Cyclotron wave amplifiers with a pump-to-signal frequency ratio of 5:1 (rather than the usual 2:1) have been built and have given a single-channel noise figure of 2 db. This paper reports on the difficulties which had to be overcome and the restrictions which had to be observed before achieving what, at the start, had looked deceptively simple. The paper begins with a discussion of the relative merits of assigning signal, idler and pump waves in various ways to beam and circuit. In the preferred arrangement the circuit carries only the pump wave; signal and idler are cyclotron waves on the beam. The validity of Tien's equations for the propagation constants and of the Manley-Rowe relations for the power-frequency ratio is demonstrated for the case of cyclotron waves. It is also shown that in a nondegenerate tube, idler noise must be extracted, no matter how high the idler frequency may be. Early experiments by C. B. Crumly, using interdigital couplers for signal and idler, gave very poor noise figures. The twisted coupler, a new structure with high coupling efficiency, was introduced but failed to reduce the high noise. It was then found that a signal carried by a cyclotron wave of low phase velocity was attenuated as it traveled along the beam, and that the noise extraction process worked poorly. This was traced to the spread of axial electron velocities, a source of trouble mentioned by C. P. Lea-Wilson and later by E. I. Gordon. P.G. Everett's measurements of this effect are described. It is then shown how the consequences of the effect may be minimized. Nondegenerate tubes are described in which the velocity spread effect is rendered almost harmless. The twisted coupler structure is particularly well adapted to experiments with synchronous waves; a brief report on such experiments is included.
Read full abstract