THE L3 SYSTEM — AMPLIFIERS 907 



than asking how many db of gain margin and how many degrees of 

 phase margin any particular return ratio displays, we ought instead to 

 inquire how quickly the apparent margins disappear as we change 

 transconductances and network impedances. The margins on all the 

 return ratios will vanish simultaneously, regardless of the apparent dif- 

 ference in original margin magnitudes. It is therefore satisfactory to 

 examine the behaviour of whichever return ratio is most easily observed. 



The design choices made in arriving at the coupling network also 

 affect the magnitude and shape of the feedback which can be obtained: 

 as mentioned above, the relative magnitude of the impedance seen 

 looking into the coupling network and the impedance from first tube 

 grid to ground determines how much of the voltage developed across the 

 beta circuit wall reach the first stage as a driving force. 



Study of the modulation products which will arise in system opera- 

 tion, both for the all telephone case and for the combined telephone-tele- 

 vision signal, led to the conclusion that optimum shaping of the feedback 

 for the L3 system would be to maximize the feedback at low frequencies 

 in order to suppress intermodulation products falling in that part of the 

 spectrum in the combined telephone-television case. Shaped feedback, 

 falling off at the higher frequencies, is also consistent with obtaining a 

 smooth and simple shape of gain-change as tubes age (known as "mu- 

 beta effect"), which is desirable from the equalization standpoint. With 

 these considerations in mind, the interstage of the input amplifier has 

 been peaked well above the transmitted band — at 11 mc — to partially 

 compensate for the input potentiometer term, and to help in achieving 

 this smooth shape of mu-beta effect. If flat feedback over the band were 

 the objective, it would also be necessary to design the grid-cathode ad- 

 mittance of the second tube so that the parasitic grid-cathode capacity 

 would be absorbed in a flat impedance versus frequency, but in this 

 case the desired shaping of the second tube feedback is attained by taking 

 advantage of the way in which the grid-cathode capacity naturally limits 

 the high-frequency feedback on this stage. 



The loop feedback in the output amplifier is similarly shaped, for the 

 same modulation and equalization reasons. The use of the double triode 

 circuit in this amplifier is an unusual feature. This connection of two 

 triodes, sometimes referred to as the "cascode circuit," has appeared in 

 many contexts in recent years, usually to serve some other purpose than 

 here. It serves as a superior output stage in the L3 amplifier largely be- 

 cause the effective transconductance w^hich can be obtained is about 3 db 

 higher than that of a pentode of the same grid-cathode spacing. The 

 effective transconductance, ignoring for the moment the division of out- 



I 



