Traynor Custom Special 100 has ~10kHz oscillation on PI

[Delta362]

I'm trying to figure out how to resolve a 10kHz 50mV peak to peak oscillation that is appearing on the plates of the phase inverter tube as well as can be heard thru the speaker. I tried a few different preamp tubes with all presenting the same waveform/tone. I tried a handful of increasing sizes of capacitors across the plates starting with 47pF, but could only get it to disappear when I added a 10nF cap across the plates of the PI tube. I am skeptical that it required such a high value to tame the oscillation.

I pulled the power tubes and PI tube and scoped the B+ power supply thru each dropping resistor until I got the PI tube plates. It remains clean there until I install the PI tube. Then it's at the PI plates and makes its way to the grids of the power tubes. The incoming B+ plate, screen and splitter voltages remain clean.

Any suggestions what else to look for or other things to try? Is the 10nF across the PI plates too large of capacitance? Thanks!

[Helmholtz]

Does the Presence or "Resonance" setting have an influence?
Does the oscillation stop when you disconnect the NFB loop, e.g. by lifting one end of R166 or C111?

10nF between the PI plates certainly is too large.

Is the OT original?

[Delta362]

The OT appears to be original. The amp came in with a complaint of not being loud enough. I found a handful of the preamp tubes sockets that were lacking solder on some of the pins and reflowed all tube sockets (preamp and power tubes). I'll report back in a few minutes on the rest of your suggestions. Thank you!

[Delta362]

The resonance pot has no effect on it. Turning up the presence pot seems to reduce the amplitude of the oscillation. With one leg of R166 lifted, I get the same thing - resonance has no effect, while presence turned up lowers the amplitude of the oscillation.

[Helmholtz]

Doesn't looh like typical power amp oscillation.

What happens if you short the power amp input to ground at PA IN (not the PI grid pin 2!).

[Delta362]

No change with shorting PA IN to ground

[Helmholtz]

Ok, now connect a 0.1µ cap from PI input grid pin 2 to ground.

[Delta362]

Seems to get slightly larger in amplitude with the added cap. I've attached a waveform of what I'm seeing with the presence at minimum and no cap added between pin 2 of PI and ground. Hopefully I described it properly before as oscillation.

The photo is the waveform at the speaker output using an 8 ohm dummy load. Pin 1 and 6 of the PI show the same waveform.

[Helmholtz]

Scope pic shows a 10kHz signal + 3rd harmonic.

Now disconnect one end of C77.

[Delta362]

Still there, but drops it to 10mV pk-pk with same shape.

[Helmholtz]

Sorry, too late for me now.

[Delta362]

Sorry about how long it took to get back to you. I had to pull the board to access the cap. Thank you for your help today!

[g1]

Best to first make sure you have the exact schematic as there are many revisions, even a 100H and 100H2 version. I have a V3.00 schem for the 100h and a V4.00 schem for the 100H2.
Both mention some changes to R166 among other things.
For Yorkville, first 3 digits of the ser.# are usually the date code. 1st digit being year (ex. 7 for 2007), next 2 digits being month. There should also be rev. numbers on the board.
Sometimes repairs have been done with wrong revision of schem. and you end up with incorrect parts in a unit.

Not saying it's likely to be an issue, but you might as well start off on the right foot.

[Delta362]

My board is M600 2V00 which matches the schematic, but thank you for letting me know about a few variations. The schematic lines up with what I am working on.

Per the serial number, this is 704 = April 2007.

I did find heat damage to R183 and replaced it, but that is linked to the footswitch control.

Given there are ribbons between boards, I tried moving them to impact the oscillation, but didn't seem to make any difference. I am working with the bottom off, though.

[g1]

I'm wondering why the presence is decreasing it rather than increasing. Does presence otherwise seem to be doing what it should?
Was the 'not loud enough' problem duplicated or resolved?