Power amp input sensitivity

[naciketas]

@Jana: if cutoff were to happen I should have seen clipping on one side of the sinewave only, as I said it's almost perfectly simmetrical all over the range of the volume pot.
@Martin: I thought that grounding one grid wouldn't affect primary Z.
I'm at work now, pi has 0.47uf input capacitors, 470k Rg, 470R Rk, 15k tail, 82k/100k Rp, 47pf across plateS. 0.22 uf coupling caps.
PA has 22k Rg, 500R Rs, 260R Rk bypassed with 2200uf.
NFB 8ohm 100k/4k7 no presence.
Sorry, I Forgot to thank you guys for the interest!

[martin manning]

I thought that grounding one grid wouldn't affect primary Z.

Not true. Grounding one grid will cut the load on the other tube in half. What primary Z (a-a) are you using? Do you have separate cathode resistors? A schematic would REALLY help.

[naciketas]

6k6 primary, separate cathodes, schem asap.

[martin manning]

In Class A push pull with a 6k6 a-a load, each tube is working at 3k3, which puts the load line way below the knee. I'm not surprised it starts clipping at -7 to -8V Vg1. When you ground out the signal to one tube and go SE, the other one sees a 1650 ohm load, which might be pretty well centered at 55mA, but it will produce a lot of second order harmonic.

[naciketas]

"Addio sogni di gloria"
Went back in the basement and took some other measurements, my God it's true!
When the two are working, the swing goes up to 300v and when one is grounded, it goes down to 200v.
So this is a dead end street, thanks for your patience and collaboration.

[martin manning]

Maybe another way to go about it is to keep it always in Class A p-p, and switch the screen voltage. Use 200V Vg2 (where you have it) for the dirty mode, and reduce it to ~125V to clean it up. A MOSFET VVR would be fine for this job, and you could change modes using a footswitch. Do the experiment of lowering the screen voltage with a resistor first to see if this idea will work for you. A lot of hardware for a 5W amp, though.

[matt h]

(deleted)

[naciketas]

Cents are always welcome!
VVR already implemented scaling the whole B+ rail.
I took other measurements that confirm what has been said before about primary impedance shifting and 6k6 being too low: when in SE mode, the voltage swing goes up again doubling the impedance (16 ohm cab in the 8 ohm speaker jack); in PP mode when using 6v6 or EL34, it takes about double the bias voltage to start clipping the signal as expected.
A mad idea appeared on my mental orizon: why not use a dpdt relay to ground one grid of the pair and to double primary impedance changing OT secondary tap?

[martin manning]

Sure that would work, but it won't solve your limited headroom problem. I like the idea of changing the screen voltage if it has the desired effect.

[Firestorm]

I build parallel SE amps using the multitap OTs from Hammond. That way you can max the output power for different tube types. But mine are only incidentally clean. That is, you have to set them right to stay clean. Class A is a bitch. If your voltage is right, the output will stay clean, but I build mine to get dirty.