Chapter Four

Pulse Modulation Systems

4.1 Introduction
In continuous wave modulation some parameter of a sinusoidal carrier is varied continuously in
accordance with the message signal and in pulse modulation some parameter of a pulse train is
varied in accordance with the message signal. We may distinguish two families of pulse
modulation systems: analog pulse modulation and digital pulse modulation. In analog pulse
modulation, a periodic pulse train is used as the carrier wave, and some characteristic of each
pulse (e.g., amplitude, duration, or position) is varied in continuous manner in accordance the
corresponding sample value of the message signal. On the other hand, in digital pulse modulation
the message signal is represented in a form that is discrete both in amplitude and time, thereby
permitting its transmission in digital form as a sequence of coded pulses.

Pulse Amplitude Modulation (PAM) systems are analog systems and share a common problem
with AM and FM systems; i.e. each of these systems are extremely sensitive to the noise present
in the receiver.

4.2 Sampling Theorem

A signal whose spectrum is band-limited to B Hz [G (f) = 0 for |f|>B] can be reconstructed
exactly from its samples taken uniformly at a rate R>2B Hz. The minimum sampling frequency
is  = 2 .

To prove the sampling theorem, consider a signal ( ) (shown in Fig. 4.1) whose spectrum is
band limited to B Hz. Sampling ( ) at a rate of   can be accomplished by multiplying ( )
by an impulse train  ( ), consisting of unit impulses repeating periodically every  seconds,
where  = 1/ . This results in the sampled signal ̅ ( ) shown in Fig. 4.1d. The sampled signal
consists of impulses spaced every  second. The nth impulse, located at =  , has
strength ( ), the value of ( ) at =  . Thus,

̅( ) = ( )  ( ) = ∑ () ( − ) (4.1)

Because the impulse train  ( ) is a periodic signal of period  , it can be expressed as a Fourier
series. The trigonometric Fourier series is:

 ( )=

1 + 2 cos( ) + 2 cos(2 ) + 2 cos(3 ) + ⋯"  = 2# (4.2)

Therefore,

̅( ) = ( )  ( )

1|Page
 
=  ( ) + 2 g(t)cos( ) + 2 g(t)cos(2 ) + 2 g(t)cos(3 ) + ⋯" (4.3)


To find the Fourier transform of ̅ ( ), we take the Fourier transform of the right-hand side of
equation (4.3), term by term. So, the spectrum &̅ () consists of &() repeating periodically with
period  = 2# as shown in Fig. 4.1e. Therefore,

&̅ () =

∑'
()' &( −  ) (4.4)


If we are to reconstruct ( ) from ̅ ( ), we should be able to recover &() from &̅ (). This is
possible if there is no overlap between successive cycles of &̅ (). Fig. 4.1e shows that this
requires

 > 2

Thus as long as the sampling frequency  is greater than twice the signal bandwidth B, &̅ ()
consists of non overlapping repetitions of &(). When this is true, Fig. 4.1e shows that ( ) can
be recovered from its sample ̅ ( ) by passing the sampled signal ̅ ( ) through an ideal low pass
filter of bandwidth B Hz. The minimum sampling rate  = 2 required to recover ( ) from its
sample ̅ ( ) is called the Nyquist rate for ( ), and the corresponding sampling interval  =
1+ is called the Nyquist interval for ( ).
2

Fig. 4.1 Sampled signal and its Fourier Transform

2|Page
Noting that the Fourier transform of the delta function  ( −  ) is equal to exp (−/2# ),
we can write the Fourier transform of ̅ ( ) as follows:

&̅ () = ∑'

3()' g(nT2 )exp (−/2# ) (4.5)

Since Ts=1/2B

&̅ () = ∑'

3()' g(n/2B)exp (−/#/) (4.6)

And from equation 4.4

&̅ () =  &() +  ∑'

()' &( −  ) (4.7)
56

Hence under the following two conditions:

1. &() = 0 89 || ≥ 

2.  = 2

We find from equation 4.7 that

&() = <= &̅ (), || < 


(4.8)

Substituting equation 4.6 into 4.8, we may also write


&() = <= ∑'
3()' g(n/2B)exp (−/#/) || <  (4.9)

Therefore if the sample values g(n/2B) of a signal ( ) are specified for all n, then the Fourier
transform &() of the signal is uniquely determined by using the discrete-time Fourier transform
of equation 4.9. The signal ( ) itself is also uniquely determined by the sample values g(n/2B)
for−∞ <  < ∞. That means, the sequence g(n/2B) has all the information contained in ( ).

Consider next the problem of reconstructing the signal ( ) from the sequence of sample
values g(n/2B).
'
( ) = A &()exp (/2# ) B
)'

'
=
1
( )=A C g(n/2B)exp (−/#/) exp (/2# ) B
)= 2 3()'

Interchanging the order of summation and integration:

 =
( ) = ∑'
3()' g(n/2B) D exp (/2#( − /2)) B (4.10)
<= )=

3|Page
 3
g(t) = ∑'
3()' g E<FG sinc(2Bt − n) , −∞< <∞ (4.11)

Equation 4.11 provides an interpolation formula for reconstructing the original signal ( ) from
the sequence of sample values of g(n/2B), with the sinc function sinc (2Bt) playing the role of
an interpolation function.

We may therefore state the sampling theorem for strictly band-limited signals of finite energy in
two equivalent parts:

1. A band-limited signal of finite energy, which has no frequency components higher than B
Hz, is completely described by specifying the values of the signal at instants of time
separated by 1/2B seconds.
2. A band-limited signal of finite energy, which has no frequency components higher than B
Hz, may be completely recovered from knowledge of its samples taken at a rate of 2B
samples per second.

Aliasing for Real signals

Actual signals are not frequency limited

• Always some residual HF (over-Nyquist) components

• HF signals are folded to baseband by sampling
• Aliasing noise

The amount of aliasing noise is related with

• Input signal frequency spectrum (modified by antialias filter)

• Sampling frequency 

The effects of aliasing in practice can be removed by using a low pass anti-aliasing filter to
attenuate the high frequency components of the signal that are not essential to the information
being conveyed by the signal and the filtered signal is sampled at a rate slightly higher than the
Nyquist rate.

So the reconstruction filter is a low-pass with a pass band extending from –B to B, which
is itself determined by the anti-aliasing filter.
The filter has a transition band extending (for positive frequencies) from B to  − ,
where  is the sampling rate.

4|Page
Oversampling

• Sampling only slightly above the Nyquist limit requires steep antialiasing filters
(expensive)
• Another choice: sampling at a rate far higher than the Nyquist limit (oversampling)
- Relaxed specification on the anti-alias input filter, but higher bit rate (more
samples/s)

4.3 Pulse Amplitude Modulation

Pulse amplitude modulation (PAM) is an engineering term used to describe the conversion of the
analog signal to a pulse type signal where the amplitude of the pulse denotes the analog
information.

There are two classes of PAM signals: PAM that uses natural sampling (gating) and PAM that
uses instantaneous sampling to produce a flat-top pulse.

Natural Sampling

Definition: If ( ) is an analog waveform band limited to B Hz, the PAM signal that uses the
natural sampling (gating) is

̅ ( ) = ( )( ) (4.12)

Where
'
− 
( ) = C I
J
()'

is a rectangular wave switching waveform and  = 1/ ≥ 2.

The spectrum for a naturally sampled signal is

2L3(MK⁄ )
&̅ () =  ∑'
K
()' MK⁄ &( −  ) (4.13)
 

Instantaneous Sampling (Flat-Top PAM)

Definition: If ( ) is an analog waveform band limited to B Hz, the instantaneous sampled

signal is given by

̅ ( ) = ∑'
()' ( ) ℎ( −  ) (4.14)

Where h (t) denotes the sampling pulse shape, and for flat-top sampling the pulse shape is

5|Page
 1, | | < J⁄2 S
ℎ( ) = I P Q = R
J 0, | | > J⁄2

where J ≤  = U VB  ≥ 2.


The spectrum of a flat-top PAM signal is

&̅ () =

() ∑'
()' &( −  ) (4.15)


where

sin (#J)
() = J
#J

Fig. 4.2 PAM signal with natural sampling

6|Page
 Fig. 4.3 PAM signal with flat-top sampling

Other Forms of Pulse Modulations

Pulse-duration modulation (PDM): also referred to as pulse-width modulation, where samples of

the message signal are used to vary the duration of the individual pulses in the carrier.

Pulse-position modulation (PPM): where the position of a pulse relative to its un-modulated time
of occurrence is varied in accordance with the message signal

7|Page
Fig. 4.4 illustrating two different forms of pulse-time modulation for the case of a sinusoidal
modulating wave, a) Modulating wave b) Pulse carrier c) PDM wave, d) PPM wave

8|Page