1 n=1

and the coefficients in the linear expansion above can be computed through the following equations:
Z T
1
C10 = √ f (t)dt (3.8a)
T 0
Z T
2
C1n = √ f (t)cosnω0 tdt n = 1, 2, ... (3.8b)
T 0

Page 25
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Z T
2
C2n = √ f (t)sinnω0 tdt n = 1, 2, ... (3.8c)
T 0

The above expansion, due to Fourier, is referred to as the Trigonometric Fourier Series expansion of a
signal f (t). A more common form for the trigonometric Fourier series expansion of a signal is given below.

∞
X ∞
X
f (t) = a0 + an cosnω0 t + bn sinnω0 t (3.9)
n=1 n=1

where Z T
1
a0 = f (t)dt (3.10a)
T 0
Z T
2
an = f (t)cosnω0 tdt n = 1, 2, ... (3.10b)
T 0
Z T
2
bn = f (t)sinnω0 tdt n = 1, 2, ... (3.10c)
T 0

By a simple trigonometric manipulation we get that

∞ p
X −bn
f (t) = a0 + a2n + b2n cos(nω0 t + θn ) θn = arctan( ) (3.11)
n=1
an

The above equation puts f (t) in the desired form. In other words, we have now expressed f (t) as a linear
combination of sinusoids, and as a result we can represent this signal f (t) by plotting the amplitude of each
one of its sinusoids at the corresponding angular frequency location. This will give rise to the amplitude -
frequency plot of f (t). Observe though that in equation (3.11) each sinusoid involved in the expansion of
f (t) has a phase associated with it. Hence, in order to completely describe the signal f (t) we need to draw
another plot that provides the information of the phase associated with each sinusoid. This plot is denoted
as the phase - frequency plot and it corresponds to a plot of θn at the location ωn .
It is worth mentioning that the aforementioned expansion of f (t) in terms of sinusoidal signals of frequencies
that are integer multiples of the frequency ω0 is often referred to as the expansion of f (t) in terms of its
harmonic components. The frequency ω0 is called the fundamental frequency, or first harmonic, while
multiples of the first harmonic frequency are referred to as the second harmonic, third harmonic, and so
on. The corresponding signals in the expansion are named first harmonic component, second harmonic
component, third harmonic component, and so on. The component a0 in the above expansion is called the
DC component of the signal f (t).
p
Since we shall frequently be interested in the amplitudes, a2n + b2n ’s, and phases, θn ’s, of our signal f (t) it
would be simpler to obtain these directly from f (t), rather than by first finding an and bn . In order to achieve
this, let us consider the functions
1
φn (t) = √ ejnω0 t (3.12)
T
where ω0 = 2πT . The above set of functions is a complete, orthonormal set. Therefore, any function f (t),
defined over an interval of length T , can be expanded as a linear combination of the φn ’s. In particular,
∞ ∞
X X 1
f (t) = Cn φn (t) = Cn √ ejnω0 t (3.13)
n=−∞ n=−∞ T

The coefficients Cn in the above expansion are chosen according to the rules specified by the Orthogonality
Theorem. That is,

Z T
1
Cn = f (t) √ ejnω0 t dt (3.14)
0 T

Page 26
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

The above equation is denoted as the Exponential Fourier Series expansion of the signal f (t). Actually, in
most instances, the exponential Fourier series expansion of a signal is provided by the following equation:

∞
X
f (t) = Fn ejnω0 t (3.15)
n=−∞

where

Z T
1
Fn = f (t)e−jnω0 t dt (3.16)
T 0

Our original goal has now been accomplished, since we can now derive the amplitude - frequency plot of
f (t) by plotting the magnitude of Fn , with respect to frequency, and we can also derive the phase - frequency
plot of f (t) by plotting the phase of Fn with respect to frequency. Note that, in general, the Fn ’s are complex
numbers and a complex number Fn , can be written as the product of its magnitude (|Fn | times ej∠Fn ). One
can show that the coefficients of the exponential Fourier series expansion of f (t) and the coefficients of the
trigonometric Fourier series expansion of f (t) are related as follows:

F0 = a0 (3.17a)
1
Fn = (an − jbn ) 1 ≤ n < ∞. (3.17b)
2
1
F−n = (an + jbn ) 1 ≤ n < ∞. (3.17c)
2
The following are some comments:

• The exponential Fourier series and the trigonometric Fourier series expansion were introduced for a
signal f (t) defined over the interval (0, T ). Actually, the formulas provided are valid for any signal f (t)
defined over any interval of length T (the starting and the ending points of they interval are immaterial,
as far as the length of the interval is equal to T ).

• We expanded so far a signal f (t), defined over an interval of length T , in terms of a linear combination
of sinusoids (Trigonometric Fourier Series expansion), or complex exponentials (Exponential Fourier
Series expansion). It is worth noting, that the trigonometric or exponential Fourier series expansion of
a signal defined over an interval T is a periodic signal of period T ; that is, it repeats itself with period
T . Hence, if the signal of interest f (t) is not periodic the trigonometric or exponential Fourier series
expansion of the signal is only valid for the interval over which the signal is defined.

• Due to the periodicity nature of the Fourier series expansion, Fourier series is primarily used to rep-
resent signals of periodic nature. Signals of aperiodic nature can also be represented as a "sum" of
sinusoids or complex exponentials, but this "sum" is in reality an integral and it is designated by the
name Fourier Transform. The Fourier transform of an aperiodic signal will be introduced later as an
extension of the Fourier series of a periodic signal.

Example 1
(a) Evaluate the trigonometric Fourier series expansion of a signal, which is depicted in Fig. 3.6.
The interval of interest is (− π2 , π2 ). Hence, T = π and as a result ω0 = 2π
T = 2. The trigonometric Fourier
series of a signal f (t) is therefore of the form

∞
X ∞
X
f (t) = a0 + an cos2nt + bn sin2nt (3.18)
n=1 n=1

Page 27
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Fig. 3.6: Plot of f (t) = cost for [ −π π

2 , 2 ] and zero elsewhere

Z T
1 2
a0 = f (t)dt
T − T2
Z π
1 2
(3.19)
= costdt
π −π
2

2
=
π

Z T
2 2
an = f (t)cos2ntdt
T − T2
Z π
2 2
= costcos2ntdt
π −π
Z
2
π
(3.20)
2 2
= [cos(2n − 1)t + cos(2n + 1)t]dt
π −π
2

2 (−1)n+1 (−1)n
= [ + ]
π 2n − 1 2n + 1

Z T
2 2
bn = f (t)sin2ntdt
T − T2
π
(3.21)
Z
2 2
= costsin2ntdt
π −π
2

=0

The last equation is a result of the fact that f (t) is even and sin2nt is odd. Consequently, f (t)sin2nt is odd,
and whenever an odd function is integrated over an interval which is symmetric around zero, the result of
the integration is zero. Based on the above equations we can write:

∞
2 X 2 (−1)n+1 (−1)n+1
f (t) = + { [ + ]}cos2nt (3.22)
π n=1 π 2n − 1 2n + 1

Page 28
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Also, if we write out a couple of terms from the above equation we get:

2 2 2 2
f (t) = + [1 + cos2t − cos4t + cos6t + ...] (3.23)
π 3 15 35

(b) Find the exponential Fourier series of the signal f (t) in Fig. 3.6.
We can proceed via different paths. We can either find, directly, the exponential Fourier series expansion of
f (t) by applying the pertinent formulas, or we can use the trigonometric Fourier series expansion already
derived to generate the exponential Fourier series expansion. We choose the latter approach, because it is
easier. Recall that

∞
2 X
f (t) = + an cos2nt (3.24)
π n=1

where

2 (−1)n+1 (−1)n+1
an = [ + ] (3.25)
π 2n − 1 2n + 1

Due to Euler’s identity we can write

ej2nt + e−j2nt
cos2nt = (3.26)
2

Using Euler’s identity in equation (3.26) we get

∞ ∞
2 X an j2nt X an −j2nt
f (t) = + e + e (3.27)
π n=1 2 n=1
2

Notice also that the exponential Fourier series expansion of the signal f (t) has the following form

∞
X ∞
X ∞
X
f (t) = Fn ej2nt = F0 + Fn ej2nt + F−n e−j2nt (3.28)
n=−∞ n=1 n=1

Comparing, one by one, the terms of equations (3.27) and (3.28) we deduce the following identities:

F0 = a0 (3.29a)
an
Fn = (3.29b)
2
an
F−n = (3.29c)
2
In particular,

2 2 j2t 2 −j2t 2 j4t 2 −j4t

f (t) = + e + e − e − e (3.30)
π 3π 3π 15π 15π

Page 29
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Fig. 3.7: Line Spectrum of f (t) of Fig. 3.6

Fig. 3.8: Plot of the RHS of equation (2.27)

Based on equations (3.29) we can plot the amplitude - frequency, and the phase - frequency plots for the
signal f (t). In the case where the coefficients Fn are real we can combine these two plots into one plot; this
plot is denoted as the line spectrum of f (t). The line spectrum of a signal f (t) corresponds to the plot of
Fn ’s with respect to frequency. The line spectrum of the signal f (t) in this example is depicted in Fig. 3.7.
(c) Plot the right hand sides (RHS’s) of equations (3.23) and (3.30).
The plot is shown in Fig. 3.8. Note that the RHS’s of equation (3.23) and (3.30) is equal to the signal f (t)
(see Fig. 3.6) in the interval (− π2 , π2 ). Also the RHS’s of equation (3.23) and (3.30) are periodic with period
π. hence the plot of Fig. 3.8 is produced by reproducing the plot of Fig. 3.7 every π units of time.
Example 2
Evaluate the trigonometric Fourier series expansion of the periodic signal which is depicted in Fig.
3.9.
Let us concentrate on one period of the signal, that is the interval (− π2 , π2 ). Knowing that T = π we conclude
that ω0 = 2π
T = 2. The trigonometric Fourier series of a signal f (t) is therefore of the form

∞
X ∞
X
f (t) = a0 + an cos2nt + bn sin2nt (3.31)
n=1 n=1

Fig. 3.9: Plot of signal f (t)

Page 30
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

where

Z T
1 2
a0 = f (t)dt
T − T2
π
(3.32)
Z
1 2
= f (t)dt
π −π
2

=0

Z T
2 2
an = f (t)cos2ntdt
T − T2
Z π Z π Z π
2 4 2 4 2 2
= (−1)cos2ntdt + (1)cos2ntdt + (−1)cos2ntdt (3.33)
π π
−2 π π
−4 π π
4

sin( nπ
2 )
=2 π
n2

Z T
2 2
bn = f (t)sin2ntdt
T − T2 (3.34)
=0

The last equation is a result of the fact that f (t) is even and sin2nt is odd. Consequently, f (t)sin2nt is odd,
and whenever an odd function is integrated over an interval which is symmetric around zero, the result of
integration is zero. Based on the above equations we can write:

∞
X sin( nπ
2 )
f (t) = [2 π ]cos2nt (3.35)
n=1
n2

Also, if we write out a couple of terms from the above equation we get:

4 1 1 1
f (t) = [cos2t − cos6t − cos10t + cos14t + ...] (3.36)
π 3 5 7

3.3.3 Power - Parseval’s Relation

The power Pf of a real signal f (t) is defined as follows:

Z T
1 2
Pf = lim f 2 (t)dt (3.37)
T0 →∞ T0 − T2

It is not difficult to show that if the signal f (t) is periodic with period T , then its power Pf can be computed
as follows

Z T
1 2
Pf = f 2 (t)dt (3.38)
T − T2

Page 31
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Parseval’s relation for periodic signals says that the power content of a periodic signal is the sum of the
power contents of its components in the Fourier series representation of the signal. In particular, if the
exponential Fourier series coefficients of f (t) is given by the following equation:

∞
X
f (t) = Fn ejnω0 t (3.39)
n=−∞

then, Parseval’s theorem says that the power Pf of f (t) is equual to

∞
X
|Fn |2 (3.40)
n=−∞

In the above equation |Fn |2 is the magnitude square of the complex number Fn . An alternative expression
of Parseval’s theorem for periodic signals, says that the power of a periodic signal f (t) is equal to

∞
1X 2
a20 + (a + b2n ) (3.41)
2 n=1 n

where the an ’s and bn ’s in the above equation are the trigonometric Fourier series coefficients of the periodic
signal f (t).
Besides being an alternative way of calculating the power of a periodic signal, Parseval’s relation allows us to
calculate the amount of power of a periodic signal that is allocated to each one of its harmonic components.
Example 3
(a) Find the power contained in the first harmonic of the periodic signal f (t) of Example 1.
From the trigonometric Fourier series expansion of f (t) (see equation (3.23)) we observe that the first
harmonic of f (t) is equal to

4
cos(2t) (3.42)
3π
4 2
[ 3π ]
It is not difficult to show that the amount of power contained in the first harmonic is equal to 2 as
Parseval’s relation predicts.
(b) Find the power contained in the first and second harmonics of the periodic signal. f (t) of Example
1
From the trigonometric Fourier series expansion of f (t) (see equation (3.23)), we observe that the first and
second harmonic of f (t) are equal to

4 4
cos(2t) , cos(4t) (3.43)
3π 15π
respectively.
4 2 4 2
[ 3π ] [ 15π ]
It is not difficult to show that the amount of power contained in the first two harmonics is equal to 2 + 2
as Parseval’s relation predicts.
(c) Find the power contained in the DC component of the periodic signal f (t) of Example 1.
From the trigonometric Fourier series expansion of f (t), (see equation (3.23)) we observe that the DC
component of f (t) is equal to π2 . Hence, the amount of power contained in the DC component is equal to
( π2 )2

Page 32
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

3.4 Pre-lab Questions:

1. Find both the trigonometric and the exponential Fourier series expansion of the signal f (t) in Figure 3.9.
Plot amplitude verses frequency for the signal. Identify the first three harmonics and their amplitudes. Find
the power content from the time domain representation of the signal and from the first three harmonics of
the Fourier series expansion, and make appropriate remarks.
2. What is the difference between an energy signal and a power signal? Give two examples of each.

3.5 Implementation
In this lab you will be introduced to the relationship between pulse shapes and their spectra. There are
some important concepts to recognize when performing the experiment, such as:
1. When a pulse becomes narrower in the time domain its energy becomes more spread out in frequency.
2. The envelope of the spectrum of a periodic signal is the Fourier transform of the pulse shape.
To illustrate the first concept given above, consider the Fourier Transform of a rectangular pulse shown
in Fig. 3.11. It can be found from the frequency domain graph (Fig. 3.11) that 90% of the energy is
concentrated within the main lobe in frequency. As the pulse width is decreased in time, or T decreases,
the width of the main lobe in the frequency domain spreads, thus spreading out the energy of the signal.
To illustrate the second concept, recall that when two signals are convolved in the time domain, their trans-
forms are multiplied in the frequency domain.
Any periodic signal can be represented mathematically as the basic pulse shape convolved with a time
sequence of pulses. Thus, the spectrum of the periodic signal can be derived as shown in Figure 2.12.

3.6 Procedure:
3.6.1 Sine Wave
In this section you will observe when the signal frequency of a sine wave is shifted, the spectral line will also
shift in frequency.
Time Domain

a. Set the Function Generator for a 1 KHz sine wave output. Set the amplitude to 0.5 volts peak with zero
DC offset.
b. Connect the Channel 1 output of the Function Generator to Channel 1 input of the Oscilloscope using a
BNC to BNC cable. Turn on the Channel 1 output of the Function Generator.
c. View the time domain waveform on the Oscilloscope.
Hint for RS RTM 3004 Oscilloscope: See EXPERIMENT 1, Section 5.1.c
Frequency Domain

a. View the frequency domain waveform of the input signal on the Oscilloscope.
Set the timebase to 2s. Wait for the sampling process shown on the Pre touch button to reach Post 100%
until both time domain and frequency domain are displayed for the signal. Adjust the center frequency to
1KHz.( i.e try to adjust the stop frequency to 2KHz)
Hint for RS RTM 3004 Oscilloscope: See EXPERIMENT 1, Section 5.2.a
b. Determine the amplitude and frequency displayed for the spectrum of the sine wave.

Page 33
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Use the measure button Meas to measure the amplitude and frequency displayed. When the right menu
on he screen appears choose Type by clicking on Vertical button then Amplitude. Similarly by clicking on
Horizontal button then Frequency. On the lower left corner the measurement value will appear.
c. Save the Screen Image to your USB Flash Drive to include in your lab report.
d. Change the Function Generator frequency to 5 KHz. Determine the amplitude and frequency displayed.
(i.e don’t forget to adjust the right sampling rate and frequency ranges as in Time domain part)
e. Save the Screen Image to your USB Flash Drive to include in your lab report.
Sampling Windows
A sine-wave signal is initially measured as an output signal from the signal generator.
Set the following on the function generator:

FUNCTION Sinusoid

FREQUENCY 20 kHz

AMPLITUDE 500 mV (Vpp)

• Set the timebase for the oscilloscope to 2ms and select a maximum signal segment for the FFT.
• Select the minimum frequency band for these settings and verify that the spectrum of the sinusoidal
signal is displayed correctly.
• Press the FFT button to open the menu again.
• Display the result in dBV.

The RTC3004 offers four different window functions.

a. Explain why the window functions are needed, and explain the differences between the various functions.
(For detailed information please see the RTC3004 manual)

b. Compare how each of the window functions affects the spectrum

3.6.2 Square and Rectangular Waves

A rectangular wave contains many sine wave harmonic frequencies. When the reciprocal of the duty cycle
(either the positive or negative duty cycle) is a whole number, the harmonics corresponding to multiples of
that whole number will be missing. For example, if the duty cycle is 50%, then 1/0.5 = 2. Thus, the 2nd,
4th, 6th, etc. harmonics will be missing, i.e. zero. A rectangular wave with a 50% duty cycle is a square
wave. For the square wave, the magnitude of the harmonic will be inversely proportional to the harmonic’s
number. For example, if the magnitude of the 1st harmonic is A, then the magnitude of the 3rd harmonic is
A/3, the 5th harmonic’s magnitude is A/5, etc.
a. Set the Function Generator for a 1 KHz square wave output.
b. Determine the amplitude and frequency displayed for the 1st, 3rd, 5th, and 7th harmonics in the spectrum
of the square wave. Identify the zeros of the spectrum envelope (i.e. the frequency locations at which
harmonic amplitudes are zero). For this question you should use the cursors. Press the Cursor button.
Then choose the Source to the Spectrum of the required channel. Choose Type to Vertical. A cursor

Page 34
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

will appear move it over the required harmonic in order to measure its amplitude and frequency. You can
achieve that by either moving the cursor manually or clicking on the buttons Next Peak, Prev. Peak to move
the cursor to different harmonics. On the lower right corner the cursor measurement will appear. Be sure
to adjust the measurement type through using the right type of cursor being Horizontal or Vertical.
c. Save the Screen Image to your USB Flash Drive to include in your lab report.
d. The duty cycle of a pulse is measured as illustrated in Fig. 3.10. Switch the Function Generator to Pulse
and change the duty cycle of the 1 KHz rectangular wave to 10
e. Save the Screen Image to your USB Flash Drive to include in your lab report.
f. Change the duty cycle of the 1 KHz rectangular wave to 25%. Determine the amplitude and frequency of
the 1st, 3rd, 5th, 6th, and 7th harmonics. Determine the location of zeros.
g. Save the Screen Image to your USB Flash Drive to include in your lab report.
h. Change the duty cycle of the 1 KHz square wave to 90%. Determine the amplitude and frequency of the
1st, 2nd, 3rd, 4th, and 5th harmonics. Determine the location of zeros.
i. Save the Screen Image to your USB Flash Drive to include in your lab report.

Fig. 3.10: Duty cycle of a square wave signal

3.6.3 Triangular Wave

The triangular wave, like the square wave, only contains odd numbered harmonics. The magnitude of
the harmonics is inversely proportional to the square of the harmonic’s number. For example, if the 1st
harmonic’s magnitude is A, then the 3rd harmonic’s magnitude will be A/9, the 5th harmonic’s magnitude
A
will be 25 , etc.
a. Set the Function Generator for a 1 KHz triangular wave (Ramp) output. Determine the amplitude and
frequency of the 1st, 3rd, and 5th harmonics.
b. Determine the location of zeros.
c. Save the Screen Image to your USB Flash Drive to include in your lab report.

3.7 Calculations and Questions:

a. Compare the waveforms you obtained for rectangular waves on the basis of harmonic location, zero
location, and peak spectrum amplitude (i.e. first harmonic level). Explain the results on the basis of the
pulse width and signal frequency.
b. Use your data to show that the spectrum of a triangular wave is equal to the spectrum of a square wave
squared.
c. The spectrum of the square wave should decay proportional to ω1 (where ω = 2πf ). The spectrum of the
triangular wave should decay proportional to ω12 . Verify this using your measurements.

Page 35
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Fig. 3.11: Rectangular function and its frequancy representation(Fourier transform)

Page 36
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

Fig. 3.12: Mathematical representation of a periodic signal in the time and frequency domains

Page 37
EEL 3552C Lab Manual 3. EXPERIMENT: PERIODIC SIGNAL SPECTRA

3.8 Hints & Comments

Students need to distinguish between two ways of calculating the power.

Trigonometric series:

• For the DC level, they need to use the coefficient squared of the amplitude.
• For other harmonics, they should used half the coefficient squared of the amplitude.

Exponential series:

• For both DC and other harmonics, they should use coefficient squared of the amplitude.

When finding the spectrum on the oscilloscope (amplitude vs frequency), the amplitudes should always be
positive because these are the RMS values of the coefficients of the exponential series harmonics.

Page 38
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

4 EXPERIMENT: Low Pass Filter

4.1 Objective
To observe some applications of low-pass filters and to become more familiar with working in the frequency
domain.

4.2 Equipment:
The equipment used in this experiment are:

Oscilloscope: Rohde & Schwarz RTM 3004

Function Generator, Tektronix AFG 3022B
Digital Multimeter, Tektronix DMM 4050
Triple Output Power Supply, Agilent E3630A
Op Amp module TL084
Bring a USB Flash Drive to store your waveforms.

4.3 Theory
4.3.1 Systems
A communication system consists of three major components: the transmitter, the channel and the receiver.
The transmitter and the receiver are comprised of a cascade of black boxes that accept input signals,
produce output signals, and they are referred to as systems. This section is devoted to understanding
what a system does, and clarifying the various types of systems used in the transmitter and receiver for a
communication system.
Definition: A system is a rule for producing an output signal g(t) due to an input signal f (t).
If we denote the rule as T [.], then

g(t) = T [f (t)] (4.1)

An electric circuit with some voltage source as the input and some current branch as the output is an
example of a system.
Note that for two systems in cascade, the output of the first system forms the input to the second system,
thus forming a new overall system. If the rule of the first system is T1 and the rule of the second system
is T2 , then the output of the overall system due to an input f (t) applied to the first system is equal to g(t),
such that

g(t) = T2 [T1 [f (t)]] (4.2)

There are a variety of classifications of systems that owe their name to their properties. In this section we will
focus only in the classification of systems into the linear versus nonlinear categories and the time-invariant
versus time-variant categories.

Page 39
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

If a system is linear then the principle of superposition applies. The output of a system with rule that
satisfies the principle of superposition exhibits the following property:

T [a1 f1 (t) + a2 f2 (t)] = a1 T [f1 (t)] + a2 T [f2 (t)] (4.3)

Where a1 and a2 are arbitrary constants. A system is nonlinear if it is not linear. A linear system (friendly
system) is usually described by a linear differential equation of the following form:

an (t)g n (t) + an−1 g n−1 (t) + ... + a1 (t)g 1 (t) + a0 (t)g 0 (t) = f (t) (4.4)

Where g(t) designates the output of the system, while f (t) designates the input of the system. In the above
equation g k (t) denotes the k-th time derivative of the function g(t).
Virtually every system that you considered in your circuit classes (e.g. first order R - C , R - L circuit, second
order R - L - C circuits) is an example of linear systems. Any circuit that has components, whose v - i curve
is nonlinear (e.g., a diode) is likely to be a nonlinear system.
Another useful classification of systems is into the categories of Time-Invariant versus Time-Varying sys-
tems. A system is time-invariant if a time-shift in the input results in a corresponding time-shift in the
output. Quantitatively, a system is time-invariant if

g(t − t0 ) = T [f (t − t0 )] (4.5)

For any t0 and any pair of f (t) and g(t), where f (t) denotes an input to the system and g(t), denotes its
corresponding output. A system is time-varying if it is not time-invariant.
As we emphasized before, a linear system is described by a linear differential equation of the form shown
in equation (4.4). If in the differential equation (4.4) the coefficients ak (0 ≤ k ≤ n) are constants and not
functions of time, then we are dealing with a linear time-invariant system. Otherwise, we are dealing with a
linear time-varying system. Examples of linear time-varying systems from your circuit classes were circuits
consisting of R, L and C components and involving one or more switches that were ON or OFF at special
instances.

4.3.2 The Convolution Integral

The convolution integral appears quite often in situations where we deal with a linear time invariant system
(LTI). If this system is excited by an input f (t) and the impulse response of the system is h(t), then the
output g(t) of the system is equal to the convolution of f (t) and h(t), as illustrated below

Z ∞ Z ∞
g(t) = f (t) ∗ h(t) = f (t − τ )h(τ )dτ = h(t − τ )f (τ )dτ (4.6)
−∞ −∞

Note that the impulse response of a linear, time-invariant system is defined to be the output of the system
due to an input that is equal to an impulse function, located at t = 0. In this manual we are going to
demonstrate the convolution of two rectangular pulses.

4.3.3 Convolution Examples

Consider two rectangular pulses, f1 and f2 , depicted in Fig. 4.1. We are going to find the convolution of
these two rectangular pulses. We designate the convolution by f3 (t). That is

Page 40
 EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.1: Two rectangular pulses to be convolved

Z ∞
f3 (t) = f1 (t) ∗ f2 (t) = f1 (t − τ )f2 (τ )dτ (4.7)
−∞

We want to compute the convolution of f1 (t) and f2 (t). It seems that we have to compute the convolution
integral for infinitely many instances of time t (see equation (4.7)). A more careful observation though allows
us to distinguish the distinct t-range over which the convolution integral needs to be evaluated. One way of
finding the distinct t-range is by remembering that equation (4.7) tells us that we need to slide the rectangle
f1 with respect to the stationary rectangle f2 and identify the product of these two pulses over the common
interval of overlap. It is not difficult to show that the sliding rectangle f1 (actually f1 (t − τ )) is nonzero over
the τ -interval (t − 2, t − 1). We now distinguish five cases:

Case A: f1 (t − τ ) is to the left of f2 (τ ), and f1 and f2 do not overlap (see Fig. 4.2).
In order for Case A to be valid we have to satisfy the constraint that t − 1 < 3 or t < 4 . Then it is easy
to show that

f3 (t) = 0 (4.8)

Case B: f1 (t − τ ) is to the left of f2 (τ ), and overlap partially from the left (see Fig. 4.3)
In order for Case B to be valid we have to satisfy the constraint that 3 < t − 1 < 4 or 4 < t < 5 . Then
is easy to show that

Z t−1
f3 (t) = dτ = t − 4 (4.9)
3

Case C: f1 (t − τ ) is completely overlapping with f2 (τ ) (see Fig. 4.4).

In order for Case C to be valid we have to satisfy the constraint that 4 < t − 1 < 5 or 5 < t < 6 Then it
is easy to show that

Z t−1
f3 (t) = dτ = (t − 1) − (t − 2) = 1 (4.10)
t−2

Page 41
 EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.2: Convolution under Case A

Fig. 4.3: Convolution under Case B

Case D: f1 (t − τ ) is to the right of f2 (τ ), and f1 overlap f2 partially from the right (see Fig. 4.5).
In order for Case D to be valid we have to satisfy the constraint that 5 < t − 1 < 6 or 6 < t < 7. Then
it is easy to show that

Z 5
f3 (t) = dτ = 5 − (t − 2) = 7 − t (4.11)
t−2

Case E: f1 (t − τ ) is to the right of f2 (τ ), and f1 and f2 do not overlap (see Fig. 4.6).
In order for Case E to be valid we have to satisfy the constraint that 6 < t − 1 or t > 7. Then is easy
to show that

f3 (t) = 0 (4.12)

Hence, combining all the previous cases, we get




0 f or t<4
t − 4 f or 4<t<5



f3 (t) = 1 f or 5<t<6 (4.13)

7−t f or 6<t<7





0 f or t>7

A plot of the function f3 (t) is shown in Fig. 4.7.

Based on our previous computations we can state certain rules pertaining to the convolution of two rectan-
gles. These rules can be proven following the technique laid out in the previous example. These rules can
be used to find the convolution of two rectangle pulses, without actually having to compute the convolution.

Page 42
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.4: Convolution under Case c

Fig. 4.5: Convolution under Case D

Fig. 4.6: Convolution under Case E

Fig. 4.7: Function f3 (t) ( the result of the convolution of f1 (t) and f2 (t)

Page 43
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.8: Convolution of two Rectangular Pulses

4.3.4 Golden Rules for the Convolution of two Rectangle Pulses

Consider a rectangular pulse f1 (t) with amplitude over the interval [x1 , x2 ], and another rectangular pulse
f2 (t) with amplitude over the interval [y1 , y2 ]. Denote their convolution by f3 (t). Then,

1. The function f3 (t) is a trapezoid.

2. The starting point of the trapezoid is at position x1 + y1 .
3. The first breakpoint of the trapezoid is

(a) At position x1 + y2 if pulse f2 (t) is of smaller width than pulse f1 (t).

(b) At position x2 + y1 if pulse f1 (t) is of smaller width than pulse f2 (t).

4. The second breakpoint of the trapezoid is

(a) At position x2 + y1 if pulse f2 (t) is of smaller width than pulse f1 (t).

(b) At position x1 + y2 if pulse f1 (t) is of smaller width than pulse f2 (t).

5. The end point of the trapezoid is at position x2 + y2

6. The maximum amplitude of the trapezoid is

(a) Equal to A1 A2 (x2 − x1 ) if the f1 (t) is of smaller width.

(b) Equal to A1 A2 (y2 − y1 ) if the f2 (t) is of smaller width.

A plot of f1 (t), f2 (t) and f3 (t) is shown in Fig. 4.8. The technique, described above, to compute the
convolution of two rectangular pulses can be generalized, in a trivial way, to compute the convolution of
arbitrary shaped, finite-duration pulse.
All of the above rules pertaining to the convolution of two rectangular pulses can be proven rigorously.
They have been verified above for an example case. In the special case where the pulse (rectangle) f1 (t)
and f2 (t) are defined to be nonzero over an interval of the same width convolution f3 (t) turns out to be a
triangle. Rules 2 and 5, stated above, for the convolution of two rectangular pulsed extend for the case of
the convolution of arbitrary shaped, finite-duration pulses.
Some useful properties of the convolution are listed below. These properties can help us compute the
convolution of signals that are more complicated than the rectangular pulses.

Page 44
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

1. Commutative Law:

f1 (t) ∗ f2 (t) = f2 (t) ∗ f1 (t) (4.14)

2. Distributive Law:

f1 (t) ∗ [f2 (t) + f3 (t)] = f1 (t) ∗ f2 (t) + f1 (t) ∗ f3 (t) (4.15)

3. Associative Law:

f1 (t) ∗ [f2 (t) ∗ f3 (t)] = [f1 (t) ∗ f2 (t)] ∗ f3 (t) (4.16)

4. Linearity Law:

[αf1 (t)] ∗ f2 (t) = α[f1 (t) ∗ f2 (t)] (4.17)

Where α in the above equation is a constant.

4.3.5 Impulse Function

The impulse function showed up in the discussion of the convolution integral and linear time-invariant sys-
tem. We said then, that the convolution integral allows us to compute the output of a linear time-invariant
system if we know the system’s impulse response. The impulse response of a system is defined to be the
response of the system due to an input that is the impulse function located at t = 0.
The impulse or delta function is a mathematical model for representing physical phenomena that take
place in a very small time duration, so small that it is beyond the resolution of the measuring instrument
involved, and for all practical purposes, their duration can be assumed to be equal to zero. Examples of such
phenomena are a hammer blow, a very narrow voltage or current pulse, etc. In the precise mathematical
sense, the impulse signal, denoted by

δ(t)

is not a function (signal), it is a distribution or a generalized function. A distribution is defined in terms of its
effect on another function (usually called "test function") under the integral sign.
The impulse distribution (or signal) can be defined by its effect on the "test function" φ(t), which is assumed
to be continuous at the origin, by the following relation:

(
Z b
φ(0) f or a < 0 < b
φ(t)δ(t) =
a 0 otherwise

This property is called the sifting property of the impulse signal. In other words, the effect of the impulse
signal on the "test function" φ(t) under the integral sign is to extract or sift its value at the origin. As it is
seen, δ(t) is defined in terms of its action on φ(t) and not in terms of its value for different values of t.

Page 45
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.9: Function sequence definition of impulse function: (a) rectangular pulses; (b) triangular
pulses; (c) two-sided exponentials; (d) Gaussian pulses.

One way to visualize the above definition of the impulse function is to think of the impulse function as a
function determined via a limiting operation applied on a sequence of well-known signals. The defining
sequence of signals is not unique and many sequences of signals can be used, such as

1. Rectangular Pulse:

1 τ τ
δ(t) = lim [U (t + ) − U (t − )]
τ →∞ τ 2 2

2. Triangular Pulse:

1 |t|
δ(t) = lim [1 − ]
τ →0 τ τ

3. Two-sided exponential:

1 −2 |t|
δ(t) = lim e τ
τ →0 τ

4. Gaussian Pulse:

Page 46
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.10: Function δ(t)

1 −π( t )2
δ(t) = lim e τ
τ →0 τ

The above functions are depicted in Fig. 4.9.

The impulse function that we have defined so far was positioned at time t = 0. In a similar fashion we
can define the shifted version of the impulse function. Hence, the function δ(t − t0 ) designates an impulse
function located at position (time) t0 , and it is defined as follows:

(
Z b
φ(t0 ) f or a < t0 < b
φ(t)δ(t − t0 ) =
a 0 otherwise

The impulse function has a number of properties that are very useful in analytical manipulations involving
impulse functions. They are listed below:

• Area (Strength):
The impulse function δ(t) has unit area. That is

Z b
δ(t − t0 )dt = 1 where a < t0 < b
a

• Amplitude:

δ(t − t0 ) = 0 f or allt 6= t0

• Graphic representation:
See Fig. 4.10.
• Symmetry:

δ(t) = δ(−t)

Page 47
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

• Time Scaling:

1
δ(at) = δ(t)
|a|

• Multiplication by a time function:

f (t)δ(t − t0 ) = f (t0 )δ(t − t0 )

• Relation to the Unit Step Function:

The unit step function is the function defined by

(
1 f or t ≥ t0
U (t − t0 ) =
0 f or t < t0

It is not difficult to see that

Z t
δ(τ − t0 )dτ = U (t − t0 )
−∞

And that

d
δ(τ − t0 )dτ = U (t − t0 )
dt

Similar to the definition of δ(t) we can define δ (1) (t), δ (2) (t), ..., δ (n) (t) the generalized derivatives of δ(t) by
the following equation:

( n
b d
(−1)n dt
Z
(n) n φ(t)|t=0 f or a < 0 < b
δ (t)φ(t)dt =
a 0 otherwise

We can generalize this result to

(
b dn
(−1)n dt
Z
(n) n φ(t)|t=t0 f or a < t0 < b
δ (t − t0 )φ(t)dt =
a 0 otherwise

For even values of n , δ (n) (t)is even, and for odd values of n , δ (n) (t) is odd.

Page 48
 EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

4.3.6 The Frequency Transfer Function

Consider a linear-time invariant system, which is described by the following differential equation.

M K
X dm g(t) X dk f (t)
am = bk (4.18)
m=0
dtm dtk
k=0

In the above equation, f (t) represents the input to the system, and g(t) represents the output of the system.
0
The expression d dtf 0(t) denotes 0-th derivative of f (t); same convention holds for g(t).
Assume for a moment that the input f (t) to the above system is equal to ejωt . Then, we prove that the
output of the system is equal to g(t) = H(w)ejωt with

PK
bk (jω)k
H(ω) = PMk=0 (4.19)
m
m=0 am (jω)

In actuality we are contending that if the input to the system is a complex exponential function ejωt then the
output of the system will be the same exponential function times a constant H(ω) which depends on the
input parameters ω and the system parameters (the a’s and b’s) when we are referring to H(ω) as being
a constant we mean that it is independent of time. The constant H(ω) is called the transfer function of
the system. When the Fourier transform is introduced it will be easy to prove that the transfer function
of a system is the Fourier transform of the impulse response of the system. We have already stated that
the impulse response of a linear time invariant system describes the system completely. We can make
a similar statement about the transfer function of a linear, time-invariant system. That is, if we know the
transfer function of a LTI system we can compute the output of this system due to an arbitrary input. Since,
we are still operating in the context of Fourier series expansions, and since Fourier series expansions are
most appropriate for periodic signals, consider for a moment a periodic input f (t) applied to the system of
equation (4.18). Obviously, f (t) has a Fourier series expansion whose form is given below.

∞
X
f (t) = Fn ejnω0 t (4.20)
n=−∞

It is not difficult to demonstrate that in this case the output g(t) of our system will be equal to

∞
X
g(t) = H(nω0 )Fn ejnω0 t (4.21)
n=−∞

Equation (4.21) validates our claim that the frequency transfer function of a linear time invariant system is
sufficient to describe the output of the system due to an arbitrary input (at least for the case of an input
which is periodic). After the Fourier transform is introduced we will be able to extend this result to aperiodic
inputs as well. Looking at equation (4.21) more carefully we can make a number of observations:

2π
Observation 1: If the input to a linear time-invariant system is periodic with period T0 = ω0 then the output of this
system is periodic with the same period.
Observation 2: Only the frequency components that are present in the input of a LTI system can be present in the
output of the LTI system. This means that a LTI system cannot introduce new frequency components
other than those present in the input. In other words all systems capable of introducing new frequency
are either nonlinear systems or time-varying systems or both.

Page 49
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.11: Input signal f (t) to the filter in "Frequency Transfer Function Example"

Fig. 4.12: Transfer function of the filter in "Frequency Transfer Function Example"

4.3.7 Frequency Transfer Function Example

Let f (t) be a signal as shown in Fig. 4.11. This signal is passed through a system (filter) with transfer
function as shown in Figure 4.12. Determine the output g(t) of the filter.
We first start with the Fourier series expansion of the input. This can be obtained by following Fourier series
formulas.

∞
4 X (−1)n 2 5
f (t) = cos[2π(2n + 1)105 t] = ej2π10 t
π n=0 2n + 1 π
(4.22)
2 5 2 j6π105 t 2 −j6π105 t 2 j10π105 t 2 −j10π105 t
+ e−j2π10 t − e − e + e + e + ...
π 3π 3π 5π 5π
To find the output of the system we need to find the output of the system due to every complex exponen-
tial involved in the expansion of f (t). A closer inspection of H(ω) though tells us that only the complex
exponential within the passband of the filter will actually produce an output (the passband is the range of
frequencies from −12π105 rads to 12π105 rads.) Hence, using the already derived formula that gives the
output of an LTI system due to a periodic input, we can write:

Page 50
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

2 j(2π105 t+ π ) 2 −j(2π105 t+ π )
g(t) = e 2 + e 2
π π (4.23)
2 5 π 2 5 π 2 5 π 2 5 π
− ej(6π10 t+ 2 ) − e−j(6π10 t 2 ) + ej(10π10 t+ 2 ) + e−j(10π10 t+ 2 )
π π π π

4.3.8 Filters
Filtering has a commonly accepted meaning of separation - something is retained and something is re-
jected. In electrical engineering, we filter signals usually frequencies. A signal may contain single or multiple
frequencies. We reject frequency components of a signal by designing a filter that provides attenuation over
a specific band of frequencies, and we retain components of a signal through the absence of attenuation
or perhaps even gain over a specified band of frequencies. Gain may be defined as how much the input is
amplified at the output.
Filters are classified according to the function they have to perform. Over the frequency range we define
pass band and stop band. Ideally pass band is defined as the range of frequencies where gain is 1 and at-
tenuation is 0, and stop band is defined, as the range of frequencies where the gain is 0 and the attenuation
is infinite.
Filters can be mainly classified as low pass, high pass, band pass and band stop filters.
A low pass filter can be characterized by the property that the pass band extends from frequency ω = 0 to
ω = ±ωc where ωc is known to be the cutoff frequency. (See Fig. 4.13a)
A high pass filter is the complement of a low pass filter. Here the previous frequency range ω = 0 to
ω = ±ωc is the stop band and the frequency range from ω = ωc to positive infinity and from ω = −ωc to
negative infinity is the pass band. (See Fig. 4.13b)
A band pass filter is defined as the one where frequencies from ω1 to ω2 are passed while all other fre-
quencies are stopped. (See Fig. 4.13c)
A band stop filter is the complement of the band pass filter. Frequencies from ω1 to ω2 are stopped here
and all other frequencies are passed. (See Fig. 4.13d)
In Fig. 4.14 below a practical (non-ideal) 6th order Butterworth low-pass filter is shown.
In this the various filter characteristics such as pass band, cut-off frequency are clearly indicated. In Fig.
4.15 we are illustrating how the filter characteristics change with the order of the filter.
The following Fig. 4.15 shows how the order of a filter changes the frequency response. The higher the
order, the closer the filter behaves as ideal.
Filters are key components in communication systems. An ideal filter will only allow a specified set of
frequencies to pass from input to output with equal gain. However, no filter is ideal, and as a result there
are many types of filters, that are used in modeling an ideal filter closely in one or more aspects. Filters
have many applications besides separating a signal form surrounding noise or other signals. Some of these
applications are listed below.
1. Integration of a signal
2. Differentiation of a signal
3. Pulse shaping
4. Correcting for spectral distortion of a signal
5. Sample and hold circuits
Two Butterworth low pass filters (LPF’s) will be used in this lab: a first order and a fourth-order Butterworth
filter. The general transfer function of a Butterworth LPF is:

Page 51
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.13: The magnitude of the frequency transfer function of low-pass, high-pass, band-pass, band-stop
and all-pass. In all cases the responses are ideal.

Fig. 4.14: 6th order Low-pass Butterworth Filter

Page 52
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.15: Butterworth Filters (How the frequency response changes with order)

A
|H(ω)|2 = ω 2n (4.24)
(1 + ω0 )

Where ω0 = the filter cutoff frequency and

n = the order of the filter
The actual transfer functions of the filters that are going to be used in this lab are:
1st order LPF:

1
|H(s)| = (4.25)
(1 + RCs)

4th order LPF:

R2 R4
(1 + R1 )(1 +
R3 )
|H(s)| = (4.26)
((RC)4 s4 + (3 − ( R1R+R
1
2
)RCs + 1)(RC)2 s2 + (3 − ( R1R+R
1
2
)RCs + 1))

1
where ω0 = RC for both filters.
Two applications for low pass filters will be illustrated in this lab:
1. A LPF used as an integrator
2. A LPF used as a first harmonic isolator (square-to-sine conversion).
Integrator
1
The Laplace representation of integration is s (where s = jω). A first order low pass filter can approximate
an integrator when ω >> ω0 ( VVin = 1s )
ω0

th
For an n order low-pass filter, the transfer function will approximate n integrators in series at ω >> ω0
( VVin = ( s1 )n )
ω0

For simplicity, in this lab a first order LPF will be used to perform a single integration on a signal.
Square-to-sine-Converter
Any periodic signal can be expressed as an infinite sum of orthogonal sinusoids. For orthogonality, each
sinusoid (or "harmonic") must have a different frequency, and frequency must be an integer multiple of the
frequency of the periodic signal.

Page 53
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.16: The harmonics of a square wave form of frequency fs

For example, if the frequency of a square wave is fs , then the nth harmonic will have a frequency of nfs
(where n is any integer from one to infinity). You will see that some of these harmonics will have zero or
negative amplitude.
The harmonics of a square wave of frequency fs are shown in Fig. 4.16 (Not to scale); the DC component
of the square wave is assumed to be equal to zero.
If the square wave excites the input of the low pass filter such that the first harmonic is in the pass band of
the filter transfer function and the remaining harmonics are outside of the pass band, then the output of the
low pass filter will be a single-tone signal (sinusoid).

4.4 Simulation
We will use MATLAB simulation to see the response of our filter circuits. For simulation you will need the
transfer functions of both filter circuits shown in Fig. 4.20 and Fig. 4.21. The transfer functions were given
earlier in this manual (See equation 4.25 and 4.26.)
In both filters and R = 1.6kΩ and C = 0.01F . For the second filter R1 = 10kΩ, R2 = 12kΩ, R3 = 10kΩ,
R4 = 1.5kΩ and C = 0.01F . With the above values for resistors and capacitors the transfer functions will
have the following form (notice that the coefficients are in descending order of s).

1
H1 =
1 + (RC)s

2.2 × 1.5
H2 =
(RC)4 s4 + 2.65(RC)3 s3 + 3.48(RC)2 s2 + 2.65RCs + 1

Construct a frequency array as following.

w=[0:100:100000];
Construct the coefficient vectors as follows:
b=[coefficients of numerator separated by comma];
a=[ coefficients of denominator separated by comma];
Use the MATLAB ’freqs’ function to generate frequency response of the corresponding filter.
Note: The freqs function is not available in all MatLab installations. It is installed in all EE Laboratories.
Plot the frequency response of both filters in a single graph using the following a MATLAB function.
Apply the above procedure for both filters. Note that the frequency response you obtained from the MATLAB
simulation that is gain versus angular frequency in radian. You can also plot gain versus frequency (Hz). In
ω
that case you have to use the f = 2π vector as the parameter in plot function.

Page 54
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.17: The simulated gain versus frequency (Hz) plot for both filters (2nd order and 4th order)

The simulated gain versus frequency (Hz) plot for both filters (2nd order and 6th order) is shown above as
Fig. 4.17.
Compare the response of two filters. After the hardware experiment compare the experimental results with
the simulated results.

4.5 Prelab Questions:

1. Calculate the Trigonometric Fourier Series of a square wave with the parameters shown in Fig. 4.18
(assume it is periodic with periodic T ):

Fig. 4.18: A periodic square wave with period T

You may find it easier to add a DC level of A volts. This will only change your spectrum at f = 0, where you
can subtract the DC back out.
2. Assuming an ideal filter with a cutoff frequency of ωc (ωc = 2π
T , show graphically, in the frequency domain,
why the first harmonic of the square wave should be the only harmonic at the output.
3. Suppose you have an input signal with frequency components from 0 to 2 KHz. Is it possible to design a
filter, which will produce an output with frequency components 4 to 6 KHz? Why or why not?
4. What is the difference between dB/octave and dB/decade?

4.6 Procedure
4.6.1 Components
In this lab, we will use Op Amp module TL084, which is shown in Fig. 4.19.

Page 55
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.19: Pin layout of OpAmp module TL084

Fig. 4.20: First order low-pass filter

4.6.2 Filters
a) Build the first order low-pass filter shown in Fig. 4.20.
b) Build the fourth order low-pass filter shown in Fig. 4.21. Keep both circuits built for the remainder of the
lab.

4.6.3 Calculating Filter Parameters

Cutoff Frequency
a) From the function generator, obtain a 0.2 volt, low frequency (10 or 100 Hz range) sine wave. It is
important to use a sine wave because there is only one harmonic in a sinusoid, and you only want to test
the transfer function at one point in frequency at time. Connect the function generator output to the filter
input.
b) Connect the filter output to either the oscilloscope or to the DMM4050 Digital Multimeter. When using the
DMM, select dB measurements.
c) Vary the input frequency by a factor of 2 or 3 either way and make sure the output level doesn’t vary
significantly. If it doesn’t, then your input frequency is within the pass band. When you know you are within
the pass band, measure the output voltage level (or dB level). This will be your reference level. (i.e. Use

Page 56
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

Fig. 4.21: 4th -order low-pass

Meas button and choose type Peak Peak on its menu) The Vpp value will appear at the lower left corner of
the oscilloscope screen.
d) Vary the input frequency until the output level drops 3 dB below the reference level. You can achieve that
by keeping the measure on Vpp and rotating the knob on the functional generator while the cursor is on the
frequency measurement. On the oscilloscope, -3 dB is equivalent to the voltage level reduced by a factor
of .707 (or, at −3dB point,Vout (−3dB) = .707(Vout (ref )). Note this frequency. This is your cutoff frequency.
N.B. Be sure to fit the two scales manually by knobs in case Autoset doesn’t set well.
Op-amp voltage +15V , it can reach +20V
if clipping voltage with Autoset then manually change timebase.
Passband Gain
a) Using the function generator, generate a 0.2 volt sine wave at a frequency that you know to be within the
filter passband. Connect the function generator output to the filter input.
Vo Vo
b) Measure the output voltage (0V). The filter passband gain is equal to Vin = 0.2
Rate of Rolloff
a) Use the function generator to generate a sinusoid at a frequency that is higher than the cutoff frequency
of the filter (maybe by a factor of 2 or so). Record the input amplitude (Vin ) or measure in dB on the digital
multimeter (you may want to use an input amplitude of 1 volt or more since the output will be attenuated).
b) Connect the function generator output to the filter input. Measure the filter output in volts (Vo ) or dB.
Calculate VVin
o
and convert to dB (20log VVin
o
) or subtract the input dB level from the output dB level. This will
give you the attenuation, in dB, at your selected frequency.
c) Repeat steps 1 and 2 with a frequency that is a factor of 2 higher than the first frequency used. If you
take the difference between the two calculated values of attenuation, in dB, this will be your rate of rolloff in
dB/octave. If you want to convert, (x) dB/octave corresponds to (x ∗ (20/6)) dB/decade .

4.6.4 Square Wave Spectrum

a) Set the function generator for a square wave of 0.2 volts amplitude and 1 kHz frequency. Connect the
output of the function generator to Channel 1 input of the oscilloscope.
b) Use the Oscilloscope to display the spectrum analysis. You should see the first six nonzero harmonics
of the square wave displayed. Record the amplitude (in dBV) of the first six nonzero harmonics and the

Page 57
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

frequency of each harmonic. Be sure to adjust the timebase to 100 ms or above, for a display that can
show you the needed spectrum. Choose measure frequency by pressing the Meas button and changing
the Type, choose Cursor by pressing its button. A menu will appear, on Source choose FFT: Spectrum.
Then press on Next Peak to measure the frequency at that harmonic.

4.6.5 Integrator
a) Generate a square wave of amplitude 5V (peak) and frequency = 10 × ωc (ten times the cutoff frequency).
Apply this signal to the input of the filter from Fig. 4.20 (the first order filter)
b) Sketch the output of the filter from the oscilloscope, recording signal peak-to-peak amplitude. You can
achieve that by pressing the Meas button and adjusting the Type to Peak Peak by plugging the Basic menu
button. The peak to peak amplitude measurement will appear on the bottom left corner of the oscilloscope
screen.

4.6.6 Square-to-sine Conversion (First Harmonic Isolation)

a) Obtain a 0.2 volt peak square wave with frequency f = 0.75 × fc from the function generator.
b) Apply the signal from (a) to the input of the filter in Fig. 4.20 Sketch the output signal in both the time and
frequency domains, scaling axes.
c) Repeat b) with the filter in Fig. 4.21.

4.7 Calculations and Questions

1.Convert the values you measured for the square wave in Section 4.5.4 in dBV to volts and compare to
your Fourier Series Coefficients calculated in the Prelab. Recall that the measured amplitudes are RMS
amplitudes, so the values must be multiplied by the square root of 2.
2. Compare the expected results for filter gain and filter roll off rate to your measured results.
3. Using the waveform shown in question (1), Integrate (in time) to show that the integral of a square wave
is a triangular wave. What peak amplitude do you calculate? Recall that you have to solve an indefinite
integral to get an answer in functional form.
4. When the first harmonic of the input signal is well beyond the filter cutoff frequency, a first order low pass
filter cutoff frequency, a first order low pass filter approximates an integrator (shown in ’Integrator’ section
of lab introduction). Show that this transfer function, when applied to your square wave, will closely predict
the output amplitude you measured. Note that this transfer function can be reduced to a constant multiplied
by the integral of the input signal. Also note that your DC level may vary from that calculated, because the
DC level will depend on at what point on the square wave the integration began.

Page 58
EEL 3552C Lab Manual 4. EXPERIMENT: LOW PASS FILTER

4.8 Hints & Comments

• Students need to answer all the questions for the two types of filters (First Order Low Pass Filter and
Fourth Order Low Pass Filter).

• When implementing the Fourth Order Low Pass Filter, students should use the Op-Amp module
TL084. This module consists of four Op-Amps. The module needs to be stimulated with a DC volt-
age. A typical DC voltage value is +15V to +VCC Pin 4 and −15V to −VEE Pin 11 from the DC Power
Supply.

• When computing the Roll Off value of the filter, the frequencies used should be double the cut off
frequency of the filter or more.

• When using the Low Pass Filters at a high frequency, above the cut off frequency (110 kHz) with an
input signal of a Square wave, the output should be a triangular wave.

• When using a higher order Low Pass Filter, the output result should be smoother (better resolution).

Page 59
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

5 EXPERIMENT: Amplitude Modulation

5.1 Objective
To understand the amplitude modulation.

5.2 Equipment:
The equipment used in this experiment are:

Oscilloscope: Rohde & Schwarz RTM 3004

Function Generator: Tektronix AFG 3022B
Digital Multimeter, Tektronix DMM 4050
Triple Output Power Supply, Agilent E3630A
Op Amp module TL084
Bring a USB Flash Drive to store your waveforms.

5.3 Theory
5.3.1 From Fourier series to the Fourier Integral
The Fourier series is a means for expanding a periodic signal in terms of complex exponentials. This
expansion considerably decreases the complexity of the description of the signal, and simultaneously, this
expansion is particularly useful when we analyze LTI systems.
We can extend the idea of the Fourier series representation from periodic signals to the case of nonperiodic
signals. That is the expansion of a nonperiodic signal in terms of complex exponentials is still possible.
However, the resulting spectrum is not discrete any more. In other words, the spectrum of nonperiodic
signals covers a continuous range of frequencies. To be able to demonstrate our point, consider for a
moment the periodic signal shown in Fig. 5.1.

Fig. 5.1: A periodic signal fT (t)

We call this periodic signal fT (t) to remind ourselves that we are dealing with a periodic signal of period T .
We can write:

(
−T
fT (t) 2 < t < T2
f (t) = (5.1)
0 otherwise

Page 60
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

where f (t) is equal to the periodic signal fT (t) over the interval ( −T T
2 , 2 ) and zero otherwise. Obviously,
fT (t) has a Fourier series expansion, such that

∞
X
fT (t) = Fn ejnω0 t (5.2)
n=−∞

With

Z T
1 2
Fn = fT (t)e−jnω0 t dt
T − T2
Z T
(5.3)
1 2
−jnω0 t
= f (t)e dt
T − T2

Also, since fT (t) and f (t) are equal over the interval (− T2 , T2 ) we can write that

∞
X
f (t) = Fn ejnω0 t
n=−∞
∞
(5.4)
X1
= F (ωn )ejωn t
n=−∞
T

where

F (ωn ) = T Fn (5.5)

and

ωn = nω0 (5.6)

Let us now define

2π
∆ω = ωn+1 − ωn = ω0 = (5.7)
T

Then, from equation (5.7) we get

∞
X ∆ω
f (t) = F (ωn )ejωn t (5.8)
n=−∞
2π

Let now T → ∞ , or equivalently, ∆ω → 0. It is easy to see that the equation (5.8) becomes:

Z ∞
1
f (t) = F (ω)ejωt dω (5.9)
2π −∞

Page 61
 EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Also, it is not difficult to show that

Z ∞
F (ω) = f (t)ejωt dω (5.10)
−∞

Equations (5.9) and (5.10) are referred to as the Fourier Transform pair, that allows us to transform the
aperiodic time signal f (t) into the frequency signal F (ω), and vice versa. Based on the aforementioned
discussion we can now formally introduce the Fourier transform theorem.

5.3.2 Fourier Transform Theorem

If the signal f (t), satisfies certain conditions known as the Dirichlet conditions, namely,

1. f (t) is absolutely integrable on the real line, that is,

Z ∞
|f (t)|dt < ∞ (5.11)
−∞

2. The number of maxima and minima of f (t) in any finite interval of the real line is finite.
3. The number of discontinuities of f (t) in any finite interval of the real line is finite.

Then, the Fourier transform (FT) of f (t), defined by

Z ∞
F (ω) = f (t)e−jωt dt (5.12)
−∞

exists and the original signal can be obtained from its FT by following the rule

Z ∞
1
f± = F (ω)ejωt dω (5.13)
2π −∞

where

(
f (t) if f (t) is continuous at t
f± (t) = f (t+ )+f (t− ) (5.14)
2 if f (t) is not continuous at t

Observation 1: The Fourier transform, F (ω) of a signal f (t), is, in general, a complex function. Its magnitude |F (ω)|
and phase ∠F (ω) represent the amplitude and phase of various frequency components of f (t).
Observation 2: If the independent variable in the FT of a signal f (t) is chosen to be the frequency f instead of the
angular frequency ω. we have the following Fourier transform pair:

Z ∞
X(f ) = x(t)e−j2πf t dt < ∞ (5.15)
−∞

and

Z ∞
x(t) = X(f )ej2πf t df (5.16)
−∞

Page 62
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.2: A periodic signal fT0 (t)

5.3.3 Fourier Series to Fourier Transform Example

Let fT0 (t) be a periodic signal depicted in Fig. 5.2.
The Fourier series expansion of this signal, of period T0 , is given by the following equation.

∞
X
fT0 (t) = Fn ejnω0 t (5.17)
n=−∞

where

Z T0
1 2
Fn = f (t)e−jnω0 t dt (5.18)
T0 −
T0
2

and

2π
ω0 = (5.19)
T0

We can compute Fn ’s as follows:

Z T0
1 2
Fn = f (t)e−jnω0 t dt
T0 −
T0
2
Z T0
1 2 2πnt
= f (t)e−j T0
dt
T0 T
− 20
Z τ
1 2 2πnt
(5.20)
= f (t)e−j T0
dt
T0 − τ2
1 T0 nπt nπt
= ( )[e−j T0 − ej T0 ]
T0 −jn2π
sin( nπτ
T0 ) τ
= nπτ
T0 T0

The line spectrum of the signal fT0 (t) is shown in Fig. 5.3 for Tτ0 = k −1 . Note that the separation of
two consecutive lines in the line spectrum of f (t) is equal to 2π T0 . Hence, as T0 gets larger and larger
(i.e., as fT0 (t) becomes less and less periodic-looking) the line spectrum of fT0 (t) becomes denser and
denser. Also, note that the envelope of line spectrum of fT0 (t) is the Fourier transform (within a constant
of proportionality) of the aperiodic pulse of width τ , centered at t = 0. It can be shown that most of the

Page 63
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.3: The line spectrum of the signal fT0 (t)

power of fT0 (t) is located within the frequency interval [− 2π 2π

τ , τ ] This indicates to us that as we make
the signal faster in the time domain (by decreasing the duration τ of the pulses) the frequency content of
the signal gets expanded in the frequency domain. This statement is a manifestation of what is called in
communication theory the "time-bandwidth" product. It can be shown that the time- bandwidth" product is
constant; that is if we expand the signal in the time domain, we shrink the frequency content of the signal in
the frequency domain, and vice versa.

5.3.4 Fourier Transform Properties

The Fourier transform of an aperiodic signal is as useful as the Fourier series of a periodic signal. As a
reminder, the Fourier series of a periodic signal allows us to compute the output of a LTI system due to
a periodic input, without having to compute the convolution. Furthermore, the Fourier series of a periodic
signal provides us with information pertaining to the frequency content of the signal. Similarly, the Fourier
transform of a signal allows to compute the output of a LTI system due to an input, without having to compute
the convolution. Also, the Fourier transform of the signal contains information about the frequency content
of the signal. Equations (5.12) and (5.13) or equations (5.15) and (5.16) are rarely used for the computation
of the Fourier transform of signals. Normally, the Fourier transform of a signal is computed by utilizing
Fourier transforms of well known signals (see Table 5.1) and Fourier Transform Properties (see Table 5.2).

• The Linearity Property says that the Fourier Transform (FT) is a linear operation and if we know the
FT of two signals we can easily compute the FT of any linear combination of these two signals.

• The Scaling Property says that by contracting (expanding) the signal in the time domain (through
multiplication with an appropriate time constant) results in the expansion (contraction) of the FT of
the signal (through multiplication with the inverse of the constant); this is another ramification of the
"time-bandwidth" product principle. Note that contraction (expansion) of the signal in the time domain
corresponds to making the signal faster (slower).

• The Delay Property illustrates something that you might have suspected all along. Delaying a signal
in the time domain does not change the frequency content of the signal. The frequency content of the
signal is determined by looking at the magnitude of the FT of the signal. The delay property reiterates
that a time shift will produce a phase shift in the FT of the signal but it will leave the magnitude of the
FT unaltered.

• The Convolution Property assures us that we do not need to perform convolution to produce the
output of a LTI system due to an arbitrary input; simply multiplying the FT of the input and the FT of
the impulse response of the system allows us to compute the FT of the output of the system.
• The Time Differentiation and Time Integration properties demonstrate that making the signal faster/slower
(differentiate/integrate) in the time domain results in an expanded/contracted frequency spectrum
(FT).

Page 64
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

• The Duality Property provides us with a new FT pair, every time a FT pair is computed. In particular,
if F is the FT of f , the duality property says that F would have a FT that is proportional to the mirror
image of f with respect to the ω = 0 axis.
• Finally, the Amplitude Modulation property tells us that we can shift the frequency content of a
signal at an arbitrary frequency by multiplying the signal with a sinusoid. AM systems are based on
this property of the FT.

5.3.5 Introduction to Communication Systems

The primary function of most communication systems is to transmit information from point A to point B.
Consider, for example, the situation where we have to transmit human voice from point A to point B and the
channel between points A and B is the air. The human voice at point A is first converted, via a microphone,
to an electrical signal. Then, we have to transform this electrical signal into an electromagnetic wave and
send it to destination B. The way to achieve that is by using an antenna. Knowing though that the maximum
frequency content of a voice signal is around 15KHz we come to the conclusion that the antenna required
must be many miles long! A reasonable size antenna could adapt the signal to the channel only if the
signal’s lower frequency is higher than 800 KHz. Hence, in our case, where a voice signal needs to be
transmitted using a reasonable-size antenna it suffices to raise the frequency content of the voice signal to
the vicinity of 800KHz. To accomplish the aforementioned goal we ought to multiply the voice signal with a
sinusoidal signal. The result of this multiplication is to latch our (the voice signal) f (t) on the amplitude of
the sinusoid, or to modulate the original amplitude of the sinusoid with f (t). This procedure gives rise to a
class of systems that are called amplitude modulation systems.
We distinguish four major classes of amplitude modulation systems:

• Double Sideband Suppressed Carrier Systems (DSB-SC)

• Single Sideband Suppressed Carrier Systems (SSB-SC)

• Double Sideband Large Carrier Systems (DSB- LC)
• Vestigial Sideband Suppressed Carrier Systems (VSB-SC)

5.3.6 Double Sideband Suppressed Carrier System (DSB-SC)

Basically double sideband suppressed carrier (DSB-SC) modulation arises by multiplying the information
signal f (t) and the carrier wave c(t), as follows:

m(t) = c(t)f (t) = Ac cos(ωc t)f (t) (5.21)

Consequently, the modulated signal m(t) undergoes a phase reversal whenever the formation signal f (t)
crosses zero as indicated in Fig. 5.4. The envelope of a DSB-SC modulated signal is therefore different
from the information signal.
The Fourier transform of m(t) can easily be determined from the Fourier transform properties of Table 5.2.
That is,

1
M (ω) = Ac [F (ω − ωc ) + f (ω + ωc )] (5.22)
2

For the case when the baseband signal f (t) is bandlimited to the interval [−ωm , ωm ], as in Fig. 5.5(a),
we thus find that the spectrum of m(t) is as illustrated in Fig. 5.5(b). Except for a change in scale factor,

Page 65
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.4: The information signal f (t) and modulated signal m(t)

Fig. 5.5: (a) Spectrum of baseband signal. (b) Spectrum of DSB-SC modulated wave

Page 66
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.6: Coherent detection of DSB-SC modulated wave

Fig. 5.7: Illustrating the spectrum of a product modulator out put with a DSB-SC modulated wave as input

the modulation process simply translates the spectrum of the baseband signal by ±ωc . Note that the
transmission bandwidth required by DSB-SC modulation is equal to 2ωm , where ωm stands for the maximum
frequency content of the signal f (t).
The baseband signal f (t) can be uniquely recovered from a DSB-SC wave m(t) by first multiplying m(t)
with a locally generated sinusoidal wave and then low pass filtering the product, as in Fig. 5.6. It is assumed
that the local oscillator signal is exactly coherent or synchronized, in both frequency and phase, with the
carrier wave c(t), used in the product modulator to generate f (t). This method of demodulation is known
as coherent detection or synchronous demodulation.
It is instructive to derive coherent detection as a special case of the more general demodulation process
using a local oscillator signal of the same frequency but arbitrary phase difference φ, measured with respect
0
to the carrier wave c(t). Thus, denoting the local oscillator signal by Ac cos(ωc t + φ) and using equation (5.1)
for the DSB-SC wave, we find that the product modulator output in Fig. 5.7 is

0
e(t) = Ac cos(ωc t + φ)m(t)
0
= Ac Ac cos(ωc t)cos(ωc t + φ)f (t) (5.23)
1 0 1 0
= Ac Ac cos(2ωc t + φ)f (t) + Ac Ac cos(φ)f (t)
2 2

The first term in equation (5.23) represents the DSB-SC modulated signal with a carrier frequency of 2ωc ,
whereas the second term is proportional to the baseband signal f (t). This is further illustrated by the
spectrum (Fourier Transform) E(ω) shown in Fig. 5.7, where it is assumed that the baseband signal f (t)
is limited to the interval [−ωm , ωm ]. It is therefore apparent that the first term in equation (5.23) is removed
by the low-pass filter in Fig. 5.6, provided that the cut-off frequency of this filter is greater than ωm but less
than 2ωc − ωm . This is satisfied by choosing ωc > ωm . At the filter output we then obtain a signal given by

Page 67
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

1 0
o(t) = Ac Ac cos(φ)f (t) (5.24)
2

The demodulated signal o(t) is therefore proportional to f (t) when the phase error φ is a constant. The
amplitude of this demodulated signal is maximum when φ = 0, and is minimum (zero) when φ = ± π2 .
The zero demodulated signal, which occurs for φ = ± π2 , represents the quadrature null effect of the
coherent detector. Thus the phase error φ in the local oscillator causes the detector output to be attenuated
by a factor equal to cos(φ). As long as the phase error φ is a constant, the detector output provides an
undistorted version of the original baseband signal f (t). In practice, however, we usually find that the phase
error φ varies randomly with time, due to random variations in the communications channel. The result is
that at the detector output, the multiplying factor cos(φ) also varies randomly with time, which is obviously
undesirable. Therefore, provision must be made in the system to maintain the local oscillator in the receiver
in perfect synchronism, in both frequency and phase, with the carrier wave used to generate the DSB-SC
modulated signal in the transmitter.

5.3.7 Quadrature-Carrier Multiplexing

The quadrature null effect of the coherent detector may also be put to good use in the construction of the
so-called quadrature-carrier multiplexing or quadrature-amplitude modulation (QAM). This scheme
enables two DSB-SC modulated waves (resulting from the application of two physically independent infor-
mation signals) to occupy the same channel bandwidth, and yet it allows for the separation of two information
signals at the receiver output. It is therefore a bandwidth conservation scheme.
A block diagram of the quadrature-carrier multiplexing system is shown in Fig. 5.8.
The transmitter part of the system shown in Fig. 5.8(a), involves the use of two separate product modulators
that are supplied with two carrier waves of the same frequency but differing in phase by −90 degrees. The
transmitted signal m(t) consists of the sum of these two product modulator outputs, as shown by

m(t) = Ac f1 (t)cos(ωc t) + Ac f2 (t)sin(ωc t) (5.25)

where f1 (t) and f2 (t) denote the two different information signals applied to the product modulators. Thus
m(t) occupies a channel bandwidth of 2ωc centered at the carrier frequency ωc , where ωm is the maximum
frequency content of the signals f1 (t) and f2 (t).
The receiver part of the system is shown in Fig. 5.8(b).The multiplexed signal m(t) is applied simultane-
ously to two separate coherent detectors that are supplied with two local carriers of the same-frequency but
0
differing in phase by −90 degrees. The output of the top detector is 21 Ac Ac f1 (t), whereas the output of the
0
bottom detector is 21 Ac Ac f2 (t). For the system to operate satisfactorily it is important to maintain the correct
phase and frequency relationships between the local oscillators used in the transmitter and receiver parts
of the system.
To maintain this synchronization, we may use a Costas receiver. Another commonly used method is to send
a pilot signal outside the passband of the modulated signal. In the latter method, the pilot signal typically
consists of a low-power sinusoidal tone whose frequency and phase are related to the carrier wave c(t);
at the receiver the pilot signal is extracted by means of a suitably tuned circuit and then translated to the
correct frequency for use in the coherent detector.

5.3.8 The Single Sideband Suppressed Carrier System (SSB-SC)

The DSB-SC system discussed so far sends the signal m(t) over the channel with Fourier transform shown
in Fig. 5.9. One important observation regarding the Fourier transform of m(t) is that is consists of two

Page 68
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.8: Quadrature-carrier multiplexing system. (a) Transmitter. (b) Receiver.

Page 69
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.9: An amplitude modulation suppressed-carrier transmission

sidebands, the upper sideband and the lower sideband. There is no need to send both sidebands over
the channel, since the FT of the information signal can be constructed from either the upper or the lower
sideband. The single sideband suppressed carrier (SSB-SC), that we intend to discuss in the following,
takes advantage of the above observation by sending only one of the sidebands over the channel.
Hence, the SSB-SC system differs from the DSB-SC system because just before the modulated wave m(t)
is ready to head-off for the channel, one of the sidebands is chopped off. The block diagram of an SSB-
SC system is shown in Fig. 5.10. Furthermore, the sequence of Fourier transform plots shown in Fig.
5.11 (upper sideband) demonstrate that the system works. The only problem with the block diagram of the
system shown in Fig. 5.10 is that it requires the utilization of a "perfect filter" that eliminates completely the
frequencies a little bit below ωc and passes without distortion the frequencies a little bit above ωc (we are
referring only to the positive frequencies of the upper sideband signal). To relax the requirement for such a
"perfect" filter we will consider another way of generating the upper sideband version of the signal m(t).
For the sake of simplicity consider first the case where the information signal is sinusoidal (i.e., f (t) =
cos(ωm t)). Then, the Fourier transform of m(t) = cos(ωm t)cos(ωc t) is given in Fig. 5.12. From Fig. 5.12, it is
easy to identify the upper sideband of m(t) in the frequency domain; also, the time-domain representation
of this upper sideband is readily defined as follows:

1
m+ (t) = cos((ωc + ωm )t) (5.26)
2

or

1 1
m+ (t) = cos(ωc t)cos(ωm t) − sin(ωc t)sin(ωm t) (5.27)
2 2

In Fig. 5.13 we depict a way of constructing the upper sideband signal of equation (5.27) without having
to resort to the design of perfect filters. In Fig. 5.13 there is a system designated by a block with the −90
notation inside it. We know what this block does to an input signal that is sinusoidal; it shifts the phase of the
sinusoidal signal by −90 degrees. We want to be able to characterize this block in the frequency domain.

Page 70
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.10: Block diagram of a SSB-SC system

Fig. 5.11: FT’s at various stage of the SSB-SC system of Fig. 5.10

Page 71
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.12: Fourier Transform of m(t) = cos((ωm + ωc )t)

To do so let us identify the Fourier transforms of an input signal to this block, equals to cos(ωm t) and the
output signal produced by this block, equals to cos(ωm t − 90) = sin(ωm t) . Obviously,

F T [sin(ωm t)] = −jπδ(ω − ωm ) + jπδ(ω + ωm ) (5.28)

F T [cos(ωm t)] = πδ(ω − ωm ) + πδ(ω + ωm ) (5.29)

Based on the above equations it is straightforward to derive the result that

F T [sin(ωm t)] = −jsignum(ω)F T [cos(ωm t)] (5.30)

where

(
1 ω≥0
signum(ω) = (5.31)
−1 ω<0

Equation (5.30) tells us that the transfer function of the block designated by the notation −90 is equal to
−jsignum(ω). Now we are ready to generalize the construction of an upper sideband signal using the block
diagram of Fig. 5.13 for the case where the information signal f (t) is of arbitrary nature (i.e., not necessarily
sinusoidal). In Fig. 5.14, we provide a block diagram for the construction of the upper sideband signal of a
modulated wave m(t) when the information signal f (t) is arbitrary. Comparing Fig. 5.13 and Fig. 5.14 we
see that they are identical, where in both figures the block designated by the notation −90 corresponds to a
system with transfer function equal to −jsignum(ω).
In Fig. 5.15 we show, in a pictorial fashion, why the block diagram of Fig. 5.14 works. What is worth noting
is that the output of a −90 degree shift system is called the Hilbert transform of the input. Furthermore,
if we designate by the input to a −90 degree shift system the output (Hilbert transform) of this system is
designated by fˆ(t). Finally, it is easy to show that

1
fˆ(t) = f (t) ∗ (5.32)
πt

Page 72
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.13: Block diagram of a system that generates the upper sideband of m(t) where f (t) = cos(ωm t) and
c(t) = cos(ωc t)

Fig. 5.14: Block diagram of a system that generates the upper sideband of m(t)

Page 73
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.15: FT at various stages of the block diagram of Fig. 5.14

Page 74
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.16: DSB-LC Signals for two information signals f1 (t) and f2 (t)

5.3.9 Double Sideband Large Carrier (DSB-LC) System

The main idea behind this system is money. Its claim to fame comes from the use of a very cheap receiver
called peak or envelope detector. To understand this detector let us first look at the modulated signal m(t)
for two examples of information signals f (t). In Fig. 5.16 we show m1 (t) for an information signal f1 (t) and
m2 (t) for an information signal f2 (t). The difference between the two information signals f1 (t) and f2 (t)
is that f1 (t) is always positive, while f2 (t) assumes positive and negative values. If we now look at the
positive peaks of m1 (t) we see that a line through them produces the information signal f1 (t). On the other
hand if we look at the positive peaks of m2 (t) a line through them produces the information signal f2 (t)
only when f2 (t) is positive and it produces the negative of f2 (t) when the signal is negative. The reason
that the positive peaks of m1 (t) and m2 (t) are important is because we can design an inexpensive receiver
that traces these positive peaks. Hence, when the signal f (t) is always positive, this receiver will be able to
reproduce the information signal f (t) from m(t); this is not the case though when the signal m(t) assumes
positive and negative values. Unfortunately, information signals can be positive or negative. To remedy the
problem of a negative information signal we can add a constant to the signal that is larger than or equal to
the most negative signal value. Hence, now

m(t) = [A + f (t)]cos(ωc t) = Acos(ωc t) + f (t)cos(ωc t) (5.33)

The Fourier transform of is shown in Fig. 5.17 and it is provided by the following expression.

1
M (ω) = Aπ[δ(ω − ωc ) + δ(ω + ωc )] + [F (ω − ωc ) + F (ω + ωc )] (5.34)
2

By looking at Fig. 5.17 we can justify the name DSB-LC for this communication system. The Fourier trans-
form of the modulated wave m(t) has both of the sidebands (upper and lower). Furthermore, the presence
of the pure carrier Acos(ωc t) is evident by the presence of the two impulses in the Fourier transform of m(t).
Consider now the special case where the information signal f (t) is of the form.

f (t) = Kcos(ωm t) (5.35)

Based on our previous discussion, the maximum amplitude of the pure carrier that we need to add to the
DSB-SC modulated signal to make it a legitimate DSB-LC signal must satisfy the following inequality:

Page 75
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.17: FT of a DSB-LC signal

A≥K (5.36)

Then,

k
m(t) = A(1 + cos(ωm t))cos(ωc t) (5.37)
A

We usually define

K
m= (5.38)
A

the modulation index of the system. Since m’s maximum value is one and its minimum value is zero, it is
often given in percent. In Fig. 5.18 we show the DSB-LC signal for the case of an information signal that is
sinusoidal and for various m values (i.e., m = 1, m < 1 and m > 1). The case m > 1 is not allowed because
then we are not going to be able to recover the information signal f (t). In Fig. 5.19 we show a modulated
wave for an arbitrary information signal f (t) and a modulation index m < 1 or m > 1.
As we emphasized before, the information signal f (t) can be recovered from a DSB-LC signal by utilizing a
peak (envelope) detector. An envelope detector consists of a diode and a resistor-capacitor (RC) filter (see
Fig. 5.20). The operation of this envelope detector is as follows.
On a positive half-cycle of the input signal, the diode is forward biased and the capacitor C charges up
rapidly to the peak value of the input signal. When the input signal falls below this value, the diode becomes
reverse-biased and the capacitor C discharges slowly through the load resistor Rl . The discharging process
continues until the next positive half-cycle. When the input signal becomes greater than the voltage across
the capacitor, the diode conducts again and the process is repeated. We assume that the diode is ideal,
presenting resistance rf to current flow in the forward-biased region, and infinite resistance in the reverse-
biased region. We further assume that the AM wave applied to the envelope detector is supplied by a
voltage source of internal impedance Rs . The charging time constant (rf + Rs )C must be short compared
with the carrier period 2π
ωc , that is

2π
(rf + Rs )C << (5.39)
ωc

Page 76
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.18: Effects of varying modulation indices

Fig. 5.19: Importance of sufficient carrier in DSB-LC waveform

Page 77
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.20: Envelop detector. (a) Circuit diagram (b) AM wave input (c) Envelop detector output

Page 78
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

so that the capacitor C charges rapidly and thereby follows the applied voltage up to the positive peak
when the diode is conducting. On the other hand, the discharging time constant Rl C must be long enough
to ensure that the capacitor discharges slowly through the load resistor Rl , between positive peaks of the
modulated wave, but not so long that the capacitor voltage will not discharge at the maximum rate of change
of the modulating wave, that is

2π 1
<< Rl C << (5.40)
ωc ωm

where ωm is the information signal bandwidth. The result is that the capacitor voltage or detector output is
nearly the same as the envelope of the AM wave, as shown in Fig. 5.20.
One of the serious flaws of a DSB-LC system is that it wastes transmitter power. To illustrate that consider
the DSB-LC signal of equation (5.33). It is easy to show that

p
P ower of m(t) = P ower of {Acos(ωc t)} + P ower of {f (t)cos(ωc t)} + P ower of { 2Af (t)cos(ωc t)}
(5.41)

2
The power of Acos(ωc t) is equal to A2 , while the power of f (t)cos(ωc t) can be shown to be equal to the
one-half the power of f (t); the latter result is valid under the legitimate assumption that p
the bandwidth of
the information signal is much smaller than the carrier frequency ωc . Finally, the power of 2Af (t)cos(ωc t)
is equal to zero under the assumptions that

(i) the DC value of f(t) is zero.

(ii) the bandwidth of the information signal is much smaller than the carrier frequency.

A2 P ower of f (t)
P ower of m(t) = + (5.42)
2 2

For the special case where f (t) = Kcos(ωm t) we can conclude that

A2 A2 m 2 A2 m2
P ower of m(t) = + = (1 + ) (5.43)
2 4 2 2
2
It is worth noting that from the power dedicated to transmit the modulated signal, A2 goes to the transmission
of the pure carrier and the rest goes to the transmission of the useful signal. If we define the efficiency η
of our system as the ratio of the power dedicated to the transmission of the useful signal versus the power
dedicated to the transmission of the modulated signal we conclude that the efficiency of our DSB-LC system
(for a sinusoidal information signal) is equal to

m2
Ef f iciency = η = (5.44)
m2 + 2

The best we can do is to let η assume its maximum value; this happens when m = 1 Then,

η = 0.33 (33%) (5.45)

Page 79
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

So 67% of the available transmitter power goes to the carrier (alarming quantity!) Usually, we do not know
K so we might operate at less than 33% efficiency.
The notion of the modulation index can be extended to all signals, as long as they are normalized so that
their maximum negative value is unity. To be more specific consider the DSB-LC signal corresponding to
an information signal whose maximum negative value is different than unity. Then,

m(t) = Acos(ωc t) + f (t)cos(ωc t)

f (t)
= Acos(ωc t) + |min{f (t)}| cos(ωc t) (5.46)
|min{f (t)}|
|min{f (t)}| 0
= A[1 + f (t)]cos(ωc t)
A
0 0
where f (t) is the normalized version of f (t). Note that f (t) has minimum value of −1. In the above
equation we define the modulation index m to be equal to the ratio

|min{f (t)}|
(5.47)
A

Actually, the aforementioned definition is applicable independently of whether f (t) has minimum value of
unity or not.

5.3.10 Vestigial Sideband Suppressed Carrier (VSB-SC)

The stringent frequency-response requirements on the sideband remover of a SSB-SC system can be re-
laxed by allowing a part, called vestige, of the unwanted sideband to appear at the output of the modulator.
Thus, we simplify the design of the sideband filter at the cost of a modest increase in the channel band-
width required to transmit the signal. The resulting signal is called vestigial/-sideband suppressed carrier
(VSB-SC) system. The suppressed carrier name sterns from the fact that one more time no pure carrier is
sent over the channel.
To generate a VSB-SC signal, we begin by generating a DSB-SC signal and passing it through a sideband
filter with frequency response H(ω) as shown in Fig. 5.21. In the time domain the VSB-SC signal may be
expressed as

m(t) = [f (t)cos(ωc t)] ∗ h(t) (5.48)

where h(t) is the impulse response of the VSB-SC filter. In the frequency domain the corresponding ex-
pression is

1
M (ω) = [F (ω − ωc ) + F (ω + ωc )]H(ω) (5.49)
2

To determine the frequency-response characteristics of the filter, let us consider the demodulation of the
VSB-SC signal m(t). We multiply m(t) by the carrier component and pass the result through an ideal
lowpass filter, as shown in Fig. 5.22. Thus the product signal is

e(t) = m(t)cos(ωc t) (5.50)

Page 80
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.21: Generation of a VSB-SC signal

Fig. 5.22: Demodulation of the VSB-SC signal m(t)

or, equivalently

1
E(ω) = [M (ω − ωc ) + M (ω + ωc )] (5.51)
2

If we substitute for M (ω) from equation (5.49) into equation (5.50), we obtain

1 1
E(ω) = [F (ω − 2ωc ) + F (ω)]H(ω − ωc ) + [F (ω + 2ωc ) + F (ω)]H(ω + ωc ) (5.52)
4 4

The lowpass filter rejects the double frequency components and passes only the components in the fre-
quency range |ω| < ωm . Hence the signal spectrum at the output of the ideal lowpass filter is

1
O(ω) = F (ω)[H(ω − ωc ) + H(ω + ωc )] (5.53)
4

We require that the information signal at the output of the lowpass filter be undistorted. Therefore, the
VSB-SC filter characteristic must satisfy the condition:

H(ω − ωc ) + H(ω + ωc ) = constant f or ω ≤ ωm (5.54)

The condition is satisfied by a filter that has the frequency response characteristic shown in Fig. 5.23. We
note that H(ω) selects the upper sideband and a vestige of the lower sideband. It has an odd symmetry
about the carrier frequency ωc , in the frequency range ωc − ωa < ω < ωc + ωa , where ωa is a conveniently

Page 81
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.23: VSB filter characteristics

Fig. 5.24: Frequency response of VSB filter for selecting the lower sideband of the message signals

selected frequency that is some small fraction of ωm , i.e., (ωa << ωm ). Thus, we obtain an undistorted
version of the transmitted signal. Fig. 5.24 illustrates the frequency response of a VSB filter that selects the
lower sideband and a vestige of the upper sideband.
In practice, the VSB filter is designed to have some specified phase characteristics. To avoid distortion of
the information signal, the VSB filter should be designed to have a linear phase over its passband ωc − ωa <
ω < ωc + ωa

5.3.11 Why modulation?

Modulation is the process by which a property of a parameter of a signal is in proportion to a second signal.
The primary reason for using modulation in communication is:

1. To raise up the frequency of a signal to reduce the wavelength such that a relatively small antenna
can transmit or receive the signal.
2. To separate base band signals in frequency (or time) so that more than one signal can be transmitted
on the same channel.

3. To have the base band signal transformed for ease of transmission.

The degree to which a signal is modulated is measured by a modulation index, which will have a different
physical significance for each type of modulation. Every form of modulation has advantages and disadvan-
tages when compared to the others. FM has more strengths than weaknesses compared to the others.
Delta Modulation has more weakness than strengths, but all have their applications. Some modulation
techniques and their characteristics are listed in Table 5.3.

Page 82
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

5.3.12 Aliasing
Aliasing occurs in most forms of modulation when the sampling rate or the carrier frequency is not large
enough compared to the maximum message frequency.
Once the spectra have overlapped there is no way to separate the two signals. The Nyquist theorem states
that for a sampled signal, such as PAM, the sampling frequency fs must be at least twice the maximum
frequency of the signal to prevent aliasing from occurring.

5.4 Simulation
In this simulation we are going to see the amplitude modulation and demodulation with different parameters.
Let us consider a sinusoidal message signal of 100 Hz and a sinusoidal carrier in the transmitter of 1000
Hz.
f = 100;
fc = 1000;
m = 0.5;
fs = 16384;
MODULATION
1) Construct the time array as following
t = [0:1/fs:0.05];
2) Construct the message signal as
signal = sin(2 * pi * f * t);
3) Calculate the required dc offset from modulation index definition
A = abs(min(signal))/m;
4) Construct the carrier at the transmitter end as
carrier1 = sin(2 * pi * fc * t);
5) Modulated output of the transmitter is given by
x = (signal+A).*carrier1;
(Notice a dot before the multiplication sign. In Matlab it represents multiplication element by element.)
6) Find the Fourier transform of signal, carrier1 and x. Name them say S1, C1 and X respectively. Use the
ft function for this. You have to create your own ft function. The code for this function is given below. Make
sure that you saved this function as a separate MATLAB file with the same name ft.m in the Matlab working
directory.

function [X]=ft(x,fs)
n2=length(x);
n= 2 ∧(max(nextpow2(f s), nextpow2(n2)));
X1=fft(x,n)/fs;
X=X1(1:n/8);

Page 83
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

An example has been given below how you find the Fourier transform using this MATLAB function. You can
write the following line of code in your main program to find Fourier transform of a signal.
[S1]=ft(signal,fs);
For good demodulation of the received signal, the carrier at the receiver end needs to be of the same
frequency and phase as the carrier of the transmitter end. In practice this may not be the case.
7) Carrier at the receiver (demodulator) is given by the following equation.
carrier2=sin(2 * pi * (fc+delta) * t + phi);
Where delta is the frequency error and phi is the phase error. Initially assume that both of them are zero.
8) For demodulation multiply the received signal by the carrier and name it x1 and then do a low pass
filtering.
For low pass filtering use the following piece of MATLAB code.
b=fir1(24,500/fs);
(The first parameter gives you the order of the filter. The second parameter is the cutoff frequency. Here it
is 200 Hz.)
x_demod=conv2(x1,b,’same’);
Here b is the coefficient of the filter. So to find the filtered output we did the convolution of x1 and b.
9) Find the Fourier transform of carrier2 and x_demod. Name them C2 and S2.
10) Plot the following:
signal: message signal before modulation versus time
x: modulated signal versus time
x_demod: demodulated signal versus time
S1: message signal in frequency domain
C1: carrier at transmitter end in frequency domain
X: modulated signal in frequency domain
C2: carrier at the receiver end in frequency domain
S2: demodulated signal in the frequency domain
You may use following MATLAB functions to plot all of them together.
Subplot
Plot
11) Repeat the simulation using square wave carrier at both ends. What happens if you use different wave
shapes? Explain. Actually in the hardware experiments that follow you will be using a square wave carrier.
12) Repeat the above simulation with a ’sinc’ function as the message signal.
13) Repeat the above simulation with phi = pi/4, pi/2 in step 7). What happens? Explain.
What happens if delta = 200/400 Hz?

5.5 Pre-lab questions

Non sinusoidal carrier
Consider a case of double-sideband suppressed carrier amplitude modulation (DSB-SC AM), where instead
of having a sinusoid to modulate our signal we use some other periodic signal p(t) with Fourier series
expansion:

Page 84
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

∞
X
p(t) = Pn ejnω0 t (5.55)
n=−∞

1. By providing a qualitative example (i.e. assume a specific p(t), say square wave that we will be using
in our lab) show how the spectrum of the modulated signal will look like.
2. How can we demodulate a signal of this kind?

Explain and illustrate all your answers.

5.6 Implementation
By this time you already know that in amplitude modulation, the amplitude of the carrier frequency varies
with the amplitude of the desired message signal. Typically the carrier will be a cosine or some sine wave
but in this lab a square wave is used as the carrier for simplicity in implementation. Amplitude Modulation
is implemented by multiplying the message signal with the carrier as in the theory part, which can be
expressed as:

m(t) = f (t)c(t) = f (t)cos(ωc t) (F or a cosine carrier) (5.56)

The end result with sinusoidal modulation is such that the spectrum of X(f ) remains unchanged but is
removed in frequency and centered about ωc . For any general carrier, the multiplication of two signals in
time yields convolution in frequency.
A DSB AM signal is described by the equation:

m(t) = f (t)c(t) = (f (t) + A)cos(ωc t) (F or a cosine carrier) (5.57)

where A is the average value of the signal, or the DC component. Typically a message signal will have no
DC component.
When A > 0, the m(t) is a DSB-LC (Large Carrier) AM signal. In this case the DC term must be added
to the message signal. The case when A = 0 is called Suppressed Carrier (DSB-SC) AM. The modulated
index for AM is a measure of how large A is with respect to the amplitude of f (t).

|f (t)min |
m= (5.58)
A

Where f (t)min is the minimum value of the message signal before any DC is added and A is the amount of
DC added. Note that m = 100% for A = |f (t)min |.
There are two forms of demodulation used in this lab:

1. Synchronous:
The modulated waveform is multiplied by another carrier of (hopefully) identical frequency and phase
to the AM carrier. High frequency terms that result from the multiplication are removed with a filter
and only the message is left. Note that the carrier must be regenerated for an actual receiver.

Page 85
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.25: AM Modulation

Fig. 5.26: Carrier Generator

2. Envelop Detection:
When m < 100%, peak detection (sample and hold) can be used. Sample and hold means that the
amplitude of the signal at a given point in time is sampled and held until the next sample pulse arrives.

The FET used in the AM modulator and demodulator (Fig. 5.25 and Fig. 5.26) is used as a variable resistor,
where the resistance between the drain (D) and the source (S) varies proportional to the voltage applied to
the gate (G).
Since there is a square wave varying from 0 to −Vcc volts been applied to the gate, the resistance across
the source and the drain can be thought as either an open (R = ∞) when Vg = −Vcc or a short (R = 0) for
Vg = 0. Analysis of the circuit shows a gain of VVout
in
= 1 for R = ∞ (differential amplifier configuration) and
Vout
a gain of Vin = −1 when R = 0(inverting amplifier configuration).

5.7 Procedure
5.7.1 AM Modulation
1) Build the modulator shown in Fig 5.25 and Fig. 5.26.
2) Connect the oscilloscope to the function generator output (or the message oscillator output), Vin (t)
and to the modulator output, Vo (t), using direct coupling (DC) throughout. Set Vin (t) to a sinewave with
fm = 100Hz.
3) Adjust the dc offset and amplitude of the message of the message oscillator to yield a DSB-LC modulated
waveform having a modulation index m = 0.75. Sketch Vin (t) and Vo (t). Change Vin (t) to a triangular
waveform and sketch Vin (t) and Vo (t).

Page 86
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.27: AM Block Diagram

Fig. 5.28: Envelope Detector

For a better view of Vo (t) deactivate the channel that reads Vin (t) and set the timebase to 5ms. i.e., Vin (t)
can be plotted clearly at timebase equal to 2ms
4) Adjust the dc offset of the message oscillator to zero, thus producing DSB-SC. Again sketch Vin (t) and
Vo (t) first for Vin (t) sinusoid and then Vin (t) triangular.
5) Obtain once again the DSB-SC signal for Vin (t) sinusoidal and 100=mf Hz. Observe and sketch the
spectra of Vin (t) and Vo (t), using the spectrum analyzer. To see the modulated output you may adjust the
frequency display of the spectrum analyzer (i.e, FFT on oscilloscope) to the value equal to the value of
the carrier frequency. That is, use FFT button adjusted to the channel of the Vo (t) reading. The preferable
timebase is between 2ms and 5ms. The frequency of the modulated signal is between 19KHz and 24KHz.
6) Repeat part 5 for DSB-LC. Furthermore, using the spectrum analyzer, determine the modulation index,
m.

5.7.2 Amplitude Demodulation

The overall general communication system block diagram is given in Fig. 5.27. You have already built the
modulator. As a channel, use a small length of wire. There are two types of AM demodulator: the envelop
detector (no carrier waveform is required), and the synchronous demodulator (carrier waveform is needed).

5.7.3 Envelope detection

1) Connect the envelope detector of Fig. 5.28.
2) Adjust the oscillator for a DSB-LC modulated waveform using sinusoidal modulation with fm = 100Hz .
Connect Vd (t) to the oscilloscope input.
3) Set C to the minimum value of the capacitor substitution box, noting the output waveform as C is in-
creased. Sketch the optimum demodulated waveform. Compare Vd (t) with the modulating waveform Vin (t).
Comment.
4) Repeat part 3) for a triangular modulating signal.
5) Now adjust for a DSB-SC waveform, comparing Vd (t) with the modulating waveform Vin (t). Does the
envelope detector demodulate DSB-SC?

Page 87
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Fig. 5.29: Synchronous Demodulation

5.7.4 Synchronous detection

1) Connect the synchronous detector of Fig. 5.29.
2) Adjust for a DSB-LC signal. Compare Vin (t) with Vd (t). Sketch the waveforms. Has the message been
recovered?
3) Adjust for a DSB-SC signal. Again compare Vin (t) with Vd (t). Sketch the waveforms. Is it possible to
retrieve the original message? Explain.
4) Is it possible to demodulate a DSB-LC signal? A DSB-SC signal?

5.8 Questions and Calculations

1. Why must m be less than 100% for envelope detection?
2. Compare Envelope Detection to Synchronous Detection and list at least one advantage and one disad-
vantage of each.
N.B. The usual Op-Amp IC DC- voltage trigger is 15 V, some Op-Amp’s IC DC- voltage trigger can reach
20V .

Page 88
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Page 89
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Page 90
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

Page 91
EEL 3552C Lab Manual 5. EXPERIMENT: AMPLITUDE MODULATION

5.9 Hints & Comments

• When building the demodulation circuit, you need to connect it to the modulation circuit which you
should have already built

• Distinguish between the Ground G, Source S and Drain D nodes of the transistors according to the
âĂIJkey note in the lab tool kitâĂİ to match the circuit schematic provided.

In the modulation part: the modulation index m should be calculated using the following formula.
|f (t)min |
m= A

where f (t)min is the minimum voltage of the input signal and A is the offset shift needed to make the
input signal all positive

In the demodulation part: When using the spectral analysis on the oscilloscope: The modulation index m
should be calculated using equation (4.41) where the power of the signal is calculated using its spectrum
amplitudes.

A2 A2 m 2 A2 m2
P ower of m(t) = + = (1 + ) (4.41)
2 4 2 2

Using the envelope detection or synchronous detection: The role of the capacitors is to sharpen the out-
put of the demodulation so that it matches the input signal, i.e. by calibrating the capacitor values you will
make the signal thinner matching the input signal shape and thickness. By increasing the capacitor values
further, you will mismatch the original input signal.

Page 92
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

6 EXPERIMENT: Frequency Modulation

6.1 Objective
To understand the principles of frequency modulation and demodulation.

6.2 Equipment:
The equipment used in this experiment are:

Oscilloscope: Rohde & Schwarz RTM 3004

Function Generator, Tektronix AFG 3022B
Digital Multimeter, Tektronix DMM 4050
Triple Output Power Supply, Agilent E3630A
Op Amp module TL084
XR2206 VCO
LM565 PLL
Bring a USB Flash Drive to store your waveforms.

6.3 Theory
6.3.1 Introduction
A sinusoidal carrier c(t) = Acos(ωc t + θ0 ) has three parameters that can be modified (modulated) according
to an information signal f (t).

1. Its amplitude A which leads us to the class of systems designated as amplitude modulating (AM)
systems.
2. Its frequency ωc , which leads us to a class of systems designated as frequency modulating (FM)
systems.
3. Its phase θ0 , which leads us to a class of systems designated as phase modulating (PM) systems.

We have already discussed the class of AM systems. In the sequel we focus on the class of FM and PM
systems. Note that we can write that

c(t) = Acos(ωc t + θ0 ) = Acos(θ(t)) (6.1)

where θ(t) is often called the angle of the sinusoid. That’s why FM and PM systems are sometimes referred
to as angle modulating systems.

Page 93
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

6.3.2 Preliminary notions of FM and PM Systems

Consider the carrier c(t) = Acos(ωc t + θ0 ). We can write

c(t) = Acos(θ(t)) (6.2)

where we call θ(t) the instantaneous phase of the carrier. If we differentiate θ(t) with respect to time t we
get a time function that we designate by ωt and we call it the instantaneous frequency. That is

dθ
ω(t) = (6.3)
dt

It is easy to see that the above definition of instantaneous frequency makes sense if we apply it to a pure
carrier c(t) = Acos(ωc t + θ0 ), because then we get

ω(t) = ωc (6.4)

From equation (6.3), above, we see that if you have the instantaneous phase of a sinusoid, you can compute
its instantaneous frequency by differentiation. Furthermore, if you know the instantaneous frequency ωt of
a sinusoid you can compute its instantaneous phase by integration as follows:

Z t
θ(t) = ω(τ )dτ + θ0 (6.5)
0

Obviously, if we start with ω(t) = ωc we get

θ(t) = ωc t + θ0 (6.6)

Phase and frequency modulations are techniques that modify the instantaneous phase and frequency,
respectively, of a sinusoid in a way dictated by an information signal f (t).

6.3.3 Phase Modulation

Here the information signal f (t) is placed as a linear term in the instantaneous phase of the carrier. That is

θ(t) = ωc t + θ0 + kp f (t) (6.7)

where kp is a constant of the modulating device. Hence, the PM modulated signal is equal to

mp (t) = Acos(ωc t + θ0 + kp f (t)) (6.8)

Page 94
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

6.3.4 Frequency Modulation

Here the information signal gets inserted as a linear term into the instantaneous frequency of the carrier.
That is,

ω(t) = ωc + kf f (t) (6.9)

where kf is a constant due to the modulator. In this case the instantaneous phase is equal to

Z t
θ(t) = ωc t + θ0 + kf f (τ )dτ (6.10)
0

and as a result, the FM modulated signal looks like

Z t
mf (t) = Acos(ωc t + θ0 + kf f (u)du) (6.11)
0

A plot of FM and PM signals is shown in Fig. 6.1.

Fig. 6.1: Examples of frequency and phase modulation

Now let us discuss FM and PM simultaneously and get a better insight into their similarities and differences.
For simplicity assume that θ0 = 0. Then,

mp (t) = Acos(ωc t + kp f (t)) (6.12)

and

Z t
mf (t) = Acos(ωc t + kf f (u)du) (6.13)
0

Page 95
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Let us take mp (t) and find its instantaneous frequency ω(t). Indeed,

df (t)
ω(t) = ωc + kp (6.14)
dt

The above equation tells us that in the PM case the instantaneous frequency has a linear term proportional
to the derivative of the information signal f (t). So, we can say that the PM case is in reality the FM case but
with an information signal being the derivative of the actual information signal. In other words, if we have a
device that produces FM signals we can make it to produce PM signals by giving it as an input the derivative
of the information signal. The above equation also tells us that the FM case is in reality a PM case with the
modulating signal being the integral of the information signal. Also, if we have a device that produces PM,
we can make it to produce FM by providing to it as an input the integral of the information signal.
Hence, what it boils down to is that we need to discuss either PM or FM and not both. We choose to focus
on FM, which is used to transmit baseband analog signals, such as speech or music. PM is primarily used
in transmitting digital signals.
Our primary focus in the examination of FM signals will be the analysis of its frequency characteristics.
Although it has been a straightforward task to find the Fourier transform of an AM signal the same is not
true for FM signals. Let us again consider the general form of an FM signal

mf (t) = Acos(ωc t)cos(kf g(t)) − Asin(ωc t)sin(kf g(t)) (6.15)

Rt
where g(t) = 0 f (u)du. If we are able to find the FT of cos(kf g(t)) and sin(kf g(t)) we can produce the FT
of the signal mf (t) without a lot of effort. The FT of cos(kf g(t)) and sin(kf g(t)) cannot be found for any g(t)
and in fact it has been found for few f (t)’s. To get a deeper insight consider cos(kf g(t)) and expand it in
terms of its Taylor series.

kf2 g 2 (t) kf4 g 4 (t) kf6 g 6 (t)

cos(kf g(t)) = 1 − + − + ... (6.16)
2! 4! 6!

If the signal f (t) is known, its Fourier transform F (ω) is also known, and the Fourier transform of g(t) can
be computed. In particular, from well known Fourier transform properties we can deduce that

g 2 (t) → G(ω) ∗ G(ω)

g 4 (t) → G(ω) ∗ G(ω) ∗ G(ω) ∗ G(ω) (6.17)
.. .
. → ..

Based on the above two equations (6.16) and (6.17) we can state that to compute the Fourier transform of
cos(kf g(t)) for an arbitrary signal g(t) becomes a formidable task (we need to compute a lot of convolutions).
Furthermore, it seems that the bandwidth of cos(kf g(t)) is infinite (note that every time we multiply a signal
with itself in the time-domain we double its bandwidth in the frequency domain). Hence if the bandwidth
of g(t) is ωm , the bandwidth of g 2 (t) is 2ωm , the bandwidth g 4 (t) is 4ωm , and so on. In reality though not
all the terms in the Taylor series expansion cos(kf g(t)) contribute equally to the determination of the signal
cos(kf g(t)). Notice that in the Taylor series expansion of cos(kf g(t)) the coefficients multiplying the powers
of g(t) get smaller and smaller. This observation will leads us into the conclusion that the bandwidth of the
terms cos(kf g(t)) and sin(kf g(t)) is indeed finite, and as a result the bandwidth of the FM signal mf (t) is
also finite.
To investigate the bandwidth of an FM signal thoroughly we will discriminate two cases of FM signals: The
case of Narrowband FM and the case of Wideband FM.

Page 96
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

6.3.5 Narrowband FM
Consider again the FM signal mf (t) given by the following equation.

mf (t) = Acos(ωc t)cos(kf g(t)) − Asin(ωc t)sin(kf g(t)) (6.18)

The terms for which FT is difficult to evaluate are: cos(kf g(t)) and sin(kf g(t)). Each one of these terms
has a Taylor series expansion involving infinitely many terms. Let us see what happens if each one of these
terms is approximated only by their first term in the Taylor series expansion. Then,

cos(kf g(t)) ≈ 1
(6.19)
sin(kf g(t)) ≈ kf g(t)

Obviously, if we make the above substitutions in equation (6.18) we get

mf (t) = Acos(ωc t) − Akf g(t)sin(ωc t) (6.20)

The advantage of the above equation is that we can evaluate its FT, and consequently the FT of mf (t). It is
not difficult to see that the bandwidth of mf (t) is approximately equal to 2 times the bandwidth of f (t)(f (t)
is the information signal). Hence, when the above approximations are accurate we are generating an FM
signal whose bandwidth is approximately equal to the bandwidth of an AM signal. Since, in most cases an
FM signal will occupy much more bandwidth than an AM signal, the aforementioned type of FM signal is
called narrowband FM. In the sequel, we are going to identify (quantitatively) conditions under which we
can call an FM signal narrowband. These conditions will be a byproduct of our discussion of wideband FM
systems.

6.3.6 Wideband FM
To illustrate the ideas of wideband FM let us start with the simplest of cases where the information signal is
a single sinusoid. That is,

f (t) = acos(ωm t) (6.21)

Then, the instantaneous frequency of your FM signal takes the form:

ω(t) = ωc + kf acos(ωm t) (6.22)

Integrating the instantaneous frequency ω(t) we obtain the instantaneous phase θ(t):

kf a
θ(t) = ωc t + sin(ωm t) (6.23)
ωm

Consequently, the FM modulated wave is

kf a
mf (t) = Acos(ωc t + sin(ωm t)) (6.24)
ωm

Page 97
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

k a
The quantity ωfm is denoted by β and it is referred to as the modulation index of the FM system. Let us now
write the above expression for the signal mf (t) in a more expanded form.

mf (t) = Acos(ωc t)cos(βsin(ωm t)) − Asin(ωc t)sin(βsin(ωm t)) (6.25)

To simplify our discussion, from now on, we will be referring to the quantity cos(βsin(ωm t)) as term A and to
the quantity sin(βsin(ωm t)) as term B. Terms A and B are the real and the imaginary part of the following
complex exponential function

ejβsinωm t (6.26)

2π
which is a periodic function with period ωm . The above function can be expanded as an exponential Fourier
series, as follows:

∞
X
ejβsinωm t = Cn ejnωm t (6.27)
n=−∞

where the coefficients Cn are calculated by the equation:

Z π
ωm ωm
Cn = ej(βsinωt−nωt) dt (6.28)
2π − ωπm

Let us now make a substitution of variables in the above equation. In particular, let us substitute ωm t with
x. Then,

Z π
1
Cn = ej(βsinx−nx) dx (6.29)
2π −π

The above integral cannot be evaluated in closed form, but it has been extensively calculated for various
values of β’s and most n’s of interest. It has a special name, called the nth order Bessel function of the first
kind and argument β. This function is commonly denoted by the symbol Jn (β). Therefore, we can write:

Cn = Jn (β) (6.30)

As a result,

∞
X
ejβsinωm t = Jn (β)ejnωm t (6.31)
n=−∞

often called the Bessel Jacobi equation. If we evaluate the real and imaginary parts of the right hand
side of equation (6.31) we will be able to calculate term A and term B, respectively. It turns out that if we
substitute these values for term A and term B in the original equation for mf (t) (see equation (6.25)) we
will end up with

Page 98
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

∞
X
mf (t) = Jn (β)cos((ωc + nωm )t) (6.32)
n=−∞

The property of the Bessel function coefficients (actually Property 1) that led us to the above results is listed
below. Some additional properties of the Bessel function coefficients are also listed.

1. Jn (β) is real valued.

2. Jn (β) = J−n (β) for n even.
3. Jn (β) = −J−n (β) for n odd.
P∞ 2
4. n=−∞ Jn (β) = 1

The advantage of equation (6.32) compared to the original equation (6.24) that defined the FM signal mf (t)
is that now, through equation (6.32), we can compute the FT of the signal mf (t). It will consist of an infinite
sequence of impulses located at positions ωc + nωm where n is an integer. In reality though, no matter how
big β is, the significant Jn (β)’s will be only for indices n ≤ β + 1. Hence, the approximate bandwidth of your
FM signal, when the information signal is of sinusoidal nature, is given by the following equation.

B = 2(β + 1)ωm (6.33)

In Fig. 6.2 various plots of the Bessel function coefficients Jn (β) are shown. As we can see these plots
verify our claim, above, that Jn (β) become small for indices n > β + 1. In Fig. 6.3 we show the FT of signals
mf (t) for various β values.

Fig. 6.2: Plot of Bessel function Jn (β) of the first kind

In Table 6.1 the values of the Bessel function coefficients Jn (β) are shown for various β values. We can use
the values of the Table 6.1 to evaluate the bandwidth of the signal mf (t) as follows. We are still operating
under the assumption that the information signal is of sinusoidal nature. As a result, expression (6.32)
is a valid representation of our signal mf (t). Let us now impose the criterion that for the evaluation of the
bandwidth of the signal mf (t) we are going to exclude all terms of the infinite sum with index nmax , such that
|Jn (β)| < 0.01 for n > nmax . This criterion is often called the 1% criterion for the evaluation of bandwidth.
If we find that nmax is the minimum index n that does not violate the 1% criterion then we can claim that the
approximate bandwidth of our signal, according to the 1% criterion, is:

Page 99
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Fig. 6.3: Magnitude line spectra for FM waveforms with sinusoidal modulation (a) for constant ωm (b) for
constant ∆ω

B = 2nmax ωm (6.34)

It is worth mentioning that the evaluation of bandwidth based on equation (6.33) corresponds to the band-
width of your FM signal according to a 10% criterion.
The above procedure followed for the evaluation of the FT of mf (t) can be extended to the cases where the
information signal f (t) is a sum of sinusoidal signals, or a periodic signal. In particular, if

f (t) = a1 cos(ω1 t) + a2 cos(ω2 t) (6.35)

then the phase of our FM signal mf (t) is provided by the following equation.

kf a1 kf a2
θ(t) = ωc t + sin(ω1 t)) + sin(ω2 t)) (6.36)
ω1 ω2

Omitting the details, we arrive at a representation of the signal mf (t) such that

∞
X ∞
X
mf (t) = A Jn (β1 )Jk (β2 )cos((ωc + nω1 + kω2 )t) (6.37)
n=−∞ k=−∞

k a
where β1 = ωf 1 1 and β2 = f rackf a2 ω2 . As we can see from the above equation, we now have impulses at
ωc ± nω1 , ωc ± kω2 as well as ωc ± nω1 ± kω2 .
Most of the aforementioned discussion regarding the FT and the bandwidth of an FM signal mf (t) is based
on the assumption that the information signal f (t) is a sinusoid, a sum of sinusoids, or a periodic signal. We
want to be able to derive a formula for the bandwidth of an FM signal mf (t) for an arbitrary information signal

Page 100
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

f (t). Let us revisit the approximate FM signal bandwidth formula (6.33) derived for a sinusoidal information
signal

B = 2(β + 1)ωm = 2(akf + ωm ) (6.38)

The second equality in (6.38) is obtained by substituting the value of β with its equal. Now let us pay a
closer look at the two terms involved in the evaluation of the approximate bandwidth B. The first term akf
is the maximum frequency deviation of the instantaneous frequency ω(t) from the carrier frequency ωc ; it
is often denoted by ∆ω. The second term ωm is the maximum frequency content of the information signal
f (t). Keeping these two clarifications in mind, we now define the approximate bandwidth of an FM signal
f (t) to be equal to

B = 2(∆ω + ωm ) (6.39)

where ∆ω is the maximum frequency deviation from the carrier frequency, and ωm is the maximum fre-
quency content of the information signal f (t). It is not difficult to show that for an arbitrary information signal
f (t), ∆ω = kf maxt |f (t)|. Furthermore, to find ωm we first need to compute the FT of the information
signal f (t). Hence, based on the above equation we can claim that the approximate bandwidth of an FM
signal is computable even for the case of an arbitrary info signal. It is worth pointing out that the bandwidth
formula given above has not been proven to be true for FM signals produced by arbitrary info signals; but it
has been verified experimentally in a variety of cases. Equation (6.39) is referred to as Carson’s formula
(rule) for the evaluation of the bandwidth of an FM signal, and from this point on it can be applied freely,
independently of whether the information signal is of sinusoidal nature or not.
One of the ramifications of Carson’s rule is that we can increase the bandwidth of an FM signal at will,
by increasing the modulation constant kf , or equivalently, by increasing the peak frequency deviation ∆ω.
One of the advantages of increasing the bandwidth of the FM signal is that larger bandwidths result in FM
signals that exhibit better tolerances to noise. Unfortunately the peak frequency deviation of an FM signal
is constrained by other considerations, such as a limited overall bandwidth that needs to be shared by a
multitude of FM users. For example, in commercial FM the peak frequency deviation is specified to be equal
to 75 KHz.
Let us now say a word about PM. From our previous discussions the form of a PM signal produced by a
sinusoidal modulating signal is as follows:

mp (t) = Acos(ωc t + kp acos(ωm t)) (6.40)

and the instantaneous frequency is equal to

ω(t) = ωc − kp aωm sin(ωm t) (6.41)

Hence,

∆ω = akp ωm (6.42)

That is ∆ω depends on ωm . This is considered a disadvantage compared to commercial FM, where ∆ω

is fixed. Carson’s rule is also applicable for PM systems but to find the peak frequency deviation of a PM

Page 101
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

system you need to find the maximum, with respect to time, of |f˙(t)|, where f˙(t) is the time derivative of
f (t). Actually, for PM systems
One last comment to conclude our discussion of angle modulation systems. It can be shown that the power
of an FM or PM signal of the form

m(t) = Acos(θ(t)) (6.43)

A2
is equal to 2 .

6.4 Pre-lab Questions

1) Provide a paragraph, where you compare AM and FM modulation (advantages, disadvantages).
2) Why is the use of FM more preferred than PM? Explain your answer. (Hint: Compare the frequency
deviations in both cases)
3) Give a short qualitative justification of the fact that FM has more noise immunity than AM.

6.5 Simulation
We are going to do the simulation in Simulink of Matlab. To run the Simulink, enter the ’Simulink’ command
in the MATLAB command window. It should look as it is shown in the following. Open a new window using
the left icon.

Construct the following block diagram for the FM modulation and demodulation simulation. The following
blocks will be necessary for your simulation. The paths of the blocks have been given as well.

Page 102
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Just drag and drop the blocks you need for your simulation in your work window. Click the left mouse button,
hold and drag to connect the blocks by wire. The parameters for the individual module have been given as
an example. To set the parameters of a block just double click on it.
Signal Generator
Wave form: sine
Amplitude: 1
Frequency: 1000
Discrete time VCO

Page 103
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Phase Lock Loop

Page 104
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Set the simulation stop time at say 0.002 from the simulation -> parameters menu.
Run the simulation from the simulation -> start menu. Then double click on the scopes to see the time
domain signals.
Do the simulation with square wave input signal. Is the demodulated signal the faithful reproduction of the
input signal? Why?

6.6 Implementation
In linear FM (frequency modulation), the instantaneous frequency of the output is linearly dependent on
the voltage at the input. Zero volts at the input will yield a sinusoid at center frequency at the output. The
equation for a FM signal is:

mf (t) = cos(ωc t + g(t)) (6.44)

Where ωc t + g(t) = φf (t) (instantaneous phase) and the instantaneous angular frequency is:

dg(t)
ωf (t) = ωc t + (6.45)
dt

In linear FM the instantaneous frequency can be approximated by a straight line.

ωf (t) = ωc + kf f (t) (6.46)

Where kf is a constant and f (t) is the input signal. fk()tf

Z
ωf (t) = ωc + kf f (t)
Z (6.47)
= ωc t + 2πkf f (t)dt

Z
mf (t) = cos(ωc t + 2πkf f (t)dt) (6.48)

Note that the amplitude of an FM signal never varies.

For a sinusoid input and positive kf , the modulated FM waveform will relate to the input as shown in Fig.
6.1.
Depending on the VCO, kf can be positive or negative and fc can also vary and is a function of external
timing resistor and capacitor values. The relationship between input voltage and output frequency at any
given point in time is:

fout = fc + kf Vin (6.49)

where kf is positive for Fig. 6.1 and has units of Hz/Volt. fk

Page 105
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Let f (t) = Acos(ωm t), then

akf
mf (t) = cos(ωc t + sin(ωm t)) (6.50)
fm

The peak frequency deviation from ωc is akf 2π (radian/sec), and the total peak-to-peak deviation is 2(akf 2π).
The modulation index β for this signal is:

akf
β= (6.51)
fm

Note that β will vary for each frequency component of the signal.

6.6.1 Spectrum of FM
The spectrum of an FM signal is described by Bessel functions. As shown in section 6.3.6, for a single
frequency, constant amplitude message, the spectrum of mf (t) is:

∞
X
mf (t) = Jn (β)cos((ωc + nωm )t) (6.52)
n=−∞

ak 1
R jβsinβ−J (β)
where fmf = β = FM modulation index and Jn (β) = 2π e n
dθ, which is the nth order Bessel
function eva1uated at β. Therefore, for each frequency at the input there are infinite number of spectral
components at the output with the amplitude or each component determined by the modulation index β.
The amplitude of the higher order terms will decrease, on the average, such that there will be a limited
bandwidth where most of the energy is concentrated.

6.6.2 VCO (FM modulator)

A voltage-controlled oscillator (VCO) converts the voltage at its input to a corresponding frequency at its
output. This is accomplished by a variable reactor (varactor) where the reactance varies with the voltage
across it. The varactor is part of a timing circuit, which sets the VCO output frequency. There are two
limitations for most VCO’s:

1. The input voltage must be small (usually there is an attenuating circuit at the input).
2. The bandwidth is limited for a linear frequency-to-voltage relationship.

6.6.3 PLL (FM demodulator)

The phase-locked loop (PLL) is used to demodulate FM. The phases of the input and feedback signals are
compared and the PLL works to make the phase difference between the two signals equal to zero.
Fig. 6.4 shows a basic block diagram of a PLL.
The VCO in the feedback is an FM modulator, and the center frequency can be set equal to the center
frequency of mf (t).
The equation for mf (t) is

Page 106
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Fig. 6.4: Phase Locked Loop (PLL) block diagram

Z
mf (t) = cos(ωc t + 2πkf f (α)dα) (6.53)

R
where kf f (α)dα = θ1 t and the equation of e(t) is

Z
e(t) = cos(ωc t + kd F (α)dα) (6.54)

R
where kd F (α)dα = θ2 t.
Since each signal has the same center frequency ωc , the phase comparator compares the instantaneous
values of θ1 and θ2 . The difference in phase is transformed into a DC voltage level proportional to the phase
difference and then amplified to yield F (t). This voltage is the input for the VCO in the feedback. As a result,
the difference between θ1 and θ2 will be made smaller. This process occurs continually, such that θ1 = θ2
at all times.
Substituting for θ1 = θ2

Z Z
kf
kd F (α)dα = kf f (α)dα and F (t) = f (t) (6.55)
kd

where f (t) was the original message signal and F (t) is the demodulated output.

6.7 Procedure
6.7.1 Modulation
a. Build the FM modulator shown in Fig. 6.5 (a).
b. Determine the constant kf from the following:
(1) Use your triple output power supply to apply the input voltages specified in the following table and record
the output frequency: (Take screen shots for the modulated signal with its frequency measurement) Adjust
the best signal sketch by using the scale knob. You can also use an adequate timebase value.
(2) Plot fout vs. Vin . Draw the best straight line through these points. The slope of this line is kf . Note that
kf has units of Hz/Volts. What is the measured value of kf ?
(3) According to the XR2206 data sheet, the expected voltage-to-frequency conversion gain is kf = − R0.32
1 C1
Hertz/Volts.

Page 107
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Calculate the expected value of kf . Assuming the resistors have a tolerance of ±10% and the capacitors
have a tolerance of ±20%, how do the measured and calculated values compare?
c. Add input coupling capacitor C2 to the circuit as shown in Fig. 6.5 (b). Set the modulation frequency to
fm = 2KHz. Fill in the following table using the following equations: (Take screen shots for the modulated
signal with its frequency measurement) Adjust the best signal sketch by using the scale knob. You can also
use an adequate timebase value.

∆fpeak = βfm = αkf (6.56)

∆fpeak
|Vin | = (6.57)
kf

d. Use the Vin found as the amplitude of the input signal and find ∆fpeak on the oscilloscope. Compare
∆fpeak calculated and measured. Display only the FM signal on the oscilloscope and trigger on the rising
edge of the waveform such that you see the following Fig. 6.6 (it will look like a ribbon):
This "ribbon" displays all frequencies in the FM signal at once. The minimum and maximum frequencies
can be easily detected and directly measured. Recall that ∆fpeak is only half of the peek-to-peak frequency
swing. What parameters determine the bandwidth of an FM signal?
e. View the frequency domain waveform to obtain modulation indices of 2.4, 5.52, 8.65. These are zero
carrier amplitude indices. Include calculations to verify your results in your lab report.

6.7.2 Demodulation
a. Build the FM demodulator utilizing the LM565N PLL shown in Fig. 6.7. Connect the output of the FM
modulator shown in Fig. 6.5 (b) to the input of the FM demodulator.

Page 108
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

b. Apply a sinusoidal message signal and observe the demodulated message signal. Sketch both wave-
forms. How do they compare?
An FM demodulator is typically followed by a filter to block the carrier. Setting the oscilloscope Trigger
Coupling to HF REJ will help sync on the modulation frequency.
c. Apply a square wave message signal and observe the demodulated message signal. Sketch both
waveforms. How do they compare? Why is the demodulated sinusoidal message more faithfully reproduced
than the demodulated square wave?

Page 109
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

6.8 Hints & Comments

6.9 Section 6.5 Simulation
In the simulation Matlab Simulink the oscilloscope connected to the input wave might not show the input
signal. Be sure to adjust the time instant iteration caliber on oscilloscope.

The output of the two different inputs of the filter, i.e. the sine wave or the square wave will be shown
as a sine wave. Although the output of a square wave input through this filter is not necessary a sine wave

The output of the square wave input through the filter is shown as a leveled sine wave (i.e. like a lad-
der)

6.10 Section 6.7 Procedure

Section 6.7.1 Modulation

In part (b), Vin (V olts) is a DC voltage connected from the DC power supply.

The nominal value of kf is calculated from the resistor and capacitor measured values instead of the nomi-
nal (ideal) values.

In part (c), Vin (V olts) is an AC voltage fed from the functional generator with input frequency of 2 kHz

The students should be able to see that the equipment used in frequency demodulation gives a linearized
approximation of the varying frequency of the carrier signal (by the input signal amplitude), which is a fea-
ture of the demodulation apparatus. This linear approximation is best described by a Bessel function up
to a specified order n. So, the better the demodulation equipment the more precise is the output signal
to the original input signal, and the higher the order of the Bessel function approximating this linearization
of frequency. Moreover, the less complicated the input signal used, the better is the demodulation. When
simple input signals are used we don’t need a better hardware demodulator. So, simple signals are better
demodulated than complex signals when using the same demodulator equipment. The simplicity of the
signal is a feature of its spectrum structure. The more harmonics in the spectrum of the input signal the
higher is the order of the Bessel function used for correct demodulation (i.e. the more complicated and
costly is the demodulator hardware) This is why a sine wave is better retrieved than a square wave by our
build circuits, since the spectrum of a sine wave contains one harmonic while that of a square wave contains
more harmonics.

Page 110
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Page 111
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Fig. 6.5: PLL as an FM Demodulator

Page 112
EEL 3552C Lab Manual 6. EXPERIMENT: FREQUENCY MODULATION

Page 113
 University of Central Florida

Department of Electrical & Computer

Engineering
EEL 3552
Signal Analysis & Communications

Experiment 1
Cover Page Template

Author: Instructor:
Author’s Name Professor’s Name

August 20, 2018

 Table 1: Lab Report Outline and Grading Key

Description
Section Grade (100)
——
Cover Page A cover page containing the course name and number, experiment title, names of 5
the collaborating students, name of professor and date
——
——
——
Objective A description of the scope and objectives of the experiment 5
——
——
Equipment Descriptions of all the equipments used in the experiment 5
——
——
Prelab An analysis with the needed details, calculations and proofs 15
——
——
Simulation 20
• Mainly a Matlab Simulation

• Code used should be provided

• Simulation results shown

• Simulation questions answered

• Simulation faced discrepancies discussed (not necessarily resolved or answered)

——
——
Experiment 30
• circuit description, discuss its assembly ( no need for real photograph picture
of the assembled circuit) A circuit schematic is preferred

• Detailed answers to all the questions in the experiment lab manual proce-
dure part with the needed tables, graphs and measurements and the required
analysis

• Faced discrepancies and possible solutions to them

1
• Graphs and measurements should be as described in the note below

——
——
Conclusion A paragraph or two containing a general overview of the experiment: 10

• Objective of experiment

• Lessons learned (or new methods)

• Faced discrepancies and suggested solutions

——
——
(Organization) This part is not a sectional part of the lab report it is rather a metric used by the 10
instructor to measure:

• Clarity of the report

• Organization of material and ideas

• Format structure (e.g. Clear figures and organized tables)

——
1 - Graphs and Figures: The figures should be clear with good resolution (not
necessarily colored) and should show the considered measurements (mainly from
oscilloscope) with a well presented caption and a discussion of the measurements
presented
Provide figures for each measurement you take from the oscilloscope.