Sampling Theorem and pulse

Modulation Techniques
 In modern communications, most of the information data is of
digital form such as binary data used in computer programs and
the codes for alphanumeric characters.
 Moreover, analog information including voice and video signals
are also transmitted using digital transmission techniques.
 The evolution from analog to digital transmission is the
conversion of analog information to digital representation using
sampling theorems for both baseband and bandpass signals.
 Pulse-modulation techniques such as Pulse Amplitude
Modulation (PAM), Pulse Width Modulation (PWM), Pulse
Position Modulation (PPM), Pulse Code Modulation (PCM), and
Delta Modulation (DM) are basically baseband digital signals
which are used for transmitting discrete signals.
Digital Transmission refers to transmission of digital signals
between two or more points in an electronic communication
system.
 The form of digital signals may have binary levels or discrete
levels, or analog information signals converted to digital pulses.
 Analog Transmission using Baseband Channel figure below depicts
analog transmission using baseband channel, in which an analog
information signal is sent over a wireline (wired)channel with no
modulation.
 An example of analog signal transmission over baseband channel
is a public address system using twistedpair wire as a channel,
and mainly comprises of a microphone, an audio amplifier, and a
speaker.
 • Analog Transmission using Bandpass Channel Figure below depicts
analog transmission using bandpass channel such as wireless
channel, in which an analog information signal is sent using
modulation.

 An example of analog signal transmission over bandpass channel

is broadcast radio and television systems.
 • Digital Transmission over Digital Channel Figure below depicts
digital transmission of digital data over digital channel. The
digital communication channel can handle digital pulse signals
directly after applying appropriate line encoding signaling
formats.

 An example of digital transmission over digital channel is

transmission of digital data such as a data file from PC over
digital communication channel (usually wireline).
 • Digital Transmission over Analog Channel Figure below depicts
transmission of digital data over analog channel. Digital data signals or
digitized analog signals are modulated onto an analog carrier signal at
the transmitter end and demodulated at the receiver end.

 An example of digital transmission over analog channel include

transmission of digitized speech signals over analog channels (usually
an ordinary telephone line, or a radio channel which requires a
modulation process).
The function of modulation and demodulation is performed by a
device known as modem (modulator + demodulator). It may be
noted that modem can be used to interface digital data source to
analog channel, and analog channel to digital data destination.
Advantages of digital transmission:
There are number of inherent advantages of digital transmission over
analog transmission of analog signal (after digitization process) or digital
data.
 Resistant to Additive Noise Digital communication systems are more
resistant to additive noise because they use signal regeneration rather
than signal amplification as in analog transmission systems.
 Immunity to External Noise Digital transmission signals are inherently less
susceptible to interference (caused by external noise) than analog
signals.
 Ease of Multiplexing Digital signals are better suited for processing and
combining using multiplexing techniques. Even analog signals are
processed using digital signal processing methods, which includes band
limiting the signals with filters and amplitude equalization.
 Convenient to Store Data It is much convenient to store and forward
digital signals. Transmission rate of digital signals can be easily varied
to different operating environments and interface requirements with
various types of equipment.
 Ease of Evaluation and Measurement The received digital pulses are
evaluated during a predefined precise sample interval, and a decision is
made whether the received digital pulse is above or below a specified
threshold level.
 More Suitable for Processing Digital transmission is more convenient for
processing, regeneration, and amplification of digital signals. The
digital signal processing includes filtering, equalizing, phase shifting,
and storing of digital data.
 Use of Signal Regenerators Digital transmission systems use signal
regeneration rather than signal amplification as in analog transmission
systems. Digital regenerators sample the received noisy signals, then
reproduce an entirely new digital signal with same SNR as that of
original transmitted signal. So digital signals can be transmitted over
much longer distances without signal deterioration.
 Improved Performance Digital signals can be easily measured and
evaluated. It enables to compare the error performance in terms of
bit-error rate (BER) of one digital system to another. Moreover,
transmission errors can be accurately detected and corrected.
 Digital signals are less susceptible to external noise, which is
mainly due to the fact that digital signals do not require
evaluating the precise value of amplitude, frequency, or
phase to determine its logic condition as with analog
signals. Instead, received pulses are evaluated during a
precise time interval based on the predefined threshold level.
 Comparison of Digital Transmission and Analog Transmission:
 Transmission of digitally encoded analog signals requires significantly
more bandwidth, additional hardware for encoding and decoding,
precise time synchronization between transmission and reception, and
suffers from incompatibility with existing analog transmission
facilities.
Design Goals of Digital Communication System:
The goals of the system designers for any digital communication system
may include the following aspects:
 To minimize required system bandwidth
 To maximize transmission bit rate
 To minimize required power, or equivalently, to minimize required
bit-energy-to- noise power spectral density (Eb /N0)
 To minimize probability of bit error, or simply probability of error, Pb
 To provide reliable user services (minimum delay and maximum
resistance to interference)
 To minimize system complexity, computation load, and system cost.
Sampling Theorem:
The sampling theorem is the fundamental principle of digital
communications. It provides the basis for transmitting analog
information signal by use of digital transmission techniques.
 • The analog information signal such as speech or video signals is
sampled with a pre-determined train of narrow rectangular
pulses in such a way so as to closely approximate the
instantaneous sampling process.
 • In the sampling process, a continuous-time varying analog
signal is converted into a discrete-time signal by measuring the
signal amplitude level at periodic instants of time.
Sampling Theorem for Baseband Signal:
 A baseband signal is an analog information signal which is band-
limited having a finite energy.
In Time domain:
“A baseband signal having no frequency components higher
than fm Hz may be completely recovered from the knowledge of
its samples taken at a rate of at least 2 fm samples per second,
that is, sampling frequency fs ≥ 2fm.”

Nyquist sampling rate:

The minimum sampling rate fs = 2fm samples per second is
called the Nyquist sampling rate.
In frequency domain:
“A baseband signal having no frequency components higher
than fm Hz is completely described by its sample values at
uniform intervals less than or equal to 1/(2fm) seconds apart,
that is, the sampling intervalTs ≤ 1/(2fm) seconds.”
 Then these samples uniquely determine the analog information
signal, and the analog signal may be reconstructed from these
samples with no distortion.
 The maximum sampling interval Ts = 1/(2fm) seconds is called
Nyquist sampling interval.
Significance of Nyquist Sampling Theorem:
 The Nyquist sampling theorem establishes the minimum
sampling rate (fs) that must be used to transmit analog signal in a
given digital communication system.
 As per Nyquist criterion, the minimum sampling rate must be
equal to twice the highest frequency present in the input analog
signal in order to ensure reliable reconstruction of the
information signal at the receiver.
 Thus, Nyquist criterion is just a theoretically sufficient condition
which may allow an analog signal to be reconstructed completely
from a set of uniformly spaced discrete-time samples.
 Theoretically, a band-limited signal has zero value of Fourier
transform beyond the highest spectral frequency fmHz.
 Theoretically, the Fourier transform of a band-limited signal
extends from – ∞ to + ∞
 But practically the Fourier transform provides a finite bandwidth
because higher-order terms can be neglected without any
significant error.
 The statement of sampling theorem for baseband signal can be
proved using the frequency convolutional property of the Fourier
transform.
 Consider a band-limited signal, having no frequency components
beyond fm Hz, that is, s(t) = 0 outside the frequency range (– fm
< f < fm).
 In frequency domain, this can be represented as S( f ) = 0; for
| f | > fm.
 The sampling of band-limited signal, s(t) may be viewed as the
multiplication of s(t) with a periodic train of unit impulse
function s𝛿 (t), defined as

 where Ts is the sampling period and 𝛿(t) is the unit impulse

function.
 To satisfy the Nyquist criterion, let
 Where, is the sampling frequency.

 The Fourier transform of the impulse train 𝛿 T (t) is given as

 This indicates that the Fourier transform of an impulse train is
another impulse train, but the values of the periods of the two
impulse trains are reciprocal of each other.
 The output signal so(t) is a sequence of impulses located at
uniform intervals of Ts seconds, having amplitude levels equal to
that of input signal s(t) at the corresponding instants.That is,

 Using the frequency convolutional property of the Fourier

transform, the time-domain product of so(t) can be transformed
to the frequency-domain convolution.
 The convolution of any function with an impulse function simply
shifts that function, that is,
 Figure below depicts the corresponding waveforms in time
domain as well as in frequency domain of sampling theorem
using the frequency convolution property of the Fourier
transform.
 It is evident that S(f) will repeat periodically without overlapping,
provided fs ≥ 2fm
1
≥ 2𝑓𝑚
𝑇𝑠

where Ts is the uniform sampling interval.

 Sampling rate or sampling frequency fs should meet the condition
fs ≥ 2fm samples/second with minimum sampling rate fs = 2fm
samples/sec.
 In fact, this is the statement of the sampling theorem for
baseband signal.
Sampling Theorem for Bandpass signal:
If an analog information signal containing no frequency outside the
specified bandwidth W Hz, it may be reconstructed from its samples at a
sequence of points spaced at 1/(2W) seconds apart with zero-mean squared
error.
 • The minimum sampling rate of (2W) samples per second, for an
analog signal bandwidth of W Hz, is called the Nyquist rate.
 • The reciprocal of Nyquist rate, 1/(2W), is called the Nyquist
interval, that is, Ts = 1/(2W).
Quadrature sampling of Bandpass signal:
The Inphase and Quadrature components of bandpass signal
are known as Quadrature sampling of bandpass signal.
Quadrature sampling of bandpass signal can be obtained by
multiplying inphase and quadrature components by cos (wct) and sin
(wct), followed by low-pass filtering, thereby retaining the low-
frequency components only.
 • Consider a bandpass signal s(t) whose Fourier transform S( f )
exists only in a band of frequencies 2 W, say, centered about ± fc,
where fc is the carrier frequency.
 • In most of the communication signals, the signal s(t) is a
narrowband signal such that its bandwidth 2W is small compared
with fc.
 • Let the pre-envelope of such narrowband signal s(t) be
expressed in the form, where
Ƹ represents the complex envelope of the signal.
 𝑠(t)
 • The spectrum of s+(t) is limited to the frequency band
(fc –W)≤f≤(fc + W).
 • Applying the frequency-shifting property of the Fourier
transform, the spectrum of the complex envelope 𝑠(t) Ƹ is limited
to the band (-W)≤f≤(+W). and centered at the origin.
 • This implies that the complex envelope 𝑠(t)Ƹ of a bandpass signal
s(t) is a low-pass signal.
 • The spectrum of inphase and Quadrature components is limited
between – W and +W, where 2W is the maximum bandwidth of
given bandpass signal.
 Figure below shows the spectrum of inphase and Quadrature
components of bandpass signal.

 In standard form, the original bandpass signal s(t) can be

expressed as

 where sI(t) the inphase component, and sQ(t) is the Quadrature

component of the bandpass signal.
 Figure below depicts a simplified block diagram for generation of
inphase and quadrature samples.
 The multiplication of the low pass sI(t) with cos (wct) and sQ(t)
with sin (wct) represent linear forms of modulation.
 Since the carrier frequency fc is sufficiently large, the resulting
band-pass signal s(t) is referred to as a passband signaling
waveform, and the mapping from sI(t) and sQ(t) into s(t) is
known as passband modulation.
 The sum frequency components are then suppressed by means of
appropriate low-pass filter (LPF).
Reconstruction of Bandpass signal:
 • To reconstruct the original bandpass signal from its
Quadrature-sampled version, firstly the inphase and Quadrature
components are reconstructed from their respective samples by
using independent reconstruction filters.
 • Then the inphase component, sI(t) is multiplied by cos (2𝜋fct)
and the Quadrature component, sQ(t) is multiplied by sin (2𝜋fct).
 • The resultant signals are added to give the desired bandpass
signal. Figure below depicts the reconstruction process of the
band-pass signal.
 The recovered band-pass signal s(t) is given as
Practical Aspects of Sampling:
What happens if sampling rate exceeds Nyquist rate?
 The sampling rate must be fast enough so that at least two
samples are taken during the period corresponding to the
highest-frequency spectral component present in the analog
information signal.
 The minimum sampling rate is known as Nyquist rate.
 An increase in sampling rate above the Nyquist rate increases the
width of the guard band between two adjacent samples. This
makes filtering operation easier.
 But it extends the minimum bandwidth required for transmitting
the sampled signal.
What happens if sampling rate Nyquist rate?
 When the sampling rate is reduced (sampling at too low a rate
called undersampling), such that fs < 2fm, spectral components
of adjacent samples will overlap and some information will be
lost.This phenomenon is called aliasing.
 Thus, the Nyquist rate, fs = 2fm, is the minimum sampling rate
below which aliasing occurs.
 In order to avoid aliasing, the Nyquist criterion fs > 2fm must be
satisfied.
What are practical difficulties in reliable reconstruction of the
sampled analog information signal in sampling process?
 In the statements of sampling theorem for either base-band or band-
pass analog signal, it is assumed that it is strictly band-limited.
 This means that there is no spectral frequency component outside the
highest frequency fm Hz or the specified bandwidth W Hz.
 However, an analog signal cannot be finite in both time and frequency.
 Therefore, it implies that the analog signal must have infinite time
duration, ranging from – ∞ to + ∞, for its frequency spectrum to be
strictly band-limited.
 Practically, it is generally required to analyze a finite portion of the
analog signal, in which case the frequency spectrum cannot be then
strictly band-limited.
 Consequently, when an analog signal of finite duration is sampled, an
error in the reconstruction of
Aliasing:
The phenomenon of the presence of high frequency component in
the spectrum of the original analog signal is called aliasing or
simply foldover.
 When aliasing occurs, some desirable information content is
inevitably lost in the sampling process.
Describe the phenomenon of aliasing with the help of waveforms.
 Consider an analog signal s(t) whose spectrum S(f) decreases with
increasing frequency without limit.
 The spectrum S𝛿 (f) of the discrete-time signal s𝛿 (t), resulting
from the use of ideal sampling, is the sum of S(f) and an infinite
number of frequency shifted replicas of it of the form of
 S𝜹(f)
 The replicas of S(f) are shifted in frequency by multiples of the
sampling rate fs.
Figure below shows the spectrum of a finite-energy analog signal
before sampling and composition of spectrum of discrete-time
signal after sampling for two replicas of S(f) at fs and – fs.
 It is quite clear that the use of a low-pass reconstruction filter,
with its pass-band extending from – fs/2 to + fs/2, where fs is
the sampled frequency, does not yield an undistorted version of
the original analog information signal.
 It results into the portions of the frequency-shifted replicas
folded over into the desired frequency spectrum.
Aliasing Distortion:
“The absolute error between the original analog signal and the
signal reconstructed from the sequence obtained by sampling is
termed as aliasing distortion or foldover distortion, or simply
aliasing error.”
 • If fs < 2 fm, aliasing distortion will occur.
 Figure below shows an illustration of aliasing distortion.
 If fs = 2fm, the resultant sampled signal is just on the edge of
aliasing. In order to separate the signals sufficiently apart, the
sampling frequency fs should be greater than 2 fm, as stated by the
sampling theorem.
 If aliasing does take place, the interfering frequency component,
called as aliasing frequency, will be at a frequency
fa = fs – fm
 where fa is the frequency component of the aliasing distortion
(Hz), fs is the minimum Nyquist sampling rate
 (Hz), fm is the maximum analog input (baseband or modulating)
frequency (Hz).
There are certain corrective measures for sampling of an analog
information signal in order to overcome the practical difficulties
encountered in sampling and recovery process.
 A practical procedure for the sampling of an analog signal whose frequency spectrum
is not strictly band-limited involves the use of the following corrective measures:
 Prior to sampling, a low-pass pre-alias filter of sufficient higher order is
recommended to be used. This will attenuate those high-frequency spectral
components of the analog information signal that do not contribute significantly to the
information content of the analog signal.
 The filtered analog information signal (by pre-alias filter) is
recommended to be sampled at a rate slightly higher than that
determined by the Nyquist rate, that is, greater than 2fm Hz where fm Hz is the 3
dB high cut-off frequency of the pre-alias filter.
 With such a sampling rate, there are gaps each of width (fs – 2fm) Hz between the
frequency-shifted replicas of the analog signal. These frequency gaps are generally
refereed as guard bands.
 • In practice, a low-pass filter (pre-alias filter) is used at the front end of the impulse
modulator (used for sampling). This enables to exclude the frequency components
greater than the required maximum frequency component of the information signal.
 • Thus, the application of sampling process allows the reduction of continuously
varying information waveform to a finite limited number of discrete levels in a unit
time interval.
 Accordingly, the reconstruction filter at the receiver end is designed to
satisfy the following characteristics:
 • The passband of the reconstruction filter should extend from zero to
fm Hz.
 • The amplitude response of the reconstruction filter rolls off gradually
from W Hz to (fs – 2 fm) Hz.
 • The guard band has a width equal to (fs – 2 fm) Hz which is nonzero
for (fs > 2 fm) Hz.
Classification of Pulse Modulation Techniques:
 An analog signal can be transmitted using digital-processing
techniques depending upon the nature of transmitting medium
such as
 • Analog communication channel
 • Digital communication channel
 Analog communication channel cannot carry digital pulses, so
there is a need of modulation and demodulation (modem) after
digitization process.
 Digital communication channel can carry digital pulses, so the
digitized analog signal can be directly transmitted digitally.
Pulse Modulation:
When any one characteristics (such as amplitude, width or position) of a
relatively higherfrequency carrier signal comprising of discrete pulses is
varied in accordance with the amplitude of the analog modulating signal, it
is called pulse modulation.
 To transmit an analog information data using digital signals and
digital-transmission techniques, pulse modulation is necessary.
 Pulse modulation essentially consists of sampling analog
information data at regular intervals and then converting those
samples into discrete pulses.
 These pulses can then be transmitted from a source to a
destination over a physical communication channel.
Pulse Amplitude Modulation:
When the amplitude of a relatively higher frequency carrier signal
comprising of discrete pulses is varied in accordance with the amplitude of
the analog modulating signal, it is called pulse amplitude modulation
(PAM).
 It is seen from the figure that the amplitude of a pulse coincides
with the amplitude of the analog information signal. PAM
waveform resembles the original analog signal depending on the
frequency of the pulse carrier signal.
 • In single-polarity PAM, a fixed dc level is added to the
information analog signal which ensures that the pulses are
always positive.
 • In double-polarity PAM, the sampled pulses are positive when
the information analog signal has positive amplitude, and it is
negative when the information analog signal has negative
amplitude.
Methods of Sampling:
There are three distinct methods of sampling in PAM:
 Ideal sampling—an impulse at each sampling instant
 Natural sampling—a pulse of short width with varying
amplitude
 Flat-top sampling—sample and hold (like natural sampling)
but with fixed amplitude value
 In each case the sampling rate must be at least twice the highest
frequency contained in the analog signal, according to the
Nyquist criterion.
 In ideal sampling, an arbitrary analog signal is sampled by a train
of impulses at uniform intervals, Ts. An impulse (having virtually
no pulse width) is generated at each instant of sampling.
Natural sampling refers to PAM signals when top of the
sampled pulse retain their natural shape during the sample
interval.
 In natural sampling, an arbitrary analog signal is sampled by a
train of pulses having finite short pulse width occurring at
uniform intervals. The amplitude of each rectangular pulse
follows the value of the analog information signal for duration of
the pulse.
Disadvantages:
 • It is difficult for an analog-to-digital converter to convert the
natural sample to a digital code.
 • In fact, the output of analog-to-digital converter would
continuously try to follow the changes in amplitude levels and
may never stabilize on any code.
Flat Top sampling refers to PAM signals when tops of the
sampled pulse remain constant during the sample interval.
 In flat-top sampling, an arbitrary analog signal is sampled by a
train of pulses having finite short pulse width occurring at
uniform intervals.
 The amplitude of each rectangular pulse is retained as the value
of the analog information signal at the leading edge of the pulse.
Disadvantages:
 The use of flat-top PAM samples results into amplitude
distortion.
 There is delay by Tb /2, where Tb is the width of the pulse, that
results into lengthening of the samplesduring transmission.
 At the receiver, amplitude distortion as well as delay causes
errors in decoded data.
Aperture Effect:
In Flat Top PAM signals, the high frequency contents of the analog signal
are lost which results into distortion known as the aperture effect.
 The aperture effect occurs in flat-top sampling. It is due to the
presence of finite pulse width at the instant of sampling in pulse-
amplitude amodulated signal. In fact, the sampling process in
flat-top sampling introduces aperture error because amplitudes
of analog signal changes during the sample pulse width.
How to Compensate that:
 • This distortion due to aperture effect can be minimized by
passing the demodulated signal through an equalizer.
 • The equalizer has the effect of compensating for the aperture
effect.
 • However, the amount of equalization required in practical
applications is usually small.
Generation and Demodulation of PAM signals:

 • An analog information signal is passed through low-pass filter

which removes any high-frequency components present in it.
 • The sampling circuit also receives a train of pulses from pulse-
sequence generator.
 • The output is pulse amplitude signal.
 At the receiver, the original analog information signal can be
recovered using a low-pass filter, also called reconstruction filter.
 The resultant output waveform is very close to the original signal
waveform.
Mathematical Analysis of Reconstruction of PAM signal:
 • The analog signal s(t) can be reconstructed in time-domain from
its sampled version Ss(t).
 • It may be recovered in frequency domain by passing its sampled
version Ss(f) through a low-pass filter (LPF) with a cut-off
frequency, fm.
 For a signal s(t) sampled at Nyquist rate, fs = 2fm, we have
 where Sa(fmt) is the transfer function of a low-pass filter with
amplitude Ts and bandwidth 2fm.
 Figure below illustrates the process of recovering signal from
sequence of sampled signal.
 The sampled function Ss(t) can be considered as a sum of
impulses located at sampling instants nTs, n = 1.2.3….., having
amplitude level equal to sample value sn at that instant, as shown
in Figure below.
Sample and Hold Circuit for Signal Recovery:
 • In natural sampling or flat-top sampling, the spectrum of the sampled
signal is scaled by Tb/Ts, where Tb is the duration of the sampling-
pulse, and Ts is the sampling period.
 •Typically, the Tb/Ts ratio is quite small.
 • A simple sample-and-hold circuit is used for signal reconstruction.
Figure below shows a general concept of sample-and-hold circuit for
recovery of original analog signal from its sampled signal.
 • Sample-and-hold circuit generally comprises of an amplifier of unity
gain with low output impedance, an electronic switch, and a capacitor
with an assumption that the load impedance is large.
 • The electronic switch is timed precisely to close only for the small
duration Tb of each sampling pulse.
 • During this interval, the capacitor rapidly charges up to a voltage
level equal to that of the input sample
 • When the switch is open, the capacitor retains its voltage level until the
next closure of the switch happens.
 Sample-and-hold circuit, in its ideal form, produces an output waveform that
represents an approximation of the original analog information signal.
 PAM is simple to generate and detect.
 Due to variation in the amplitude of PAM signal, peak
transmitted power does not remain constant.
 The effect of additive white Gaussian noise on PAM signal is
maximum because the amplitude variations of the PAM pulses
represents the information too.
 Transmission bandwidth required for a PAM signal is quite large
as compared to highest frequency component present in the
input analog information signal.
 Pulse Width Modulation (PWM):
When the width of relatively higher frequency carrier signal comprising
of discrete pulses is varied in accordance with the amplitude of the analog
modulating signal, it is called pulse width modulation (PWM).
 • The amplitude of the pulse remains constant and does not carry any
information.
 • PWM provides better noise immunity and allows use of amplitude limiters.
 • It is also known as pulse duration modulation (PDM).
 • In pulse width modulation, the information signal modulates the width of
the pulse signal according to its instantaneous value while keeping the
amplitude and position of the pulse constant.
 • The larger the sample value, the wider the corresponding pulse and vice
versa.
 • PWM is a non-linear form of modulation as compared to PAM which is
linear form of modulation.
 • If the information signal is slowly varying, that is, sampled at a fast rate
compared to the Nyquist rate, and then the adjacent pulses have almost the
same width.
Advantages & Disadvantages:
PWM has certain distinct advantages such as
 It is easier to detect.
 It has very good noise immunity.
 There is no need of synchronization.

PWM has some disadvantages such as

 Its signal power varies due to variable pulse width.
 The transmission bandwidth required for a PWM signal is larger than
that of PAM signal.
 Due to the randomness of the width, it is not suitable for time-division
multiplexing.
The concept of PWM is mostly used in the design of switching-mode power supply
(SMPS). Due to this, PWM finds application in motor control circuits and robots
which requires regulated power supply with precision.
Generation of PWM signal:

 • The information signal is applied to the non-inverting (+) input terminal of

the comparator device (IC 710 or equivalent).
 • A sawtooth signal operating at the carrier frequency is applied to the
inverting (–) input of the comparator.
 • The maximum amplitude level of the information signal should be less than
that of the sawtooth signal.
 • When sawtooth signal is at its minimum level, the non-inverting input of
the comparator is at higher level and the output of the comparator is
positive.
 • When the sawtooth signal rises with a constant slope and crosses input
signal level, the inverting input of the comparator is at higher level and the
output of the comparator will be negative.
 • Therefore, the duration or the width of the output pulse generated, for
which the output of the comparator remains high is dependent on amplitude
of the information signal.
The transmission bandwidth of PWM signal is given by

 Where tr is the rise time in seconds which should be very much

smaller than the time period of the sawtooth signal.
Demodulation of PWM signal:
 The principle of operation of demodulation of PWM signal can
be simply described by averaging of PWM signals, followed by an
averaging low-pass filter.
 A ramp signal is started at the positive edge and stopped at the
negative edge of PWM signal. This operation is called time-
averaging of the incoming signal.
 The resultant ramp signal will have different heights in each cycle
depending on the corresponding pulse width of incoming PWM
signal.
 Thus, the output is directly proportional to the pulse width
which in turn represents the amplitude of the information signal.
 The output of low-pass filter will follow the envelope which is
basically approximate replica of the information signal.
 Pulse Position Modulation:
When position of a relatively higher frequency carrier signal
comprising of discrete pulses is varied in accordance with the
amplitude of the analog modulating signal, it is called Pulse
Position Modulation (PPM).

 In PPM, the position of the pulses which is changed with respect

to position of reference pulse.
 The amplitude and width of the pulses remains unchanged and
does not carry any information.
 PPM provides better noise immunity and allows use of amplitude
limiters.
PPM has certain distinct advantages such as
 It is easier to detect.
 It has very good noise.
 PPM signal power is constant.

PPM has some disadvantages such as

 PPM signals do need synchronization.
 The transmission bandwidth required for a PPM signal is also
quite large.
 The randomness in position in PPM do not make it suitable for
time-division multiplexing.
 The relationship between PPM and PWM can be seen from the fact that
while the position of the pulse varies in PPM, the location of the trailing edge
of the pulse varies in PWM.
 The PWM signal is applied to an inverter which reverses polarity of the
input pulse signals.
 The output of the inverter is then applied to a differentiator.
 When the PWM signal input make transition from high to low, the output of
differentiator consists of positive spikes.
 When the PWM signal input make transition from low to high, the output of
differentiator consists of negative spikes.
 These spikes are fed to a positive edge triggered fixed-width pulse generator.
 Pulses of fixed width are generated when a positive spike appears at its input.
 The occurrences of falling edges of PWM signal are proportional to
amplitude of the original information signal.
 Therefore, the delay in occurrence of these fixed width pulses are
proportional to the amplitude of the input information signal at that instant.
 The final output is PPM signal where positions of the pulses carry original
information signal.
Demodulation of PPM signal:
 A PPM demodulator circuit can be realized using transistor and
RC combination for ramp generation and filtering.
 • As an alternate arrangement for PPM demodulator is to
convert PPM signal to PWM signal, followed by use of PWM
demodulator to recover the information signal.
 • In PPM demodulator, the ramp signal starts at one positive
edge of the pulse and stops at the positive edge of the next pulse.
 • Thus the gap between positive edges of two successive pulses
received in PPM signal determines the height of the ramp signal
generated.
 • This, in turn, closely follows the amplitude of the information
signal.
 • This signal is then applied to a low-pass filter which filters out
the envelope information as demodulated signal.
 Pulse Code Modulation
Pulse code modulation is a type of signal encoding technique in which the
analog information signal is sampled and the amplitude of each sample is
approximated to the nearest one of a finite set of discrete levels, so that both
amplitude and time are represented in discrete form.
 With PCM, the continuously-varying analog signals are
converted into pulses of fixed amplitude and fixed duration.
 PCM is a binary system where a pulse or no pulse within a
prescribed time slot represents either a logic 1 or logic 0
condition.
PCM is used in digital telephone systems (trunk lines) and is also the
standard form for digital audio in computers and various compact disc
formats, digital videos, etc. PCM is the preferred method of communications
within Public Switched Telephone Network (PSTN) because with PCM it is
easy to combine digitized voice and digital data into a single, high-speed
digital signal and transmit it over either coaxial or optical fiber cables.
The operation of PCM system is briefly described below.
 • The band-pass filter (BPF) limits the frequency of the input
analog information signal to standard voice-band frequency range
of 300 Hz – 3000 Hz.
 • The sample-and-hold circuit periodically samples the analog
input signal and converts these samples to a multi-level PAM
signal.
 • The analog-to-digital (A/D) converter converts the PAM
samples to parallel PCM codes.
 • PCM codes are converted to serial binary data in the parallel-
to-serial converter and then presented to the transmission line as
serial digital pulses.
An analog signal is converted to a pulse code modulated signal through
three essential processes-sampling, quantization and encoding.
• The transmission line repeaters are placed at prescribed distances
to regenerate the digital pulses and enable to remove interference,
if any, due to channel noise.
• In the PCM receiver, the serial-to-parallel converter converts
serial pulses received from the transmission line to parallel PCM
codes.
• The digital-to-analog (D/A) converter generates sequence of
quantized multi-level sampled pulses, resulting in reconstituted
PAM signal.
• The hold circuit is basically a low-pass filter (LPF) to reject any
frequency component lying outside its baseband.
• It converts the recovered PAM signal back to its original analog
form.
How PCM is different from PAM??
 The combined operation of sampling and quantizing generate a
quantized PAM waveform.
 In fact, PAM is a train of pulses whose amplitudes are restricted
to a number of discrete magnitudes. So no further encoding is
done in PAM.
 In PCM, each quantized level is represented by equivalent
number of bits and the code-word thus formed is transmitted.
 The output of PCM is in the coded digital form having digital
pulses of constant amplitude, width and position.
PCM Encoding and Coding Efficiency:
 PCM encoding is a process to translate the discrete set of sample
values into a particular arrangement of discrete events, referred
to as a code.
 • This particular arrangement of symbols (comprising of binary
logic 0 or 1 in a code) to represent a single value of the discrete
set is called a codeword.
 • As an instance, in a binary code, each codeword consisting of m
bits will represent total 2m distinct discrete sample values.
 • That is, for m = 4 bits, there will be 16 distinct discrete sample
values, ranging from 0000 to 1111.
 Thus, the number of discrete levels is given by M= 2m
 Where M is the number of levels, and m is the number of bits
per sample
 The most important feature of PCM systems lies in its ability to
control the effects of channel noise and distortion. This is made
possible by the regenerative repeaters that reconstruct the PCM
waveforms.
There are three basic functions performed by a regenerative
repeater:
• Equalization Amplitude and phase distortions in PCM signals are
introduced due to imperfections in the transmission characteristics
of the channel. The equalizer reshapes the received distorted
pulses.
• Timing The timing circuit derives a periodic pulse train from the
received pulses. These are used for sampling the equalized pulses at
that instant of time where SNR is maximum.
• Decision making The decision-making device is enabled when the
amplitude of the equalized pulse plus noise exceeds a
predetermined threshold value.
These pulses are then regrouped into predefined code-words as
regenerated PCM signal.
Benefits of regenerative repeater:
 • The accumulation of noise and distortion is completely
removed provided it is not too severe to cause an error in
decision-making process.
 • Delay as well as jitter is also minimized into the regenerated
pulse.
 • Digital regenerators sample noisy signals and then reproduce an
entirely new digital signal with same SNR (generally 10–12 dB)
as original transmitted signal.
Received Signal Reconstruction:
• The received pulse is regenerated in a similar way as done in one
of the regenerative repeaters used in the transmission path.
• The decoding process involves generating a pulse whose
amplitude is the linear sum of all the pulses in the codeword.
• The decoder output is passed through a low-pass reconstruction
filter whose cut off frequency is equal to the information signal
bandwidth.
• An equalizer is used to correct for the aperture effect which
occurs due to flat-top sampling in the sample-and-hold circuit.
• Assuming error-free transmission path, the recovered analog
signal may contain only quantization error introduced by the
quantization process in the PCM transmitter.
Transmission Bandwidth of PCM:
The signaling rate in PCM transmission is the multiplication of the
number of bits per sample and the number of samples per second.
That is,
 Where, fb is the PCM transmission signaling rate in bps, m is the
number of bits per sample, and fs is the sampling rate or number
of samples per second.
 By definition, the transmission bandwidth in PCM system should
be greater than or equal to half of the signaling rate.That is,
 • This expression clearly shows that increasing the number of bits
per sample increases the transmission bandwidth in PCM.
 • The actual transmission bandwidth will be slightly higher than
calculated above.
 • In practical PCM transmission systems, additional bits would be
needed to detect and correct errors as well as to ensure
synchronization between transmitter and receiver, which would
further increase the effective transmission bandwidth.
 Quantization of Signal
The conversion of an analog sample of the information signal into
discrete form is known as quantization. Thus, an infinite number of
possible levels are converted to a finite number of conditions.

 • The peak-to-peak range of the input sample values is subdivided

into a finite set of decision levels or decision thresholds.
 • The output is assigned a discrete value selected from a finite set
of representation levels.
 • A quantizer is memory less in the sense that the quantizer
output is determined only by the value of a corresponding input
sample only.
 Quantization Interval: Quantization interval is defined as the
difference in magnitude levels between adjacent steps.
 Quantum Overload Distortion: Quantum overload distortion occurs
when the magnitude of the sample exceeds the highest
quantization interval.
 Resolution: The resolution is defined as the voltage level of the
minimum step-size, which is equal to the voltage of the least
significant bit of PCM code.
 • It is also the minimum voltage (other than 0 V) that can be
decoded by digital-to-analog converter in the PCM receiver.
 • The process of quantization involves rounding off sample values of an
analog signal to the nearest permissible level of the quantizer.
 • As a result, the quantization error is produced.
Quantization error is defined as the difference between
rounding off sample values of an analog signal to the nearest
permissible level of the quantizer during the process of
quantization.
 • The quantization error is directly proportional to the difference
between consecutive quantization levels.
 • The quantization error is inversely proportional to the number
of levels for amplitude range.
With a higher number of quantization levels, a lower quantization error is
obtained. The performance of a quantizer is measured as the output signal-
to-quantization noise ratio.
Classification of Quantization Process:
 Depending upon the process of quantization to distribute
quantization levels to be uniformly spaced or not, the
quantization can be categorized as
 • Uniform quantization
 • Non-uniform quantization
A uniform quantizer is a quantizer in which the quantization levels (step
size) are uniformly spaced over the complete input range.
 • It has linear characteristics.
 • The maximum quantization error remains same over the
complete input range.
 • The signal-to-noise ratio does not remain same over the
complete input range.
A non-uniform quantizer is a quantizer in which the quantization levels (step-
size) vary according to the instantaneous value of the input signal level.
 • It has nonlinear characteristics.
 • The step size varies according to the signal level to maintain the
signal-to-noise ratio adequately high.
Uniform quantization:
The process of quantizing a discrete signal has a two-fold effect:
 • The peak-to-peak range of input samples is subdivided into a finite
set of decision levels.
 • The output is assigned a discrete value selected from a finite set of
reconstruction values that are aligned with the treads of the staircase.
Depending on the position of the origin, the uniform quantizer are of
two types:
 • Midtread Uniform Quantizer The origin lies in the middle of a tread of
the staircase type of quantization process.
 • Midriser Uniform Quantizer The origin lies in the middle of a rise of the
staircase type of quantization process.
Midtread Uniform Quantizer:
 • The separation between the decision thresholds and the
separation between the representation levels of the quantizer
have a common value referred to the stepsize.
 • If decision thresholds of the quantizer are located at 0, ±∆,
±2∆, …….., and the representation levels are located at ± ∆/2,
± 3∆/2, ± 5∆/2, ……., (∆ is the step size), the origin lies in
the middle of a riser of the staircase.
 • Midtread quantizer has odd number of quantized levels 2m – 1,
where m is the number of bits required to encode a sample.
Midriser Uniform Quantizer:

• Midrise quantizer has even number of quantized levels 2m – 1, where

m is the number of bits required to encode a sample.
Overload level
Overload level is defined as one half of the peak to peak range of the input
sample values.
The number of intervals into which the peak-to-peak excursion is
divided, is equal to twice the absolute value of the overload level
divided by the step size.

Idle channel noise

Idle channel noise is the coding noise measured at the PCM
receiver output with zero input analog signals at the PCM
transmitter end.
 • In practice, the zero input analog signal condition occurs during
silence periods in the speech.
 • The average power of idle channel noise depends on type of
quantizer used.
 • In a quantizer of the midtread type, the output is zero for zero
input analog signal level, and the idle channel is correspondingly
zero.
 • In a quantizer of the midriser type, zero input analog signal
level is encoded into one of the two innermost levels ± ∆/2.
 • Assuming that these two representation levels are equiprobable,
the idle channel noise for midriser quantizer has zero mean and
an average power of ∆2/4.
Dynamic Range
The Dynamic range of the PCM system is the ratio of the strongest possible
signal amplitude level to the weakest possible signal level (other than 0 V)
which can be decoded by the digital-to-analog converter in the receiver.

where Vmax is the maximum value that can be discerned by the

digital-to-analog converter in the receiver, Vmin is the minimum
value (also called the quantum value or resolution).
The number of bits used for a PCM code depends on the dynamic
range which are related as DR ≤(2m – 1) in case of midtread
uniform quantizer; where m is the number of bits in a PCM code
(excluding the sign bit, if any).
One positive and one negative PCM code is used for 0 V, which is
not considered for determining the dynamic range.
Expressing dynamic range in dB, we have
For values of m > 4, dynamic range can be approximated as

 For a linear PCM system, the maximum dynamic range is given

approximately by (ignoring any noise present in the analog signal
itself)

 The minimum number of bits required to achieve the specified

value of dynamic range is given by
Non-uniform Quantization:
In non-uniform quantization, the spacing between the quantization
levels is not uniform and step size varies in accordance with the
relative amplitude level of the sampled value.
 In the use of PCM for the transmission of speech signals, the
quantizer has to accommodate input signals with widely varying
power levels.
 For example, the range of voltage amplitude levels covered by
normal speech signals, from the peaks of loud speech levels to
the lows of weak speech levels, is on the order of 1000 to 1.
 So it is highly desirable that signal-to-quantization noise ratio
should remain essentially constant for a wide range of input
signal levels.
Necessity of Non-uniform Quantization:
 The largest possible quantization error is one-half the difference
between successive levels.
 Thus, the quantization error is proportionally greater for small
signal levels.
 This means that the signal-to-noise ratio varies with the signal
level and is higher for large signal levels.
 The amount of quantization error can be decreased by increasing
the number of levels, but it also increases the number of bits
required per sample.
 The only solution to have a constant signal-to-quantization noise
ratio is to adjust the step size in accordance with the input signal
amplitude levels.This is non-uniform quantization.
Advantages of Non-uniform Quantization:
 • High Average SNR value Non-uniform quantization has higher
average signal to quantization noise power ratio value than that of
in the uniform quantizer.
 • Reduced Quantization Noise RMS value of the quantizer noise
power of a non-uniform quantizer is substantially proportional to
the sampled value and hence the quantization noise is also
reduced.
Robust Quantization or Companding:
 A quantizer whose SNR remains essentially constant for a wide range
of input power levels is said to be robust.
 This necessitates that the stepsize must be small for low amplitude
signals and large for high amplitude signals.
 The provision for such robust performance necessitates the use of a
non-uniform quantizer.
 The non-uniform quantization technique employs an additional
logarithmic amplifier before processing the sampled speech signals by
a uniform quantizer.
 The operation of a non-uniform quantizer is equivalent to passing the
analog signal through a compressor and then applying the compressed
signal to a uniform quantizer at transmitter end.
 At the receiver, a device with a characteristic complementary to the
compressor, called expander is used to restore the signal samples to
their correct relative level.
 The combination of a compressor and an expander is called
a compander.
Companding is the process of compressing signal at transmitter end
and expanding the signal at receiver end to achieve non-uniform
quantization.

 Non-uniform quantization at PCM transmitter is achieved

by first distorting the original signal with a compression
characteristics which is logarithmic (as compared to linear
characteristics for no compression), as shown in below.
 For small magnitude analog signals the compression
characteristics has a much steeper slope than for large magnitude
analog signals.
 This implies that for a given change at small magnitudes of the
input signal will move the uniform quantizer through more steps
than the same change at large magnitudes of the input signal.
 Thus, the compression characteristic effectively changes the
distribution of the input signal magnitudes.
 The output of the compressor is then applied to a uniform
quantizer whose input-output characteristics are shown in Figure
7.40.
 At the receiver, an inverse compression characteristic is applied
in order to retrieve the non-distorted original analog signal at its
output.
 The input-output characteristics of expander used in the receiver
is shown in Figure 7.41.
𝝁-Law Companding:
 In the 𝜇-law companding, the compressor characteristics are
continuous, approximating a linear dependence for low input
levels and a logarithmic one for high input levels. The
compression characteristics for 𝜇 -law is given as

 where Vout is the compressed output amplitude level in volts,

 Vmax is the maximum uncompressed analog input amplitude
level in volts,
 𝜇 is the parameter used to define the amount of compression
(unitless),
 Vin is the amplitude of input signal at a particular instant of time
in volts
 Figure 7.42 shows the relationship between Vout and Vin for
different values of 𝜇.
 • The value 𝜇 = 0 corresponds to uniform quantization.
 • For a relatively constant signal-to-quantization ratio and a 40
dB dynamic range, the value of 𝜇 ≥ 100 is required.
 • The practical value of m is approximately 255.
 Figure 7.43 shows the comparison of S/N versus signal level for
both cases – without companding and with μ-law companding.
 It is observed that the signal-to-noise (S/N) ratio of PCM system
remains constant with companding.
A-law Companding:
The compression characteristics for A-law is given as

where Vout is the compressed output amplitude level in volts

Vmax is the maximum uncompressed analog input amplitude level in
volts
A is the parameter used to define the amount of compression (unit less)
Vin is the amplitude of input signal at a particular instant of time in volts
Figure 7.44 shows the relationship between Vout and Vin for different
values of A.
•The value A = 1 corresponds to uniform quantization.
•The practical value of A is approximately 100.
Give step-by-step procedure for μ-law and A-law standard companding processes for
256 quantization levels. [10 Marks]
Solution:
(1) Create the logarithmic curves for the input signal during the compression part of
companding μ-law and A-law.
(2) Calculate the linear approximation of the logarithmic curve. μ-law and A-law
calculate the linear approximation differently.
(3) Since given number of quantization levels is 256, divide the curve into 8 positive
and 8 negative segments, with 16 quantization intervals per segment.
(4) Because of logarithmic increase, each successive segment is twice the length of the
previous segment. μ-law and A-law have different segment lengths because of different
calculations of linear approximations.
(5) Use 8 bit code words for each quantization interval, thereby resulting into 256 code
words.
• The first bit of each code word represents the polarity of quantization interval.
• The next three bits represent the segment number.
• Last four bits indicate the quantization interval within the segment.
(6) The bit rate can be calculated by multiplying the sampling rate (twice the input
frequency) by the size of the codeword. For example, 8 bit code words allow for a bit
rate of 64 kbps corresponding to 4 kHz signal.