I have been puzzling over this for a few days .... it should not be necessary to do this .... this practice is undesirable for several reasons especially shortening the life of expensive components .... I restored many guitar amplifiers (mostly Fender) over the years and I found suitable AB1 operation for most 6L6 family tubes at 30 to 35 mA cathode current .... basically the bias is adjusted until the crossover 'knee' disppears and a few mA added for aging .... happy happy players
so why is it necessary in a modulator stage to move closer to class A push pull ? the guitar amp has a speaker(s) for a load .... the actual impedance model for a speaker is complex and hardly resistive BUT it is reasonably constant over the + to - swing of the output ac waveform ...otoh the single ended class C modulated stage approaches its calculated modulated impedance only on the + audio peak and is higher elsewhere .... so it appears that the higher conduction angles of the class A pp modulator MAY provide a load as well (my speculation)
My experience is that all tubes approach a zero plate current bias with non-linearity. At the other end of the dynamic range we have compression leading to saturation. I call this the 'S' curve. Some tubes can be biased close to cutoff and still be very linear. The 805 triode is one of them, as these tubes make good fidelity in class B service. Another tube that sounded terrible at a low idle current was the 8417 Beam Power tube. With this tube, the sound kept clearing up as the bias raised the DC operating point. The 8417 is unique in that its transconductance is off the charts; somewhere around 2X the other big bottles in the same power category.
The 'S' curve can be straightened to some extent with negative feedback
I have since learned that sweep tubes with the drive to G2 can be exceeding linear, and in class B with zero bias, and a low DC idling point. The 807/6L6's can be backed off some too...more so then some of the big Hi-Fi tubes that came later.
What often happens is the greater bias for low idle current creates the need for more peak grid driving voltage. Lets say as an example, you have a bias of -20V on each tube in a P-P pair running class Ab1. The driver must therefore provide 20v peak to each grid (referenced to ground). If we were to throttle back the power tubes to a lower DC operating point, perhaps moving the bias to -25V, then the driver must provide 5v peak additional drive, or a 25% increase. In many amplifiers the driver distortion will go way up unless the headroom is there.
I am attaching two files:
File 'e' is an EL-34 running 8 watts AM as a class Ab1 RF linear amplifier. The trapezoid sides are not straight indicating non linearity, and an example of the 'S' curve. As I raised the idle current, the linearity improved. The driver had the headroom. The EL-34 was just not very linear unless I increased the DC operating point. This tube ran 700V B+, and about 35MA plate current. I later used a Teslovok (from Penta) E-34L which had the big 6L6GC tri-metal plate. This tube was more stable at a higher Pd resting state.
File '130_Modulated_Trapezoid' is from a pair of old Fisher branded 7591's set for 12.5 watts carrier on AM. These tubes are extremely linear, and the trapezoid sides are very straight. With this tube, lowering the idle current is OK as it stays linear. Unfortunately the new production 7591's are down on power, and have the same linearity issue as the EL-34. On the scope plot, the high modulation percentage brought the + peaks off the scope grid, and the negative peaks made those wavelets typical of a low level balanced modulator setup.
The two tube types referenced were in a highly modified Central Electronics 20a.
Jim
Wd5JKO