Oscillations gone wild... Kenwood L-07MII

[EchoWars]

Never fought anything quite like this...

My personal amps, just acquired (cheap) because - of course - they're dead. Anyway, got to working on the first, and there's 12V of offset. Turns out that it's a simple dead zener for the input diff pairs. Replace, and now it comes out of protect.

Fire the amp up with the bias turned all the way down, and output is stable. But...turn the bias up to anything near to the 25mA spec, and it busts into oscillation after a few seconds (triangle wave, 4MHz or so, about 3V P-P). Once oscillation is started, only power down will kill it (but you knew that, huh?).

Figured a damaged semiconductor someplace, so I took the driver board out of the other amp, verified that it is working, and swapped the drivers, then the pre-drivers, then...well, you get the idea. Eventually I'd swapped every transistor on the driver board from the good board to the one with the issues. Nothing has changed.

I figure a flaky cap at this point, but this board has a slew of them. But before I go off on one more wild goose chase, is there any on this board that might be more likely to be the culprit?

[ilimzn]

You probably tried this already but looking at the schematic I really don't see anything obvious to point out, so:
- Clean the driver board thoroughly
- Check ground and power connections
- Check zobel network
A triangle wave seems to imply some sort of positive feedback and histeresis, like there is a portion of the output coupling to a non-inverting input node somewhere. I'm guessing the problem is happening without a load, so the fact that increasing idle current sets it off also suggests coupling through the power supply lines is possible.
Does the amp only oscillate or does it also amplify while doing so - i.e. superimposes the oscilation waveform onto the amplified input?

[EchoWars]

I had not tried feeding it a signal while doing all this, as I assumed that it would amplify a signal normally until the oscillations begin, so I tested and with bias turned down to a few mA and it amplifies just fine. Turn up the bias, and I get the oscillation modulating the 1KHz test signal.

Whatever it is, it is specific to this driver board, as I can put the one in from the other amp (the one that now contains all the semiconductors from the driver board that is giving me fits) and with this driver board it works fine. Amplifies normally, bias adjusts normally, and not a trace of instability.

I de-fluxed the troublesome board, and examined the traces carefully under a magnifier and I can see nothing of concern. Board looks like new, and all solder connections look very good.

Has to be a bad cap...

[ilimzn]

Hm... might sound crazy, but have you put the 'bad' driver board into the 'good' amp?

[EchoWars]

The other amp has more serious problem (I said both were dead... ) Can't swap the bad board into the other amp till I get around to fixing it too...

I have to set the Kenwood aside for a day or so, but I'll get back to it before the end of the week. That said, apparently you think the 'bad cap' idea is off the mark... ?

Every other time I've fought oscillation, I've found a damaged semiconductor of one sort or another. But the only ones not changed out on this board are De9 and De10 (STV-4H multijunction diodes), and I'm doubting that they are the issue.

Where's the 'head scratch' smiley?

[anatech]

Hi Glenn,
I'm suffering a similar problem with an MC-2505, only not nearly as bad. Low level oscillation and it's the driver board. Everything checks good.

As ilimzn suggested, the caps. If it uses box polyestors, try replacing the lot. I have found them to go open in some amps. Your diodes are probably not the cause.

Have you replaced the electrolytic caps? The ceramics?

-Chris

[EchoWars]

New electrolytics are in place.

This uses some strange-looking caps. What's wierdier, is that many of them are on very long leads, some nearly 2" long (leads insulated with tubing). They are just there...hovering above the board.

I'm ready to swap the lot of them. Problem is finding a few odd values like 4pf and 8pf. Probably use a 5pf for the 4, and a 10 for the 8. Likely the best I can do...

[Roar Malmin]

These amps can be quite hard to tame. Try putting som capacitance from base to collector on the drivers. Qe14 and Qe15, something in the vicinity of 22p to 100p.
Roar

[ilimzn]

Originally posted by EchoWars
That said, apparently you think the 'bad cap' idea is off the mark...

No I don't - in fact, now that you mention the caps on long leads, it may very well be possible you have a cap problem. Do they look like small squares dipped in epoxy?
BTW I have a Sansui AU-D9 here with a similar, but unfortunately self-destructive problem, but in my case it appears to be a case that needs a semiconductor change. It happens about every 7 days

[JohnW]

Seen this kind of problem before, try replacing Ce22 (47uf). It should be bypassed by Ce17, but you can never tell....

John