I have an early Knight kit amplifier that uses push-pull 6V6s driven
by two 6SN7s, with a 6SL7 for the phono stage. In tracing out the
circuit I noticed that in addition to the global feedback from the
output-transformer secondary, there was a feedback path from the grid
of one 6V6 to a 6SN7 cathode. The feedback signal originates at the
220k-ohm 6V6 grid resistor, passes through a 1-megohm resistor, and
winds up at the 4.7k-ohm 6SN7 cathode resistor. I couldn't figure out
why this feedback path was there, particularly since the signal at the
6V6 grid would be expected to be somewhat distorted, being inside the
global feedback loop. The feedback path causes about a half-volt
offset at the 6V6 grid (the 6V6 cathodes are at about 20 volts), so
there ought to be a good reason to incur this error voltage. I
disconnected the 1-meg resistor and the overall gain decreased by 3
dB, indicating positive feedback. The gain came back up 3 dB when I
bypassed to ground the cathode where the feedback had gone.

My guess is that the purpose of this feedback path is to increase the
gain of the 6SN7 stage, basically eliminating an electrolytic bypass
at the cost of a 1-meg resistor. Is this right? I've disconnected the
feedback because I need neither the added gain nor the output bias
asymmetry.

Brian