AT&T 518-600-016 manual PCM uses PAM as its starting point but goes further

Quantizing

Commanding

PCM uses PAM as its starting point but goes further.

PCM digitizes the information and sends the 8-bit code via PAM. PCM digitizes the PAM information by comparing each PAM sample to a point on a chart. Each point has a distinct 8-digit binary code. Therefore, it is the binary code, not the amplitude pulse, that is transmitted in PCM.

The process of converting PAM samples into discrete PCM values is known as “quantizing”. In the case of voice grade signals eight bits are commonly used to encode one PAM sample. Quantizing with eight bits allows 256 unique values to represent the analog signal being sampled. With adequate sampling of an analog voice signal, the digital product can be toll quality.

The quantizing process creates errors due in part to the “rounding off” of PAM samples into discrete PCM values. This distortion can be minimized by increasing the number of bits used to sample the analog signal or to increase the sampling rate.

Distortion also occurs in the quantizing process because equal consideration is given to all amplitude levels. Linear quantizing divides the amplitude range into equal segments which results in unnecessary quality at levels where voice signals are unlikely to occur. Commanding eliminates this distortion by using a nonlinear scale which has smaller divisions and more accuracy in areas of the scale where voice signals are most likely to occur.

Commanding is, in effect, a compression/expansion of the quantizing scale. This nonlinear scale is more accurate in areas where signal amplitudes occur most (thereby reducing quantizing distortion). The scale is less accurate in areas where signal amplitudes occur least (resulting in more quantizing distortion).

Theory of Operation 1-41