His sense is that the 1/2 power switch (triode connects the EL34s) isn't doing anything at all. He has triple checked the wiring, switch functionality, retouched all of the solder joint and all of the other standard checks to no avail.
Quiescent voltages move in the right direction between 1/2 and full power...but only by a few volts.
I have no practical experience with this set-up but believe that dynamic voltage checks will reveal whether or not it is functioning properly.
Thinking that pounding out some power chords in full power mode with the + end of the DVM on a tube plate and the - end on the screen grid (down stream of the screen resistor). And then repeating at 1/2 power setting will identify a significant difference in behavior.
Unfortunately, I keep getting into a "do loop" everytime I try to think it through to predict what the difference will be.
Hoping that someone can help lead me out of that loop!
TIA,
Dave O.