890 



THE BELL SYSTEM TECHNICAL JOURNAL, JULY 1956 



Fig. 14 — Isolator model. 



which occurs due to the scattering of the TM-Uke lono;itudinal modes, 

 decays rapidly and is not of consequence in an experiment now to be 

 described. This experiment was designed to demonstrate the nonre- 

 ciprocal nature of the longitudinal electric fields associated with the dis- 

 torted dominant mode. It also shows that the existence of these com- 

 ponents is significant as a loss mechanism for the reverse direction of 

 propagation in the isolator. The geometry employed is shown in Fig. 11. 

 The copper plate was inserted to minimize longitudinal electric field 

 components, and we may therefore expect to obtain less reverse loss than 

 in the condition of its absence. The result of this experiment was that 

 the reverse loss decreased from about 25 db* without the plate to 18 db | 

 with the plate. The forward loss was unaffected. 



E. Determination of Length 



Given a dominant mode distribution in a waveguide, attenuation will 

 be a linear function of length, once this mode has been established. Con- 

 sequently, one would expect that doubling the loss film length would 

 double the isolator reverse loss. The isolator does not exhibit this be- 

 havior, however, as is illustrated in Fig. 12. 



This occurrence might be explained by the appearance of still another 

 longitudinal mode, peculiar in form to gyromagnetic media alone, which 

 propagates simultaneously with the transverse electric mode, and is 

 essentially uncoupled to the loss material. The maximum reverse loss 



* This experiment was conducted with a different ferrite than that employed ini' 

 the eventual design. 



