34 



THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1951 



sampling frequency it is also desirable to band limit the input signal to avoid 

 undesirable distortion products due to extraneous frequency components 

 which may be present in the original wave. For a nominal 5 mc television 

 channel the sampling rate used in these experiments was 10 mc per second 

 and the input and output filters passed components of 4.3 mc and atten- 

 uated components of 5.0 mc. The sampling process produces a discrete 

 number of samples to be transmitted. For the present case this number is 

 10 million per second. 



QUANTIZED 

 NUMBER 



BINARY NUMBER 



WEIGHTED EQUIVALENT 



I I 1 1 1 I 1 I 1 ; 



1 I 1 ; 1 I 



: 1 1 1 ! 1 



16 I 8 1 4 i 2 I 1 



16 I I i 2 I i 



I I I 2 I 1 



DECODED 

 NUMBER 



= 31 



= 18 



J=l =3 



0;0i0]0;0j lojolololo 



Fig. 1 — Five-digit code groups. 



Each of these samples may have any value in a continuous range between 

 and a maximum value set by the amplitude range which the system is 

 designed to transmit. 



In binary PCM each of these amplitudes is transmitted by a code group 

 of binary digits. As an example, consider a five-digit code which is illustrated 

 in the second column of Fig. 1. Here we have five digits or on-and-off pulses. 

 The maximum number of values that can be represented by these five two- 

 position pulses is 2^ or 32 values. Examples shown are for amplitudes of 31, 

 18, 3 and 0. It is easy to see that any other integer value greater than and 

 less than 31 can also be represented by one of the combinations of pulses 

 and spaces. It is also apparent that when all the combinations have been 

 used up ne other values can be obtained. 



