TRAVELING WAVE TUBE FOR 6,000-MC RADIO RELAY 1337 



Intermodulation effects are ordinarily complicated and results are 

 jvery hard to predict from single frequency measurements on an amplifier, 

 i'or a TWT, however, one case — that in which two signals of very 

 [different amplitude are passed through the tube — can be treated simply, 

 IConsider an input to a TWT consisting of two signals at frequencies 

 l/i and /i + A/ with the signal at/i being very much larger in amplitude. 

 The composite signal applied to the amplifier will then be a signal at 

 frequency /i which is amplitude and phase modulated at a rate A/ in an 

 amount proportional to the relative magnitudes of the two signals. 

 This can be represented vectorially as shown in Fig. 41(a) and b. In this 

 figure the amplitude of the signal /i + A/ has been normalized to unity. 

 "A" thus represents the ratio of the larger to the smaller signal. The 

 locus of the resultant signal is shown by the dotted line. The single 

 rotating vector can be considered as the sum of vectors at /i + A/ and 

 /i — A/ as shown in Fig. 41(c). One set of vectors produces PM and the 

 other AM. The AM and PM vectors cancel at /i — A/ and add at /i + 



A/. 



Suppose this signal is put through an amplifier operating in com- 

 pression. For the time being let us assume this amplifier has no AM-to- 

 PM conversion. The compression in the amplifier will operate on the 

 AM sidebands of the signal but will leave the PM sidebands unaffected. 

 Let us define the quantity c as a measure of compression in the amplifier 

 by 



' = '- AVW< ^'^ 



where Vo is the output voltage, Vi input voltage, and AFo is the change 

 in output voltage for a change AVi in the input voltage. This quantity is 

 the per cent compression used in Section 4.2 divided by 100. If the signal 

 in Fig. 41 is put through the amplifier while it is in compression, and the 

 level of the signal at /i is subsequently brought back to amplitude A, 

 we would then expect to have the situation shown in Fig. 42. Each AM 

 sideband component has been multiplied by the factor (1-c). The locus 

 of the composite signal is now elliptical. Let Si and S2 be the magnitude 

 of the sidebands at /i + A/ and /i — A/ respectively. From Fig. 42 it is 

 seen that 



^1 = K + Hd - c) =1- c/2 (2) 



S9 = y2- Hil - c) = c/2 (3) 



When c = 0, the amplifier is operating in the linear region and *Si = 1, 



