864 



THE BELL SYSTEM TECHNICAL JOURNAL, JULY 1956 



plotted the distribution predicted from Fig. 9(b). The RF losses in the 

 4,000-mc tube were not neghgible, and probably account for slightly 

 smaller power output and greater proportion of higher velocity electrons. 



COMPARISON WITH COMPUTED CURVES 



Non-linear calculations of traveling wave tube behavior have been 

 made by Tien, Walker and Wolontis' and by Tien^ covering the same 

 region of parameter values as is reported here. In Figs. 2, 3, 4 and 9 are 

 shown some of their data on our coordinates. The similarity of the 

 results over much of the range is rather reassuring. It is interesting that 

 in order to make the computations it was necessary to assume two space 

 charge factors, just as was found experimentally. There are, however, 

 some significant differences: 



1. In general, the computed values give a higher ^alue of efliciency 

 than is measured, by about 25 per cent. Thus, the computations indicate 



rwm(m^ 



^ 



Hill 



zero signal 

 characteristic: 



IDEAL ^>4 



MEASURED ^j 



I 

 I 



l( 



-400 



400 800 



STOPPING GRID VOLTAGE 



I 

 1200 



■ F^^' A ZZ. '^^^\^^ current versus stopping potential. The oscilloscope curve 

 IS tor a 4,0n0-mc tube, and the other that predicted from the scale model meas- 

 urements. By integrating current as a function of velocit v for Figs. 7-10 stopping 

 potential distributions can be deduced for other conditions 



