Peavey MiniMax DC on output

[Delta362]

I have a Peavey MiniMax bass amp that has DC on the output. With no speaker attached, the + and - rails are reading solid +/-75VDC, but P6 (red positive speaker connection has -3VDC on it). If I connect a dummy load (8ohm power resistor to it), the positive rail will oscillate from +75 to some much higher DC voltage, while the negative rail oscillates from -76VDC to something near 0V. The positive speaker terminal also goes from -3VDC to 0VDC during this oscillation. This oscillation continues about every 1-2 seconds.

I've pulled the board and measured the diodes and found no shorts. I also didn't find any shorted caps, but that's not to say I didn't overlook something in my testing. Maybe U6? Any suggestions on what possible causes are or measurements to take? Thanks for your help/guidance!

See sheet 2 of attached for the power amp schematic.

MINI MAX PWR.pdf

[Jazz P Bass]

U6 is an all in one power amplifier ic.
At idle (no signal) it should not do anything as far as driving the output mosfets is concerned.
If there is not a spurious input signal observed then it comes down to ic 6.
Look at Q7/ Q8 gate pins to see if they are being driven at no signal.

[Jon Snell]

There are a number of possible reasons to have DC on the output;
The saw tooth wave is not symmetrical causing an offset,
There is a leak in an output transistor etc.
Mos Fets usually destroy themselves when they fail under excessive current/overload conditions.
Q6 is unlikely the culprit.
With your scope, is the output waveform symmetrical?

[Delta362]

Just put scope on gate for Q7 and Q8. Now my output is measuring a solid 75VDC. Also have 75VDC on Q7 gate and source. Looks ike Q7 is now perhaps shorting out.

I did notice prior to this test, that sometimes the DC offset would dissipate towards 0V after about 10 seconds, and wasn't always a consistent DC offset value, but now it's a solid 75VDC. I read 185 ohms from Q7G to speaker positive. Q7D is 1ohm to speaker red, and Q7S is 0 to speaker red.

[Jon Snell]

Maybe Q7 has failed ... change the pair.