48 BELL SYSTEM TECHNICAL JOURNAL 



desired, sufficient sensitivity can generally be obtained by wattmeter 

 or ballistic galvanometer methods. 



Graphical Separation of Losses 



Since it is generally important to distinguish between types of 

 magnetic losses, methods of analyzing the measurements have been 

 devised to accord with the degree of refinement desired. A fairly 

 simple graphical loss separation method is suitable if magnetic shielding 

 can be ignored. However, it will be seen to lead to the inclusion of 

 an additional term to account for the residual loss. If the effect of 

 eddy current shielding is also to be considered, a more complicated 

 analytical method of separation will be found necessary. 



With magnetic shielding negligible, eq. (36) reduces to the form 



77^ = ^ \Hmyifi + -=— IXmf (37) 



jLm 3 6p 



==^X5„-^° + ^%./ (38) 



3 Mm -^P 



or 



t^mfLn 



= aB^ + ef. (39) 



The last form of the expression is most suitable for routine testing 

 and design purposes. The hysteresis area constant a will be seen to 

 be intimately related to the hysteresis loop area ai previously discussed ; 

 thus 



a =1^3. (40) 



Within the limits of applicability of Rayleigh's equation, the following 



relations also apply: 



8Xmo 8q: ,..s 



a = ^ — -„ = ^ — ^. (41; 



The losses observed on any test core can be separated graphically ^^ 

 by calculating the values of Rm/fXmfLm for a fixed value of Hm at all 

 measuring frequencies, and plotting such values against frequency. 

 The slope of the resulting straight line should then give e, and the 

 intercept aB„. When this process is repeated for other values of //„, 

 a series of intercepts will be obtained, all of which would be expected 

 to yield a constant value for the loop area constant a. However, this 

 is frequently found not to be true, but if the several intercepts so 



1= B. Speed and G. W. Elmen, Trans. A. I. E. E. 40, 596 (1921). 



