Runaway tubes or tubes drawing excess current--baffling

[trombone]

This has me stumped. Harmon Kardon Citation II, rebuilt circuit, not the original.

Symptoms. Both tubes on one channel run away or are drawing excessive current.

I measure the voltage over the cathodes of the KT90s. Usually it's around a volt. But most of the time, it simply takes off in that one channel, getting over 2VDC before I turn it off. (I often do this with a Variac.) I can usually hear the little transformer I have in series with the main power transformer to boost the B+ getting louder, and this doesn't sound normal. The amp's s been on the bench for a week with this problem. It's intermittent. Sometimes I can turn it on several times in a row with no problem, but another time the problem kicks in. For a while, the problem was in one tube (or seemed to be, but I may not have caught it), but now it's definitely in both, and they rise together. I have individual bias pots, but this problem is in both tubes, so I can't see how it can be a pot that's causing this trouble.

What I've tried.

Swapping tubes from channel to channel. Problem remains in the original channel.
Replaced tube sockets on that side.
Replaced coupling caps on that side. I'm not convinced one of the originals isn't bad, but it's now happening with the replacements, so I'm guessing it isn't.
Checking for bias on the tubes when the tube is running away. It's there. About -50VDC on the coupling cap-grid connection, something like that, checking quickly. In any case, I replaced the pot to ground resistor and resoldered all bias connections, plus all those on the KT90s.
For a while, I thought maybe the input jacks were picking up an AC signal somewhere, but I can't see from where.

Hypotheses
This thing is picking up an AC signal, or something's drawing a lot of current from the tube somehow. That would suggest a short, but I haven't a clue from where. Could this be a bad output transformer?

I'm stumped. Any ideas where to look for a problem?

[SY]

You boosted the B+? That could be an issue. My first suspicion, though, given what you've seen so far and what you've already replaced, is the grid leak resistors. Replace them. If there are grid stoppers, replace them as well. If there aren't, add them. Screen stoppers as well.

Second is a possibility of oscillation; it may be worthwhile to attach a dummy load and a scope to the output, then bang on the input with different signals and see what happens.

If Jim McShane is looking in, he's worked on more of those particular amps than I have and may have more suggestions. Kevin Kennedy as well.

[trombone]

Thanks. You pushed me to check grid stoppers, and I found a wiring error, made when I replaced the sockets. I bypassed the grid stopper by mistake. Fixing that should solve that problem. The problem existed in a different form before I replaced the socket, but I'm hoping that I solved that problem alas, only to create this new one. Fingers officially crossed when I get a chance to return to it. But I wanted you to know now that I found something.

[SY]

Cool! Let us know if that fixed things and if not, we'll take another stab at it.

[trombone]

I'm getting nowhere. Fixed the wiring errors, replaced the grid snubbers, grid leak resistors, 1 screen resistor (have to get another one), nothing helps. It's still occasionally intermittent, but mostly both tubes take off. I should add that I boosted the B+ many years ago with no problem, and the other channel is fine. I also tried swapping the small tubes from channel to channel. Voltages seem normal, too. I have to say, I've never run into anything like this.

[quikie22]

Did you just replace the tubes? Could be gassy tubes.

My newly built amp seemed to intermittently runaway and melted through a couple of cathode resistors. I thought it was oscillation or over-dissipation or leaky caps. Then I noticed a bluish glow in the tubes. I baked the tubes and since then no more runaway.

[trombone]

No, these tubes have been in for a while. I haven't noticed a bluish glow, though I haven't tested for gas. As I said, I switched them between channels. Even worked a fifth tube in. No effect. I could try a set of KT88s that I have on the bad side, see what happens.

This amp uses a differential driver stage and cathode feedback. I wonder if the problem is in one of those places. I may try hooking the left driver stage to the right output tubes and vice versa. Maybe the left bias pots to the right output stage, etc.

[quikie22]

Have you checked the dissipation of G2?

If G2 over dissipates, it may push the system into runaway.

[trombone]

Sorry. What is G2? (May not apply. I'm not using the original HK circuit.)

[quikie22]

G2 = screen grid. Pin 4 of the KT90. I believe it is connected as ultra linear in the Citation 2 via a 270ohm resistor to the ultralinear tap on the OPT. Your higher B+ may have caused the G2 to overheat.