Quantization Process

The operation of quantization is represented in Fig. 5.1. Here we contemplate a

signal m(t) whose excursion is confined to the range from V L to V H. We have
divided this total range into M equal intervals each of size S. Accordingly S,
called the step size, is S = (V H - V d/M. In Fig. 5.7-1 we show the specific
example in which M = 8. In the center of each of these steps we locate
quantization levels m0 , m 1, . .. , m7 • The quantized signal mq(t) is generated
in the following way: Whenever m(t) is in the range .1 0 , the signal mq(l)
maintains the constant level m 0 ; whenever m(t) is in the range .1 1 , mq(t)
maintains the constant level m 1 ; and so on. Thus the signal mq(t) will at all
times be found at one of the levels m0,m 1 , ••• , m7 . The transition in mq(t)
from mq(t) = m0 to mq(t) = m 1 is made abrupt- ly when m(t) passes the
transition level L 01 which is midway between m0 and m 1 and so on . To state
the matter in an alternative fashion, we say that, at every
instant of time, mq(t) has the value of the quantization level to which m(t) is
closest. Thus the signal mq(t) does not change at all with time or it makes a
"quantum" jump of step size S. Note the disposition of the quantization levels in
the range from V L to V H . These levels are each separated by an amount S,
but the separation of the extremes V Land V H each from its nearest
quantization level is only S/ 2. Also, at every instant of time, the quantization
error m(t) - mq(t) has a magnitude which is equal to or less than S/2.

We see, therefore, that the quantized signal is an approximation to the original

signal. The quality of the approximation may be improved by reducing the size
of the steps, thereby increasing the number of allowable levels. Eventually, with
small enough steps, the human ear or the eye will not be able to distinguish the
original from the quantized signal. To give the reader an idea of the number of
quantization levels required in a practical system, we note that 256 levels can be
used to obtain the quality of commercial color TV, while 64 levels give only
fairly good color TV performance. These results are also found to be valid when
quantizing voice.

How Quantization helps to eliminate additive noise.

The limitation of the system we have been describing for communicating over
long channels is that once noise has been introduced any place along the
channel, its difficult to remove noise.

However, the situation is modified by subjecting a signal to the operation of

quantization. When quantizing a signal m(t), we create a new signal mq(t)
which is an approximation to m(t). However, the quantized signal mq(t) has the
great merit that it is, in large measure, separable from the additive noise.

Now let us consider that our quantized signal has arrived at a repeater
somewhat attenuated and corrupted by noise. This time our repeater consists
of a quantizer and an amplifier. There is noise superimposed on the quantized
levels of mq(t). But suppose that we have placed the repeater at a point on the
communications channel where the instantaneous noise voltage is almost
always less than half the separation between quantized levels. Then the output
of the quantizer will consist of a succession of levels duplicating the original
In Uniform type, the quantization levels are uniformly spaced, whereas in non-
uniform type the spacing between the levels will be unequal and mostly the
relation is logarithmic.
Types of Quantizers:
1. Uniform Quantizer
2. Non- Uniform Quantizer
Types of Uniform Quantizers: (based on I/P - O/P Characteristics)
1. Mid-Rise type Quantizer
2. Mid-Tread type Quantizer
In the stair case like graph, the origin lies the middle of the tread portion in Mid
–Tread type whereas the origin lies in the middle of the rise portion in the Mid-
Rise type.
Mid – tread type: Quantization levels – odd number.
Mid – Rise type: Quantization levels – even number.
 • The use of a nonuniform quantizer is equivalent to passing the message signal
through a compressor and then applying the compressed signal to a uniform
quantizer. A particular form of compression law that is used in practice is the so
called μ-law.
• We see therefore that the μ-law is neither strictly linear nor strictly logarithmic,
but it is approximately linear at low input levels corresponding to μ|m| << 1, and
approximately logarithmic at high input levels corresponding to μ|m| >> 1.

We must, of course, use a subsystem in the receiver with a characteristic complementary

to the compressor, used in the transmitter. Such a subsystem is called an expander. Ideally,
the compression and expansion laws are exactly inverse so that, except for the effect of
quantization, the expander output is equal to the compressor input if these two devices
were connected directly. The combination of a compressor and an expander is referred to
as a compander.
The use of a nonuniform quantizer is equivalent to passing the baseband signal through a compressor and
then applying the compressed signal to a uniform quantizer, followed by expander

There are two types of compression laws used in practice:

1. 𝜇 law

log⁡(1 + 𝜇|𝑚|)
|𝑣| =
log⁡(1 + 𝜇)

𝑚 and 𝑣 are the normalised input and output voltages, 𝜇 is the positive constant

The derivative of |𝑚| with respect to |𝑣| is given by

𝑑|𝑚| log(1 + 𝜇)
= (1 + 𝜇|𝑚|)
𝑑|𝑣| 𝜇

𝜇 law is neither strictly linear nor strictly logarithmic

2. A law

A|𝑚|) 1
⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡0 ≤ |𝑚| ≤
1 + 𝑙𝑜𝑔𝐴 𝐴
|𝑣| =
1 + log⁡(𝐴|𝑚|)
⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡1/𝐴 ≤ |𝑚| ≤ 1
{ 1 + log⁡(𝐴)

𝑚 and 𝑣 are the normalised input and output voltages, 𝜇 is the positive constant

The derivative of |𝑚| with respect to |𝑣| is given by

1 + log 𝐴 1
𝑑|𝑚| ⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡⁡0 ≤ |𝑚| ≤
={ 𝐴 𝐴
𝑑|𝑣|
(1 + log(𝐴) |𝑚|⁡⁡⁡⁡⁡⁡⁡⁡⁡1/𝐴 ≤ |𝑚| ≤ 1

In order to restore the signal samples to their correct relative level, an expander is used. Compression
and expansion are inverse. Combination of compressor and expander is called Compander.
Decision circuit: It compares the signal sample with the predetermined threshold, if the
sample is found to be greater than threshold then the PCM pulse is reconstructed as ‘1’, else
if the sample is found to be lesser than threshold then the PCM pulse is reconstructed as ‘0’
Comparator: It computes the difference between 𝑚(𝑛𝑇𝑠 ) and 𝑚𝑞 (𝑛𝑇𝑠 )

Quantizer: It consists of a hard limiter, whose input-output relation is a sgn function

Accumulator: Depending on 𝑒(𝑛𝑇𝑠 ), Accumulator approximates the value to 𝑒𝑞 (𝑛𝑇𝑠 ) either +∆ or −∆

𝑒𝑞 (𝑛𝑇𝑠 ) = ∆𝑠𝑔𝑛[ 𝑒(𝑛𝑇𝑠 )]

𝑚𝑞 (𝑛𝑇𝑠 ) = 𝑚𝑞 (𝑛𝑇𝑠 − 𝑇𝑠 ) + 𝑒𝑞 (𝑛𝑇𝑠 )

The quantized signal then undergoes encoding to create a Delta Modulated wave
At the receiver, Delta Modulated wave is decoded, the positive and negative pulses are passed through the
Accumulator to produce a staircase approximation 𝑚𝑞 (𝑛𝑇𝑠 ). The LPF removes the out of band noise

Delta Modulation suffers from two types of quantization noise:

(1) Slope Overload Distortion

If the quantization error is 𝑞(𝑛𝑇𝑠), then 𝑚𝑞 (𝑛𝑇𝑠 ) = 𝑚(𝑛𝑇𝑠 ) + 𝑞(𝑛𝑇𝑠 )

Substituting in the equation for 𝑒(𝑛𝑇𝑠 ),

𝑒(𝑛𝑇𝑠 ) = 𝑚(𝑛𝑇𝑠 ) − 𝑚(𝑛𝑇𝑠 − 𝑇𝑠 ) − 𝑞(𝑛𝑇𝑠 )

Thus, except for the quantization error 𝑞(𝑛𝑇𝑠 ), the quantizer input is a first backward difference of the input
signal, which equivalent to taking derivative of the input signal or taking slope of the signal. Thus, if the ∆
value is small, especially in case of signals with high rise then, the resulting waveform resembles:

(2) Granular Noise:

This occurs when the step-size ∆ is too large relative to the local slope characteristics of the input waveform
𝑚(𝑡), thereby the staircase approximation no longer complies with the original input signal.
There is a need to have a large step-size to accommodate a wide dynamic range, whereas a small step-size is
required for the low-level Slope-overload distortion signals. It is therefore clear that the choice of the
optimum step-size that minimizes the mean-square value of the quantization Hence, the step size must be
adaptive.

Delta Sigma Modulation:

Delta Modulation is an approximation to the derivative of the incoming message signal, to overcome this
drawback the message signal is integrated prior to delta modulation. The use of integration in the has the
following beneficial effects:

i. The low-frequency content of the input signal is pre-emphasized.

ii. Correlation between adjacent samples of the delta modulator input is increased
iii. Design of the receiver is simplified

A delta modulation scheme that includes integration at its input is called Delta-Sigma modulation (𝐷 − ΣM).
Figure A shows the block diagram of a delta-sigma modulation system. In this diagram, the message signal
𝑚(𝑡) is defined in continuous-time form, hence the pulse modulator consists of a hard-limiter followed by a
multiplier. Pulse modulator is fed from an external pulse generator (clock) to produce a 1-bit encoded.

The reason for investigating delta modulation is its reduced bandwidth requirements compared to PCM.
Delta modulation has 50-75 percent lesser bandwidth compared to PCM.