Negative voltage on grid of cathode biased output tube...

[Steve Conner]

My reply just got trashed somehow. So another shorter one:

Does the amp have feedback from the speaker output to the PI? Try disconnecting it, and see if the oscillations stop. If they stop, the feedback needs tuning. If they don't, or maybe even get worse, it's a layout issue.

The various paraphase and floating-paraphase inverters are notorious for poor balance at high frequencies, due to Miller effect rolloff in the second stage. That could explain why only one power tube is seeing the HF voltage. The LTPI's second stage is grounded-grid, so it has no Miller effect and the HF balance is better.

With 35V at the grid, the oscillation is probably not far short of 1kV p-p at the power tube plate, so it will couple into everything. Watch out for overheating tubes, arcing sockets, OT catching fire etc...

[alchemy]

My reply just got trashed somehow. So another shorter one:

Does the amp have feedback from the speaker output to the PI? Try disconnecting it, and see if the oscillations stop. If they stop, the feedback needs tuning. If they don't, or maybe even get worse, it's a layout issue.

The various paraphase and floating-paraphase inverters are notorious for poor balance at high frequencies, due to Miller effect rolloff in the second stage. That could explain why only one power tube is seeing the HF voltage. The LTPI's second stage is grounded-grid, so it has no Miller effect and the HF balance is better.

With 35V at the grid, the oscillation is probably not far short of 1kV p-p at the power tube plate, so it will couple into everything. Watch out for overheating tubes, arcing sockets, OT catching fire etc...

It has feedback to the cathode of the PI. Disconnecting it does not change the oscillation. Deffinately seems like there are some layout issues as I can change the amplitude of the oscillation by moving the OT secondary wires around. That and the fact that it is one of the worst jobs of lead dress I've ever seen! Because of the paraphase imbalance I had thought about rewiring it to a LTPI, but I wanted to see if I could get the paraphase sounding decent first, just because I've never heard this particular circuit and wanted to see what it sounds like. Would going back through all of the math and getting the gain balanced between the first and second paraphase stage make any difference, or is it completely unrelated to the oscillation?

The one thing I thought really odd: With an oscillation that big, I fully expected overheating tubes, etc. They are, in fact, not drawing too much current, nor is the OT even warm to the touch.

[alchemy]

I think I am just going to try and rewire a different PI. Another thing that seems problematic is that the output transformer is located next to the preamp tubes and the primaries run all the way across the chassis to the output tubes. Seeing as I cannot move the OT, what is the best way to route the primary leads? Twisted tight and run around the outer edge of the chassis as far as possible from all signal wires? And what about the center tap?