JBL/Urei 6260 weird thing happens

[Alex R]

Ok attached is a schematic of this big old stereo power amp. It had many shorted transistors right through inc. half the power amp. I replaced:

Q26 to 29, Q19, Q30, Q5 and all the power transistors with the same types.

Q20 and Q22 with ZTX 653 and 753

Q4 and Q8 with MJE15032 and 33

Brought it up on the variac with no load. Fine. The whole amp drew about 800mA from the 240v mains after the protection relay kicked in. Checked for DC on the output. None. Put it on the 8 ohm dummy load and brought it up again. Fine. Put in a signal with the volume down. Fine. Turned it up a bit, then a lot. Nice clean sine wave output, no DC offset that I can recall. Then, as I got near to full output - pop. It blew Q20, Q4, two PNP power transistors and one of the NPNs.

So I replaced all this stuff and went through the whole process again... including, alas, the bit about going pop and blowing transistors. Imagine my chagrin, as they used to say.

I monitored the mains AC current draw during the test and it came up to about 1100mA - nothing unusual. Then pop it went. It did blow a supply fuse the second time around.

What's going on do you think? Have I got the subs wrong? Or could this be a fault I haven't spotted?

[Enzo]

Let me get this straight. Sitting there at idle itdraws 800ma from the 240v mains? That means the amp is using close to 200 watts doing nothing. I'd have to guess it was way underbiased.

Are all the 0.33 ohm ballasts OK?

R15 is not open I hope?

Since it holds OK until pushed, monitor mains current and turn the bias control for least mains draw. In fact the other channel may need tweaking. Once it works, we can worry about setting the bias more precisely.

I suspect that if it is conducting this hard at idle, it is fighting itself pretty hard under heavy load.

Checking R1 especially, but also R10, R14 and CR5. SInce Q20 blew, those parts were potentially stressed.

I suspect the outputs blew because they were both side conducting at the same time. And the failure of the outputs took Q4 with them.

[Alex R]

Thanks Enzo, that's clarifying things a lot! I don't yet have a good set of 'internal parameters' for ss amps but I am developing them. You know, like "75mA per tube? That's got to be way too much!" etc.

Yes, 800mA supplying both channels. So it sounds like underbiasing, just too much conduction all round and things getting a bit too Class-A-ish as the output gets towards maximum. I don't think the 'turn the bias down till the mains current draw starts to increase on its own, then back it off a bit' approach to biasing will work for this amp, as the 800mA was quite steady. Is there another method you can recommend, like a broad-brush area for mains power drawn at idle? No problem if not as I can now guess at the sort of place the bias should be, and will scope the output for distortion anyhow.

Many thanks again for this guidance, which is allowing me to branch out in the jobs I'm taking on so is appreciated not just as web chat but as genuine practical real live helpfulness.

[Enzo]

Disconnect the power from one channel at a time to see if one draws more current than the other.

AS to current not dropping for cold bias settings, I don't know why it wouldn't work, ther is nothing odd about this amp. Perhaps the current adjust circuit is not working?

SS amps tend to be pretty much class B until you get into fancy circuits with switching power rails etc. That means each polarity is biased right to the edge of turn-on. Bias it hotter than that point, and both sides are conducting at once, which basically makes a path from pos to neg rail through the outputs. As you bias hotter andf hotter, the current should ramp up pretty quickly. On the other hand, below that point, you can turn the thing colder and colder and it will do nothing but increase crossover distortion.

SO even if the 800ma is not due to bias adjustment, the amp STILL should respond in that fashion.

[Alex R]

Ah Enzo, now that I know the reasoning behind it I realise I had got the current draw bias method wrong. If there is a wrong end to a stick I will generally seize it with enthusiasm. So... you turn the bias down till the draw is as low as it will go, then you turn it back up till you see the current increase as you turn - because you're kind of taking it out of class B where it wants to be, into a place where both sides conduct at once with no signal - past the point where the bias starts to turn on the transistors at zero signal with no swing going on. Then you go backwards decreasing the bias and the current just to the point at which the current levels out and won't go lower (because rail-to-rail conduction has stopped). Much below that point there will be a gap as the signal crosses zero when neither side conducts. You're looking for the elbow in the curve of bias volts v current draw.

Seems so obvious now but I was looking for a condition like the one I saw in the overheating Hartke we sorted a few months ago - where the current draw starts to rise all on its own (as heat builds, I now realise - a different phenomenon)... so basically I blew this amp up twice by biasing it way towards class A (as it were). Guess it serves me right that I burned up all the profit...