76 



THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1956 



Fig. 7 — Effect of regenerating in amplitude without retiming. A — Outputof 

 regenerator, no timing, firt trip. B — Output of regenerator, no timing, 10th trip. 

 Output of regenerator, no timing, 23rd trip. 



when no gating pulse is present and hence finds the gate blocked regard- 

 less of amplitude. 



To show the need for retiming the pictures shown on Figs. 7 and 8 

 were taken. These were taken with the amplitude slicer in operation but 

 with the pulses not being retimed. Figs. 7A, 7B and 7C, respectively, 

 show the output of the slicer for the first, tenth and twenty-third trips. 

 After ten trips, there is noticeable time jitter caused by residual noise 

 in the system; after 23 trips this jitter has become severe though pulses 

 are still recognizable. It should be pointed out that for this experiment 

 no noise was purposely added to the system and hence the signal-to- 

 noise ratio was much better than that which would probably be encoun- 

 tered in an operating system. For such a system we would expect time 

 jitter effects to build up much more rapidly. For Fig. 8 conditions are 

 the same as for Fig. 7 except that the pulse spacing is decreased by the 

 addition of an extra pulse at the input. Now, after ten trips, time jitter 

 is bad and after 23 trips the pulse group has become little more than a 

 smear. This increased distortion is probably due to the fact that less 

 jitter is now required to cause overlap of pulses. There may also be some 

 effects due to change of duty cycle. For Fig. 9 there was neither slicing 

 nor retiming of pulses. Here, pulse groups deteriorate very rapidly to 

 nothing more than blobs of energy. Note that there is an increase of 



i 



