l

ya
ha
G
a
ity
Ad
Chapter

Pulse Modulation

Need of analog to digital conversion, Sampling theorem for low pass signal in time domain and Nyquist
criteria, Types of sampling - natural and flat top. Pulse amplitude modulation and concept of TDM :
Channel bandwidth for PAM, Equalization, Signal recovery through holding. Pulse Width Modulation
(PWM) and Pulse Position Modulation (PPM): Generation and detection .

..

Chapter Contents
5.1 Pulse Modulation 5.10 Flat Top Sampling

5.2 Time and Band Limited Signals 5.11 Applications of Sampling Theorem

5.3 Narrowband Signals and Systems 5.12 Pulse Amplitude Modulation (PAM)

5.4 A to D Conversion 5.13 Pulse Width Modulation (PWM)

5.5 Sampling Process 5.14 Pulse Position Modulation (PPM)

5.6 Sampling Theorem for Low Pass Signals in 5.15 Comparison of PAM, PWM and PPM
Time Domain
5.7 Aliasing or Foldover Error 5.16 Multiplexing and Demultiplexing

5.8 Sampling Techniques 5.17 Time Division Multiplexing (TOM)

5.9 Natural Sampling

 PCS(Sem.4/E&Tc/SPPU) 5-2 Pulse Modulation
Definition :
5.1 Pulse Modulation :
Definition : PAM is defined as the type of pulse modulation in

l
ya
which, the amplitude of a constant width, constant

ha
Pulse modulation is defined as the type of modulation

G
position rectangular carrier is varied in proportion with
in which, the carrier is in the form of train of periodic

a
ity
rectangular pulses. the instantaneous magnitude of the modulating signal

Ad
Types: as shown in Fig. 5.1.2(c).

Pulse modulation can be either analog or digital. Analogk=

(a) i n p u t ~ :=::::--.-.
5.1.1 Analog Pulse Modulation : signal_ ~ ~L

□□□□□□□□
In the analog pulse modulation, the amplitude, width or
position of the rectangular carrier pulses is varied in
(b)Cameh I
accordance with the modulating signal. t

This will result in PAM (Pulse Amplitude Modulation),

PWM (Pulse Width Modulation) or PPM (Pulse Position
Modulation) respectively.
Thus PAM, PWM and PPM the examples of analog pulse
(<) PAM h O D D
'-J a EJ v , •t

modulation.
nn n CL.t
5.1.2 Digital Pulse Modulation :
In the digital pulse modulation, the information signal is
converted in some kind of a code which is transmitted
in the form of digital pulses.
The well known examples of digital pulse modulation (L-154) Fig. 5.1.2 : Pulse modulation

are Pulse Code Modulation (PCM), Delta Modulation

5.1.4 Pulse Width Modulation (PWM) :
(DM), Adaptive Delta Modulation (ADM), etc.

The classification of the pulse modulation system is as

. :·•,...>.'.:'.
University Questions
shown in Fig. 5.1.1.
Pulse Modulation
Q.t Reipi'E!$E!l1t and explain iri brie{ a Ji ;< • >
/ 'f~~~t!s:·_•PAM,· ··pvy_~.,•._.~ntf••·'p•·•.·

Definition :

PWM is defined as the type of pulse modulation in

Pulse Code
PAM Modulation (PCM) which, the width of carrier pulses is made to vary in
proportion with the instantaneous magnitude of the
modulating signal as shown in Fig. 5.1.2(d).
PWM is also called as Pulse Duration Modulation (PDM)
or Pulse Length Modulation (PLM).

(L-153)Fig. 5.1.1 : Classification of pulse modulation 5.1.5 Pulse Position Modulation (PPM) :
5.1.3 Pulse Amplitude Modulation (PAM):
University Ques1ions /,-.

~ Q. 1 • Represent and explain in brief, a sinuspidatsign~I

Q. 1 ?,..present and explain in brief, a sinusoidal signal by using PAM, PWM and PPM ,m9dlif?ticm
by using PAM, PWM and PPM modulation techniques. (Pee. o't,. 6 Marks)
t~chniques. (Dec. 07; 6Marks)
 •
, , . PCS (Sem. 4 / E&Tc / SPPU) 5-3 Pulse Modulation

Definition : Application of Pulse Modulation Systems :

PPM is defined as the type of pulse modulation in PAM does not have a good noise immunity. So its
which, the position of each pulse is varied in accordance practical use is restricted.

l
ya
ha
with the amplitudes of the sampled values of the PWM and PPM are used for some military applications

G
but are not used for commercial communication

a
modulating signal.

ity
applications.

Ad
In PPM the amplitude and width of the pulses is kept
PCM is the most useful method of all.
constant.
The position of the pulses is changed with respect to
5.1. 7 Need of Pulse Modulation :
the position of reference pulses. Even though the pulse modulation does not help in
The PPM pulses can be derived from the PWM pulses as frequency translation, it helps in some other aspects of
shown in Fig. 5.l.2(e). signal processing such as digital representation of
message signal.
Note that with increase in the modulating voltage the
In pulse modulation, we convert a continuous analog
PPM pulses shift further with respect to reference.
signal into a discrete signal which can be eventually
5.1.6 Pulse Code Modulation (PCM) : converted into a digital signal.
Thus some of the pulse modulation techniques are
fundamental to the digital communication field.
University Questions
Q. 1 . Repre~ent .and explain in brief, a si11usoidat signal 5.2 Time and Band Limited Signals:
by U$ing PAM, PWM and PPM modulation
techniques. (0~¢. 07, 6 IVlarks) 5.2.1 Time Limited Signal :
SPPU : Dec. 12, May 14, May 15, May 18
The analog message signal is sampled and converted to
University Questions
a fixed length, serial binary number as shown in
0,1 Oescribe with suitablf exaniple _bahdlimited and
Fig. 5.l.2(f). Umelimited signal. . (Dec. _12, May 18,_6 ~arks)
In other words a binary code is transmitted. Hence the
Q.2 Explain bandlimited and th:ne lirpited ~ig~a!~.
(May 14, May 15, 6 Marks)
name pulse code modulation.
Definition :
Difference between analog and digital pulse
The signal x (t) which has a non-zero value only over a
communication :
certain interval of time is called as a time limited signal.
For analog as well as digital pulse communication The value of the time limited signal outside this time
systems, the transmitted signal is a discrete time signal. interval is zero.
In analog pulse communication, the information is Fig. 5.2.1 shows some examples of time limited and
transmitted in the form of change in amplitude, width time unlimited signals.

or position of the rectangular carrier pulses.

So the transmitted pulsed signal is still an analog signal.

In digital pulse communication, the information is

.6
-2 0 2

transmitted in the form of codes. Codeword are formed

by grouping the digital pulses. I' "j' 't
Note that for digital pulse communication we do not
change amplitude, frequency or phase of the
0
□ •t
transmitted signal.

Thus the transmitted signal in digital pulse (a) Time limited signals
communication is a digital signal. Fig. 5.2.l(Contd ...)
.
~~
l"
-
TetltKnowledge
Fubilr.atlons
~I!!'
~p PCS (Sem. 4 / E&Tc I SPPU) 5-4 Pulse Modulation
~==:a'a.-======a-==============•·=-=========-=
Examples:

The examples of bandlimited signals are the speech

signal which occupies the band of 0 - 3.4 kHz or the

l
ya
ha
video signal ·nr1ich occupies a frequency band of

G
a
0-4.5 MHz.

ity
Ad
5.2.3 How to Obtain a Bandlimited Signal?:
We can obtain a bandlimited signal from a band
unlimited signal by passing it through a low pass or
band pass filter as shown in Fig. 5.2.3.
(b) Time unlimited signals
(0-396} Fig. 5.2.1
. r±::r.
-100 100
f

Band .unlimited -~
Lp F .
•..···------~ Bandlimited
5.2.2 Bandlimited Signals : signal ~ signal

-~-. A-.
SPPU: ,Dec. 12, May 14, May 15, May 18
University Questions
Q. 1 De~cribe with suitable exartiple • bandlirhlted and
time limited signal. (Dec. 12, May 18, 6 Marks)
(D-400) Fig. 5.2.3 : Obtaining the bandlimited signal
Q. 2 Explain bandlimited and tima limited signals.
(May 14, May 15, 6 Marks) The LPF passes the frequency components between
- 100 to 100 Hz only to produce a bandlimited signal.
Definition :
Note:
If the amplitude spectrum of a signal is non-zero only
1. A signal cannot be time limited and bandlimited
over a specific band of frequencies and zero outside this simultaneously.
band then the signal is called as a band limited signal.
2. A time unlimited signal is bandlimited in the
Fig. 5.2.2 shows some examples of bandlimited and frequency domain.
band unlimited signals. 3. The time and frequency domains are inter-related

,k
-f
, X(l
O f
,
with each other and with the help of Fourier
transform, it is possible to travel from time domain
to frequency domain and vice versa.

5.3 Narrowband Signals and Systems :

'f 5.3.1 Narrowband Signals :
-200 300
Definition :
l...
- - - - X - ( f )... 1 1 < - - ~ - - ~ - - - The signals that have a small bandwidth are called as
2 5 fkHz
the narrowband signals.
(a) Bandlimited signals
The signals that occupy a large bandwidth are called as
X(f)
the wideband signals.
Examples:
For example the speech signal that occupies the
frequency band of 0 - 3.4 kHz is a narrowband signal,
whereas the video signal that occupies the frequency
band of 0 - 4.5 MHz is a broadband signal.

As stated by the time scaling property of Fourier

(b) Band unlimited signals transform, the compression of signals in the time
(O-399) Fig. 5.2.2
.
domain will expand their frequency spectrum i.e. a
..
~~ TethKnowlellge
~~ Pub!lr.Jtlons
~•er
~~ PCS (Sem. 4 / E&Tc I SPPU) 5-5 Pulse Modulation

5.4 A to D Conversion :
wideband signal, whereas expansion in the time domain
res u Its in the compression of the frequency spectrum
-----------------------
Need:
(narrowband signal).

l
ya
ha
For all the digital communication systems, the analog to
Thus the delta function 8 (t) is an example of a

G
digital conversion is essential.

a
wideband signal whereas the de signal is a narrowband

ity
Ad
signal as shown in Fig. 5.3.1. The process of converting the analog data to digital
signal is known as digitization.

l 0 t

(a) Delta function is a wideband signal

0
Spectrum
It is also called as A to D conversion. In this conversion
process, the input analog data is converted into
equivalent digital signal as shown in Fig. 5.4.1.
1 1
~

±
/""'\. Analog
V

l
signal Digital signal

(L-218) Fig. 5.4.1 : Transformation from Analog

0 t O f signal to digital signal

(b) A DC signal is a narrowband signal In order to carry out this transformation, one has to
(D-397) Fig. 5.3.1 follow a sequence of operations such as sampling,
5.3.2 Narrowband System : quantization and encoding.

Definition : Examples:

This process is essential in all the digital communication

The system that has a small bandwidth is known as a
systems such as Pulse Code Modulation (PCM) or Delta
narrowband system.
Modulation (D.M).
The system that has a large bandwidth is called as a Block Diagram :
wideband system.
The analog to digital conversion (A/D) can be achieved
Examples:

The examples of narrowband systems are : SSBSC, ~ Q~~ntized

DSBSC and DSBFC (Conventional AM) whereas the J ~gnal

examples of the wideband systems are FM, PCM, TDM,

etc.

The FM system with smaller bandwidth is called as the

1 0 0 1
narrowband FM system (it is used for the point-to-

point communication) whereas the one used for FM

.JlILJUlJL
broadcasting is known as the wideband FM system. (E-1) Fig. 5.4.2 : Analog-to-digital conversion

The narrowband systems have higher bandwidth This system consists of three blocks namely sampler,
quantizer and encoder.
efficiency whereas the wideband systems are less
bandwidth efficient. The message signal can be analog or digital type. An
analog signal can always be converted into a digital
' SSE(' SSB
X(f) x(t) signal signal.
Sampler:
~mf .J fc
~
fc+fm The analog signal is applied at the input of the sampler.
BW=fm The sampler is a switch which samples the input signal
(D-398) Fig. 5.3.2 : SSB is a narrowband system at regular intervals of time and produces the discrete
version of the input signal.

~~ TecllKnowledge
~jl# Put1!Ji;;3tlons
 •
~ , PCS (Sem. 4 / E&Tc I SPPU) 5-6 Pulse Modulation

Quantizer: In many applications, it is easier to process discrete-

Quantization is a process of approximation or rounding time signals because it is more flexible and is often
off. preferable to processing continuous-time signals.

l
ya
ha
Quantization process approximates each sample to its This is because the discrete signals obtained due to

G
nearest standard voltage level called quantization level.

a
sampling can be easily processed by the digital

ity
Ad
We get the approximated version of the sampled signal technology systems that are inexpensive, lightweight,
at the output of the quantizer.
programmable, and easily reproducible
The number of quantization levels is finite and generally
Due to sampling, it becomes possible to use discrete-
it is a power of 2 i.e. 2, 4, 8, 16, 32 ..... .
time system technology to implement continuous-time
Encoder:
systems and process continuous-time signals.
An encoder converts each quantized sample into a
Thus, we use sampling to convert a continuous-time
separate code word of length say N bits.
signal to a discrete-time signal, process the discrete-
Thus at the output of the encoder we get digital code
words. time signal using a discrete-time system and then
convert back to continuous time.
5.5 Sampling Process : SPPU : Dec. 17
Block diagram :

University Questions Fig. 5.5.1 summarizes the sampling process.

Q. 1 Discuss the role of sampling theorem in digitization Continuous time ••·s • :\
. amp 1er
Discrete time
.
of signal. (Dec. 17, 6 Marks) analog sign(\ . > ana 1og s1gna 1

Definition : Sampling
t signal
The sampling process is defined as the process of
s(t)
converting a continuous time analog signal to a discrete
analog signal and the sampled signal is the discrete (L-156) Fig. 5.5.1 : Sampling process
time representation of the original analog signal. At the input of the sampler we apply an analog signal
Importance of sampling : and at its output we get the sampled version of the
input signal as shown.
If certain conditions are met, a continuous-time signal
can be completely represented by and recoverable from This sampled signal represents the original analog input
signal faithfully if the sampling rate is adequately large.
knowledge of its values, or samples, obtained at points
equally spaced in time. Need of sampling:

This is a surprising property that follows from a basic In the pulse modulation and digital modulation systems,
the signal to be transmitted, needs to be in the discrete
result that is referred to as the sampling theorem.
time form.
This theorem is extremely important and useful.
If the message signal is coming from a digital source
This property is used in certain applications, such as in (e.g. a digital computer) then it is in the proper for a
moving pictures, which consist of a sequence of digital communication system to be processed.
individual frames, each of which represents an
However, this is not always the case. The message signal
instantaneous view (i.e., a sample in time) of a can be analog in nature (e.g. speech or video signal).
continuously changing scene.
In such a case it has to be first converted into a discrete
When these samples are viewed in sequence at a time signal. We use the "sampling process" to do this.
sufficiently fast rate, we perceive an accurate Thus using the sampling process we convert a
representation of the original continuously moving continu:;us time signal into a discrete time signal.
scene. Input signal :
The sampling theorem also acts as a bridge between For the sampling process to be of practical utility it is
continuous-time signals and discrete-time signals. necessary to choose the sampling rate properly.
.
~I!!!' TechKnowled.ge
..
lo n ~
~Fr, 1.1t1: 11: i3 t
~·,
~- PCS (Sem. 4 / E&Tc / SPPU) 5-7 Pulse Modulation

The input signal x(t) should be a band limited signal. Q. 2 State ~nc! p~bv~ sall"lpHng theorer11 WlttisGit~t:ii~' ......
That means its spectrum should exist only between 0 •wavef9rms•ana·11"1clthe'ntaticai ~xptession:.·•••· •
(IVl~y 15, TM~rks}

l
ya
and some finite frequency say fm or W.

ha
Sampling signal : stat;theldw·pass.sampUhgth~~felllandbn1t1y ··•·

G
a
explai11it$signiflcance.··· • • (Pec.15,JtMarks) .

ity
A sampling signal s(t) is a train of unit impulses, spaced . .· .•• .· .• .. ·: .•. . < ._.·.· .• .•.

Ad
Q; 4 Discuss-the role ofsarripling theorem in digitization
by a period of Ts seconds.
•• >>rOf sigi,~l,2>: i•• , •• • • ••••• ••••••• ••••i••(Qe~,4}!,.~.,l~tl<S)
This sampling function samples the input signal at a
Q. 5 State sampling theorem and discuss its. types.
rate of "fs" samples per second.
(May 18, 6 Marks, Dec.19,7<Marks)
Therefore "Ts" represents the sampling period such that, Q. 6 State and. prove sampling theorem,
1 (Dec 1 18, . $ Marks)
Ts = f s = Sampling period ... (5.5.1)
Introduction :
1
and fs = = Sampling rate.
Ts In order to represent the original message signal
"faithfully" (without loss of information), it is necessary
Output signal :
to take as many samples of the original signal as
The output signal is the sampled version of the input
possible.
signal.
Higher the number of samples, closer is the
It is a discrete time signal which is represented by x0 (t).
representation.
Output signal is obtained by multiplying the input
The number of samples depends on the "sampling rate"
signal with the sampling signal in time dimain.
and the maximum frequency of the signal to be
.'. X0 (t) = X (t) X S (t) sampled.
Reconstruction or recovery : Sampling theorem was introduced to the
Reconstruction or recovery is the process of recovering communication theory in 1949 by Shannon.
the original signal x (t) from its sampled version x0 (t).
Therefore this theorem is also called as "Shannon's
For this the sampled signal x0 (t) is processed through sampling theorem".
an ideal low pass filter with gain T and cutoff frequency
The statement of sampling theorem in time domain, for
greater than W.
the band limited signals of finite energy is as follows :
The resulting output signal will be exactly equal x(T).
Statement:
Requirements of sampling process :
The sampling process should satisfy the following 1. If a finite energy signal x (t) contains no
requirements : frequencies higher than "W" Hz (i.e. it is a band

1. Sampled signal should represent the original signal limited signal) then it is completely determined by
faithfully. specifying its values at the instants of time which
are spaced (1/2 W) seconds apart.
2. We should be able to reconstruct the original
signal from its sampled version. 2. If a finite energy signal x (t) contains no frequency
components higher than "W" Hz then it may be
5.6 Sampling Theorem for Low Pass
Signals in Time Domain : completely recovered from its samples which are
spaced (1/2 W) seconds apart.
SPPU: May 14, May 15, Dec. 15, Dec. 17, May 18,
Dec.18,Dec.19 Combined statement of sampling theorem :A continuous
University Questions time signal x (t) can be completely represented in .its
,'. ··,> ·.-·· •· ...

Q.1 What is Nyqtiistctiterion? State sampling theorehl sampled form and recovered back from the sarnptedform if
in time ddmain: Draw
the spectrum showing . the sainplihg frequency fs c 2 W where "W' is the maxim1Jm
aliasing arjd
gµard~and. • (May Ma.rks) . 14-, 7 frequency of the continuous time signat x (f).'
~~
~F PCS (Sem. 4 / E&Tc / SPPU) 5-8 Pulse Modulation

5.6.1 Proof of Sampling Theorem : 1

T, = f s = Sampling period ... (5.6.1)
SPPU: May 15, Dec. 15, May 16, May 17, Dec .•18
1

l
ya
Un,tyersity ,Questions and fs =
T, = Sampling rate.

ha
Q. 1 State and prove sampling theorem with suitable

G
a
waveforms and mathematical expression. Procedure to be followed :

ity
(May 15, 7 Marks)

Ad
We are going to follow the steps given below to prove
Q. 2 State the<IOllV P.Eiss sampling theorem <ind bri~fly
explain its signffrcance. • (Dec.15, 6 Marks) the sampling theorem :
Q. 3 State and prove sampling theorem in time domain. Step 1 : Represent the sampling function s (t)
(May 16, 7 Marks)
mathematically.
Q. 4 State and prove sampling theorem for bandlimited
signal. (May 17, 7 Marks) Step 2: Represent the sampled signal Xo (t)
Q. 5 State and prove sampling theorem. mathematically.
(Dec. 18, 6 Marks) Step 3 : Obtain the Fourier transform of the sampled signal.

Let us now prove the sampling theorem in time domain. Step 4 : Prove that the sampled signal x3 (t) completely

The assumptions made for this proof are as follows : represent$

.
x (t).
. . . ......

Step 5 : Represent x •(t) as summation of sln~ifUnctiC>t'lS

Assumptions :
(interpolation).
Let x (t) be a continuous time analog signal as shown in Step 6 : Graphical representation of the interpolation
Fig. 5.6.1. process.

Continuous Step 7 : Actual recovery of x (t) using an ideal low pass

time filter.
analog (a)
signal
x(t) Part -1 : Spectrum of the sampled signal :

Step 1 : Represent the sampling function s (t)

mathematically :
Aunit 1
impulse
t l······ (b)
Fig. 5.6.1 shows the sampling function s (t) which is a
train used • II> train of unit impulses.
as sampling 0 T0 2T5 3T8 4T8 ............. t
function
/ x~~ ~~ The spacing between the adjacent unit impulses is T,

Sampled ,, ,. '' ,. seconds, therefore the frequency of the sampling

I 3T8 4T8
signal
0
' ✓
(c) function is equal to the sampling frequency f,.
/
/
2T5 '-, ;
/ t
' /
The sampled signal is denoted by x6 (t) and it is as
'' ~
~

shown in Fig. 5.6.1.

(D-408) Fig. 5.6.1 : Sampling of a continuous time signal x (t) The sample function s (t) can be represented
mathematically as follows :
Let x (t) be a signal with finite energy and infinite
duration. s (t) = ....... 8 (t + 2 T,) + 8 (t + T,) + 8 (t)
Let x (t) be a strictly bandlimited signal. That means it + 8 (t - T,) + 8 (t - 2 T,) + .....
does not contain any frequency components above 00

"W" Hz. .-. s (t) = L 8( t- nT,) ... (5.6.2)

n =- oo
Let s (t) be the sampling function as shown in
Step 2 : Represent the sampled signal x6 (t)
Fig. 5.6.1.
mathematically :
It is a train of unit impulses, spaced by a period of T,
seconds. Fig. 5.6.1 shows the sampled signal x6 (t) graphicallY,, It is
present only at the sampling instants i.e. T,, 2 T, etc. and
This sampling function samples the original signal at a
its instantaneous amplitude is equal to the amplitude of
rate of "fs" samples per second. Therefore "Ts"
original signal x (t) at the sampling instants.
represents the sampling period such that,
,e
~'fl PCS(Sem.4/E&Tc/SPPU) 5-9 Pulse Modulation

This is shown by the encircled points in Fig. 5.6.1. Let us i.e. Xs (t) = X (t) X s (t) ... (5.6.6)
represent the instantaneous amplitude of x (t) at the Taking the Fourier transform of both the sides we get,
various sampling points t = n T, as x (n T,).
... (5.6.7)

l
ya
This is the amplitude of the encircled points of

ha
This is because the Fourier transform of the product of

G
Fig. 5.6.1.

a
two signals in the time domain is the convolution of

ity
Looking at the sampled signal x8 (t) we can say that the

Ad
their Fourier transforms.
sampled signal is obtained by multiplying x (t) and s (t).
Substituting the value of S(f) from Equation (5.6.5) we
... (5.6.3)
get,
Substituting the expression for s (t) from
Equation (5.6.2) we get the mathematical expression for
the sampled signal x8 (t) as,
00
Where * denotes convolution. Interchanging the orders
I x (nT,) • 8 ( t - nT,) ... (5.6.4) of convolution and summation results in:
n=- oo 00

Step 3 : Obtain the Fourier transform of the sampled X8 (f) = f, I X (f) * 8 (f - nf,) ... (5.6.9)
n = -oo
signal:
The fourier transform of a train of impulses (dirac delta From the properties of delta function, we find that the
function) is given by, convolution of X (f) and 8 (f - nf,) is equal to

00
X (f - nf,).
X (f) = f 0 I 8 (f - nf0) Hence the above equation can be simplified as follows :
n = -oo
00

Here we have the similar pulse train as sampling F.T. of the sampled signal, X0 (f) = f, I X (f - nf,)
function s (t). n = -oo

Therefore the Fourier transform of the sampling ... (5.6.10)

function is given by, where X (f) = Fourier transform of the original signal
00
X (t).
s (f) = f, I 8 (f - nf,) ... (5.6.5)
n = -oo The spectrum X6 (f) of the sampled signal is plotted as

Note that f 0 has been replaced by f, in the above shown in Fig. 5.6.2.
equation.
The sampled signal in the time domain is represented
as product of x (t) and s (t).
X(f)

Spectrum of a bandlimited
signal x(t)

/
-w 0 w
(a) Spectrum of the original signal x(t)

Spectrum of the
sampled signal
atf8 = 2W

-3W

(b) Spectrum of the sampled signal x8 (t) with f 5 "' 2W, i.e. Sampling
is done exactly at Nyquist rate
(D-410) Fig. 5.6.2 : Spectrum of sampled signal
~•::w
~f"
Tecl!Knowledge
Put,:ii;;3tions
~el!
~, PCS(Sem.4/E&Tc/SPPU) 5-10 Pulse Modulation

Conclusion from Equation (5.6.10) : X0 (t) = .... X (- Ts) 8(t + Ts ) + X (0) 8 (t)
The term X (f - nfs) in Equation (5.6.10) represents the
+ X (Ts) 8 (t - Ts) + ...... (5.6.12(a))
shifted version of the spectrum X (f) of the original
00

l
ya
signal x (t).
L (n Ts) • 8 (t - nTs)

ha
X
= -oo

G
n
Thus depending on the value of "n" (which extends

a
ity
from - oo to + oo) we will get infinite number of original We can obtain another useful expression for the fourier

Ad
spectrums X (f) centered at frequencies 0, ± fs, ± 2fs, transform X0 (f) by taking the fourier transform of both
± 3fs, ± 4fs ... etc. In other words, the sides of the equation stated above as,
00
X (f - nfs) = X (f) at f = 0, ± fs, ± 2f9 ± 3fs ... (5.6.11)
L x (nT)
s e-j2nnfTs ... (5.6.13)
This concept will be clear if we open Equation (5.6.10) n=- oo

and write the terms separately as shown below. This equation is the Fourier transform of a discrete time
Now open the summation sign in Equation (5.6.10) to signal x0 (t).
get, Step 2 : Obtain the FT of input signal :

X, (f) = •••••••L
fs X (f+
+ 2fs) + L
X (ff+s
fs) +l<,
X (0 • T X;~shiftod Oghl by<,
Compare it with the definition of fourier transform of a
continuous time signal. i.e.
Spectrum X(f) 00
X (f) shifted left by f5 X (f) = f x (t) e- j 2" ft dt
X (f) shifted left by 2f5 -oo

(D-1705) As the signal is discrete, the integration sign has been

Equation (5.6.10) can also be written as : replaced by the summation sign and "t" has been
00 replaced by "nT;'.
L
X0 (f) = fs X (f) + n = _ 00 fs X ( f - nfs) ... (5.6.12) Now consider Equation (5.6.12)

n *0 00

Comment:
L
X0 (f) = fs X (f) + n = _ 00 fs X (f - nfs)
From Equation (5.6.12) we conclude that the process of n *0
uniform sampling of a signal in the time domain results
in a periodic spectrum in the frequency domain with a
period equal to the sampling rate fs. But in the range - W 5. f 5. W the second term of the
Part -2 : Prove that sampled signal x6 (t) completely above expression will not be present
represents x (t) : 1
X (f) = fs XO (f) ... (5.6.14)
Step 1 : Obtain the FT of sampled signal :

The sampled signal has been shown in Fig. 5.6.3. Substitute fs = 2 W and X8 (f) from Equation (5.6.13) to

get,
00
1
.. X (f) = 2W L X (nTs) . e-j 2nn fTs ... (5.6.15)
.. .. n=-oo

•·. This is the frequency spectrum of x (t) in terms of

0 Ts 2T5 3T5
x (nTs) i.e. the sampled signal.
\_x(T5 )8(t-Ts)
\_ Substitute Ts = 1/2 W to get,
x(O)o(t)

(D-411) Fig. 5.6.3 : Sampled signal x6 (t) 1 00

X (f) = W L X (n / 2 W) • e- j 2itn 112 w .... - W 5. f ~ W
2 n = -oo
x0 (t) can be represented in the summation form as
follows (Refer Fig. 5.6.3). ... (5.6.16)
~~
~~ PCS (Sem. 4 / E&Tc / SPPU) 5-11 Pulse Modulation

Note,: .•Th.is ~qµ,tipn sl19vvs that th~ spectrum of x {t) is Sampled signal \Reconstruction\ Original signal
•• $ai'ne 8$ foe; spectrum of Xs (t) in the. frequency filter i----.x(t)
X 0(t)
range'-Wto+W. Ideal L.P.FC

l
ya
Hence the sampled signal represents the original signal (D-413) Fig. 5.6.4(a) : Recovery using an ideal LPF

ha
x (t) successfully.

G
The ideal low pass filter is called as reconstruction

a
ity
Thus if the sample values x (n/2W) of the signal x (t) are filter.

Ad
specified for all time, then the Fourier transform X (f) of The resulting output signal will be exactly equal x(D.
the original signal is uniquely determined by using the
The filter used here is an ideal low pass filter with a
Equation (5.6.16).
brickwall frequency spectrum as shown in Fig. 5.6.4(b).
Because x (t) is related to X (f) by the inverse Fourier Spectrum of x(t)

transform, it follows that the signal x (t) is itself uniquely

determined by the sample values x (n/2W) for
- oo ::; n ::; oo.
-Is 0 •rW Frequency
In other words, the sequence of samples {x (n / 2W)}
contains all the information of x (t). Frequency response
of reconstruction filter
Thus, we have proved second part of the sampling
theorem. -W 0 +W Frequency

This equation shows that the spectrum of x (t) is same

x(t)
as the spectrum of x6 (t) in the frequency range - W to
+W.
Spectrum of filter output
Hence the sampled signal represents the original signal
x (t) successfully. .. w 0 +W Frequency

(D-414) Fig. 5.6.4(b): Operation of reconstruction

Thus if the sample values x (n/2 W) of the signal x (t) are
When the sampled signal x8 (t) is applied at the input,
specifi~d .for all time, then the Fourier transform
this filter will allow only the shaded portion in the
X (f} ofthe origin~! signaUs uniquely determined by using
spectrum of x8 (t) to pass through to the output and will
•the Equ~tion ($.q.l6). • ••
block all other frequency components.

i:i;~lJ;~;~iLft)•·J~r,fati4Jo·•·)(••·.m:Qy, the . •ioverse f:9urier Thus the frequency components only corresponding to
. . .•tt~ri$¥¢f~;•· it}FdriBWs ±k~r•iw~ §igrfo/•.)$ ,(t1····j··~·· it~§lf\,n.iqµely· x (t) will be passed through to the output and the
·,~~i~r~ih~d·•,tt'tK~··:~1£;'1;s::1·c;;·· ;···(~/;·· W)····;;~·•··~•;··•i·· ·~· original signal x (t) is recovered.

·~co> However, ideal filters are generally not used in practice

for a variety of reasons.
6J"in other wor~s the sequehce of samples'{x (n /2 W}l
In any practical application, the ideal LPF in Fig. 5.6.4(a)
c6nt~lnsallth~·1rif<>rmatidn bf x(t):'
would be replaced by a non ideal filter HQw) that
Thus we have proved first part of the sampling theorem. approximates the desired frequency characteristic with
5.6.2 Recovery using Ideal LPF : sufficient accuracy. (i.e., HOw) "" 1 for lwl < W, and
HOw) "" 0 elsewhere).
Reconstruction or recovery is the process of recovering
Due to such approximation in the low pass filtering
the original signal x (t) from its sampled version x6 (t).
stage, some discrepancy is expected to exist, between
For this the sampled signal x8 (t) is processed through
the original signal and the recovered signal.
an ideal low pass filter with gain T and cutoff frequency
We need to select the non-ideal filter based on the
greater than W.
acceptable level of distortion for the application under
The recovery process is as shown in Fig. 5.6.4(a).
consideration.
~~ TedtKnowlellge
~r" Pubiications
 • PCS (Sem. 4 I E&Tc I SPPU)
~F 5-12 Pulse Modulation

5.6.3 Interpolation: Therefore Equation (5.6.17) can be written as :

Definition : OCJ

x (t) = I x (n / 2W) sine (2 Wt - n) ... (5.6.19)

Interpolation, that is, the fitting of a continuous signal

l
n=-oo

ya
ha
to a set of sample values, is a commonly used

G
Equation (5.6.19) provides an interpolation formula for
procedure for reconstructing a function, either

a
ity
reconstructing the original signal x (t) from the

Ad
approximately or exactly, from samples.
sequence of sample values {x (n /2W)}.
5.6.4 Interpolation Formula :
The "sine" function plays the role of an interpolation
From Equation (5.6.16) we can obtain x (t) by taking the
function.
inverse Fourier transform (]FT).
Each sample x (n/ 2W) is multiplied by a delayed version
x (t) = !FT {X (f))
of the interpolation function i.e. sine function.
1 OCJ • }
= IFT { 2W n ;_ oo x (n / 2W). e-Jnfn/W Then all these resulting waveforms are added to obtain
X (t).
Using the definition of inverse Fourier transform, Graphical representation of the interpolation process :
w 1 OCJ
2 Let us re-arrange Equation (5.6.19) as follows :
x(t) = f W L x(n/2W)•e-jnfn/Wr! "ftdf
2
-W n = - oo
x (t) = n =;- x (nTs) sine 2W (t - 2~)
Interchanging the order of summation and integration 00

we get, 1
This is because
2
w = Ts
OCJ
X (t) =
x (t) = I x (nTs) sine 2W (t - nTs) ... (5.6.20)
n=-oo
OCJ
1 1
x (t) = I x (n / 2W) • 2W x [ n ] Let us expand this equation to write,
n = - oo j2n t - 2W
x (t) = x (0) sine 2 Wt + x (±Ts) sine 2W (t ± T,)
. [ e j 2n 1 (t - n/2WJ ] ':_w + x (± 2T,) sine 2W (t ± 2Ts) + ... ..... (5.6.21)

Refer Fig. 5.6.5 for the graphical representation of the

OCJ 1
X (t) = I x (n / 2W) [ n ] interpolation process.
00
n=- j4nW t - W
2 1. First term : x (0) sine 2Wt :
j 21t W(t - n/2W) - j 21t W(t - n/2W) ]
• [e -e This will have a maximum amplitude at t = 0. The
maximum amplitude is equal to the sample value x (0)
00 [ j 21t W(t - n/2W) - j 2n W(t - n/2 W) ]
e -e at t = 0.
= I x (n / 2W) • . [ n ]
n=-oo J4nW t - - - This sine function will pass through zeros at t = ± 1/2 W,
2W
± 1/4 W ... etc.
The term inside the square bracket is a "sine" function.
This is as shown in Fig. 5.6.5.
OCJ
sin (211:Wt - nn)
:. X (t) = n ;_ oo x (n / 2W) (211:Wt- nn) ... (5.6.17) 2. Second term : x (± T5) sine 2W (t ± T5) :

This sine function will have maximum amplitude at

We can simplify the equation above by using the t =±Ts.
definition of the "sine function". The maximum amplitude is equal to the sample value. x

The sine function is defined as : (± Ts) at t = ± Ts respectively.

sin (nx) Thus sine 2W (t ± T,) represents shifted sine function i.e.
sine x = ... (5.6.18)
nx "sine 2Wt" by a period ±Ts.This is as shown in Fig. 5.6.5.
 • PCS (Sem. 4 / E&Tc I SPPU) 5-13 Pulse Modulation

3. Remaining terms : We can plot all these sine functions along with the

Similarly the third term, x (± 2T5) sine 2W (t ± 2T5) sampled signal x0 (t) as shown in Fig. 5.6.5.

l
ya
represents shifted sine function "sine 2Wt" by a period

ha
of± 2T5 and so on.

G
a
ity
Ad
Sampled signal
',,,,, '.✓

Reconstructed signal x(t) [Output of filter]

(D-412) Fig. 5.6.S : Reconstruction of the original signal x (t) from its samples using the interpolation

Note that the peak amplitude of any sine function is Q. 4 What are the effects .of. under sampllrrg ? Explain
equal to the corresponding sample value x (nTJ with the help of frequency sp~ctrum. Also explain
Actual reconstruction with a low pass filter :
function of reconstruction filter. (Dec; 14; i Marks}
Q. 5 What is aliasing? How can it be avoi_ded'?
As shown in Fig. 5.6.5, the peaks of the sine pulses .(Dec; 15,·3.Marks)
represent the amplitudes of the samples. .Q; 6 Explain distortion occur in pr9c:ess ofsampHl1g and
The signal x (t) expressed in Equation (5.6.19) i.e. the its remedial solutions. (O~c. 17,7. l\/larks)
sampled signal is then passed through an ideal low pass If the signal x (t) is not strictly bandlimited and / or if
filter to recover the original signal x (t). the sampling frequency f5 is less than 2 W, then an error
This low pass filter is therefore called as the called aliasing or foldover error is observed in the
reconstruction filter. sampled signal.
5.7 Aliasing or Foldover Error : For the signal x (t) is not strictly band limited, the

SPPU: Dec. 12, May 13, May 14, Dec. 14, spectrum of signal x (t) is shown in Fig. 5.7.l(b).

The adjacent spectrums overlap if f 5 < 2 W. This is

University Questions
shown in Fig. 5.7.l(b).
Q~'1 Ostat~ '.~fuJiig~ fu~tr;;_ •. ~xplairt ·~i1a~rng. and The spectrum X0 (f) of the discrete time signal x6 (t) is
••• • • • $pJrh.if~';#~tin detail'.Withspectral tji?St~fus.
••••••• , •• • • • • • • •••• • • • •••• •• • :(Dec, 12,J~ Marks) shown in Fig. 5.7.l(b) which is nothing but the sum of

Explain aliasing>. and·. different· ways }o, .;ivc,id X (f) and infinite number of frequency shifted replicas of
aliasing. (May 13, 8 Marks) it as explained earlier.
What is Nyquist criteria ? State sampling theorem Consider the two replicas of X (f) which are centered
• .• tinie •• dbrrtaih. Draw •the spectrJrn showing
about the frequencies f 5 and - f 5.
aliasing and guardband. (May 14, 6 Marks)
 •
~ , PCS (Sem. 4 / E&Tc / SPPU) 5-14 Pulse Modulation
X(f) x(t)

l
ya
Strictly

ha
bandlimited x(t)
fj2 High frequencies in x(t)

G
(a) Spectrum of a continuous time signal x(t) Fig. S.7.2(a): Use of a bandlimiting filter to eliminate

a
(D-416)

ity
X;s(fJ

Ad
aliasing

This filter has a cutoff frequency at fc = W, therefore it

will strictly bandlimit the signal x (t) before sampling
takes place.
High frequencies in x(s) take on the identity
of lower frequency due to aHasing This filter is also called as antialiasing filter or prealias
(b} Spectrum of the sampled version of x(t) with f5 < 2W filter.

(D-415) Fig. 5.7.1 2. Using the sampling frequency f5 > 2W :

If we use a reconstruction filter with its pass-band In order to avoid alising, increase the sampling
extending from - f 5 / 2 to + f 5 / 2 then its output will not frequency f 5 to a great extent i.e. f, >» 2W.
be an undistorted version of the original signal
Due to this, even though x (t) is not strictly band limited,
X (t).
the spectrums will not overlap.
Some distortion will be present in the filter output.
A guard band is created between the adjacent
The distortion occurs due to the overlapping of the
spectrums as shown in Fig. 5.7.2(b).
adjacent spectrums as shown in Fig. 5.7.l(b). X(f)
Spectrum of original
Due to this overlapping, it is seen that the portions of signal x(t)
the frequency shifted replicas are "folded over" inside
the desired spectrum. -W: :w
Guard band
Due to this "fold over", high frequencies in X (f) are -+: :.-
reflected into low frequencies in X0 (f).

This can be understood by comparing the shaded

portions of the spectra shown in Fig. 5.7.l(a) and (b).
Definition of Aliasing :
(D-417) Fig S.7.2(b): Spectrum of a sampled signal for
The phenomenon of a high frequency in the spectrum
f, > 2W
of the original signal x (t), taking on the identity of
lower frequency in the spectrum of the sampled signal 5.7.2 Nyquist Rate and Nyquist Interval:
x0 (t) is called as aliasing or fold over error.
Effect of aliasing : University Questions

Due to aliasing some of the information contained in

the original signal x (t) is lost in the process of sampling.

5.7.1 How to Eliminate Aliasing?:

Aliasing can be completely eliminated by:

1. Using an antialiasing or prealiasing filter and
2. Using the sampling frequency f, > 2W.
Nyquist rate :
1. Using an anti aliasing filter :
The minimum sampling rate of "2 W" samples pet
In order to avoid alising, use a band limiting low pass
second for a signal x (t) having maximum frequency of
filter and pass the signal x (t) through it before sampling
"W" Hz is called as "Nyquist rate".
as shown in Fig. 5.7.2(a).
 • PCS (Sem. 4 I E&Tc I SPPU)
~J 5-15 Pulse Modulation
Nyquist interval : It is possible to use the practical low pass filter without
The reciprocal of Nyquist rate i.e. 1/2 W is called as the introducing any distortion due to the presence of the

Nyquist interval. guard bands between the adjacent frequency spectrums

l
ya
as shown in Fig. 5.7.3.

ha
Nyquist rate = 2 W Hz

G
That is why it is necessary to have f, > 2W.

a
ity
Nyquist interval = 1/2 W seconds
~}:;+- (5

Ad
~- 5.7,1 : CM~ider th~ signal {3 .. co~ ;(~00 sin
5. 7.3 Effect of Non Ideal Reconstruction iQ.OQ;:~};~JctO.;cos•.·1·200,m}. •.What•{S•tl'te,;l'l~~j~.;r,~~ror;.~S
Filter: ~ignat1· May 2000, 3 Marks

Ideal reconstruction filter : Soln.:

The highest frequency component in the given signal is,

We have to pass the sampled signal through a
reconstruction filter in order to obtain the original fm = 6000 Hz ... corresponding to the last term.

signal back from the sampled version. Nyquist rate = 2 fm = 12 kHz ...Ans.

As mentioned earlier the reconstruction filter is a low

•~*~if•t;;.?:t•lt*.(t)••==•·1Pcos· ZO{)et.~t;,;~~:~,tfi~,k.·~9,t}s
pass filter. It is expected to pass all the frequencies in th~rn1ntm~111$arrlpling l'~Ie.·? May 2000, 8 Marks
the range of (- W to + W) Hz. Soln.:
This is because the original signal x (t) is band limited to To calculate the minimum sampling rate:
"W"Hz.
X (t) = 10 COS 2000 nt • COS 8000 7tt
Therefore the frequency response of a reconstruction
= 5 cos ((2000 + 8000) nt J
filter should be as shown in Fig. 5.7.3.

Ii''
+ 5 cos [(8000 - 2000) nt]

?(.::• x (t) = 5 [cos 10000 nt + cos 6000 nt]

:·=::,

Thus the two frequency components present in the

:/:;_ .:
signal x (t) are f1 = 5000 Hz and f 2 = 3000 Hz.
-w w Frequency
Therefore the minimum sampling rate is given by:
{D-418) Fig. 5.7.3: Frequency response of an ideal low pass

filter used as a reconstruction filter f,(min) = 2X f1 = 2 X 5000 Hz

This is the frequency response of an ideal low pass filter. = 10 kHz ... Ans.

Practical reconstruction filter : •et 5.7.3 ; Ffricf Jhe N· .• •0{sfr~t~,~nd ~yqQi$,li~t~(\i~IJor the
However, it is not possible to practically realize an ideal
low pass filter.
,t~;1!f!~Jlh;o
Soln.:
1
7tt)ti~{(1000~;.•···.•······
Dec.O.··
2000, 4 Marks

Therefore a practical low pass filter with a frequency

x (t) = 10 cos (4000 nt) cos (1000 nt)
response as shown in Fig. 5.7.4 is used.
X(f)
= 5 [cos 5000 nt + cos (3000 nt)]
x (t) consists of two frequency components,
Frequency f 1 = 2500 Hz and f 2 = 1500 Hz.
Nyquist rate = 2 x f1 = 5 kHz ...Ans.

and Nyquist interval = 1/5 kHz = 0.2 msec ... Ans.

(-f5-W) -f5 (-f5 ~W) -Vy (15 ~ W) f5 Frequency
~--+---... :+Guard band ~x.. $(7;4£ t)etermine the• NyqUist rate in~Nyquislinterv~I
A'mplitude response of
reconstruction filter
forttiefollo~i~g signal:
(-f5 +W) -W o w (f.-W) Frequency x(t) = 5 cos (2000 t) + 7 sin (7000 t). May 01, 4 Marks
(D-419) Fig. S.7.4: Amplitude response of a practical Soln.:
reconstruction filter The expression for x(t) shows that there are two
frequency components in this signal.

*
~fi"
TechKnowled.g~)
Pub!icitlor.;,

l
~.
~,- PCS (Sem. 4 / E&Tc / SPPU)

2000
5-16 Pulse Modulation

f 1 = 2 n = 318.3 Hz and f 2 = 1114 Hz. Part I : Spectrum of sampled signal :

3
X(t) = COS (101t X 10 t) = COS ( 31415.92 t)
Nyquist rate = 2 f 2 = 2 x 1114 Hz = 2228 Hz ...Ans.

l
ya
Take the F.T. of both sides to get,

ha
Nyquist interval = 1 / (Nyquist rate) = 1/2228 Hz 3

G
X(f) = F [cos (2 1t x 5 x 10 t)]

a
ity
4
= 4.487 x 10- sec. = 0.4487 m sec ...Ans. 1

Ad
= [8 (f - fm) + 8 (f + fm)l
''S:itSttS: Aviay~:torm'[:tO,,t;40sjn(500 . t:t30~)Li~Jq.qe
2
sarrtpled<Periodically ~nci reproduced from • these sample
values. Find the rnaximllm allowable time interval between
=21 [8 (f - 5000) + 8 (f + 5000)]

sample values; How rnany sarnple values are needed to be

The spectrum of x(t) is as shown in Fig. P. 5.7.6(a).
stored in order to reproduce one second of this waveform ?
The spectrum of ideally sampled signal X0 (f) is given by,
May 02, 5 Marks, Dec. 07, 6 Marks
00 00

Soln.:
X5 (f) = f 5 L X (f- n f,) = 8000 L X (f- 8000 n)
1. The input waveform is x (t) = 10 + 10 sin n = -oo n = -oo
(500t + 30°).
X0 (f) = 8000 [... X (f + 8000) + X (f) + X (f - 8000)
The first term represents a de shift whereas the second
term is a sinewave of frequency, + X (f-16000) ... l
500 X5 is as shown in Fig. P. 5.7.6(a).
fm = h = 79.58 Hz.
As we pass the sampled signal through a LPF having a
2. Therefore the minimum sampling rate is given by:
passband from O to 4 kHz, the spectrum of LPF output is
f 5 (min) = 2 fm = 2 x 79.58 = 159.16 Hz.
as shown in Fig. P. 5.7.6(a).
3. The maximum allowable time interval between the X(f)
sample values is given by, Spectrum of
1 1 x(t)
T, (max) = =
159.16
f, (min)

T, (max) = 6.28 msec ... Ans.

4. The number of samples needed to be stored to
produce 1 sec. is given by,
1 sec
Number of samples = 6_ msec = 159.16 samples ... Ans.
28

Ait◊;it1fwave of· 1. Volt

Ex.$.7.'6 :' 5. and . kHz•
freqlieh~x.js
sampled at sampling frequency fs = 8 kHz. Sampling is ideal.
Output ofthe Sampler ate then passed through an ideal low
pass filter of O Hz to 4 kHt bandwidth. The output of LPF is
observe<:! or, CR9:•[)rc:1wthi~ qutput wayf:J shape pn graph
kHz
paper fo the scale comparing Jt with input wave shape.
lnterprete
.·., .I
the result : Also
, .. , .,_.
draw.the
. • •.
spectrum
• .:: • • •
at •...the. input,
outputof sampler.~nd.~tthe output·offilterfor·the given case
ofcbsine• 5 kHz wa"e on the !;Jraphpapet and to· the scale.• •

Dec. 05, 10 Marks

Soln.:
3
Given: X(t) = COS (2 1t X 5 X 10 t)
(a)
:. fm = 5 kHz, f, = 8 kHz.
Fig. P. 5.7.6(Contd ...)
Sampling - ideal, Filter passband 0-4 kHz.
~er
~~ PCS(Sem.4/E&Tc/SPPU) 5-17 Pulse Modulation

Soln.:
It has been given that x (t) ::: cos 200 rct + 0.25 cos 700
rct and f, = 400 Hz. The spectrum of the ideally sampled

l
ya
ha
signal is given by,

G
a
ity
Ad
(b) Output signal n=- oo n = -oo
(E-658) Fig. P. 5.7.6
= 400 X (f) + 400 X (f ± 400)
Part II : Waveform of the output signal :
+ 200 X (f ± 800) + ... ... (1)
The spectrum of output of LPF is as shown in The spectrum of x (t) is as shown in Fig. P. 5.7.?(a) which
Fig. P. 5.7.6(a). Mathematically it is expressed as, contains two frequency components f 1 = 100 Hz and
f 2 = 350 Hz.
XO (f) = 4000 [8 (f + 3000) + 8 (f - 3000)]
3 The second term in Equation (1) shows X (f) centered
Take the IFT to get x0 (t) = 4000 cos (2n x 3 x 10 t).
about ± f, = 400 Hz.
The output signal x0 (t) if observed on CRO, looks as
The spectrum shifted towards right (see Fig. P. 5.7.7(b))
shown in Fig. P. 5.7.6(b).
consists of four frequency components.
Ex. 5.7.7: The signal x(t) ::: cos 200 nt + 0.25 cos 700rct is The Spectrum of low pass filter frequency response is
sampled at the ra.te of 400 .samples per second. Sc1mpled shown in Fig. P. 5.7.7(c).

waveform is then passed through an ideal low pa$$ fUter with It allows only the frequency components of 50 Hz and

200 Hz bandwidth. Write an expression for filter>output 100 Hz to pass through to the output.

Sketch the frequency spectrum ofsarnpledwavefoim., Output = 400 cos (400 nt) + 200 (100 nt)

May 07, May 15, 8 Marks Note that cos (400 rct) is an undesired component.

(a) Spectrum of x(t)

Xs(f)

t ___,::::::i'---::::
+-'-~ :: :::____._:
-750 -500 -fs -350-300
-400
::::r__._::: :i___._:
:;: :___.:::::

-100-50
l
:: :: 1___._::::
0
2
400

:q~:i___._:::::;::_:
:~____._:
50 100
:: :: :i..___.:::::
300 350 fs
: r::_ :t: :_::
:: ::::___._::::

400
500
:r~::-~ f(Hz)
:: ::
750

(b) Spectrum of ><s(t)

I H(f) I
Response of an ideal low pass filter

-200 0 200

(c) Response of an ideal low pass filter

(E-660) Fig. P. 5.7.7

...,.~ TechKnowledge
~r' rut1itr.;3tlor:s
9~
~!'° PCS (Sem. 4 / E&Tc / SPPU) 5-18 Pulse Modulation

Ex. s. 7.8 : $Pacify th$ NyquJstrate and I\JyqJisfintervatfot Soln. :

each ofthefoUowingSignals:. ,• . ,> Given: Signal g (t) = 10 cos (40 nt) cos (400 nt)

l
1. g(t)= sine {2op t) 2. g(t) = sine~ <200 t)

ya
= s [cos (44o nt) + cos (36o TCt)J

ha
: ·: , , ·.. :. ·,· ·2 :. fmax = 220 Hz, fmin = 180 Hz
3. g(t) =sine (200.t) + sine (200 t)

G
Dec. 08, 6 Marks

a
ity
Soln.: 1. Nyquist rate :

Ad
1. g(t) = sine (200 t) :
sin (200 TC t) sin (2TC x 100 t) = 2 x 220 = 440 Hz ...Ans.
sine (200 t) =
200 TC t = 200 TC t
2. Cut off frequency of an ideal reconstruction filter :
:. f = 100 Hz
fc = 220 Hz ... Ans.
Nyquist rate = 2 x 100
3. Spectrum of the sampled signal :
= 200 Hz ... Ans. G(f)
1
Nyquist interval = = 5 mS ...Ans. J>:peotrum of g(!J
200
2
2. g(t) = sinc (200 t) :
-220 0 220
2
sin (200 n t)
sinc2 (200 t) =
200 TC t
2 1- cos 20
But sin 0 = L iciru~Mm~I •f
2
2 0
sin (200 TC t) 1- cos (2 x 200 TC t) -220 0 220 280 500 720
200 TC t = 2 X 200 TC t
(E-1342) Fig. P. 5.7.9: Spectrum of the sampled signal
1 COS (2TC X 200 t)
= 4. Lowest sampling rate if the signal is a bandpass
400 n t 400 TCt
signal:
f = 200 Hz
The bandwidth B of this signal is given by,
DC term f = 0 Hz
B = fmax-fmin = 220-180 = 40 Hz
Nyquist rate =2 x 200 Hz= 400 Hz ...Ans.
2 fM
1 fs (min) = -k
And Nyquist interval = = 2.5 ms ... Ans.
400
FM = fmax = 220 Hz
2
3. g(t) = sine (200 t) + sinc (200 t) :
fM 220
f2 = 200 Hz, f1 = 100 Hz
k = s = 40 = 55

Nyquist rate = 2 x f2 The value of k is 5

2 X 220
= 2 x 200 = 400 Hz ...Ans. f5 (min) = - -- = 88 Hz ... Ans.
5
Nyquist interval = 1
400 = 2.5 ms ... Ans. 5.8 Sampling Techniques :
SPPU: May 18, Dec. 18, Dec. 19
•• . . ' . . . .

sampfed>aHhe>tate of 500 samples/sec.... • Questions

University • - .-; .-:<?.:\//?}t·//:\tr:--'
1. •.• D~t~rrp1~;-~iNyqui~trate, )i} State sampling. theorem and dis9u&s.~ify~~f •••
2. Calculate:
•
the
.·.. •
<cut··off frequency·: or•·•
. ..
ideaf.·retottstrtrctiOT:I'
.. ·:·/>~
, :.,:;: , (May 18, 6 Mark$, Dec~ 1~i i:M~l'l<$) ·
filter. :&/a • $t;fe and explain· different ty~es of$~h,pHn9 and
3. Draw the spectrum of resulting sampled signal..· •• .• i<Jr~Wthe spectrum of sampledoutplltfof;sampling/
.. ,. • •
frequency less than, equal to an;cf gt,eate( than
4. If g(t) is coosidereq to be a band pas.~ ~i9i,aj, (,t~te1;r:p!Q,ij maximum frequency of analog sjgna/(fhr one any
the.lowest permissible sampling rate. Dec. 11, 8 Marks "type of sampling). JOec.18, 7' ~rks)
~s
~F PCS(Sem.4/E&Tc/SPPU) 5-19 Pulse Modulation

Sampling techniques can be divided into two categories 3. The sampled signal x6 (t) is expressed as,
namely: CX)

1. Ideal sampling 2. Practical sampling x6 (t) = I X (nT5) 8 (t - nT5)

l
ya
n = -cc

ha
5.8.1 Ideal Sampling : 4. Spectrum of the ideally sampled signal is given by,

G
a
ity
CX)
SPPU: Dec. 11, May 15, Dec. 17, May 18,

Ad
Dec.18,May19,Dec.19 X5 (f) = fs L X (f - fs) ... (5.8.1)
n=-cc
University Questions
5. The spectrum of the ideally sampled signal consists
Q, r Why ts ideal sampling not . us$ct fpf pr_~btical
of the spectrum of the original signal. i.e. X (f) and
appHcation ? Draw the. circuit fof flat fop sampling
its infinite number of replicas centered at the
method and explain with the wavef<:>rms:
frequencies , ± f5, ± 2f5, ± 3f5 ••• etc.
(Oec-,11, .tn1narf(s)
Q.2 Explain the types of sampling withw~yef'orrns. •• . Disadvantages of ideal sampling :
{M~yJ~, . .~ . MarksJ
1. The disadvantage of ideal sampling is that due to

• ;::iy1~-~?J:::~1;~~
~lscti~$ l~:
q, 4.· ·State·. sarnpHng >thebrem and: o/~$;
very narrow samples, the transmitted power is very
small and the S/N ratio is low. Thus the ideally
(May 18, 6 Marks, Delkl9, f Marks) sampled pulses may get lost in the background
Q. 5 •State and explain different type$ of sampHng and noise.
draw the spectrum of sampled 01;1tputfor·sampling 2. Ideal sampling is not possible to achieve
frequency less than, equal to and greater than practically, because it is practically impossible to
maximum frequency of analog signal (for any one
have pulses of widths approaching zero. Therefore
type of sampling): {Oecdt(7Marks)
practically natural or flat top sampling is used.
The sampling technique described in the previous Ideal sampling was used only to prove the
section, which uses the unit impulse train as a sampling sampling theorem.
function is called as ideal or instantaneous or impulse
sampling. 5.8.2 Practical Aspects of Sampling and
Signal Recovery :
Features of ideal sampling :

The waveform of an ideally sampled signal is as shown The practical sampling techniques are different from the
in Fig. 5.8.1. ideal sampling in the following ways :

1. In practical sampling methods, the duration of the

sampling pulses is finite and the amplitude of the
pulses is also finite.

2. Practical sampling methods use the practical low

pass filters for reconstruction. Guard band between
the adjacent spectrums in the sampled signal is
(D-428) Fig. 5.8.1 : An ideally sampled signal x6 (t) necessary to avoid distortions. Ideal filters are not
1. The sampling function s (t) is a train of impulses. used.
The duration of each impulse is extremely short 3. The signal x (t) which is to be sampled is not strictly
(can be approximated to zero). bandlimited. Due to this there are problems faced
while deciding the sampling rate f5.
2. The sampling function can be expressed
There are two popularly used practical sampling
mathematically as :
techniques. They are :
00

s (t) = I 8 (t - nT5) 1. Natural sampling or chopper sampling and

n=- oo
2. Flat top sampling.
 •
~ , PCS (Sem. 4 / E&Tc / SPPU) Pulse Modulation
5-20

5.9 Natural Sampling: The sampled signal is a train of pulses of width -c, whose
amplitudes are varying.
SPPU: May 15, Dec. 17, May 18,
Dec.18,May19,Dec.19 These pulses do not have flat tops but their tops follow

l
ya
ha
University Questions the waveform of the signal x (t).

G
a
Q.1 Explain the types of sampling with waveforms. The sampling rate is greater than or equal to the

ity
Ad
(May 15, 6 Marks) Nyquist rate.

d. 2 6e~~fifig<t9t~s\liampiihg With t'rierr merits ·and • Circuit arrangement for natural sampling :

demerits. (Dec.17, 6Marks, May 19, 7 Marks) Natural sampling is sometimes called as chopper
Q. 3 State sampling theorem and discuss its types. sampling because the waveform of the sampled signal
(May 18, 6 Marks,•oec. 19, 7 Marks) appears to be chopped off from the continuous time

Q. 4 State >and .explain different types of sampling .and signal x (t).

draw the spectrum· of sampled output for sampling The chopper arrangement is as shown in Fig. 5.9.2
fr~~~DfY.!~(;,~J~a,n_, . . . ~9.~8,l>W••••~9q 9f(~ater than where the chopper switch is being operated by the
· .•·.· ·•·•· n1~~ffillrif'.fr~q~.~niy 6f.·a;;~_igg·· siqriat{for ·any•· Orte' sampling function "c (t)".
typ~ of~arnpling)... ••• (Dec. 18, 7 Marks) c(t)JUl
!
Chopper switch
As explained earlier, the ideal sampling cannot be
implemented practically.
x(t)
1 Naturally
Waveforms: sampled signal

A more reasonable and practically feasible manner of

sampling is called as "Natural Sampling", as shown in (a) Principle of natural sampling
Fig. 5.9.1. MOSFET acting
as a switch
x(l)
Baseband ·····-·•·r
signal

x{t) Naturally
sampled signal
c(t) Jl.fl.
-----o--a············
1
c(t)
Sampling
signal (b) Generation of natural sampled signal
Switch
ON
x(t)
Naturally
~~- sampled -.. / Naturally sampled
~ signal • signal

(D-429) Fig. 5.9.1 : Process of natural sampling (c) Waveform of natural sampled signal
(D-430) Fig. 5.9.2
Looking at the waveforms in Fig. 5.9.1 we note the
Reconstruction :
following important points :
The reconstruction technique for natural sampling is
Here the sampling waveform c (t) consists of a train of
similar to that for the instantaneous sampling as shown
pulses each having a duration "-c" and separated by the
in Fig. 5.9.3.
sampling time T5•
A signal sampled at the Nyquist rate may be
The baseband or modulating signal x (t) and the
reconstructed exactly by passing the sampled signal
sampled signals (t) = c (t) x (t) are as shown in Fig. 5.9.1.
through an ideal low pass filter with cutoff at frequency
The sampled signal is obtained by multiplication of x (t) W, where W is the highest frequency component of the
and c (t). signal x (t).
 •
, , PCS (Sem. 4 / E&Tc / SPPU) 5-21 Pulse Modulation
Low pass
filter characteristics Substitute this value of en into Equation (5.9.2) to get,
Natural! 'Ide1t1T Original 00

sampleL. lo~ pas$ - ~~;~~? C (t) =

L
,A . [f ] j 21t fs n t
T sine n, e ... (5.9.4)

l
ya
signal •·•·• fdter )' x(t) n=-oo s
f

ha
---+----'~------
0 W

G
I
Substitute this expression for c (t) into Equation (5.9.1)

a
ity
Pass band'
to get,

Ad
(D-431) Fig. 5.9.3 : Reconstruction of original signal from
00
naturally sampled signal ,A " 2" r5n t
s (t) = T5
L- sine [fn ,] e
j
• x (t) ... (5.9.5)
n=-oo
5.9.1 Crosstalk in Naturally Sampled Signals:
This expression represents the naturally sampled signal
With the samples of finite duration, it is not possible to
in time domain.
completely eliminate the crosstalk generated in a
To plot the spectrum, it is necessary to take the fourier
channel which is bandlimited.
transform of the expression for s (t).
When N number of channels are to be time division
Therefore take the fourier transform of both sides of
multiplexed, the maximum sample duration is, = T/N.
Equation (5.9.5) to write,
- The width "," should be as large as possible to increase 00
,A "
the signal power but increase in "," will increase the S (f) = T L- sine (fn ,) FT {e
j 2n f5n t
. x (t)} ... (5.9.6)
5
n=-oo
possibility of crosstalk.
Use the frequency shifting property of fourier transform,
To reduce crosstalk, "," should be as short as possible.
which states that :
Thus the width "," is a compromise between two
F
contradicting requirements. ej 2n fs nt. X (t)~ X (f - fsn) ... (5.9.7)

,A oo
5.9.2 Spectrum of Naturally Sampled Signal: Therefore, S (f) = T5 I sine (fn ,) X (f- nf5) ... (5.9.8)
n=-oo
- The naturally sampled signal s (t) of Fig. 5.9.1 is
But as fn = nf5, the spectrum of a naturally sampled
obtained by multiplication of the signal x (t) and
signal is expressed as follows,
sampling function c (t). 00

S (t) = X (t) C (t) ... (5.9.1) S (f) = ~A L sine (nf5,) X (f - nf5) ... (5.9.9)
5 n=-oo
However c (t) can be expressed in the form of a
Conclusions from Equation (5.9.9) :
complex fourier series as follows :
1. The term X (f - nf5) represents the shifted version
00

C (t) = L Cnej2nf5nt ... (5.9.2) of the frequency spectrum X (f). The spectrum S (f)
n=-oo consists of X (f) and its shifted replicas as shown in
- Because c (t) is a train of rectangular pulses : We can Fig. 5.9.4(b).
obtain the value of en as : 2. These shifted replicas are observed at frequencies
TA f = ± f 5, ± 2 f 5, ± 3 f 5, ... etc.
Cn = -=rsinc [fn
0
n
3. The spectrum of x (t) is periodic in f 5 and weighted
where T = Pulse width = , in this case ,A
by the sine function. (See the term T sine (nf5,) in
s
fn = Harmonic frequency of f 5 i.e. fn = nf5
Equation (5.9.9)). Therefore the amplitude of the
n
or fn = r= nfo and = To= Ts spectrum of naturally sampled signal reduces on
0

,A both sides of Y axis as shown in Fig. 5.9.4.

.. Cn = Tsinc [fn,l ... (5.9.3)
s
~~ TecltKnomiedge
"Ii"' Put,;; r at Io 11 ,;
 ·~
~F PCS (Sem. 4 I E&Tc I SPPU) 5-22 Pulse Modulation

l
ya
ha
G
-w

a
O W

ity
(a) Spectrum of continuous time signal x(t)

Ad
zs---~-,-fil----Z\~~-:zs~'.
• • • • • • . • --"?i~f~-. . .. · . •

-2fs -fs -W O W fs 2fs

(b) Spectrum of naturally sampled signal

(D-432) Fig. 5.9.4

Merits and demerits of natural sampling : Waveforms:

1. Generation is easy. The natural sampling is rarely employed in practice.
2. We can use practical low pass filter for Instead the other practical sampling technique called
reconstruction. flat top sampling is employed in practice.
3. The amplitudes of high frequency components
In the flat top sampling technique, the analog signal
decrease therefore some distortion is introduced.
1
4. Increased SNR due to finite pulse width of the x (t) is sampled instantaneously at the rate f 5 = and Ts
sampling function and that of the sampled signal.
the duration of each sample is lengthened to a duration
5. For large values of "," there is a possibility of "," as shown in Fig. 5.10.l(b).
crosstalk.
Thus the amplitudes of these pulses are constant and
5.10 Flat Top Sampling: equal to the corresponding sampled values.

SPPU: May 15, Dec. 15, May 16, Circuit diagram :

The flat top pulses can be obtained by using the sample

and hold circuit shown in Fig. 5.10.l(a).
Q/1 •.•· t;xplain the:types ofsampljng ~ith. W~~~forms .. Sampling
<cL •> •• ••• • • .; (N,i~t~~;;~::Fvl,rk;)
x(t)
switch

Q;J praw an<i\~~tairl

• •• • •••••••••
ct circui{tdr rialtop sarylpling: . .
•••• • (Dec. 15, iMarl<s)
s(t}
Sampled
signal

Q. 3 Give.. thec;ircuit for. flattop .s~mpUni;:i. .•Explain ·its

Discharge
wor~ing. (May 16, 6 Marks) swttch
Q.4 What ls meant by "Aperture Effect" ?Howcan it b.e
reduced?· (May17,May19,6 Marks)
(a) Sample and hold circuit to obtain the flat
Describe types •of sampling with their merits and
topped samples
demerits: (bec.11,•sivfar1<s,1'v1aY19, 7Marks)
x(nT5)
State sarnpHng theorem and discuss its types.
' , ..,.,..--x(t)
(May 18, 6.M~rks,.Dec.19,7.Matks) ...
Q. 7 .·.· State. and explafn different types of. sampling and
draw the spectrum of sampled output for sampling
frequency less than, equal tb and greater than
maximum frequency ofanalog signal (for any one
(b) Flat top sampled signal
type ofsarnpting). (Oec.18, 7 Marks)
(D-433) Fig. 5.10.1

~~ TethKnowlellge
~1'8" Put1:lr.3tlons
 •
, , PCS (Sem. 4 / E&Tc / SPPU) 5-23 Pulse Modulation
h(t)
Operation of the sample and hold circuit :
The sample and hold circuit consists of two FET switches
and a capacitor as shown in Fig. 5.10.l(a).

l
ya
ha
The analog signal x (t) is applied at the input of this

G
a
circuit and the sampled signal s (t) is obtained across (D-435) Fig. 5.10.3 : Principle of generating the flat top

ity
sampled pulses

Ad
the capacitor.
A gate pulse will be applied to gate G1 at the instant of From Fig. 5.10.3 we can write that,

sampling for a very short time. h (t) = 8 (t) * h (t) ... (5.10.2)

The sampling switch will turn on and the capacitor 5.10.1 Actual Generation of Flat Top Pulses:
charges through it to the sample value x (nT5).
In Equation (5.10.2), on RHS if we replace 8 (t) by xo (t)
The sampling switch is then turned off. Both the FETs i.e. the ideally sampled signal then we get the flat top
will remain OFF for duration of "-c" seconds and the sampled signal s (t).
capacitor will hold the voltage across it constant for this Flat top sampled signal, s (t) = x6 (t) * h (t)
period.
This equation can be graphically explained as shown in
Thus the pulse is stretched to "'t" seconds. Fig. 5.10.4.
At the end of the pulse interval ('t), a pulse is applied to
G2 i.e. gate terminal of discharge FET.

This will turn on the discharge FET and short circuit the
capacitor.

The output voltage then reduces to zero. This is as

shown in Fig. 5.10.2. Ideally sampled
signal ,,

Flat top sampled signal

0 Ts 2Ts------
(b) Ideally sampled signal x;;(t) • ................,........ .
__J t___
Sampling Discharge
switch ON switch ON
(D-434) Fig. 5.10.2 : Operation of sample and hold circuit
Principle of generating the flat top sampled pulses :

From Fig. 5.10.2 it is clear that only the rising edge of

each pulse represents the instantaneous value of the
analog signal x (t).

Therefore the flat top sampled pulses can be obtained

by "convolution" of the instantaneous sample and a
pulse h (t) of duration 't.

(D-436) Fig. 5.10.4 Generation of flat topped sampled signal

This is true because convolution of any function with
Thus the flat top sampled pulses are obtained by
the delta function results in the same function.
convolution of the ideally sampled signal x6 (t) and a
i.e. X (t) * 0 (t) = X (t) ... (5.10.1) pulse train of finite pulse width h (t). The width of each
This is the replication property of the delta function. pulse is 't sec.

This property is being used for the generation of flat 5.10.2 Spectrum of Flat Top Sampled Signal':
top sampled pulses.
The flat top sampled signal is given by,
This principle is graphically explained in Fig. 5.10.3.
s (t) =x6 (t) * h (t) ... (5.10.3)

~~ TechKnowletl.ge
lltl!t"' Pu ti! ii: ;3 t inns
~~
~r PCS(Sem.4/E&Tc/SPPU) 5-24 Pulse Modulation

00
Therefore the width of s (t) is decided by h (t) and the
s (t) = L x (nTs) h (t - nTs) ... (5.10.8)
amplitude of s (t) depends on x8 (t).
n=- oo
The ideally sampled signal x8 (t) is expressed

l
ya
This expression represents the flat top sampled signal in

ha
mathematically as,
time domain, in terms of the instantaneous sample

G
a
00
values x (nTs) and train of fixed duration pulses.

ity
Ad
xii (t) = L x (n Ts) 8 (t - n Ts) ... (5.10.4)
n=- oo To obtain the spectrum of s (t), let us use the
convolution theorem which states that, convolution in
Now s (t) =x8 (t) * h (t)
time domain is transformed into multiplication of
Using the definition of convolution,
transforms in frequency domain.
00

x8 (t) * h (t) = f x8 (v) h (t - v) dv ... (5.10.5)

s (t) = x 8 (t) * h (t). ... (5.10.9)
-00
Taking Fourier transform of both the sides we get,
Note that"," has not been used deliberately instead "v" S (f) = X 8 (f) H (f) ... (5.10.10)
is being used.
From Equation (5.10.1),
Substitute Equation (5.10.4) into Equation (5.10.5) to 00

get, X8 (f) = fs L X (f - nfs) ... (5.10.11)

n=- oo
00 00

s (t) = L x (nTs) 8 (v - nTs) h (t - v) dv Therefore Equation (5.10.10) becomes,

-oo n=-oo 00

Interchanging the order of summation and integration S (f) = fs L X (f - nfs)- H (f) ... (5.10.12)
n=- oo
and rearranging we get,
This is the expression for the spectrum of a flat top
00 00

s (t) = L x (nTs) f 8 (v- nTs) h (t - v) dv ... (5.10.6) sampled signal.

n = -oo -oo
H (f) is the spectrum of h (t). As h (t) is a rectangular
Now let us use the shifting property of delta function as, pulse, its spectrum is a sine function.
00
Therefore product of the spectrums X8 (f) and H (f) is
f f (t) . 8 (t - td) dt = f (td) ... (5.10.7)
- 00 shown in Fig. 5.10.5, because S (f) is equal to the
product of Xii (f) and H (f).
Apply this property to Equation (5.10.6) to write,

Spectrum of
analog signal
x(t)

i;:~~;~ui~;! +•------1-l-_(.....
l.-+W-)--.---(-,--f--,--WJ.l,-_.....
5
IW_X__(f"Jl---~l__._1-_-W.,...)-;--(--,-f_+__.W).,.,..,__ _ _ _-+• 1
5

- ~ "'"t ----- ~. · ~
S1~~1i~~t +•----1.,...,/t--f~c;-,.!-+--:---<•---,-f,- -----+:--+.--':1":"h=---+• 1
_ _,.:....,...~•--.,,.,Jl---~-,.-,-W,.....,.-•-----,f,-
5 5
S(fl

Spectrum of flat
topped sampled
signal
0 W (f5 -W) 1/t

(D-437) Fig. 5.10.5 : Spectrum of a flat top sampled signal

~~ TethKnowiedge
lllsJ!t'8" r,.1tiiir.Jtlon~;
 •
~ , PCS (Sem. 4 / E&Tc I SPPU) 5-25 Pulse Modulation

Observations from the Fig. 5.10.5 : The aperture effect can be corrected in reconstruction

1. The signal x (t) has a flat spectrum over its entire by including an equalizer.

l
range from O to W. The transform of the

ya
5.10.4 Reconstruction of Original Signal x (t):

ha
instantaneous signal X8 (f) has been drawn below it.

G
a
The sampling frequency f 5 = l/T5 is large enough Due to the aperture effect discussed earlier, an

ity
Ad
to allow the guard band. amplitude distortion as well as a delay is introduced in
2. The spectrum of the sampling signal h (t) is a sine the flat top sampled signal.
function. This distortion can be corrected by connecting an
3. The spectrum of the flat topped signal is the equalizer after the reconstruction filter (low pass filter)
product of these two spectrum. Due to the as shown in Fig. 5.10.7.
multiplication with the sine function, this spectrum Flattop Analog signal
/Rieconstructio~ Equalizer, .. .i
sampled x(t)
goes to zero at f =1/-r. Filter
signal

5.10.3 Aperture Effect : SPPU: May 17, May 19 (D-439) Fig. 5.10.7 : Reconstruction of x (t)
•• ,; • ••:· ···< ••• ··:··,,. ••• ·: .......... ·,:· •• ;. ':":. :..-:, -<:· ::.>-:•:,::-,::::::?,-:: .. /:C.<::·-.·_:::.
University Questions 5.10.5 Equalizer :
wb~(,iliti~~rif~y ,}perture>~~:~t· ~ :~.t~nirbe The amplitude response of the equalizer is such that the
• reduced? (May 17, May 19, 6 Marks)
equivalent transfer function is 1.
Consider the spectrum of the flat topped signal. We are
interested in the portion of the spectrum upto I H (f) I X I He (f) I =1
frequency W. where He (f) = Transfer function of the equalizer
The spectrum should have been flat in this portion of and H (f) = Transfer function of the
the spectrum but it is not as shown in Fig. 5.10.5.
reconstruction filter
The shaded portion shows an error due to an effect 1 1
called "aperture effect". = I H (f) I = T sine (f n
The high frequency roll off characteristics of a typical 1 nfT
H (f) acts like a low pass filter and attenuates the upper T sin (n ft)
portion (high frequency) of the message signal
5.10.6 Merits and Demerits :
spectrum.
This loss of high frequency content is called as the 1. Better SNR due to increased signal power. This is
"aperture effect". due to the finite width ",;" of the pulses.
The aperture effect is due to the finite pulse width ",;" of
2. Generation is easy.
the sampling signal. With increase in the width 1, the
frequency 1/,; will reduce and the error will increase. 3. Practical filters can be used for reconstruction.

The aperture effect can be reduced by reducing the 4. Aperture effect introduces distortion.
pulse width ,; as shown in Fig. 5.10.6.
Aperture etTor
5.10.7 Comparison of Sampling Techniques:
Spectrum i--=
of flattop • , large SPPU : Dec. 14, May 15
sampled
signal _ ,4-_ _---1,,~_ _ _ _ _ __.._~:,.,.e.,~ - - - - - + University Questions
0 11
Spectrum r---=, Reduction in aperture effect due to a. 1 Distil'lguish between ideal sampling; natural
of fiat top
,-----_: reduction in the
sampled
pulse width "t"
sampling and flattop sampling, (Pec.14,' &;Marks)
signal -:s-01----~....__ _ _ _ _........,.....__ _ _-=-,:c.....+
Q, 2 Explain the types ofsamplingWitfr';\/av~fo;oi{
(D-438) Fig. 5.10.6 : Effect of pulse width ",;" on the >(May 15, 6 Marks)
aperture effect

'!!!!I~ Te&l1Knowlellge
~t"" Publi[ i3t ions
 •
~ ; PCS (Sem. 4 / E&Tc / SPPU) 5-26 Pulse Modulation

Table 5.10.1 : Comparison of sampling techniques

.
._i ...•. •.• .•.··· '..
Sr;
No.
· .·• / :>'Flattop Slim piing
... ::

l
ya
.·•· • ··,. ·.

ha
1. Nature of the sampling Train of impulses Train of finite duration Train of finite duration

G
a
function pulses pulses

ity
Ad
2. Circuit arrangement Uses a multiplier Uses a chopper Uses a sample and hold
circuit
3. Practical realizability Practically not Practically realizable Practically realizable
realizable
4. Waveforms Refer Fig. A Refer Fig. B Refer Fig. C
5. Sampling rate Tends to infinity Satisfies Nyquist criteria Satisfies Nyquist criteria
6. Mathematical representation 00
-r:A oo 00

in time domain s(t)=Ts L s (t) = L

n=-oo n=-oo
. i21tnf t
x (t) sine (nf5 -r:) e s

7. Frequency spectrum
f.
00 00 00

s (f) = S (f) = f5 L
n=-oo n=-oo
X (f - nfs)
8. Signal power Very low due to the use Increases with increase in Increases with increase
of impulses the pulse width -r: in pulse width 1
9. Bandwidth requirement Very high Increases with the Increases with reduction
reduction in pulse width in pulse width
10. Effect of noise Maximum Moderate Moderate

1. Pulse Amplitude Modulation (PAM) system.

;
.,. .,. 2. Pulse Width Modulation (PWM) system .
3. Pulse Position Modulatin (PPM) system.
4. Pulse Code Modulation (PCM), Delta Modulation
(DM), Adaptive Delta Modulation (ADM) systems.
Fig.A 5. Time Division Multiplexing (TDM)

5.12 Pulse Amplitude Modulation (PAM) :

SPPU: May 05, May 07, Dec. 15
University Questions

(D-440) Fig. 8 Fig. C

5.11 Applications of Sampling Theorem : lain with neath

ration ofPArvl'.
·.: .. •.•••• ·.· ,.·.. :.,,,.,::•·:.·
sO < <•i··•·
The sampling theorem is extremely important in signal • i, s. •.• y.01, ~~rks)
analysis, processing and transmission. . f;xplalq.h~w . . ·a • PAM slgp~ 9(9.net~ted, ?• m~v-~~
Hpwcan ltbedemotjulated,?·· /{P~(!; 15,7Matks)··•
It allows us to replace a continuous time signal by a
discrete time signal (sequence of samples). Principle:

This is useful in digital filters. Some other applications of In the PAM system, the amplitude of the pulsed carrier
is changed in proportion with the instantaneous
sampling theorem are :
amplitude of the modulating signal x(t).
 •
~ , PCS (Sem. 4 / E&Tc I SPPU) 5-27 Pulse Modulation

So the information is contained in the amplitude Thus the Nyquist criteria is satisfied.
variation of PAM signal.
The rectangular narrow carrier pulses generated by the
The carrier is in the form of train of narrow pulses as pulse train generator would carry out the uniform

l
ya
shown in Fig. 5.12.1.

ha
"sampling" in the multiplier block, to generate the PAM

G
If you compare the PAM system with the sampling signal as shown in Fig. 5.12.2.

a
ity
process, you will find that these two processes are Modulating signal

Ad
identical.
The PAM signal is then sent by either wire or cable or it
is used to modulate a carrier.

Types of PAM :
: Pulsed carrier
There are two types of PAM: Etr··---□---··n··---n··--·n··---n···--n··--- □
~ Time
1. Natural PAM 2. Flat top PAM 0 :+-Ts...\
Double polarity PAM signal
5.12.1 Generation of Natural PAM : + __ ••• •,.··---.,~.✓

SPPU : May 07, Dec. 15, Dec. 19 o ii-a'·L..;.......1a-_L1-_·;.,·_ _,...__...,,..._...,,,._,__ __,,._..., Time
University Questions
••.. ~I:., ~~r,:;:c~di:gra: and .,laVJoJs
generation ~f PAM: $fate 1fs··•appHcat1ons. (L-182) Fig. 5.12.2 : Waveform of natural PAM
• (May 07, 8 Marks)
The samples are placed T5 seconds away from each
•• ~xplain H6' ~ f?~t\11.signaFmay bl9igenerated ?• other.
How pan ifb¢ ~,Mpdu!atect? •· (Dec; 15, TMarks)
Q.3 . .Praw ~f,lg [~ipf~fo,: gerieration ~~ddytecti~n 6{ The "information" in the modulating signal is contained
PAM... • (De~. 19, 7 Marl<~). in the "Amplitude variations" of the pulsed carrier.

Block diagram : Therefore this system is similar to the AM system

discussed earlier.
Refer Fig. 5.12.1 to understand the generation of natural
PAM. Circuit arrangement for natural PAM :

~
Continuous
~ ~ Natural PAM is sometimes called as chopper sampled
PAM because the waveform of the sampled signal
modulating appears to be chopped off from the continuous time
signal x(t)
signal x(t).

The chopper arrangement is as shown in Fig. 5.12.3

where the chopper switch is being operated by the
pulsed carrier "c (t)".
(L-181) Fig. 5.12.1 : Generation of natural PAM MOSFET
c(t} Jµ1. Carrier acting as a
switch
The continuous modulating signal x(t), is passed
through a low pass filter. The LPF will bandlimit this 0 ----··r··-
signal to fm.
Chopper f
Natural
switch Natural
PAM RL PAM
That means all the frequency components higher than signal signal
the frequency fm are removed.
._________. ____ _J ·-
Bandlimiting is necessary to avoid the "aliasing" effect in (a) Natural PAM (b) Natural PAM with a MOSFET
the sampling process. acting as a switch

The pulse train generator generates a pulse train at a Fig. 5.12.3(Contd ...)
frequency f 5, such that f 5 ~ 2fm.
 • PCS(Sem.4/E&Tc/SPPU) 5-28 Pulse Modulation

Conclusions :
Switch Natural PAM
ON signal 1. The term X (f - n fs) represents the shifted version

l
ya
of the frequency spectrum X (f). The spectrum S (f)

ha
consists of X (f) and its shifted replicas as shown in

G
0:

a
ity
-+:' 't'- Fig. 5.12.4(b).
(c) Natural PAM signal

Ad
(L-183) Fig. 5.12.3 2. These shifted replicas are observed at frequencies f

The natural PAM signal is same as the naturally sampled = ± fs, ± 2 fs, ± 3 fs ... etc.
signal. 3. The spectrum of x (t) is periodic in f5 and weighted
-r:A
Hence the spectrum of natural PAM signal is same as by the sine function. [See the term Ts sine (n f 5 ,) in
that of the naturally sampled signal derived. It is given
by, Equation (5.12.1)].

Spectrum of natural PAM signal, Therefore the amplitude of the spectrum of natural
00 PAM signal reduces on both sides of Y axis as shown in
-r:A
s (f) = T ~ sine (n f 5-c) X (f- n f 5) ... (5.12.1)
Fig. 5.12.4.
s n=-oo

• • • • • -w • 0 • W • • • , ••
• • • . . • . (a) Spectrum of continuous time signal x(t) • • ·• • • • •

·- ~ - - - - ; f \f-_~-:_- - !--c7:l"f•2-1-H 'f

(b) Spectrum of natural PAM signal

(D-447) Fig. 5.12.4

5.12.2 Detection of Natural PAM: Demodulated PAM

--- ~gn~
SPPU: Dec. 15, Dec. 19
PAM detector
University Questions
·Ct 1 ExpJ~in no~r a PAfvl.•signal.may . . b~ generated ·? PAM signal at the detector input

H6~..tanitbe ~tnoclulated ? (D~c; 1§, 7Marks)

Dr;yl arid ~~Bl~!h generation afld de'tecttOn of
PAM". • (Dec. 19, 7 Marks)

The PAM signal can be detected (demodulated) by Time

passing it through a low pass filter. ulated output

The low pass filter cutoff frequency is adjusted to fm so
that all the high frequency ripple is removed and the
original modulating signal is recovered back.
Time
The PAM detection and the corresponding waveforms
are as shown in Fig. 5.12.5. (L-185) Fig. 5.12.5 : Detection of PAM and waveforms
,,.P° PCS (Sem. 4 / E&Tc / SPPU) 5-29 Pulse Modulation

From the waveforms, it is seen that the demodulated This is instantaneous value of x (t) at instant t = nT5
output signal is close to the original modulating signal where n = 0, 1, 2.... The sampling switch is then turned

l
(t). off.

ya
X

ha
Both the FETs will remain OFF for a duration of "1:"

G
5.12.3 Flat Top PAM:

a
seconds and the capacitor will hold the voltage across it

ity
Ad
Circuit diagram : constant for this period.
The flat top PAM can be obtained by using the sample Thus the pulse is stretched to ",:" seconds and we get a
and hold circuit shown in Fig. 5.12.6(a). pulse with a flat top.

The natural PAM is rarely employed in practice. Instead At the end of the pulse interval (1:), a pulse is applied to
the flat top PAM is employed in practice. G2 i.e. gate terminal of discharge FET.

In the flat top PAM technique, the analog signal x(t) is This will turn on the discharge FET and short circuit the
capacitor.
sampled instantaneously at the rate f 5 = ~ and the
s The output voltage then reduces to zero. This is as
duration of each sample is lengthened to a duration "1:" shown in Fig. 5.12.7.
as shown in Fig. 5.12.6(b).
Sampling
switch
x(t) s(t)
Flat top
PAM signal 'tr-
____J t__
C Sampling Discharge
switch ON switch ON
Discharge
switch (D-450) Fig. 5.12.7: Operation of sample and hold circuit

Principle of generating the flat top PAM pulses :

(a) Sample and hold circuit to obtain the flat top PAM Flat top PAM is same as flat top sampled signal
(D-449) Fig. 5.12.6 discussed in section 5.10.

•• •. ___-x(t) So the methods of generation, detection, spectrum,

• _..--Flattop aperture effect etc will be exactly the same.
PAM signal
5.12.4 Spectrum of Flat Top PAM Signal:
The spectrum of flat top PAM signal is same as that of
(b) Flat top PAM signal the flat top sampled signal. It is given by:
(D-449) Fig. 5.12.6
Spectrum of flat top PAM signal
Thus the amplitudes of these pulses are constant and co
equal to the corresponding sampled values. S (f) = f5 r X (f - n f5) • H (f) ... (5.12.2)
Operation of the sample and hold circuit: n = -co

The sample and hold circuit consists of two FET switches 5.12.5 Reconstruction of Original Signal x (t):
and a capacitor as shown in Fig. 5.12.6(a). Due to the aperture effect discussed earlier, an

The analog signal x (t) is applied at the input of this amplitude distortion as well as a delay is introduced in
circuit and the flat topped PAM signal s (t) is obtained the flat top sampled signal.

across the capacitor. This distortion can be corrected by connecting an

A gate pulse will be applied to gate G1 at the instant of equalizer after the reconstruction filter (low pass filter)

sampling, for a very short time. as shown in Fig. 5.12.8.

Flat top
•Reconstruction
The sampling switch will turn on and the capacitor PAM
filter
signal
charges through it to the sample value x (nT5).
(D-451) Fig. 5.12.8 : Reconstruction of x (t)
 •
~ , PCS (Sem. 4 / E&Tc / SPPU) 5-30 Pulse Modulation

5.12.6 Transmission Bandwidth of PAM When PAM signal travels over a communication
Signal: channel, noise gets added to it as shown in Fig. 5.12.9.
1\/oise superimposed

l
/.. on PAM signal

ya
Let us assume that ",:" is the width cf each pulse in a flat

ha
top sampled PAM and "T;' is the duration between

G
a
adjacent samples.

ity
Ad
We assume that the pulse duration "," is very small as
compared to Ts.
0
... (5.12.3) (L-190) Fig. 5.12.9 : Effect of noise on PAM signal
1 Note that the noise distorts the amplitude of PAM
But Ts = fs where fs = Sampling frequency
signal.
If W is the maximum frequency in x (t), then fs 2 2W. Since the information is contained in the amplitude, the
1 noise will contaminate the information.
Hence Ts ::; W . Substituting this in Equation (5.12.3)
2
Therefore the noise performance of PAM system is very
we get,
poor.
1
,: << ... (5.12.4) The PWM and PPM systems have a better noise
2W
performance.
For adequate pulse resolution i.e. to transmit and
receive this PAM signal without much signal distortion, 5.12.8 Merits and Demerits of Flat Top PAM:
the transmission bandwidth Br needs to satisfy the
1. Better SNR due to increased signal power. This is
follov1:ing equation, due to the finite width "," of the pulses.
1
Br -> 2, >> w ... (5.12.5) 2. Generation is easy .
3. Practical filters can be used for reconstruction.
Thus the transmission bandwidth is inversely
4. Aperture effect introduces distortion.
proportional to the pulse width "," of the PAM pulses.
Transmission bandwidth should be as small as possible. 5.12.9 Advantages and Disadvantages of
PAM:
It can be reduced by increasing the pulse width ",". But
this will increase the "aperture error" as discussed There are not many advantages of a PAM system except
earlier. for the simplicity of generation and detection. But there
The transmission bandwidth Br is much larger than the are many disadvantages.
maximum frequency content in x (t). They are as follows :
i.e. B1 >> W 1. The amplitude of PAM signal changes according to
Due to changes in amplitudes of PAM pulses, the the amplitude of modulating signal. Therefore like
transmitted power does not remain constant. AM, the effect of additive noise is maximum in

With increase in the pulse width "," the aperture error PAM. The added noise cannot be removed easily.

increases. 2. The transmission bandwidth required for a PAM

signal is too large as compared to the maximum
5.12. 7 Effect of Noise on PAM :
frequency content in x (t).
The amplitude of the pulsed carrier is being changed in 3. Due to the changes in amplitudes of PAM pulses,
proportion with the amplitude of modulating signal in the transmitted power is not constant.
PAM.
5.12.10 Applications of PAM:
Hence all the "information" is contained in the
1. PAM is used in the PAM-TDM system.
amplitude variation of the PAM signal.
2. PAM is used in the PAM-FM system.
 •
~ , PCS (Sem. 4 / E&Tc / SPPU) 5-31 Pulse Modulation

5.13 Pulse Width Modulation (PWM): 5.13.1 Generation of PWM Signal:

PPU: May 15, Dec. 15

l
ya
University Questions University Questions

ha
• :,,: : ··_. :: .. :,.:-:-· ' , ·.·

G
Q. 1 With the help of neat diagram, explainP\NM.> Q. 1 With the help of a diagrarn explal!) j::iplse ~i:9m

a
ity
modulation. Also state . its /~1j~f~~l~ge$, •

Ad
.(May 1S; 7 Marks)
,gisadyantages.~nd appli . ,~},
>Q:c?i .Describe- with: the help> of . neaf 'sketc:hes 'of
Q. 2 Describe with the help, of. ne~t ;~15etcp:~_ of
waveforms ,methods of generation of PDM/PWM
and PPM. (Dec. 15, 7 IVlarks)
waveforms methods of gen~r~tiol'l ,Q[~e>,r,~11tV~'.M
and PPM. ••• (Qep.1,~JfM~~:ks)
Principle:
Q. 3 With help of waveforms explafn<fl9~)~M and
The other type of a pulse analog modulation is the PPM can be generated? (May1$;'i~J~,rks}
Pulse Width Modulation (PWM).
Q. 4 Discuss PWM generation anc:Ldet,~q~i •. iL
In PWM, the width of the carrier pulses varies in
proportion with the amplitude of modulating signal. Q:5 .C>taw ,and explain generatibrt;a~~-~~~atf~;~:~f
The waveforms of PWM are as shown in Fig. 5.13.1. PWM and PPM. (Oec.18';7'£1,1arksL
s(t)
Carrier sampling pulses Block diagram :
Ol-~~-'---'--l--'-~..i...J'-J.--1,-.....,_-'.....___._.__T
➔ime
The block diagram of Fig. 5.13.2(a) can be used for the
x(t) Modulating signal
generation of PWM as well as PPM.
Comparator
Modulatingo-------.i
signal x(t) > - - - - - - P W M output

PWM signal
Monostabi~ PPM output
0 1-f-L---!'--........-"--'--'--'-'---........- .........___.........__.__ __
Time
(D-455) Fig. 5.13.2(a) : PWM and PPM gene.rator
(D-454) Fig. 5.13.1 : PWM signal [Trail edge modulated signal]
Operation:
As seen from the waveforms, the amplitude and the A sawtooth generates a sawtooth signal of frequency f 9
frequency of the PWM wave remains constant. Only the therefore the sawtooth signal in this case is a sampling
width changes.
signal.
That is why the "information" is contained in the width It is applied to the inverting terminal of a comparator.
variation. This is similar to FM.
The modulating signal x (t) is applied to the non-
As the noise is normally "additive" noise, it changes the inverting terminal of the same comparator.
amplitude of the PWM signal.
The comparator output will remain high as long as the
At the receiver, it is possible to remove these unwanted instantaneous amplitude of x (t) is higher than that of
ampiitude variations very easily by means of a limiter
the ramp signal.
circuit.
This gives rise to a PWM signal at the comparator
As the information is contained in the width variation, it output as shown in Fig. 5.13.2(b).
is unaffected by the amplitude variations introduced by
Waveforms:
the noise.

Thus the PWM system is more immune to noise than The waveforms for the PWM generator are as shown in

the PAM signal. Fig. 5.13.2(b)

 • PCS (Sem. 4 / E&Tc / SPPU)
~J 5-32 Pulse Modulation
Sawtooth
x(t) waveform It produces a train of constant amplitude, constant
width pulses.
Time
These pulses are synchronized to the leading edges of

l
ya
ha
PWM output the regenerated PWM pulses but delayed by a fixed

G
(comparator output)

a
interval.

ity
Time

Ad
PPM The regenerated PWM pulses are also applied to a ramp
output
Time generator.
(D-456) Fig. 5.13.2(b) : Waveforms At the output of it we get a constant slope ramp for the
Note that the leading edges of the PWM waveform duration of the pulse.
coincide with the falling edges of the ramp signal. The height of the ramp is thus proportional to the
Thus the leading edges of PWM signal are always widths of the PWM pulses.
generated at fixed time instants. At the end of the pulse a sample and hold amplifier
However the occurrence of its trailing edges will be retains the final ramp voltage until it is reset at the end
dependent on the instantaneous amplitude of x (t). of the pulse.
Therefore this PWM signal is said to be trail edge
The constant amplitude pulses at the output of
modulated PWM.
reference pulse generator are then added to the ramp
5.13.2 Detection of PWM Signal : signal.
SPPU : May 18, Dec. 18 The output of the adder is then clipped off at a
University Questions threshold level to generate a PAM signal at the output
Q. 1 Discuss PWM generation and detection in detaiL of the clipper.
(May 18,7 Marks)
A low pass filter is used to recover the original
Q, 2 Draw and·explain•generation and regeneration .of
PWM and PPM. (Dec; 18, 7 Marks) modulating signal back from the PAM signal.
Block diagram : The waveforms for this circuit are as shown in
The block diagram of PWM detector is as shown in Fig. 5.13.4.
Fig. 5.13.3. ®
PWM Input to the
Operation:
NoiseI-__,__ __.._:;___ _....__ detector I
_,.,....,......_'""""..._...,__,___.,.....
The PWM signal received at the input of the detection 0

circuit is contaminated with noise. @ i Flegi~e~ated

This signal is applied to pulse generator circuit which

0
regenerates the PWM signal.
PWMsignal ©
+ Noise Ramp
.....,_ _.,.Ramp
Barnparitt 0
; pedestal, Reter.ence
@ pulse
©aneratq(• generator
output
0 ~i+----1>1I I I
tlefereric'e'. DeTay Delay
·······pulse •
;gene@I9A
output ®
Ou1pu1of
(D-457) Fig. 5.13.3 : PWM detection circuit ,:;: the adder

Thus some of the noise is removed and the pulses are 0

squared up. Clipper output
®
PAM
The regenerated pulses are applied to a reference pulse pu1seg l----'-..L..L...-.....1--1..--..,_.i...._ _s;:::::i.._ _ __

generator.
(D-458) Fig. 5.13.4 : Waveforms for PWM detection circuit

~~
. TetltKnowledge
_,., f- 1.i ti: I 1: ;3 t
-
IO r: S
 • PCS (Sem. 4 / E&Tc I SPPU)
~J 5-33 Pulse Modulation

5.13.3 Frequency Spectrum of PWM Wave :

Fig. 5.13.5 shows the spectrum of a PWM signal for a sinusoidal modulating signal with a frequency fm.

l
The spectrum shows that the modulating frequency fm and many of its sidebands are present.

ya
ha
Amplitucle .

G
a
ity
Ad
(D-459) Fig. 5.16.5 : Spectrum of PWM signal

5.13.4 Advantages of PWM : 5.14 Pulse Position Modulation (PPM):

1. Less effect of noise i.e. very good noise immunity.

2. Synchronization between the transmitter and University Questions

receiver is not essential. (Which is essential in Describe With the help • of neat sketches of
PPM). waveforms methods of generation .of PDM/PWM
andPPM. • (Dec.15, 7 Marks)
3. It is possible to reconstruct the PWM signal from a
noise contaminated PWM, as discussed in the Principle:

detection circuit. Thus it is possible to separate out In PPM the amplitude and width of the pulsed carrier
signal from noise (which is not possible in PAM). remains constant but the position of each pulse is

5.13.5 Disadvantages of PWM : varied in proportion with the amplitudes of the sampled
values of the modulating signal.
1. Due to the variable pulse width, the pulses have
The position of the pulses is changed with respect to
variable power contents. So the transmitter must
the position of reference pulses.
be powerful enough to handle power
The PPM pulses can be derived from the PWM pulses as
corresponding to the maximum width pulse. The
shown in Fig. 5.14.1.
average power transmitted can be as low as 50% of
this maximum power.
---:------------
2. In order to avoid any waveform distortion, the
bandwidth required for the PWM communication is
Time
large as compared to BW of PAM. PWM
pulses
o 1L-:-J.,,,.L-..,-.c.1r-~L--~~~:::--~~....u:=--....L~~
5.13.6 Applications of PWM: SPPU : Dec. 14
PPM
pulsesl--+~.;....i-4-_..l...lL---l'-'~....La.......La.......t'-"--.LU-
University Questions
O
QC1 V\lith the help of a diagram explain pulse< width (D-460) Fig. 5.14.l: PPM pulses generated from PWM signal
• modulation; Also state its c:1dvantages,.
Note that with increase in the modulating voltage the
disadvantages and application. (Dec; 1,4.; 7 Marks) PPM pulses shift further with respect to reference. ,
1. In the optical fiber communication. The vertical dotted lines drawn in Fig. 5.14.1 are treated
2. In some military applications. as reference lines to measure the shift in position of
PPM pulses.

~\-TethKnowledge
~t"' Putt:li:3tlons
 PCS(Sem.4/E&Tc/SPPU Pulse Modulation

The leading edge of each PPM pulse coincides with the

trailing pulse of a PWM pulse.
The PPM pulses marked 1, 2 and 3 etc. in Fig. 5.14.1 go

l
ya
away from their respective reference lines.

ha
5.14.2 Demodulation of PPM:

G
This is corresponding to increase in the modulating

a
ity
SPPU : May 08, Dec. 18
signal amplitude.

Ad
Then as the modulating voltage decreases the PPM University Questions
pulses 4, 5, 6, 7 come progressively closer to their ·,r:t' •gxpl~i~ the ge~eration and detection of PPrv1·w~~r
respective reference lines. >with Waveforms. (May 08,·8 Marks)

5.14.1 Generation of PPM Signal: Q. 2 Draw and explain generation and regeneration of
SPPU: Dec. 08, Dec.15, May 16, Dec. 18 PWM and PPM. (Dec.18,7 Marks)

itlnive{sity Questions Block diagram :

Q, 1 pr<?.IN .. and.·· explain with block . sche111atic . the
The PPM demodulator block diagram is as shown in
>9!ifo$r~!!999f pyl~~ positj9n• mqdulatiqn t~chniq(Je,••·•.
· ;/, .•.<)' ·••· .•,.··· > .•.• • • ....... (riekos, 411/l~rk~) Fig. 5.14.3.

ci.'f:,. · ·.••·· •.· .· '.. with }he •

helJ . of• n~~f S~{:ltch~$:cC>t !' ·Pulse
,· ••• wayeforrils methods of generation o( PDM/PWM • PPM
pulses generator
~ntjPP~t (Dec. 15, 7Marks)
PWM Modulating
Q, 3··. W~h help ..ofwaveforms explain how. PWM and Q demodulator signal
• iPP:M can be generated. (May 16, 6 Marks)
Reference
a. 4 Oi~W and ~xplain generation and regeneration of pulse
:pWMandPP!\11. (Dec.18, 7 Marks) generator

Block diagram : (D-462) Fig. 5.14.3 : PPM demodulator circuit

The PPM signal can be generated from PWM signal as Operation:
shown in Fig. 6.5.2(a).
The operation of the demodulator circuit is explained as
The same block diagram has been repeated in follows:
Fig. 5.14.2 as shown.
The noise corrupted PPM waveform is received by the
Comparator
Input PPM demodulator circuit.
signal
x(t) PWM . Monostable. . PPM The pulse generator develops a pulsed waveform at its
'multivibrator' signal
output of fixed duration and apply these pulses to the

Sawtooth reset pin (R) of a SR flip-flop.

signal
The PWM signal can be demodulated using the PWM
(D-461) Fig. 5.14.2 : Generation of PPM signal
demodulator.
Operation:
A fixed period reference pulse is generated from the
The PWM pulses obtained at the comparator output are
incoming PPM waveform and the SR flip-flop is set by
applied to a monostable multivibrator.
the reference pulses.
The monostable is negative edge triggered.
Due to the set and reset signals applied to the flip-flop,
Hence corresponding to each trailing edge of PWM
we get a PWM signal at its output.
signal, the monostable output goes high.
This is same as the one discussed in section 5.13.
It remains high for a fixed time decided by its own RC
components. 5.14.3 Advantages of PPM: SPPU: May 09
Th11s as the trailing edges of the PWM signal keep
University Questions
shifting in proportion with the modulating signal x(t),
the PPM pulses also keep shifting as shown in Q; 1 What are the advantages ani:I disadvantages of
Fig. 5.13.2(b). PPM? (May 09, .6 Marks)
~•e
~" PCS (Sem. 4 / E& Tc/ SPPU) 5-35 Pulse Modulation
1. Due to constant amplitude of PPM pulses, the Sr.
Parameter PAM PWM PPM
information is not contained in the amplitude. No.

Hence the noise added to PPM signal does not 3. Bandwidth Low High High
requirement

l
ya
distort the information. Therefore it has good noise

ha
immunity. This is same as that explained for PWM 4. Noise Low High High

G
a
in section 5.13. immunity

ity
Ad
2. It is possible to reconstruct PPM signal from the 5. Information Amplitude Width Position
noise contaminated PPM signal. This is also is contained variations variation variation
possible in PWM but not possible in PAM. in

3. Due to constant amplitude of pulses, the 6. Transmitted Varies with Varies with Remains
transmitted power always remains constant. It does power amplitude variation in constant
not change as it used to, in PWM. of pulses width
7. Need to Not Not needed Necessary
5.14.4 Disadvantages of PPM : SPPU: May09
transmit needed
synchronizin
University Questions
g pulses
Q. 1 What are the advantages and disadvantages of
8. Complexity Complex Easy Complex
PPM? (May 09, 6 Marks)
of generation
1. As the position of the PPM pulses is varied with and
respect to a reference pulse, a transmitter has to detection
send synchronizing pulses to operate the timing 9. Similarity Similar to Similar to Similar to PM
circuits in the receiver. Without them the with other AM FM
demodulation won't be possible to achieve. modulation
systems
2. Large bandwidth is required to ensure transmission

non □ o n nod
10. Output
of undistorted pulses.
waveforms
(D-463)
5.14.5 Applications:

PPM is used in some military applications. 5.15.1 Comparison of Pulse Modulation and
CW Modulation :
5.15 Comparison of PAM, PWM and PPM :
Sr.
Parameter
cw Pulse
SPPU: Dec. 12, May 14, May 17, May 19, Dec. 19
No. Modulation Modulation
University Questions 1. Carrier Sinusoidal Train of
Q. 1 Distingufsh between PAM, PWM and PPM. rectangular
(Dec. 12, May 14, 8 Marks) pulses.

Q. 2 Compare PAM, PWM and PPM. 2. Types of AM. FM, PM PAM, PWM,
modulation PPM, PCM, DM
(May 17, Dec. 19, 6 Marks)
systems
Q. 3 Compare PAM, PWM and PPM with waveform.
3. Digital Not possible Possible
(May 19, 6 Marks) modulation
Sr. 4. Principle of Amplitude, Amplitude, width
Parameter PAM PWM PPM
No. operation frequency or or position of the
1. Type of Train of Train of Train of phase of the pulsed carrier is
sinusoidal varied in
carrier pulses pulses pulses
carrier is proportion with
2. Variable Amplitude Width Position changed in instantaneous
characteristic proportion with message signal.
of the pulsed instantaneous
message
carrier
signal.
$ .•
~IW' Tech Knowledge
l!il)'8" r, IJ h; ! 1: ,~ti O I'•;
~e"
~F PCS (Sem. 4 / E&Tc / SPPU) 5-36 Pulse Modulation

Sr.
Parameter
cw Pulse Demultiplexing:
No. Modulation Modulation Demultiplexing is the process of separating the signals
5. Performance FM and PM PWM and PPM from a multiplexed signal. It is exactly the opposite
in presence perform well perform well but

l
ya
of noise. but AM does PAM does not process of multiplexing.

ha
G
not. PCM and DM Demultiplexer :

a
also do well.

ity
An electronic circuit which carries out demultiplexing is

Ad
6. Applications Broadcasting Satellite
known as a demultiplexer.
(radio and TV) communication,
communication At the receiving end, of communication link, a
between
demultiplexer is used to separate out the signals into
spaceships and
earth stations. their original form.
7. Bandwidth AM needs Large BW - The operation of demultiplexer is exactly opposite to
requirements smaller BW required due to
that of a multiplexer.
FM needs a pulsed nature of
large BW carrier. Need of Multiplexing:
8. Cost and Simpler and Costly and In the applications like telephony, there are large
simplicity. less costly complex.
number of users involved.

5.16 Multiplexing and Demultiplexing: It is not possible to use a separate pair of wires from
each subscriber to all the other subscribers.
Definition of multiplexing :
This is very expensive and practically impossible.
Multiplexing is the process of simultaneously
transmitting two or more individual signals over a single Instead if we use the principle of multiplexing, then we
communication channel. can use a common communication medium such as a
Due to multiplexing it is possible to increase the coaxial cable or an optical fiber cable to carry telephone
number of communication channels so that more
signals originated from a number of subscribers.
information can be transmitted.
The same principle is applicable to every application in
Multiplexer :
which many signals originating from different sources
An electronic circuit which carries out multiplexing is
known as a multiplexer. are to be sent over a single communication medium.

The concept of a simple multiplexer is illustrated in 5.16.1 Types of Multiplexing :

Fig. 5.16.1.
Different types of multiplexing are :
1. Frequency Division Multiplexing (FDM).
Multiple Original
input input 2. Time Division Multiplexing (TDM).
signals signals 3. Wavelength Division Multiplexing (WDM).
4. Orthogonal FDM (OFDM)
Single
communication channel Fig. 5.16.2 shows the classification of multiplexing
(Wire or Radio) techniques.
(D-499) Fig. 5.16.1 : Concept of multiplexing
Multiplexing
The multiplexer receives a large number of different
input signals.
Multiplexer has only one output which is connected to
the single communication channel. Analog Multiplexing Digital Multiplexing

The multiplexer combines all input signals into a single

composite signal and transmits it over the TDM
communication medium.

Sometimes the composite signal is used for modulating

a carrier before transmission. (L-106) Fig. 5.16.2 : Classification of multiplexing techniques
 • PCS (Sem. 4 / E&Tc / SPPU)
~F 5-37 Pulse Modulation

Generally the FDM and WDM systems are used to deal The TDM system is as shown in Fig. 5.17.2.
Low pass
with the analog information whereas the TDM systems Inputs filters
are used to handle the digital information. x,(t)~ . Synch_ro~_1zed

l
ya
x,<t)
In FDM many signals are transmitted simultaneously

ha
X3(t)

G
where each signal occupies a different frequency slot

a
xN(t) ator 1

ity
within a common bandwidth.

Ad
In TDM the signals are not transmitted at a time, instead (D-513) Fig. 5.17.2: PAM/TDM system
they are transmitted in different time slots. Operation :
CDM and SDM are used in wireless mobile The operation of the system is as follows :
communication systems. The multiplexer here is a single pole rotating switch or
commutator.
5.17 Time Division Multiplexing (TDM) :
It can be a mechanical switch or an electronic switch. It
SPPU: Dec. 17 rotates at fs rotations per second.
University Questions
• a.1··•· •\t\lhafi~ hrr:ie;qiyision multipleX1pg.•?. .• Exptain,ti9W
• all◊e~tioh of time ~lots in TDM dE;pends 0~ ~it rate'.
·~. ~
{bec.17,7Marks} __::]'""•_
·••, L_____·_.
__ • ·_•_·- 1/f - - - - - - · • - f ' I • ~ •., t
5
Definition : - One revolution of commutator -+I
(a) Sampling of the first input
TDM is the multiplexing technique in which many Multiplexed X:,
PAM wave X3
signals are sent in a sequential manner but they occupy
the same band in the frequency spectrum.
Fig. 5.17.1 explains the concept of TDM.
One frame al -i,/i/Nf8 j-
Ts= 1/f8
Signal 2 Signal 3 Signal 4 Signal 1 Signal 2 ............... .
(b) Multiplexed PAM signal, transmitted on the
--+ transmission channel
: + - - - - - O n e frame----->< Time
(D-514) Fig. 5.17.3
(D-510) Fig. 5.17.1: Principle of TDM
As the switch arm rotates, it is going to make contact
In TDM all the signals to be transmitted are not with the position 1, 2, 3 or N for a short time. To these
transmitted simultaneously. Instead, they are contacts are connected the N analog signals which are
transmitted one-by-one. to be multiplexed.
Thus each signal will be transmitted for a very short Thus the switch arm will connect these N input signals
time. one by one to the communication channel.
One cycle or frame is said to be complete when all the Waveforms:
signals are transmitted once on the transmission The waveform of a TDM signal which is being
channel.
transmitted is as shown in Fig. 5.17.3.
As shown in the Fig. 5.17.1 one transmission of each
It shows that the rotary switch samples each channel
channel completes one cycle of operation called as a
during each of its rotations. Each rotation corresponds
"Frame".
to one frame. Hence 1 frame is completed in Ts seconds
The TDM system can be used to multiplex analog or
where Ts = 1/fs.
digital signals, however it is more suitable for the digital
signal multiplexing. At the receiver, there is one more rotating switch or
commutator used for demultiplexing.
5.17.1 PAM -TOM System:
It is important to note that this switch must rotate at the
Block diagram : same speed as that of the commutator 1 at the
The TDM system which is going to be discussed now, transmitter and its position must be synchronized with
combines the concepts of PAM and TDM both. commutator 1 in order to ensure proper demultiplexing.
 e
~~ PCS ,(Sem. 4 / E&Tc / SPPU) 5-38 Pulse Modulation

The same principle of multiplexing can be used for

multiplexing more number of signals.
Interleaving :

l
ya
+
-----Time

ha
On the multiplexer side the commutator-1 opens in I· TJN TJN ·I

G
front of a connection, that connection has the

a
One revolution of commutator

ity
I· ·I

Ad
opportunity to send its bit on to the channel. N pulses in (1 /f5 ) =Ts sec.

This process is called as interleaving. (D-515) Fig. 5.17.4: Calculation of number of pulses per

second for PAM-TDM system

5.17.2 Signaling Rate (r):
5.17 .3 Transmission Bandwidth:
The signaling rate of a TDM system is defined as the
The minimum transmission bandwidth of a PAM-TDM
number of pulses transmitted per second. It is denoted
channel is given by,
by "r".
1
BT =
Let us now derive the expression for the signaling rate 2 signaling rate
of the PAM-TDM system.
1
Let W = Maximum frequency of all the input signals x1 Minimum transmission bandwidth BT :::: x 2 NW
2
to XN.
Minimum transmission bandwidth BT = NW ... (5.17.4)
Therefore as per Nyquist criteria, the sampling
5.17 .4 Advantages of TOM :
frequency f, :::: 2 W. Therefore the speed of rotation of
the commutators is f, rotations per second with 1. Full available channel bandwidth can be utilized for
f,:::: 2 W. each channel.

As shown in Fig. 5.17.4, one revolution of commutators 2. Intermodulation distortion is absent.

corresponding to one frame contains one sample from 3. TDM circuitry is not very complex.
each input signal. 4. The problem of crosstalk is not severe.
:. 1 Revolution ⇒ 1 frame ⇒ N pulses ... (5.17.1) 5.17.5 Disadvantages of TOM:
1 frame period is (1/f,) i.e. T, seconds. Therefore in "Ts"
1. Synchronization is essential for proper operation.
seconds "N" number of pulses are transmitted. Hence
2. Due to slow narrowband fading, all the TDM
the pulse to pulse spacing within the frame is given by,
channels may get wiped out.
T, 1
Pulse to pulse spacing = N = Nf, ... (5.17.2) 5.17.6 Applications of TDM:
As the period of one pulse (ON + OFF) is (1/Nf,) 1. Multiplexing of digital signals.

seconds, the number of pulses per second is given by, 2. Digital telephony.

Number of pulses per second = Nf, 3. Satellite communications.

This is nothing but the signaling rate. 4. Fiber optic communication.

:. Signaling rate of a TDM system= r = Nf, pulses/second. 5. Wireless communication applications.

But as f, ~ 2 W. 5.17.7 Examples on TOM:

Ex. 5.17.1 : A signal x 1 (t) is bandlimited to 3 kHz. There are
Signaling rate of a TDM system = r:::: 2NW pulses/second
three more signals x2 (t), x3 (t) and x4 (t) which are
... (5.17.3)
bandlimited to 1 kHz each. These signals are to be
A TDM system is supposed to have its signaling rate as transmitted by a TOM system.
high as possible. (a) Design a TOM scheme where each signal is sampled at
It is evident from the expressions above that the its Nyquist rate.

signaling rate can be increased by increasing the (b) What must be the speed of the commutator?
sampling rate f, and/or the number of input signals N. (c) Calculate the minimum transmission bandwidth of the
channel.
 ~er
~, PCS (Sem. 4 / E&Tc/ SPPU) 5-39 Pulse Modulation

Soln.: Soln.:
(a) Table P. 5.17.1 shows different message signal with The number of channels N = 6
corresponding Nyquist rates.
Bandwidth of each channel, W = 5 kHz

l
ya
Table P. 5.17.1

ha
Minimum sampling rate = 2 x 5 kHz = 10 kHz

G
Message signal Bandwidth Nyquist rate Signaling rate = Number of bits per second

a
ity
(t) = 6 x 10 kHz = 60 Kbits/sec. ... Ans.

Ad
X1 3 kHz 6 kHz
Xz (t) 1 kHz 2 kHz Minimum channel bandwidth to avoid cross talk in
X3 (t) 1 kHz 2 kHz PAM/ TDM is,
X4 (t) 1 kHz 2 kHz BT = NW = 6 x 5 kHz = 30 kHz ... Ans.

If the sampling commutator rotates at the rate of 2000 Ex. 5.17.3 : Sketch a channel interleaving scheme for time
rotations per second then the signals x2 (t), x3 (t) and division multiplexing the following PAM signals : Five 4 kHz
x4 (t) will be sampled at their Nyquist rate. telephone channels and one 20 kHz music channel. Find the
But we have to sample Xi (t) also at its Nyquist rate pulse repetition rate of the multiplexed signal and estimate
which is three times higher than that of the other three. the minimum system bandwidth required.
In order to achieve this we should sample x1 (t) three Soln.:
times in one rotation of the commutator. Each telephone channel of bandwidth 4 kHz must be
Therefore the commutator must have atleast 6 poles sampled at Nyquist rate i.e. 2 x 4 kHz = 8 kHz. using a
connected to the signals as shown in Fig. P. 5.17.1. TDM commutator.
x 1 (t)
The 20 kHz music channel must be sampled at 40 kHz
2000 rotations/ sec. (Nyquist rate) hence a separate sampler is required.
The sampled signals are applied to two 4-bit A-D
converters to obtain the equivalent digital signals.
These signals are finally multiplexed using a multiplexer
as shown in Fig. P. 5.17.3.
5 Telephone channels
4kHz••--J~

4 kHz
(D-519) Fig. P. 5.17.1 TDM
commutator
4 kHz
(b) The speed of rotation of the commutator is 2000 40 K
samples/sec . .....__ __.
rotations/sec. 4 kHz ..
320 K
(c) Number of samples produced per second is calculated 4 kHz ••_ _:;~ samples
/sec.
as follows: 20kHz
music
x1 (t) produces 3 x 2000 = 6000 samples/sec. channel
x2 (t), x3 (t) and x4 (t) produce 2000 samples/sec. each.
(D-521) Fig. P. 5.17.3: PAM-TDM system for Ex. 5.17.3
Number of samples per second = 6000 + (3 x 2000)
The TDM commutator output has a pulse repetition rate
= 12000 samples/sec. of 40K samples/sec as there are 5 channel and sampling
Signaling rate = 12000 samples/sec. rate is 8 kHz.
(d) The minimum channel bandwidth is, Similarly the output of the separate sampler has a pulse
1 repetition rate of 40 K samples/sec.
BT =
2signaling rate = 12000/2
The outputs of A-D converters have pulse repetition
BT = 6000 Hz ... Ans.
rates of 40 x 4 = 160 K samples/sec.
Ex. 5.17.2: Six message signals each of bandwidth 5 kHz Therefore pulse repetition rate at the output of •a
are time division multiplexed and transmitted. Calculate the multiplexer is 160 + 160 = 320 K samples/sec.
signaling rate and the minimum channel bandwidth of the Pulse repetition rate of the system = 320 kHz.
PAM/ TOM chann7el. Bandwidth required=bit rate = 320 kHz ... Ans.
 G
PCS (Sem. 4 / E&Tc / SPPU) 5-40 Pulse Modulation

Ex. 5.17.4: One analog waveform W 1(t) is bandlimited to

3 kHz and another W2 (t) is bandlimited to 9 kHz. These two
j· Review Questions I
signals are to be sent by TOM over a PAM type system.
Q. 1 Define the sampling process and explain its

l
ya
1. Determine the minimum sampling frequency of each

ha
signal. necessity in communication system.

G
2. Design a TOM commutator and decommutator to Q. 2 State and prove the sampling theorem for a low

a
ity
accomodate the signals.

Ad
pass bandlimited signal.
3. Draw some typical waveforms for W 1 (t) and W 2(t) and
sketch the corresponding TOM-PAM waveforms. Q. 3 What do you understand by the word bandlimited ?
Soln.: Q. 4 Explain the term aliasing and its effects.
The minimum sampling rate for W1 (t) will be 6 kHz and
Q. 5 How can we avoid aliasing ?
that for W2 (t) is 18 kHz.
Q. 6 Explain the natural PAM.
The TDM commutator and decommutator is as shown
in Fig. P. 5.17.4(a). Q. 7 Explain the flat top PAM.
Commutator
w1(t) - - - - - =G) at speed Q. 8 Draw and explain the generation of flat_ top PAM
, 6000 rps
signal.
Q. 9 Compare natural and flat top PAM.
Q. 10 Explain the detection of PAM.
Q. 11 Define PAM and explain its generation and
detection.
Q. 12 Why is PAM not used for communication
(D-1498) Fig. P. 5.17.4(a): Commutator arrangement applications?
Note that Wi(t) will be sampled at a rate which is 3 Q. 13 State the disadvantages of PAM.
times higher than the speed of rotation of the
commutator. This is because points 2, 3, 4 are 0. 14 Explain the generation and detection of a PWM
connected to W2 (t). signal.
The commutator rotates at a speed of 6000 revolutions Q. 15 State and explain the merits and demerits of PWM
per second. transmission.
Due to the arrangement shown in Fig. P. 5:17.4(a) both Q. 16 With the help of neat circuit diagram explain the
the input signals are sampled at their minimum
generation and detection of a PPM signal.
sampling rates.
Q. 17 State the merits and demerits of a PPM
The waveforms of W1(t), Wi(t) and the PAM-TDM signal
transmission.
are shown in Fig. P. 5.17.4(b).
Q. 18 Compare PAM, PWM and PPM systems.
Q. 19 Explain the effects of the finite pulse width of the
sampling function.
Q. 20 What is "aperture effect" and explain how to reduce
it.
Q. 21 Compare the ideal, natural and flat top sampling
techniques.
Q. 22 Explain the principles of time division multiplexing.
Sampling
function Q. 23 Why is it necessary to use time division multiplexing
while transmitting PAM signals?
Q. 24 Why is synchronization needed in TOM system ?
TDM-PAM
waveform
Q. 25 Describe how transmission distortion of a TD,M
signal can cause cross-talk between two adjacent
channels.
(D-1499) Fig. P. S.17.4(b)
□□□