REGENERATION OF BINARY MICROWAVE PULSES 73 



tude for most of the 12.8Msec period but falling off somewhat at the end 

 of the period. This-train locks in the oscillator which oscillates at a con- 

 stant amplitude over the whole period and at a frequency of 20 mc. 

 Timing pulses obtained from the cathode circuit of the oscillator tube 

 pro^'ided the timing waves for most of the experiments. For the experi- 

 ment where a 40-mc timing wave was required it was obtained from the, 

 20 mc train by means of a frequency doubler. For this case it is necessary 

 for the output of the timing wave generator to remain constant in ampli- 

 tude and fixed in phase for the 512-pulse interval between synchronizing 

 pulses. 



In spite of the stringent requirements placed upon the timing equip- 

 ment it functioned well and maintained synchronism over adequately 

 long periods of time without adjustment. 



PERFORMANCE OF REGENERATOR 



Performance of the regenerator under various conditions is recorded 

 on the accompanying illustrations of recovered pulse envelopes. The 

 first experiment was to determine the effects of disturbances which arise 

 at only one point in a system. Such effects were simulated by adding 

 disturbances along with the group of pulses as they were fed into the 

 circulating loop from the modulator. This is equivalent to having them 

 occur at only the first repeater of the chain. 



Some of the first experiments also involved the use of extraneous 

 pulses to represent noise or distortion since these pulses could be syn- 

 chronized and thus studied more readily than could random effects. In 

 , Fig. 4A the first pulse at the left represents a desired digit pulse with 

 ' its amplitude increased by a burst of noise, the second pulse represents 

 ' a clean digit pulse, and the third pulse a burst of noise. This group is at 

 1 the input to the regenerator. Fig. 4B shows the same group of pulses 

 ' after traversing the regenerator once. The pulses are seen to be shortened 

 due to the gating, or retiming, action. There is also seen to be some ampli- 

 tude correction, i.e. the two desired pulses are of more nearly the same 

 j amplitude and the undesired pulse has been reduced in relative ampli- 

 tude. After a few trips through the regenerator the pulse group was 

 rendered practically perfect and remained so for the rest of the twenty- 

 four trips around the loop. Fig. 4C shows the group after 24 trips. In 

 'another experiment pulses were circulated for 100 trips without deteri- 

 oration. Nothing was found to indicate that regeneration could not be 

 repeated indefinitely. 

 Figs. 5 A and 5B represent the same conditions as those of 4 A and 4B 



