Here's a thought you might kick around regarding the phase inverter and it's balance issue: Convert it from it's present status as a simple paraphase inverter to a floating paraphase inverter. This would be very simple to do, and would cause the inverter to automatically balance regardless of what particular tube is used in the inverter position.
To make the conversion, the grid return resistor for the top set of output tubes should be increase from 220K to 240K. Then, R5 (originally 20K), should be increased to 100K. Finally, un-ground the grounded end of the grid return resistor for the bottom set of output tubes, and connect it to the junction of the upper output tube grid return resistor and the new value R5 resistor. That's it.
With the new configuration, the signal supplied to the bottom inverter grid now contains an error signal that automatically causes the inverter outputs to remain in balance, regardless of what tube might be used. It may need very slight tweeking, but the balance should be very close with these simple changes -- and likely much closer than the original design ever was.
In addition, the previous suggestions to reduce the inverter plate resistors to 220K is excellent. It's never a good idea to have plate resistors greater in value than the following stage grid resistors. Reducing them to 220K puts them on par with the grid return resistors for the output stage. In addition, reducing the value of the 100K B+ dropping resistor for the inverter stage is also an excellent idea. I would target a B+ operating voltage in the range of about 270 vdc for the inverter stage when all is said and done. Both of these moves will increase the drive quality and capability of the inverter stage significantly.
With these changes, and assuming the inverter has been changed to a floating type design, I would also reduce both cathode resistors for the inverter tube to 2.7K. This would allow proper linearity for the new operating parameters, and also allow the stage to produce the increased drive it is now capable of.
Also -- and again assuming conversion to a floating type design -- I would bypass the bottom inverter section's cathode resistor with a small 20 mfd cap like the upper section's cathode resistor is. This will further maximize inverter performance, and help insure balance regardless of actual inverter tube balance or condition.
Finally, when bias voltage versus peak input voltage is examined for the top inverter section, be sure to include the peak value of the NFB voltage inserted into this section in the calculation. This voltage adds on top of the dc bias present to make up the total bias for the tube under full power conditions. With the values shown, the peak value of the NFB voltage is over .90 volts from the 4 ohm tap, and 1.30 volts from the 8 ohm tap (if that is where the FB is taken from), based on a full power output voltage of 12.24 vac rms from the 8 ohm tap. In either case, this puts the total bias for the top inverter tube at least at 2.20 volts. Therefore, since the peak value of the full power input signal was less than 2.0 volts, it's unlikely the inverter stage was clipping when operating in its original form.
Dave
