I 



THE FIELD DISPLACEMENT ISOLATOR 885 



A. The saturation magnetization {4:TrMs) and the applied magnetic 

 field (Hnc). 



B. The ferrite height. 



C. The thickness (5) of the ferrite and its distance (b) from the nearest 

 side wall. 



D. The placement of the resistance material and its resistivity (p). 



E. The length of the ferrite (^). 



A. 4:TrMs and Hoc 



Theoretically, minimum forward loss occurs with a true null at the 

 face of the loss film and has been given in the condition \ fXr\ < \ kr\. 

 Although this inequality is required in the full height slab analysis, 

 lexperiment (Fig. 7) indicates the low loss region to be so broad as to 

 extend well into the low field, or | /Ur | > \K\ region. 



There is inherent loss in the ferrite so that a more accurate statement 

 of the bandwidth of operation is that in which the losses in the film are 

 of equal order to the ferrite losses at the band edges. Even discounting 

 ferrite losses, it will be shown in Section IV that we have a good analytic 

 basis for the observed broadness of the low loss region. In general, there- 

 fore, we need not be as restrictive as the null analysis of Section JIB 

 would imply. It is not surprising then that optimum operation actually 

 occurs in the region | /x^ | > \ kr\. There are several reasons why this may 

 be so: 



1. Shift of operation occurs due to the partial height nature of the 

 ferrite slab. 



2. Reverse loss has a peak in the low field region, requiring a compro- 

 mise of low forward loss and high reverse loss for best isolation ratios 

 (see Fig. 8). 



3. Optimum compromise between low ferrite loss and low film loss 

 must be made. 



The internal magnetic field, determining | ju^ | and | Av |, differs from 

 the applied field by the demagnetization of the ferrite slab. Although not 

 ellipsoidal, it may nonetheless be considered to have an average demag- 

 netization which has been computed, for this case, to be 460 oersteds. 

 A further complication in knowledge of the internal field is the proximity 

 effect of the pole pieces. This latter correction was obtained experi- 

 mentally and, all in all, it was determined that the internal field for 

 optimum operation was of the order of 300 oersteds. For the given 

 ferrite and the I'ange of frequency of operation, this internal field corres- 

 ponds to the condition that \ ij.,- \ > | Av |, as stated above. 



Taking all effects into account, it was found that optimum permanent 

 magnet design occurred for an air gap field of 660 oersteds. 



