452 BELL SYSTEM TECHNICAL JOURNAL 



factor and the other based on actual tube characteristics, is provided 

 by a consideration of the second order modulation in the case of large 

 external plate resistance. 



The ratio of second harmonic to the fundamental, when ix is assumed 

 invariable, comes out proportional to 



{R + i?o)-^ 



which shows that the ratio tends toward zero as R is made indefinitely 

 great, a condition approximated in voltage amplifiers. According to 

 this expression, the distortion would be eliminated by increasing the 

 external plate resistance. That this is not really so is demonstrated 

 by the analysis above which gives for the same ratio 



lim ( 7^ ) ^ (^02 + iJrb'io - fJibu)lboi. 



The ratio therefore approaches a constant value diff'erent from zero 

 as R is indefinitely increased. The second harmonic level referred to 

 the fundamental is about 40 T.U. down with a 101-D tube, which is 

 prohibitively large distortion for certain classes of work such as multi- 

 channel amplification used in carrier telephony; for a 104-D tube the 

 level is about 32 T.U. down on the fundamental. 



In order to bring out some important points involved in the theory, 

 we shall discuss them in connection with experimental data on a 

 standard type of tube (101-D) which are due to Mr. A. G. Landeen. 

 The method used in measuring the current components is described 

 in his paper on current analysis in the Bell System Technical Journal 

 for April 1927. 



Output Currents of a Representative Tube 



Fig. 4 shows the calculated effect of varying the plate resistance on 

 the fundamental, second, and third harmonic currents produced by a 

 representative 101-D tube, which are indicated by circles, triangles, 

 and crosses, respectively. In this drawing the values of the coefficients 

 as calculated by Mr. J. G. Kreer are plotted as ordinates, and the 

 external plate resistances are plotted as abscissae. The agreement 

 with the values obtained from experiment, and shown by the full 

 lines, is seen to be rather close and within the limits of accuracy of 

 the measurements except perhaps for the third harmonic at high 

 load resistances. The coefficients used in the calculation of the 

 quantities d, d, Cz were obtained by graphical methods, which 

 consisted in determining tangents to curves derived from ^ and Rq 

 measurements. The precision obtained is sufficient for our present 



