phase inverter has too much gain

[JoeK]

I'm building a 2 watt amp. It has a plexi preamp mated to a 12AX7 phase inverter and 6SN7GT power amp (one triode each side) The power amp is from AX84 http://ax84.com/static/corepoweramps..._Schematic.pdf
The amp is working except for the overdrive produced in the phase inverter. Note I didn't use a master volume as per the schem (shouldn't matter?). With a 100mv ac sine wave input I get a clean output at the phase inverter until about volume on 1. phase inverter input is still clean sine.
Voltages on phase inverter at idle are 227vdc on plates, 66vdc on cathodes, 39vdc on grids, 65vdc after 470r tail resistor, 11vdc after 22k resistor.
Clean sine wave at speaker and input voltage to PI is 270mvAC, output voltage to power amp grids is 8VAC. If I have the volume on 50% I have 50VAC at power tube grids which seems way over the top. Any suggestions or ideas much appreciated TIA.

[frus]

What's the DC voltage at the power tube grids (assuming you biased them following the procedure mentioned in that pdf)?

[JoeK]

Neg 12.7VDC bias on the power tube grids. I measured the plate current at test points and clean output with volume on 1 is tiny 1/8th watt as to be expected with full volume putting out about 1-2 watts. Just seems to be the phase inverter has the wrong bias but I've gone over the components and hookup a few times (pretty sure I've missed something though ). Could I remove the feedback or presence to isolate sections?

[frus]

no, the phase inverter seems to be OK
are you sure you don't have a linear volume pot?
how did you measure power output?

[dwmorrin]

The PI gain should be about 30 per side. 270mV in, 8V out is good. 50V out means 1.6V in (this was "volume at 50%").
The question is, at what input voltage (and let us know P-P or rms) does the PI output look distorted on the scope? This would be where the PI has run out of headroom.
The PI does not have to have a gain of 30. You could lower that. But you should make sure this that the PI really is the problem.

[eschertron]

If you want the distortion to come from the preamp before the PI, figure out what the AC voltage is coming out of the last preamp stage (pretty close to HV if you read peak-peak?) and then decide what the voltage you WANT going into the PI to get the voltage on the grids correct. You can put in a voltage divider, some kind of "voicing" circuit, or a pot to get the gain where you want it to be. If I'm reading the schematic correctly, this is all where that "master volume" would have been.

[dwmorrin]

My tube amp load line math is sketchy at best (tying to learn)... so maybe someone more experienced can double check me, but these are the rough guesses I've come up with:
PI input max, 2V P-P for clean output. This should give you 2 60V P-P (clean) outputs.
Each 6SN7, I calculate a max input of 12V P-P before clipping. (I used the 16Ω tap and hammond's website for transformer info... and like I said - I am a total newbie with load lines, so don't trust my math!)

That means your PI input needs to be below 400mV P-P to keep the power amp from clipping.

Maybe you can post your preamp's output voltages (or just the schematic) and then we can offer suggestions to lower the gain of the PI to give you a wider range of clean.

Installing the master volume will work to lower the preamp's output. This could be a trim if you don't want to clutter the front panel. This is where you can make a design decision: throw away the preamp voltage, and re-amplify it? or lower the gain of the PI to match the preamp to the power tubes. The latter approach is more appealing to me.

[eschertron]

I see a Mu factor of 20 for the 6SN7, so I'd be tempted to think that around double the bias voltage (and a little more to drive the output tubes - if desired) would be OK from the PI. I'd start with 30vpp if I was designing this. If the gain of the PI is 30 (each half) then that would be 1vpp at the grids. Not too far off from 400mv (yes, I agree that 400mv will avoid clipping the 6SN7) but very different from slamming the PI with the full clipped output of a preamp tube stage! What's directly in front of the PI? Tone stack? CF? Gain stage? This will determine just what kind of voltages that are currently feeding the PI.

I recently did a small-wattage amp using a 12AT7 as the power tube. I didn't make it push-pull, simply paralleled the triodes for a SE class A design. I not only allowed, but depended on massive clipping in the 12AT7, because I have only two stages in front of it. But then, I wasn't modeling an existing preamp sound, so it simply sounds like it sounds...

[loudthud]

The Monkey Wrench in the works of this amp is the feedback around the PI and output stage. You can't really measure the input and output of the PI to determine it's gain unless you measure the input differentially between the grids of the PI, or you disconnect the feedback. An amp with this much gain needs a Master Volume.

[J M Fahey]

I don't see a problem there, values measured are logic.
You calculated/measured/expected

2 60V P-P (clean) outputs.

at the PI plates.
Well, that may be true without the 6SN7 plugged in, but as soon as it appears you have diodes (grid/cathode) clamping that to around 12Vpk.

If anything , your Plexi preamp is somewhat too hot , since it was originally designed to drive EL34 and earlier, in its Tweed incarnation, 6L6 which are even harder to drive.

[JoeK]

Thanks for all your replys. I built a plexi 6v6 push pull before http://mhuss.com/Plexi6V6/Plexi6V6c.gif and I was hoping this could be implemented with the 6SN7 but thought the grid voltage would need to be tamed a bit. The amp was being squeezed into a small chassis so I opted for no master volume control. I'm thinking I might try a voltage divider before the phase inverter.
At 50% volume on the bright gain input the feed to the phase inverter is clean sine wave 18VAC, the 6SN7 grids are getting 52VAC and I'm measuring 9ma current each triode and 5VAC across the 16ohm dummy load. So 5vac X 5vac divide by 16ohms gets 1.5 watts.