REGENERATION OF BINARY MICROWAVE PULSES 



79 



Fig. 11 — The regeneration of band-limited pulses. A — Input to regenerator, 

 first two groups. B — Output of regenerator, first two groups. C — Output of re- 

 generator, 24th trip. 



that is for each trip the input to the regenerator was as shown at the right 

 of Fig. lOA and the output as shown by Fig. lOB. It is important to 

 note that Fig. 12 represents the frequency characteristic of a single hnk 

 of the simulated system. The pictures of Fig. 11 show the same experi- 

 ment but this time with a different code group. Any code group which 

 we could set up with our five digit pulses was transmitted equally well. 

 In order to determine the breaking point of the experimental system, 

 broad-band noise obtained from a traveling-wave amplifier was added 

 into the system as shown on Fig. 2. The breaking point of the system is 

 the noise level which is just sufficient to start producing errors at the 

 output of the system.* The noise is seen to be band-limited in exactly 

 the same way as the signal. With the system adjusted to operate properly 

 the level of added noise was increased to the point where errors became 

 barely discernible after 24 trips around the loop. Noise level was now 

 reduced slightly (no errors discernible) and the ratio of rms signal to rms 

 noise measured. Fig. 13A shows the input to the regenerator for the 23rd 

 and 24th trips with this amount of noise added. Note that the noise has 



* The type of noise employed has a Gaussian amplitude distribution and there- 

 fore there was actually no definite breaking point — the rate at which errors Oc- 

 curred increased continuously as noise amplitude was increased. The breaking 

 point was taken as the noise level at which errors became barely discernible on 

 the viewing oscilloscope. More accurate measurements made in other experiments 

 indicate that this is a fairly satisfactory criterion. 



