REFLEX OSCILLATORS 571 



reception of the signal might yet fail to oscillate at the desired frequency. 

 The convenience of simple replaceability of the local oscillator became in 

 this instance a cross for its designer, since an exchange of tube would fre- 

 quently eliminate the effect. This led to the obvious conclusion on the part 

 of the user that the oscillator was defective, in spite of the fact that the pre- 

 sumably defective tube had passed the test specifications and would operate 

 satisfactorily in some other receiver. The plain fact, in the light of later 

 knowledge, was that the tubes were being improperly used, so that the 

 usual range of manufacturing variations was not tolerable. 



The appreciation of this fact led to a new approach to the problem of 

 coupling the beating oscillator to a waveguide. In early designs the oscil- 

 lator was decoupled from the load by withdrawing the probe from the wave- 

 guide. This presented the oscillator with an uncontrolled admittance with 

 disastrous results in many cases. The new approach, proposed by the 

 group working with Dr. H. T. Friis at the Bell Telephone Laboratories, was 

 that of designing the receiver so that the beating oscillator could be oper- 

 ated into an essentially fixed impedance. The crystal was in this case 

 loosely coupled to form a part of this load, so that variations in its impedance 

 and the impedance looking toward the TR tube were largely masked. A 

 great many further refinements in the design of the receiver have since 

 been proposed, but this basic principle of defining the load into which the 

 oscillator is required to operate is fundamental to all. In the interests of 

 simplicity of use it appeared to be desirable to endeavor to pre-plumb the 

 coupHng of the oscillator to the wave guide. The tube designer in this 

 instance found himself perforce, as so frequently occurs in dealing with 

 micro-waves, a circuit designer — an instructive and illuminating experience 

 which might happily be reversed. 



The wave guide coupling was made separate from the tube, both to 

 preserve the plug in feature of the tube and to maintain its interchange- 

 ability with the 723A/B. As a further simplification it was desired that the 

 coupling should require no adjustments. A convenient fixed load admit- 

 tance to present to this coupler is the characteristic admittance of the wave 

 guide, since this can readily be maintained fLxed over a wide band. The 

 problem, then, is the apparently straightforward one of transforming the 

 guide admittance to the admittance which the oscillator requires for opti- 

 mum performance. Actually, the problem is complicated by the fact that 

 the optimum admittance will vary throughout the band. The electronic 

 admittance varies with frequency even for a fixed drift angle, because the 

 modulation coefficient of the gap varies as the oscillator is tuned. The 

 losses of the resonator also vary with frequency, both because of skin effect, 

 the depth of penetration of the high frequency currents changing with fre- 

 quency, and because the circulating currents in the resonator are a function 



