Driver Bypass Cap - Value?

[batsong]

i like the sound with the input and output cathode unbypassed. Will this do damage?
What value should I replace the driver bypass when switching fro eletrolytic to film? I've read 10uf or 22uf though I don't understand why.

[gingertube]

The input stage cathode bypass cap is basically doing nothing except perhaps shunting any mains frequency hum leaking from the heater to the cathode to ground. Leave it off and if necessary (to get rid of hum) elevate the 6SN7 heater supply to say +50Volts.

The 2A3 cathode bypass cap is however doing quite a bit to reduce the effective rp which is in series with the output transformer primary and reflects to the secondary as increased output impedance. I would keep the 100uF/160V cap, perhaps experiment with some better quality electrolytics or look for a film cap of the SAME value.

The driver stage cathode bypass cap is also doing a lot since that 27K would effectively reduce the gain to 1 without it. The driver is not ideal anyway - 6SN7 running about 2.5mA, 6SN7 works best with about 3 times that current BUT this would require some circuit changes with higher power rating resistors.
For the bypass cap (with the same circuit values) you could get away with a 20uF film cap but I would'nt suggest going lower than that. A quick calc might suggest that since the cathode resistor is 27K you could use a much lower value capacitor. What is not apparent when looking at the circuit is that there is a "looking into the cathode" impedance contributed by the tube which is effectively in parallel with the cathode resistor. This "looking into the cathode" impedance in this case is about 1.5 K Ohms - Guru's correct me if I'm leading him astray but I think it is:
Rk = (RL + rp)/u+1
Where RL is the anode load (27K)
rp is the internal "dynamic resistance" of the tube - about 6 or 7 K for 6SN7 at 2.5mA
u = mu = approx 20 for 6SN7

That gives rk of about 1K5

This approx 1K5 in parallel with 27K gives about 1.4K as the effective cathode "resistance" which we are trying to bypass with that capacitor.
20uF the gives a "corner frequency" of about 5.5Hz. Because this defines a "shelving" response someone is bound to object to me calling it a "corner frequency", so call it the "half maximum gain frequency" - you get the drift.

From that "back of an envelope" calc. you may actually get away with 10uF.

Cheers,
Ian

[batsong]

Thanks Ian. It was really nice of you to go through all of the trouble to answer my question.
No problem with hum, even with 105db drivers, so the input bypass can stay out.
The 2A3 cathode bypass irks me a little, taking it out resulted in better frequency extension with the speakers I'm using (16ohm coaxial with a 15khz rolloff). I will look into better caps.
And for the driver stage, it seems that 10 would be enough, but 20uf gives some room to move.