FEEDBACK AMPLIFIER DESIGN 



437 



A similar analysis can evidently be made for any amplifier. In the 

 particular circuit shown by Fig. 11 the slope of the asymptote, in units 

 of 6 db per octave, is the same as the number of tubes in the circuit. 

 The slope can evidently not be less than the number of tubes but it may 

 be greater in some circuits. For example if C5 and Ce were omitted in 

 Fig. 11 and N^ and TVe were regarded as degenerating into resistances 

 the asymptote would have a slope of 24 db per octave and would lie 

 below the present asymptote at any reasonably high frequency. In 

 any event the asymptote will depend only upon the parasitic elements 

 of the circuit and perhaps a few of the most significant design elements. 

 It can thus be determined from a skeletonized version of the final 

 structure. If waste of time in false starts is to be avoided such a 

 determination should be made as early as possible, and certainly in 

 advance of any detailed design. 



The effect of the asymptote on the overall feedback characteristic is 

 illustrated by Fig. 12. The curve ABEF is a reproduction of the ideal 



50 



40 



^ 20 





-10 



-20 



0.5 0.6 0.8 



3 4 5 6 8 10 



20 30 40 50 



Fig. 12 — Combination of asymptotic characteristic and ideal cutoff characteristic. 



cutoff characteristic originally given by the solid lines in Fig. 10. It 

 will be recalled that the curve was drawn for the choice k = 5/3, which 

 corresponds to a phase margin of 30° and an almost constant slope, for 

 the portion DEF of the characteristic, of about 10 db per octave. The 



