SA-8800 Oscillation?

[QSilver]

Went to switch it on today and it took longer than usual to come out of protection. Then I heard a slight crackling noise from inside the amp. Turned it off, and smoke came out from the front right corner of the amp. Damn.

So I took it to the bench. R80 had burned up. I changed it and replaced Q16 and Q18 with KSC1690 and KSA1220 - I've seen MTF recommend them on another threat for the 8800.

I then powered back up and the DBT stayed bright Powered off and removed the output transistors just in case they were blown but if they weren't, to protect them.

Powered back on, and the bulb goes dim, and the amp comes out of protection. So I started probing with my scope and found that there is a high pitched waveform near 20kHz on almost all the transistors on the right channel. The left working channel is quiet as ever when comparing. I can trace the waveform back through from the driver stage down to the base of Q14. Here, it looks like a sawtooth. Afterwards it looks similar to what I have attached.

The pre/power amp jumpers are removed.

I've not really dealt with oscillation before and especially with an NSA...

As a precaution I've dropped the idle current down to minimum on both channels. Maybe I'm getting this oscillation because I've removed the output transistors? I'm not sure if the oscillation can travel back through the amplifier or it it really is starting at the first stage...

 

[QSilver]

Looking at it again today, I checked the output transistors first. Q3 (2SC2525) tests blown open on my transistor tester but not on a DMM. The rest test ok on the transistor tester but the best thing is likely replace them all. I was thinking of using MJL3281 and MJL1302. 30Mhz but that's acceptable for the output stage?

I've pulled the driver transistors, they test ok, but with them removed, the amp is still giving me this saw tooth wave...

 

[rcs16]

Hard to say exactly what is causing the oscillations. Usually something to do with too much phase shift margins in the amp causing instability. Too much phase margin usually causes oscillations. Using KSA1381/KSC3503 might be better as they are faster devices than KSA1220/KSC1690
Since MJL3281 and MJL1302 are about the fastest you can get these days worth a try.

Maybe I'm getting this oscillation because I've removed the output transistors?

yes that is very much a possibility.

 

[QSilver]

I reconnected Q4 and Q8 (Q3 has blown) and the amp came on and was stable. I then powered down and then fitted the MJLs in place of the 2SC2525's and the 2SA1075's.

Powered up and still stable. I have one problem though. The idle current. I can get to 56mV just fine. But the second trimmer does not lift the current past 65mV... I tested the left channel amplifier which still has its original power transistors in but has had the same transistors replaced on it as the right channel and it can be adjusted to 70mV easily just after half way on the pot.

Right channel just cant get past 65mV with the trimmer all the way around. Does the circuit need further modification to accommodate these new outputs?

 

[markthefixer]

I’ve replaced them with MJL1302 and MJL3281, along with changing all the other transistors in the power amp section for both left and right channels with the exception of the driver transistors. The left channel still has the original output transistors for now but besides that is identical to the right amplifier with the new output transistors. The problem I have is that I cannot get the second stage of the biasing up to 70mV. I can set the bias easily with the left channel but not the right. The first stage (to 56mV) is easily adjustable but the second adjustment does not lift the idle past 62-65mV. Have I done something wrong? I’ve seen others using these transistors and no issues with idle current.

Ok, found the thread. Haven't been in much - pretty wrapped up with non AK stuff....

Really don't want to reward a private message for help, BUT I've been gone for a while, and this thread has been sitting waiting for the weekend and half the week.

when I say "replace all the SILICON in the amplifier" I do mean transistors AND DIODES.
and that already assumes 100% resistor testing with one leg lifted. Some surprising sneak
paths with damaging currents are formed when components pop.

I suspect D20 or D22 , both are used for their voltage drops, 1n4148 are fine. Is VR10 (70mv adj) at it's minimum or maximum resistance?
Next would be D16 and D18, all part of the Q28, Q30 current sources for the NSA circuit.

Also Q32 and Q34 are part of a power limiter circuit that could be misbehaving.

With oscillation, I would 100% test all the caps, first for shorts, then for capacitance.

 

[QSilver]

Thanks for your help Mark. I should have changed the diodes... I've changed D20 and D22, there was no change. Changed D16 and D18 and also found a tiny solder bridge across the C-E of Q20. My fault, I should have seen that... I removed it and now I can get 70mV on the second idle adjustment trimmer at just over halfway for a test. I've not tried this off a DBT yet... I think the left channel is now unstable.

After powering down the right channel flat lines. Now the left breaks out into all sorts of noise and waveforms. In each channel I've changed every transistor with exception to Q27 to Q30 as I was unsure of a suitable replacement. The right channel has new outputs and I changed Q12 and Q10 for gain matched pairs of KSC1845 and KSA992 respectively. I'll try changing the outputs in the left channel too, to keep the amp symmetrical - if the right has blown it might not be long till the left does anyway.

I've run out of 1N4148, more on the way I'll replace all the 1S1555's with those. Can I also replace 1S2471 with these?

For D11-D14 (10E2), 1N4004 will work?

Will 6V zener's be acceptable for the MZ-061 zeners?

 

[QSilver]

So I have now changed the outputs on the left channel too as I had some really strange breakdown when it powered down and popping. After changing the outputs, I still had the problem. I pulled the drivers out and changed them for KSC3505 and KSA1381. This stopped the breakdown but I had some oscillation in the driver stage. I then changed all the 1S1555's with the 1N4148's, the MZ-061's with 1N5233B's and 10E2's with 1N4004.

The left channel seemed mroe stable but on adjusting the bias, throws itself into oscillation. At the same time, the right channel will go into oscillation too. The right channel still has the original driver transistors but they are all looking VERY tired. The one's a pulled from the left are heavily discoloured - even the lettering and front of the case has changed colour.

One thing I have noticed. My board is GWH-128 not GWH-115.

My board has what I think are FOUR STV-2H's. Where R83, 84, 85 and 86 are located, I have STV-2H - its also marked on the silkscreen as being the STV.

I did let the amp warm up on DBT but once the heatsink got warm it broke into oscillation (maybe thermal runaway?)

Have I messed up with driver transistors?

 

[QSilver]

Picture showing the state of the drivers I pulled from the left channel, and the "oscillation" I can see around the driver stage of the amplifier.

 

[QSilver]

Actually, having read the data sheets, maybe the KSC3503/KSA1381's are no good as they have very little current handling capabilities compared to the KSC2690A/KSA1220A? So I should probably switch to those but then why is it oscillating so much?

 

[markthefixer]

I repeat:

With oscillation, I would 100% test all the caps, first for shorts, then for capacitance.

because I have seen nothing addressing this. there are some critical power supply bypasses IIAC on the board, as well as the collector to base anti-oscillation feedback caps to kill the gain at ridiculously high frequencies beyond the just scary frequencies these things can operate at.

 

[QSilver]

Sorry, I should have checked.

I've replaced the electrolytic capacitors, but I have now removed and checked all the other capacitors. I didn't find any shorts on the capacitors. I removed them to test.

C55: 58.8 pF
C53: 53.9 pF
C51: 0.01 uF
C49: 0.01uF
C47: 0.0047uF
C45: 40.3 pF
C43: 3300 pF
C41: 47.7 pF
C39: 48 pF
C37: 48 pF
C35: 225 pF

I found some strange behaviour too. The oscillation starts in the driver stage and on power up, the idle current is around 50mV and rises. I tried to probe the amplifier to try and find where the oscillation might be starting. I put my probe on the base of Q41 and the idle current dropped to 0.6mV and didn't come back until I turned it off and back on again. Oscillation is unchanged.

 

[QSilver]

Starting to wonder if I've screwed up with transistors somewhere... here's what I've changed for what...

Q7, Q8: 2SC1775A ---> KSC1845EBU
Q9, Q10: 2SA979 ---> KSA992FBU x2 (gain matched)
Q11, Q12: 2SC2291 ---> KSC1845EBU xs (gain matched)
Q13, Q14: 2SA750 ---> KSA992FBU
Q15, Q16: 2SC1915 ---> KSC2690AYS
Q17, Q18: 2SA905 ---> KSA1220AYS
Q19, Q20: 2SC1400 ---> KSC1845EBU
Q21, Q22 Q23, Q24: 2SA750 ---> KSA992FBU
Q25, Q26: 2SC1400 ---> KSC1845EBU
Q27, Q28: 2SA904A ---> KSA992FBU
Q29, Q30: 2SC1914A ---> KSC1845EBU
Q31, Q32: 2SC945A ---> KSC1845EBU
Q33, Q34, Q35, Q36: 2SA773A ---> KSA992FBU
Q37, Q38: 2SC945A ---> KSC1845EBU
Q39: 2SC1913 ---> KSC3503
Q41: 2SA913 ---> KSA1381

 

[markthefixer]

Try it with the outputs disconnected. The amp will DC balance fine without them. Don't touch the idle adjustments.

There still are caps I see that are not on your list - c61/c62 and others as I spot checked from the schematic.

 

[QSilver]

I tested
C61: 13nF
C62: 11nF

There are four capacitors fixed to the bottom of the amp board. I tested these and they all measured 4.7nF each.

I haven't tested the right channel's capacitors yet but I will do that next and post the results.

I lifted the legs on the outputs to disconnect them. All outputs are now disconnected. I checked the collector Q10 (that connects to Base of Q14) and the same point on the left amplifier (Q9) with my scope and I've attached the images.

The scope was set to the 1-5kHz range on 1/10 attenuation. The top trace is the left channel, the bottom trace is the right channel.

Next, I checked the Bases of the driver transistors, same range but the attenuation set to 1/300.

I finally checked the output of the amplifiers, attenuation back to 1/10.

 

[QSilver]

Ok, so I tihnk I have now tested all the rest of the capacitors on the amplifier board.

C56: 53.8 pF
C54: 56.9pF
C52: 10.3 nF - This one moved between 10.2 to 10.5nF while I left it in the tester.
C50: 10.21nF
C48: 4.96nF
C46: 39.8 pF
C44: 3420 pF
C42: 48.1 pF
C40: 47 pF
C38: 47.2 pF
C36: 226 pF

I made a mistake in the capacitor number on my previous cap post, I've edited that now.

 

[QSilver]

Have I missed any more capacitors?

 

[QSilver]

I left this for a little while to think about what to do. I got back to it yesterday after getting some more parts. I got some more diodes to replace the Zeners with (6.2V).

I had missed a few diodes when I was replacing some (D12, D4, D2, D24, D26) Replaced all of those so now every diode except for the STV-2's has been changed. I powered up to check and the left channel is now STABLE.

The Right channel isn't so messy on the scope, but I think I might've had an issue with my scope that I've now fixed (dirty contacts between boards). Also got a new, better DMM.

So the left channel drivers, Q40 and Q42. I checked the frequency with the frequency coutner on my DMM and Q40 comes out at around 19kHz. Q42 measures at 14kHz.

They have been changed to KSC3503 and KSA1381.

I checked the Base voltages too which seem too high...

Q40 Base: 2.1V
Q42 Base: 3.1V

This is the amp channel that originally blew, taking out R80 and an output transistor.

 

[QSilver]

Did some more testing and for a while, I could stop the oscillation with a cool air spray around Q40 and Q42, and then start it again with cold air on to C44. I swapped over C44 and C43 but this made no difference. Swapped the back and now I cant so easily stop the oscillation.

Now both amp channels are oscillating again. Between 5 to 20kHz. The left channel will sometimes completely stop oscillating. The right channel never stops but the oscillations will get much smaller if I start to probe other stages of the amplifier. I'm at a loss as to where to go next....

 

[rcs16]

Sounds like you have the design in instability, maybe a quick fix is to over compensate the amp a bit more, by increasing the dominate pole in the feedback caps, C45,46, start by increasing them 1 step size at a time, They are originally 39pF, so start by adding a few pF in parallel with the existing ones, could try adding 10pF or a bit more or less and see what happens. If you can get it stable then see if you can adj offset and biasing per the manuals procedures. Once it is stable you can run a test square wave through it and look at the step response looking for ringing at the transition points.

 

[QSilver]

Can you recommend a type of capacitor? I don't have anything that small. Will ceramic 10pF capacitors work ok?

Any idea what I might have done to destabilise it? Or why it has destabilised in the first place?

I had a thought of pulling the power from one amp at a time to see if one is setting the other offer off. None of these issues could be caused by a power supply fault could they?