1072 THE BELL SYSTEM TECHNICAL JOURNAL, SEPTEMBER 1956 



(a) 



AJ^ 



(0 



(d) 



rsa 



(e) 



Tl^ 



Plate II — (a), repeater input, no interference; (b), regenerated output with 

 input (a); (c), sinusoidal interference; (d), repeater input, signal (a) plus sinusoi- 

 dal interference (c); (e), regenerated output of (d). 



3.1 Performance of Repeaters in Tandem 



Plate III shows the results when certain phase modulated codes are 

 transmitted through a series of repeaters in tandem. The regenerated 

 signal from each successive repeater is transmitted over 2.3 miles of 

 equalized 19 gauge line. One code which has two out of a possible eight 

 pulses present has most of the phase jitter removed after passing through 

 the three additional repeaters. The other fixed code shown contains four 

 out of a possible eight pulses. The jitter is removed much more rapidlj^ 

 with this code, after passing through two repeaters it is regenerated al- 

 most perfectly. The reason for the difference in the regeneration of the 

 two codes is variations in the amplitude of the timing wave. In any period 

 of time the energy delivered to the tank circuit is proportional to the 

 number of regenerated pulses in that interval. The amplitude of the 

 timing wave for a fixed code with two pulses of the eight will be half 

 the one produced by the code having four pulses out of eight present. 

 The average number of pulses in a normal PCM signal will be half the 

 maximum possible pulses. The timing wave should then average the 

 same as that produced by the fixed code having four out of a possible 

 eight pulses present. The phase jitter of the random code should be re- 

 moved as quickly as it was with this fixed code. This is confirmed by 



