430 THE BELL SYSTEM TECHNICAL JOURNAL, MARCH 1957 



where p — \y \ M,/co normalized saturation magnetization 



a- = I 7 I H,/cc normalized static magnetic field in the ferrite 



I 7 I = 2.8 mc per oersted, gyromagnetic ratio 



CO = operating frefjueney 



It appears that some cavity techniques measure the eigenvalues of (1), 

 Xm + K and Xm — K, directly, which are seen to be 



Xm±K= -^ (2) 



Suhl and Walker show that a loss term may be introduced by replacing 

 0- by (7 + 7a:(sgn p) in (2).* Separating real and imaginary parts we get 



> , r _ p{a ± 1) . . 



^" ■ ~ (cr ± ly + a' ^^ 



„ , apisgn p) , 



For the determination of a from measurements it is convenient to define 

 a loss tangent 



Xm " ± k" ^ a(sgn p ) 



^ dz K (T ± 1 



5± = 7—. T = T-T— <^^) 



Typical curves for Xm ± k and 8^ assuming p = 0.5 and a = 5 X 10^' 

 are shown on Figure l.f For the purpose of describing and comparing 

 experimental results, it may be convenient to distinguish among various 

 regions of H^ as indicated on the graph because a difTerent measurement 

 technique may be required for accurate measurements in each region. 



3. METHODS FOR MEASURING MAGNETIC PROPERTIES 



Three measurement methods have been reported in the literature all 

 of which employ the detuning and change in 1/Q of a resonant cavity by 

 a small ferrite sample. Ya,n Trier used very thin long cylindrical samples 

 in a coaxial cavity. Artman and Tannenwald" emploj^ed small spheres, 

 and we used thin discs^ both placed close to the endwall of a cylindrical 

 degenerate cavity excited by a TEm mode (Figs. 2 and 3). Recently, 

 Berk and Lengvel suggested the use of a cylindrical post at the center 



* By definition sgn p = +1 for p > and sgn p = — 1 for p < 0. 



t Since it is customary to use Xm + k for the designation of the resonance line 

 this notation has been used here. Consequentl}', p and a shovild be assumed nega- 

 tive. 



