760 THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1953 



to megacycles in Fig. 20 by using the relation, 



A/ = (QS) -^ mc, 



where Q was found to equal 130 by the independent measurement de- 

 scribed in section 4.3.1 and /o taken as 3800 mc. The vertical power 

 scales of the theoretical curves are not entirely independent of the 

 experimental plots in as much as one of the latter, namely the plot for 

 (Qkf = 2.4, was used in determining the location of the load lines within 

 the spiral diagram. 



The curves presented in Fig. 20 are largely self-explaining. They in- 

 dicate good agreement between theory and experiment. What disagree- 

 ment there is, may be traced primarily to the assumptions involved in 

 the small signal klystron theory not being fully met in practice. Thus, 

 the slight discrepancy between Figs. 20(a) and (aa) may be due to the 

 drift angle not being linearly related to repeller voltage or possibly due 

 to the phase angle of the electronic admittance being affected by the 

 magnitude of the gap voltage.^^ These factors, however, are eliminated 

 in the mode plots (c) and (d), which for this reason exhibit better mode 

 symmetry and very close agreement with theory. 



4.4 Performance Data 



The experimental curves of Fig. 20 were obtained with particular 

 values of (Qkf and Qs/Q for which the input admittance of two coupled 

 resonators had been computed and plotted earlier. As pointed out before, 

 these values were chosen merely because they facilitated comparison 

 between theory and experiment; they were not to be regarded, however, 

 as representing optimum conditions. Whereas these curves had been 

 obtained for a fixed secondary Q (namely equal to half the primary one), 

 with changes in coupling-iris width producing the desired variations in 

 (Qk) , the demonstration of optimum performance which follows was 

 pursued along different lines. Here, the coupUng iris was opened to its 

 maximum width (0.500") and the secondary Q adjusted until the con- 

 ditions shown by the oscillograms of Fig. 21 were obtained.* These oscil- 

 lograms indicate a number of mode shapes useful in applications re- 

 quiring power output to be essentially independent of frequency. It is 



* In adjusting the circuit parameters for a flat topped mode it should be borne 

 in mind that the variation in power with frequency is a function of QJc whereas 

 the actual bandwidth varies inversely with Q, . For a maximum flat hand, there- 

 fore, Q, should be chosen as small as possible consistent with a value of (Qnk)^ 

 of about 0..35. In practice the lowest value of Q, which can be used will be deter- 

 mined by the highest value of k obtainable with a given coupling iris. 



