S-B] 



TEST BY LOSSES. 



J 55 



50 





 will generally be found > 



< 10 " 



9.2, 



8. Curves for Core Loss. The readings of the wattmeter in 

 the open-circuit test (after corrections are made, if there are 

 an y> 3 a ) gives the core loss 

 plus a small RI Q Z loss due to 

 the heating effect of the ex- 

 citing current. The latter loss 

 can be computed and deducted 

 from the wattmeter reading; 



o 20 



FIG. 3. Watts core loss for varying 

 voltage at different frequencies ; 2 K.W. 

 transformer, loo-volt coil. 



negligible.* 



The curves showing the 

 change of core loss with volt- 

 age can be plotted on ordinary 

 cross-section paper as in Fig. 

 3; a derived curve, showing 

 the variation of core loss with frequency, being plotted as in 

 Fig. 4. It is much better, however, to use a logarithmic scale 

 for ordinates and abscissae, in which case the curves become 

 (within limits) practically straight lines. For this purpose, it 

 is convenient to use logarithmic cross-section paper.f 



Above normal voltage, as higher densities are reached, the 

 curve tends to bend upwards, due to the fact that the hysteretic 

 exponent (which has a value of about 1.6 up to 10,000 gausses) 

 becomes greater. Transformers with improved iron are run 

 at higher densities, so that at normal frequency this bend may be 

 reached at normal, or even below normal, voltage. 



9. For each frequency plot a curve on logarithmic paper 



* (8a.) Although any Rio 2 loss should be deducted for obtaining true 

 iron loss, for the calculation of efficiency it is better not to make such a 

 deduction but to include the Rio 2 loss with the iron loss Wo. 



t This paper can be obtained from the Cornell Cooperative Society, or 

 Andrus and Church, Ithaca, N. Y. 



The same results can be obtained on plain paper by plotting the loga- 

 rithms of the observed quantities a laborious process or by using a 

 slide rule as a scale. 



