TRANSATLANTIC CABLE POWER SYSTEM 



153 



50 



40 



in 30 



_J 



LU 



m 

 u 



LU 



Q 20 



z 

 < 



O 10 



180 



140 



(0 

 LU 

 LU 



tr 

 100 o 



LU 

 Q 



60 UJ 



z 

 < 



LU 



in 

 < 



I 



Q. 



20 



- 



-20 



-60 



01 0.02 0.05 0.1 0.2 0.5 1 2 5 10 20 50 100 200 500 



FREQUENCY IN KILOCYCLES PER SECOND 



Fig. 6 — DC amplifier gain and phase, experimental model. 



data were obtained by opening the feedback loop at the control grid of 

 the first stage, applying normal dc bias plus a A'ariable frequency ac 

 signal to the grid of the tube, and measiu'ing the magnitude and relative 

 phase of the return signal. 



The corresponding characteristics with the compensating network in 

 place are also shown in Fig. 6. The compensating network effectively 

 puts a relatively low-impedance shunt across the interstage network at 

 the higher frequencies, resulting in a "step" in the gain characteristic. 

 A secondary effect is the phase shift in the transition region. The calcu- 

 lated "corner frequencies" are 2,800 and 195 cps, respectively, chosen on 

 the basis of the criteria (1) little effect on regulator gain at 100 or 120 

 cps, the most prominent rectifier ripple frequency, and (2) a gain step 

 of something above 20 db with no appreciable contribution to the phase 

 shift at frequencies above 30 kc. The calculated loss at 120 cps is 1.2 db 

 with a maximum phase shift of about 60 degrees at the median frequency. 

 These results agree quite well with the measured data plotted in Fig. 6. 



As indicated in Fig. 6, the phase margin at the gain crossover frequency 

 of 55 kc was somewhat over 100 degrees for the experimental model on 

 which these measurements were made. The gain margin could not be 

 measured readily but is clearly substantial. On production units, larger 

 wire sizes and longer lead lengths resulted in lesser, but still satisfactory 

 stability margins, as shown in Fig. 7, the phase margin being somewhat 



