Deluxe 112 oscillates

[Alex R]

Hi all, not been in here for ages but this one has stumped me. I tried to attach the service manual, hope it worked.

OK so this Fender Deluxe 112 solid state came in with the complaint of intermittent loud squealing. Tried it out and indeed it did squeal with the volume over about 3. Opened it up and R4 and R5 were burned open, these lift the circuit ground from the chassis, so something seems to be trying to leak current to ground. (I imagine the two resistors are 47R as on the schem though the parts list says 470R.) Fitted 2 x 47R and changed the main caps to keep the internet happy - and now the amp lights up the bulb limiter. Output transistors aren't shorted.

Thought I'd lift the secondaries to check the transformer. Took off the two AC secondary coil ends (CP3 and CP5), it still lights up the limiter bulb. Took off the centre tap and the limiter goes out. Now here's an odd thing, when I put the centre tap back on, the limiter bulb doesn't light. Secondaries back on and it lights up, secondaries off but CT still on it still lights, CT off it goes out, CT back on it stays out, secondaries on and it comes back and so on. Like the CT ends up on one of the (unpowered) rails but comes off it with everything disconnected. The amp doesn't seem to be oscillating when this happens. Some artefact of caps charging/discharging I guess.

Emitter resistors are ok, so are opamp supply zeners. All diodes and zeners in the amp are good.

I looked for shorts in the MC1436 but couldn't find any, but I might swap one in and see what that does.

Any ideas?


deluxe_112.pdf

[mozz]

470R would be 470 ohms.

[Enzo]

Please assure us you do NOT have a load connected during these tests.

What wattage bulb are you using?

I would believe 47 ohms for the two resistors before I'd believe 470.

I find it hard to come up with a scenario that burns those resistors. They connect the pot brackets to ground, so to get them burnt we have to have a situation where the brackets and common are not at the same potential.

[Mick Bailey]

I had a late-model Sessionette in for repair last year. It had an elevated ground - I think that was a 47R resistor with a cap in parallel. Can't remember the cap value, but I do recall it being 450v. Anyhow, the resistor was open and the cap sooty and also open. I spent a lot of time thinking how that could have happened. The only situation where I've personally had mysterious component burnt out is when we've been hit nearby by lightning. We've had three strikes in the past 6 years that have caused damage to equipment in the house. It's also taken out my electric fence energizers (including cutting the wire) and hit my oil tank remote transmitter. The actual direct strikes have been about 200 metres away, but close enough to cause problems.

[Enzo]

Chassis to common elevation - as on many Marshalls - can burn up resistors if something is connected with a hot chassis. But I don't see how that could happen here.

[g1]

As Enzo asked, do you have the load connected when powering up on the bulb? If so, disconnect load. If it powers up dim, check that there is zero DC on output and if none, reconnect load while still powered up.

[TomCarlos]

And this may not have anything to do with your specific problem.... but...

I remember seeing posts (on this site and others) about this particular amp with oscillation related to the reverb circuit and/or reverb tank. So in addition to removing the speaker load (as suggested previously), you might consider disconnecting the reverb tank too. See if you can troubleshoot to get to a stable amp after dealing with the burnt components. And once that happens, I would connect the reverb tank and retest before connecting speaker.

Just trying to be cautious...

[Alex R]

Many thanks for the helpful replies. In answer - yes I too think they are 47R not 470R; no load was connected, and the bulb is 60 watt.

So it was either a mains spike or a mystery. Do we think that it is not really possible for oscillation to have caused the burned resistors? In which case one wonder whether the sequence of events was, spike > signal earths burned open > oscillation. Question being what else might have been damaged? I think I'll get an OPA445 to sub in for the MC1436 driver opamp and see where that gets me, interested to hear of any other suggested approaches. Thanks again to all.

[Alex R]

Well friends this thing seemed to settle down in the end, but clearly I hadn't found the root problem because it came back to me after a few weeks having shorted Q1 (TIP142 output darlington) after working ok for several shortish sessions. Still acting weird. I'll list my observations hoping someone can see some link or other. Problem looks a bit different now (ok could be a different problem but anyway my guarantee covers it so I have to sort it!). Those 47R raised earth resistors are fine this time.

I've replaced Q1 and Q2, R74 and R75, CR5-12, R70-73, C39 and C40, CR13 and CR14. Makes no difference.

It draws too much current through the outputs because the bias is around 3v base to emitter on both outputs. This is also observable with the outputs removed, when the amp will tolerate full mains power (otherwise light bulb limiter and variac keep it under control.

All this is with no load connected, though when I connected one experimentally with the limiter on and the variac 50% it made no difference, still trying to burn itself up. With everything limited it's 0.3v or so across the emitter resistors.

It balances pretty well, pin 6 on the MC1426 as well as the speaker out remain close to zero and the rails are correct and equal. Supply to the MC1436 is a bit high at +-35v but if I dial it down till the rails are 40v instead of 45v it's correct, just a high wall voltage I guess. I don't have a spare MC1436 but other single high-voltage-supply opamps in there make no difference (I fitted a DIL socket).

CR10 and CR12 are 3.9v zeners, right? I've already found one error in the parts list - I'm not a good enough engineer to be sure but wouldn't that tend to keep the bias up above 3v? That's the only possibility I can think of, that those zeners are the wrong value on the parts list.

Any other ideas? Service manual with schematics is in the OP.

[Alex R]

... alas I chucked the old zeners away after the first time I replaced them so I can't check that.

Oh I also removed the power amp in and headphone jacks and swapped them around in case something was going on in the headphone wiring. No difference, jacks seem fine.

[g1]

The zeners are not for biasing, look at CR5,6,7, & 8. Two diode drop for top and bottom sides should make approx. 1.2V B to E for each ouput device.
Bias string is R70, R71, CR5, CR6, CR7, CR8, R72, R73.
U1 pin6 should be about zero volts DC. Verify that and then measure voltage drop across each of those diodes in the string.

[Alex R]

The zeners are not for biasing, look at CR5,6,7, & 8. Two diode drop for top and bottom sides should make approx. 1.2V B to E for each ouput device.
Bias string is R70, R71, CR5, CR6, CR7, CR8, R72, R73.
U1 pin6 should be about zero volts DC. Verify that and then measure voltage drop across each of those diodes in the string.

Ah thank you, that really helps, my theory isn't quite as good as it should be, now I have a line of enquiry to follow. I'll be back in the shop on monday and will report findings, thanks again!

[Alex R]

Okeydokey I fixed it thanks to g1, with further help from this thread https://music-electronics-forum.com/...-sounding-fine, especially moswell's final post.

OK so after the last post I took out Q1 and Q2 and found that despite looking ok on the component tester on my Hameg scope, and the junctions beeping the diode tester on my Fluke 77, they were leaky. That's why the bases looked so very high - leakage, mainly. New set of outputs and new bias diodes then. That fired up ok but ran hot and got hotter. The bias diodes CR5,6,7 and 8 sit inside the heat sink so the bias voltage they supply drops with heat. I put a bit of heat transfer paste in with them but it still ran hot and after 5 mins or so, current draw started to creep up.

I took a look at the thread above and thought, the new diodes are not quite cutting it - I guess they've got better at standing the heat. I tested on to see and the reduction in voltage drop with heat is still distinct, but I guess not quite enough. I tried the recommended 47R resistor across one diode per side, but this cooled everything down too much, 220R trimmers across one diode per side and I could dial in the bias I wanted, then fit permanent resistors. This worked well. Thank you moswell!

I also fitted the zobel network recommended in that thread, to try to guard against the return of the osdcillation that started all this.

The one symptom I'm not sure about is with power transistors removed the bias voltage still looked high. I'm going to assume I got that wrong somehow.

Not my most profitable job but a good learning experience - it's good to go and that makes me happy!