Hello, my Audiolab 8200p power amp goes into self-oscillation on one channel and I'm not very familiar on how to diagnose such behavior so decided to ask here for advice.
If the misbehaving channel has 0 bias, there is no oscillation (with no load, no input signal)
if I increase bias current for more than ~10mA (~5mV on the r148/r248 resistor) it goes into self-oscillation. The scope shows it as around 3.4Mhz sinusoidal.
If there is an input signal with no bias, no load it produces a clean sinusoidal output, with resistive load on output (8ohm) it shows severe symmetric oscillation on the pos/neg peaks of sinusoidal signal no matter the bias setting.
When it oscillates I can see the oscillation everywhere in the signal path including the entry point of input diff pair.
The other channel is fine and doesn't behave like this producing clean signal.
I checked the 60v rails, and +- 12v for the opamp and they look fine. By the way I don't fully understand what the opamp does - is it a sort of servo circuit?
How would you approach such defect? Any advice?
The schematic is attached for the reference.
Thank you!
If the misbehaving channel has 0 bias, there is no oscillation (with no load, no input signal)
if I increase bias current for more than ~10mA (~5mV on the r148/r248 resistor) it goes into self-oscillation. The scope shows it as around 3.4Mhz sinusoidal.
If there is an input signal with no bias, no load it produces a clean sinusoidal output, with resistive load on output (8ohm) it shows severe symmetric oscillation on the pos/neg peaks of sinusoidal signal no matter the bias setting.
When it oscillates I can see the oscillation everywhere in the signal path including the entry point of input diff pair.
The other channel is fine and doesn't behave like this producing clean signal.
I checked the 60v rails, and +- 12v for the opamp and they look fine. By the way I don't fully understand what the opamp does - is it a sort of servo circuit?
How would you approach such defect? Any advice?
The schematic is attached for the reference.
Thank you!
Hello Onokon I don't envy your task of trying to stop the oscillation.
I agree with you, I think that the opamp is a DC servo circuit. Needed because there isn't a capacitor in series with the feedback resistor R115 and the Dual FET, Q101, matching is unlikely to be good enough.
Beyond probing all around the circuit trying to find the source I can only suggest checking the output stage electrolytic power supply decoupling capacitors C113 and C114. They might have dried out, any low value electrolytic may have failed due to time and heat.
Also prolonged oscillation can take out the Zobel network so R149 might have failed due to excessive power dissipation.
There are a number of low value capacitors around the circuit so stability may be dependent on transistor types if any have been replaced.
I agree with you, I think that the opamp is a DC servo circuit. Needed because there isn't a capacitor in series with the feedback resistor R115 and the Dual FET, Q101, matching is unlikely to be good enough.
Beyond probing all around the circuit trying to find the source I can only suggest checking the output stage electrolytic power supply decoupling capacitors C113 and C114. They might have dried out, any low value electrolytic may have failed due to time and heat.
Also prolonged oscillation can take out the Zobel network so R149 might have failed due to excessive power dissipation.
There are a number of low value capacitors around the circuit so stability may be dependent on transistor types if any have been replaced.
First questions always have to be...
Is this your own amp and it has developed this specific fault in normal use or is it a recent acquisition with an unknown history?
Also is the amp 'all original' to the best of your knowledge or has it been worked on and parts replaced?
Is this your own amp and it has developed this specific fault in normal use or is it a recent acquisition with an unknown history?
Also is the amp 'all original' to the best of your knowledge or has it been worked on and parts replaced?
Thank you for the suggestions, I'll check all the elements named, special attention to electrolytic caps, opamp and it's surrounding components.
The history of the amp is unknown, but carefully looking at the PCB I can say with confidence that it has all the original components and not touched by anyone.
I more or less understand how the amp works, except it's servo circuit.
My current plan is to check some obvious suspects - electrolytic caps, then what, I don't know, maybe try to isolate the output emitter follower with it's bias circuit from the input diff pair to make them work is isolation and try to understand what half causes the oscillation?
I will update the thread if I will pinpoint the reason.
The history of the amp is unknown, but carefully looking at the PCB I can say with confidence that it has all the original components and not touched by anyone.
I more or less understand how the amp works, except it's servo circuit.
My current plan is to check some obvious suspects - electrolytic caps, then what, I don't know, maybe try to isolate the output emitter follower with it's bias circuit from the input diff pair to make them work is isolation and try to understand what half causes the oscillation?
I will update the thread if I will pinpoint the reason.
Last edited:
The DC servo circuit appears to be similar to that shown in 'Audio Power Amplifier Design, Sixth edition, by Douglas Self, Figure 23.5.
I think that you can disable the DC servo by disconnecting R162 which isolates the opamp output from the FET gate, Q101 pin 6 (via R122). The output will have a high DC offset but I think at should be less than +/- 500 mV which should be OK for testing.
I think that you can disable the DC servo by disconnecting R162 which isolates the opamp output from the FET gate, Q101 pin 6 (via R122). The output will have a high DC offset but I think at should be less than +/- 500 mV which should be OK for testing.
Fair enough 🙂 It is an unusual fault in itself but would possibly be expected if output and driver transistors had been substituted. Be absolutely sure device markings look identical between channels.
The servo should be playing no part in the oscillation (but never the less check the opamp output pin to make sure it is clean DC there). R162 is actually part of the feedback return network and appears electrically in parallel with R115 (110 ohm). If isolated it will decrease the closed loop gain of the amp and that would worsen stability margins.
Ordinarily that symptom can suggest a poor design but given this is a mass produced item and providing you are 100% sure all parts are correct then you have to look at obscure things.
That would definitly be something to check. Beyond that make sure the grounds are correct with no cracked/damaged print.
Also check the caps C106 and C107 and R121. Make sure they connect to where they should (no broken print).
Make sure C103 and R104 at the input are correct and have not been altered.
Also check C104.
From your description, sounds like output stage is oscillating. Check all the soldering around it and make sure the screws are tight. Check R149 4R7 isn't open circuit. Also check C115. Has anybody changed the output or driver transistors?
I raised the pcb and see some faint traces of flux around 2SA1216 output transistors and R148 resistor in the problematic channel, so maybe it was tampered in some way after all. It can be factory touches, but I'm not so sure now.
All the output transistors have markings 38P or 38Y, but the transistor that has flux traces has a marking 26P.
Would it be a good idea to swap output transistors between channels to see if they cause the problem I wonder.
I checked Zobel resistor and capacitor, resistor seems fine, the C115 cap measures as 40nF (esr 8KOhm at 1khz) in circuit in the problematic channel, as 0.1uf (esr 200 Ohm at 1khz) in the good channel. This is suspicious, I will pull it to measure more carefully out of circuit. It's a poly cap so maybe it's a problem not with the cap itself, but with something around it.
All the output transistors have markings 38P or 38Y, but the transistor that has flux traces has a marking 26P.
Would it be a good idea to swap output transistors between channels to see if they cause the problem I wonder.
I checked Zobel resistor and capacitor, resistor seems fine, the C115 cap measures as 40nF (esr 8KOhm at 1khz) in circuit in the problematic channel, as 0.1uf (esr 200 Ohm at 1khz) in the good channel. This is suspicious, I will pull it to measure more carefully out of circuit. It's a poly cap so maybe it's a problem not with the cap itself, but with something around it.
Last edited:
Certainly sounds strange. It would take a lot of energy to destroy such a cap but I suppose rail to rail oscillation could do that.
I would expect original transistors to have the same markings between channels. Markings such as 38Y etc could be groupings (gain, vbe etc) and so if one stands out as different it might have been replaced.
Just dab another cap across it to confirm if its bad or not... but it sounds duff from what you say.
I would expect original transistors to have the same markings between channels. Markings such as 38Y etc could be groupings (gain, vbe etc) and so if one stands out as different it might have been replaced.
Just dab another cap across it to confirm if its bad or not... but it sounds duff from what you say.
The datasheet for 2SA1216 says that those markings is a lot number, so it's not a grade of the transistor (thankfully).
Yes, maybe...... although that can be a way of grouping devices on the basis those from the same 'batch' will be broadly similar. With an unknown history you have to look at all possibilities such as a seemingly correct part actually being a fake replacement and so on.
Change the cap and see where that gets you first.
Change the cap and see where that gets you first.
Replaced C115, the only suitable cap with required 10mm spacing I have is a noname X2 cap, should be good enough for testing.
Powered up the amp, set idle bias to recommended 22mv, no oscillation.
Connected 8R resistors to outputs, no oscillation.
Turned gen, 0.5khz to 25khz square wave - no oscillation, at 25khz the wave is still relatively square with no ringing.
Should I consider myself lucky? Or there are any other tests that should be done to make sure the oscillation is not there?
Anyway, thank you all for directing me the right way to diagnose this!
So, to summarize my experience, to start dealing with oscillations in commercial devices the first steps are:
- check zobel network
- check nfb networks
- check that your output devices and drivers weren't replaced with biscuit cookies by previous users.
Good to know.
Powered up the amp, set idle bias to recommended 22mv, no oscillation.
Connected 8R resistors to outputs, no oscillation.
Turned gen, 0.5khz to 25khz square wave - no oscillation, at 25khz the wave is still relatively square with no ringing.
Should I consider myself lucky? Or there are any other tests that should be done to make sure the oscillation is not there?
Anyway, thank you all for directing me the right way to diagnose this!
So, to summarize my experience, to start dealing with oscillations in commercial devices the first steps are:
- check zobel network
- check nfb networks
- check that your output devices and drivers weren't replaced with biscuit cookies by previous users.
Good to know.
I guess we will never know what has happened. Normally it is always the resistor that fails. I think you just have to take this at face value as a 'one off' type of issue.
If the amp had been deliberately over driven by someone testing at high frequency (lets say testing at 50kHz and over and at very high output) then that is when Zobel networks become stressed as they are not rated for that kind of 'non valid' use. Again though, its always the resistor that fries...
Its a one off 🙂
Oscillation in commercial amps is nearly always caused by substitution of semiconductors in the driver and output stages and is pretty common in very old equipment (I'm thinking 1970's and 80's designs) where transistors of that era had relatively poor performance at high frequency and that characteristic was used to good effect as part of the compensation scheme in the design. Substitute with modern parts and the design becomes unstable and requires extra compensation adding.
If the amp had been deliberately over driven by someone testing at high frequency (lets say testing at 50kHz and over and at very high output) then that is when Zobel networks become stressed as they are not rated for that kind of 'non valid' use. Again though, its always the resistor that fries...
Its a one off 🙂
Oscillation in commercial amps is nearly always caused by substitution of semiconductors in the driver and output stages and is pretty common in very old equipment (I'm thinking 1970's and 80's designs) where transistors of that era had relatively poor performance at high frequency and that characteristic was used to good effect as part of the compensation scheme in the design. Substitute with modern parts and the design becomes unstable and requires extra compensation adding.