576 



BELJ. SYSTEM TECHNICAL JOURNAL 



Thus 



iVr:/= 10 Logio(p^) 



In practice an amplifier is almost always measured between im- 

 pedances that are pure resistances. These impedances are set up 

 by variable resistance networks so designed that when the current 

 in one mesh of the input network is equal to the current in a mesh of 

 the output network the gain of the amplifier may be read from the 

 settings of the dials and switches controlling the networks. An 

 indicating device which may be switched from the input network to the 

 output network indicates when the currents in the two meshes men- 

 tioned are equal. Where sufficient energy is available a thermo- 



wAvEi TO be: 



ANALYZED 



ATTENUATOR 



RESONANT 



AMPLIFICR 



€=70 



OSCILLATOR 



ATTENUATOR 



RESONANT 



O O — K> 



Fig. 1 — Circuit for harmonic analysis 



couple is used as the indicating device. It follows that measurements 

 are then always in terms of r.m.s. values of the currents employed. 

 So long as these currents are approximately single frequency the 

 r.m.s. value of the whole wave is very nearly equal to the r.m.s. 

 value of the fundamental. In all the experiments described here 

 gain measurements were made by thermocouple and as will be seen 

 later the proportions of harmonics were such that unless the amplifier 

 was overloaded the discrepancy between the r.m.s. measurements 

 made and those that might be made by methods taking account of 

 the fundamental only was usually less than 1%, and in the cases of 

 extreme overload less than 5*^^ . 



The harmonic analyses were made by an electrical analyzer whose 

 principles of operation arc indicated in Fig. 1.^ In operating this 



1 The princii)les of this analyzer arc fully described in a pajx^r by Mr. A. C. Landeen 

 to be published in the B. S. T. J. 



