JCM 2000 DSL 100 bias thermal runaway

[PremiumPlus]

Have a JCM2000 DSL 100 that is giving me problems with bias control. The bias voltage on one pair of tubes, the pair next to the phase inverter, starts low (more negative) and then as the amp heats up the bias voltage collapses slowly. The other pair of tubes remains stable. It's really sensitive to temperature, so much so that if I simply blow on the pc board the bias voltage returns. I've replaced 5 220 k resistors, three 47uf caps, resoldered a ton of joints including the power tube sockets, used new tubes, etc. I changed the 33K resistor to 39K to give me some more negative voltage headroom. I've got 5 hours in this stupid bias circuit and I am ready to build a simple bias circuit like Fender used back in the day and be done with it...unless anyone has seen this before and can suggest something. I haven't seen anything like this before, and I feel silly asking for help but this one is kicking my butt...

[drewl]

Common problem due to bad resistors in the bias circuit and the circuit board becoming conductive.
here's a page about the problem and what you can do to fix it.

The Marshall TSL122 JCM2000 Repair/Mods Page

[PremiumPlus]

Lots of great info, drewl. Thank you!

[drewl]

If you read the whole thing you see what a nightmare it can be.
Some I've been lucky and only had to rebuild the bias circuit.
The last one I had the pc board between pins 4 & 5 of the output tubes had to be removed as it became conductive once the amp warmed up.

On some of the output tubes the trace for the bias to pin 5 runs under the trace for the screen pin 4, that whole trace had to be removed and 5.6K grid resistors soldered right on tube socket pin 5.

[PremiumPlus]

Appreciate the heads up...I'm going to get some flameproof R's today for the bias circuit and change the grid resistors from 220k to 5.6k while I'm in there.

Was the pc board burned on the one you found conductive, or did it develop conductivity by another cause? I've seen boards get conductive due to burning and carbonization but never just on their own, at least without damage of some sort. Any idea of what the root cause of that could be? I mean...it's fiberglass, and so I'm guessing it might be a conformal coating or dye in the fiberglass that becomes conductive...?

[Amp Kat]

There is also a 1 ohm current resistor between cathode and ground that will cause thermal runnaway so make sure it isn't open but sounds like the conductive board is the issue if you can blow on it and get it to stop. This would cause it no matter what.

[jamesmafyew]

Was the pc board burned on the one you found conductive, or did it develop conductivity by another cause? I've seen boards get conductive due to burning and carbonization but never just on their own, at least without damage of some sort. Any idea of what the root cause of that could be? I mean...it's fiberglass, and so I'm guessing it might be a conformal coating or dye in the fiberglass that becomes conductive...?

Whatever the specific problem material, yes, it actually gets conductive when it gets hot. The problem seems to manifest more as the amp gets older, and it's worse in the combo versions where the tubes are underneath the chassis. In successive revisions of the mainboard, you can see that Marshall changed from carbon film to metal film resistors, then around issue 6 the PCB material changed dramatically. It's like they thought the issue was resistors with negative temp coefficients, so they changed resistors types, but then the problem really turned out to be the PCB. It's wild but it's true.

If the 1-ohm current-sense resistors are open, as Amp Kat mentioned, then you will see the voltage on the bias test points rising way out of the mV range. If the readings at the test points look normal (as in, somewhere between 0-150mV), then the 1R are fine.