72 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1956 



groups rather than individual pulses through the system. These were 

 derived from the pulse group generator which is capable of delivering 

 any number up to 5 pulses for each short input pulse. These pulses are 

 about 15 milli-microseconds long at the base and spaced 25 milli-micro- 

 seconds apart. The amplitude of each of these pulses can be adjusted 

 independently to any value from zero to full amplitude making it pos- 

 sible to set up any combination of the five pulses. These are the pulses 

 which are used to gate, or modulate, the output of the 4-kmc oscillator. 



The total delay around the waveguide loop including TW tubes, etc.,' 

 was 0.4)usec or 400 milli-microseconds. This was sufficient to allow time 

 between pulse groups and yet short enough that groups could circulate 

 24 times in the available 9.8jLtsec interval. This can be seen from Figs. 

 3C and 3D. The latter figure shows an expanded view of circulating 

 pulse groups. The pulses in Group 1 are inserted into the loop at the 

 beginning of each gating cycle, the remaining groups result from circu- 

 lation around the loop. 



When all five pulses are present in the pulse groups the pulse repeti- 

 tion frequency is 40 mc. (Pulse interval 25 milli-microseconds). For this 

 condition timing pulses should be supplied to the regenerator at the rate 

 of 40 million per second. These pulses are supplied continuously and not 

 in groups as is the case with the circulating pulses. See Fig. BE. In order 

 to maintain time coincidence between the circulating pulses and the tim- 

 ing pulses the delay around the loop must be adjusted to be an exact 

 multiple of the pulse spacing. In this experiment the loop delay is equal 

 to 16-pulse intervals. Since timing pulses are obtained by harmonic 

 generation from the quenching frequency as will be discussed later this 

 frequency must be an exact submultiple of pulse repetition frequency. 

 In this experiment the ratio is 512 to 1. 



Although the above discussion is based on a five-pulse group and 

 40-mc repetition frequency it turned out that for most of the experi- 

 ments described here it was preferable to drop out every other pulse, 

 leaving three to a group and resulting in a 20-mc repetition frequency. 

 The one exception to this is the limited-band-width experiment which 

 will be described later. - 



For all of the experiments described here timing pulses were derived 

 from the 78-kc quenching frequency by harmonic generation. A pulse 

 with a width of 25 milli-microseconds and with a 78-kc repetition fre- 

 quency as shown in Fig. 3B supplied the input to the timing wave gen- 

 erator. This generator consists of several stages of limiting amplifiers all 

 tuned to 20 mc, followed by a locked-in 20-mc oscillator. The output of 

 the amplifier consists of a train of 20-mc sine waves with constant ampli- 



til 



