In an ADC, over sampling is a technique used to

To reduce the Quantization noise two things are done.

Oversampling and Noise shaping.

Explanation:

• An analog signal first undergoes the process of sampling before it is applied to ADC for conversion into a digital signal.
• Oversampling is a process in which an analog signal is sampled at a sampling frequency that is much greater than the Nyquist rate.

• Oversampling is:

  fs' = K × fs , ( K > 1)

• A Sigma-Delta ADC is an example of an ADC that employs oversampling.
• It is the same as “Delta modulation”. That is 1 sample is represented with 1 bit.
• In oversampling correlation increases and Quantization noise decreases.

Delta Modulation: 

A signal is Delta modulated to convert its analog information into a binary sequence, i.e., 1s and 0s. Here only one bit is required to represent the analog input. An over-sampled input is taken to make full use of the signal correlation. A stair-case approximated waveform will be the output of the delta modulator with the step-size as delta (Δ).

Delta Modulator has two main drawbacks:

1. Slope overload distortion: When the rate of rising of the input signal x(t) is so high that the staircase signal can not approximate it. We will see consecutive pulses of the same polarity in the output for such type of distortion. 

2. Granular noise: When the step size is too large compared to a small variation in the input signal (input is almost constant). Here, the output of the Delta modulator will be a sequence of alternate +ve and -ve pulses. 

∴ We can conclude that if the consecutive pulses are of opposite polarity during any time interval then the signal is almost constant during that interval.