ECE 316

2005
c by Bruno Korst-Fagundes Fall 2005

Experiment # 4

Frequency Modulation

1.5 1

1
−→ K_f + 2*pi COS A −→ 0.8

0.6

0.4
0.5

0.2

0
Message Hz/Volts cosine FM Signal 0

−0.5
Signal integrator −0.2

−0.4

−0.6
−1

Carrier −0.8

−1.5 −1
0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.5 1 1.5 2 2.5 3 3.5 4
−3 −3
x 10 x 10

1 Purpose

In Experiment # 3, you designed and built an AM modulator, and hopefully had the opportunity
to see how an envelope detector works in demodulating an AM signal. In this experiment, another
widely used modulation technique will be introduced: Frequency Modulation (FM). As the ampli-
tude of the sinusoidal carrier wave was modulated in AM, this time the instantaneous frequency
of a sinusoidal carrier wave will be modified proportionally to the variation of amplitude of the
message signal.

A three-step process will allow you to become more familiar with FM:

• First, a frequency modulation system will be created and simulated using Simulink.

• Second, an FM demodulator based on a Phase Locked Loop (PLL) will be designed and
simulated, using the previous modulator as the “signal source”.

• Third, an FM modulator will be implemented on a TMS320C6711 DSP platform. The input

(message) signal to the modulator will come from the signal generator.

You will be required to change various system parameters and observe the consequences of the
changes.

2 Background Reading and Preparation

Frequency Modulation (FM), as well as Phase Modulation (PM), are types of Angle (Exponential)
Modulation. These are covered in Chapter 5 of [2], or Chapter 3 of [1]. If you have the time, take a
look at the theory behind Phase-Locked Loops (PLL), as in pp. 184-186 of [2], and its use on FM

1
demodulators, as in pg. 236 of [2] or Section 2.14 of [1]. A good explanation about the role played
by the Bessel Function on the frequency domain representation of FM signal is found on pg. 115
of [1].

Also before coming to the lab, you should complete the lab preparation and hand it to the T.A..

3 Experiment

As it was mentioned above, this experiment is divided in three parts: the design and simulation of
an FM modulator in Simulink; the design and simulation of a PLL-based FM demodulator; and
the download and test of the designed modulator on the DSP platfom. For this third step, the
incoming signal will be produced by the signal generator.

All results are to be reported in the spaces provided in this outline. Make sure that
the T.A. verifies every result you record.

3.1 Designing and Simulating an FM Modulator

Based on the FM signal equation and the block diagram for an FM modulator that you have
submitted with your lab preparation (AHA!), build your model in Simulink. Follow similar
procedures for input sources and output sinks to what you have used in previous experiments. In
particular, do not forget the following details:

• The input to an FFT-based scope must be buffered, with a buffer size that is an integer
multiple of the FFT size (could be the same size). It makes no sense to attempt to perform
an FFT on a single sample, or buffer a smaller number of samples than the length of the
FFT;
• A greater FFT length means a better resolution for the frequency domain representation of
the signal (take 1024 points as a reasonable length). Frequencies which fall at an integer
multiple of fs /N will be resolved exactly. At this point you should be able to tell when the
results are producing relevant information or insufficient information;
• All blocks that require a sampling frequency (or sampling period) to be defined must have
the same sampling frequency. Since the daughtercard utilized in the lab makes use of a
48KHz sampling frequency, you should preferably use this number throughout your simulation
exercise. Feel free to experiment with different numbers during the simulation, but do not
forget to change it back to 48KHz prior to downloading your model to the target hardware.
• The frequency of a commercial FM carrier wave in practice is much higher than the one
utilized in this experiment. Your favourite FM radio station operates between 88MHz and
108MHz. One must keep in mind, however, that in the lab a limitation is imposed by the
sampling frequency utilized by the CODEC daughtercard.

Let us assume, hypothetically, that the straight-forward approach to “build the model from the
equation on the book” did not work. If it did, make sure you show the TA and get some bonus

2
marks. Now, if it did not, you should re-work the equation. Maybe bringing some elements “into
the integral” will simplify your block diagram. Do not despise the factor 2π on the constant kf if
you are working with ω (in rad/s). One of the reference textbooks may create some confusion for
not explicitly showing the 2π factor.

Use as input signal a sine wave with amplitude 0.5 (this would mean a 1Vpp simulated signal) and
1500Hz, a carrier frequency of 6000Hz and a sensitivity factor of 3600Hz/V (that’s the kf constant
on [2]). You should observe a time domain output signal similar to the one you drafted on your lab
preparation sheet. Include in your model a Freq Domain Scope. In this lab you will concentrate
on understanding what goes on in the frequency domain. You may remove the time domain scope
from your model. Make sure your frequency domain display gives you a good picture. Your model
should look like the picture below.

Figure 1: Suggested FM Modulator for Simulation

• Rewrite your FM equation (write it in the box below) and draw the new block diagram for the
FM modulator (it’s given, if you want)

For the next questions, use the graph presented below, when necessary.

3
 Bessel Functions of the First Kind
1

Jo(β)

J1(β)

J2(β) J3(β)
0.5
J4(β) J5(β) J6(β) J7(β)

−0.5
0 1 2 3 4 5 6 7 8 9 10
Modulation Index (β)

• The output signal observed in the time and frequency domains should look like the pictures
below. Present the calculation for the deviation ratio β, the values for the carrier and two
frequency components on each side of the carrier, and the bandwidth according to Carson’s
Rule.

Figure 2: FM in Time Domain and in Frequency Domain

4
 Write your answer below.

• Change the amplitude of the message signal to 1. Draw what you observe in the frequency
domain. For crying outloud, what happened to your carrier? Explain.

• Your input signal is a 1 Vp sinusoid. Is there any other input signal frequency for which there
will not be a carrier component in the frequency domain? Present some values for those
frequencies (use the Bessel Function graph given and test them using the simulation).

5
 • Bonus:If you set your β to be << 1, what will be the bandwidth? How about for β >> 1 ?

3.2 Designing and Simulating an FM Demodulator

Following the block diagram of the PLL-based FM demodulator you have drawn on your preparation
sheet, build a new model in Simulink. The input signal for the demodulator simulation will be
the FM-modulated signal from your modulator. Your demodulator should be made of a PLL and
an output low-pass filter. The PLL, as you have drawn on your lab preparation, is made of a
Voltage-controlled oscillator, a multiplier and a low-pass filter. It is not always necessary to have
a second filter after the PLL. You may be able to avoid the use of a second filter by using a loop
filter with a sharper roll-off at the cutoff frequency of interest. To make your life easier, you can
create a “subsystem” with your modulator (VCO), and utilize this new block to implement your
demodulator. To do this, select the blocks on yor model that implement the VCO, and click on
edit/create subsystem.

The constants and gains that you have to adjust for the VCO are the same as the ones used on
the modulator (the VCO is a frequency modulator). Make the filter a low-pass FIR of order 30,
designed using the windowing method with a Kaiser window. The cutoff frequency is the frequency
of your modulating signal. The output filter is identical to the filter in the loop. Note that
if you are looking for a loop filter with a sharper roll-off, you may need to redesign it to have
a higher order, or utilize a different window (you are using a Kaiser window) or yet to go for
another design method altogether. Feel free to explore with different filters, if time allows. The
loop filter will separate the “detected” signal from all other frequency components, wich arise from
the multiplication of the incoming (FM) signal with the one output by the VCO.

You can observe this by attaching Freq.Domain Scopes to your demodulator at appropriate prob-
ing points. Also, attach a time-based scope to the output. Now run your system and observe the
output scopes.

Make it work first, and show the T.A.. Make sure the sinusoid you see on the output of the
demodulator has the same frequency as the one input to the modulator.

• The intention behind this step is to hinder the detection performed by the PLL. Vary the
frequency of your carrier on the modulator by about 200Hz. Is your demodulator working
properly? If it is, increase the frequency until your output is no longer a sinusoid. Present a
brief explanation of what happened, in your own words.

6
 • Similarly to the previous step, bring the carrier frequency back to the proper value and this
time vary the frequency of your modulating signal (again, on the modulator) by 200Hz up
and down, starting at 2000Hz. You probably noticed that the output signal has been attenuated.
How would you fix this attenuation at your receiver? (yes, this is an easy mark)

3.3 Building and Running Your Model

Find yourself a station which has the Tektronix TDS2021 Oscilloscope. You may partner up with
another group to do this part. In this part of the experiment you will download your FM modulator
onto the DSP target hardware. Your model should now look like the picture below.

Figure 3: FM Model to be downloaded onto the target hardware

7
The input to your modulator will be a 100 Hz, 3.4Vpp sinusoid from the signal generator. Make
sure you adjust your signal generator to use the H igh Z Load setting (Agilent Signal Generator).
Remember, this 3.4Vpp amplitude value spans the whole range of the ADC on the target hardware.
If you use 1Vpp, for instance, you will not be working the full range of your modulator.

After you have created your model, build and run it on the target hardware (by pressing on
“incremental build”). One channel will display the input (or message) signal and the other the FM
signal modulated with the input.

Your FM signal will vary 3600 Hertz per Volt of the input signal, having 6000Hz (carrier) as a centre
frequency. Before you think you have “something that looks like AM” on the scope, remember that
there is an analog low-pass filter on the board, which attenuates progressively above approximately
5KHz. This is to say that the amplitude variation you see on the output in time-domain
is caused by the Audio Daughtercard of the target hardware. Remember, FM changes
instantaneous frequency, not amplitude.

• On one channel you have the message signal and on the other channel you have the FM
signal. Set your oscilloscope to perform an FFT on the FM-modulated signal. Input a 1 Vpp ,
1KHz sine wave. Use the cursors on the scope to identify the components. Draw below the
spectrum of your FM signal.

• Change the input signal to just under 1 Hz (yes, under ONE cycle per second). Is your
modulator doing what it should? What should the variation be? You will only be able to see
this in the frequency domain.

8
4 The Extra Mile

If you are brave enough, try out viewing your favourite radio station on the Spectrum Analyzer.
You will need to know the frequency of its carrier (in MHz) and a make-shift antenna. Set the
spectrum analyzer to start at frequency 88.1MHz and end at 108MHz. By doing this you will be
able to see all FM radio stations operating in Toronto. Choose one, and center the reading on the
spectrum analyzer on it. Since by regulation every FM radio station has about 200MHz for its
operation, you can change the span on the spectrum analyzer to be 200MHz. If you’d like to listen
to what is being displayed on your spectrum analyzer, turn on an FM radio and listen to it.

5 Conclusion

In this experiment, you were required to design and simulate an FM modulator and demodula-
tor, and to implement an FM modulator on a DSP platform. As a result, became familiar with
the theory behind Angle Modulation, and with practical details pertaining to Voltage-Controlled
Oscillators and Phase-Locked Loops as they were implemented on a DSP platform. All of these
concepts are widely used in the study of communication systems.

Obs: The books cited in this outline contain relevant material for the student to
better understand the topics related to the experiment. The student should note that
by reading the course textbook only, one should be able to successfully perform the
experiment. This is to say that it is not necessary for the student to purchase all the
references.

References

[1] S. Haykin Communication Systems, 4th Edition. Toronto: John Wiley & Sons, Inc., 2001.

[2] B. P. Lathi, Modern Digital and Analog Communication Systems, 3rd Edition. New York: Ox-
ford University Press, 1998.