Hifi amplifier biasing issues and output stage shorting?

[xavier60]

A problem with the diode idea is that the drivers will still be damaged if there is a bias problem. 100 ohm resistors would be safer.
I usually just put a jumper across the B-E pads of the NPN output BJT. Both output BJTs removed.

[Renate]

Yeah, the diode idea is riskier, but I still think that it's a better approximation to how the circuit is supposed to work.
I'd rather throw a 0.5 A amp fuse inline with one (or both) diodes.

[Yamin]

Some further developments/discoveries, after some more digging around. I have soldered the diodes as suggest (1N4148)

I noticed that C6040 and C5050 (caps across the base’s of OT & drivers) reads slightly higher ESR value compared to a good channel. It was around 1.9Ohms. My meter indicates that cap is good <47 uF cap. Its sort of in the good zone so to speak. It says ‘audio’ on the capacitor’s label - I am pretty sure that I would not be able to source one of those from the local shop. First of all could this be an issue enough to blow the OT if so could I replace it with a non-‘audio’ capacitor?

I took out the driver transistors out of circuit and measured the resistances and compared it to new transistors and they were identical. So I am planning to keep the old one, mainly because I am not sure whether the new transistors I got from the local shop is a genuine part. What are your thoughts on keeping the old drive transistors as they are measuring ok.

I took out Q6010 out of circuit to measure it - there was thermal compound on it but it had more of a silicone like consistency. The compound had dripped onto the leads of the transistor. There was no dripping on to the leads of the same transistors on other channels. The compound when measured was not conductive like how some glues becomes conductive after time. Since 2SC1740S are obsolete the replacement suggest by people were 2SC2240. I measured resistance measurement with the one I took out and compared it to a new 2SC2240 and they were identical as well except for one measurement. Measured across the C and E using an analog meter at x10k (the highest my meter goes to) red probe on C and black probe on E - the old one appears to be more ‘conductive’ than the new one. When the probes are reversed there is no ‘conduction’ on either one. I did clean the compound off the leads before taking the readings.

I haven’t powered the amp yet to measure the voltages as I have to partly assemble the unit to do so. Will continue tomorrow.

Thanks again for the time.

[Yamin]

Right so I powered up with the amp, with the didoes in place. Left it for maybe half an hour so nothing blew

So I replaced them with transistors borrowed from yet another unused channel and I wanted to monitor anything happening on the bias. So I connected some jumper wires from the test point and had  series 100 ohm resistors on each wire. Powered up the amp - and it seems to follow what the SM described. However at one point my heart sank when I saw 254mv! I quickly turned off the amp. I touched the 'test' resistors and they seem to be slightly hot as well :S. But after thinking about it for a while and checking whether anything had shorted I realised that it was due to the fact that my test probes weren't making proper contact with the test point. PHEW!
So I powered back the amp again and monitored the biasing and it followed almost exactly what the SM describes. 2.5mV on startup and after couple of minutes goes to around 9 to 11mV and stabilizes. So is it safe to say that the biasing is ok?
Another thing I want to remark is I measured something around 55mV of DC offset on the right channel and something around 30mV on the left channel. There are no adjust points for the DC offset. My questions are:
1.Does DC offset got anything  to do with the biasing or biasing circuitry?
2.Are the above mentioned values acceptable for the DC offset?

[xavier60]

I would call 50mV borderline acceptable.
The input pair transistors should be passing about the same current to minimise offset.
Any residual offset should only be due to transistor mismatches.

Edit: DC offset and idle current biasing usually don't interact.

[Audiorepair]

I wouldn't be happy with 50mV DC offset, I would consider that to be an unsolved problem.


Well, in light of the fact that the amp has blown up twice, that is.

[Yamin]

A problem with the diode idea is that the drivers will still be damaged if there is a bias problem. 100 ohm resistors would be safer.
I usually just put a jumper across the B-E pads of the NPN output BJT. Both output BJTs removed.

I also wanted to ask - you recommended using 100 ohm resistor instead of of the didoes (in the same place)?
Could you please explain about the jumper across on Base to emitter scenario as well, its on both the output transistors right? In that case I assume there is not 100 ohm resistor right?


Follow up on my last post - could the DC offset blow up the output transistors, maybe something in the crossover circuit of the speakers shunting to ground and putting a short across the terminals?

Thanks

[Yamin]

I wouldn't be happy with 50mV DC offset, I would consider that to be an unsolved problem.

Could that be taking out my output transistors? 
I measured the 50mV on the right channel which wasn't shorting out initially.

What would be an acceptable range? I know ideally it should be 0mV.

Thanks

[xavier60]

A problem with the diode idea is that the drivers will still be damaged if there is a bias problem. 100 ohm resistors would be safer.
I usually just put a jumper across the B-E pads of the NPN output BJT. Both output BJTs removed.

I also wanted to ask - you recommended using 100 ohm resistor instead of of the didoes (in the same place)?
Could you please explain about the jumper across on Base to emitter scenario as well, its on both the output transistors right? In that case I assume there is not 100 ohm resistor right?


Follow up on my last post - could the DC offset blow up the output transistors, maybe something in the crossover circuit of the speakers shunting to ground and putting a short across the terminals?

Thanks

Yes, the 100 ohm resistors would go in the same place as the diodes.
It doesn't matter witch B-E pads are jumpered but the NPN like I said is best for the readings making more sense.
Using voltage  measurements, calculate the current through Q5000 and Q5010.

[xavier60]

The input differential pair,Q5000 and Q5010, current should be close to 1.5mA each.
8mA each for the VA transistors.

[Renate]

We've kind of been going on the presumption that it's a bias failure.
Since only the top output transistor blew (is blowing), might it be a DC balance failure going positive and dumping current into the speaker?
When it blew did it just go silent or was there a "whump" or bad hum for a second?

How about checking leakage in C5040?
When you remove it, you can check your output DC offset without it.
(With that cap removed you have reduced your amplifier for both DC and audio to a gain of 1.)

Still, even if C5040 was shorted it would only amplify the DC offset of the input pair by 28 times.

Maybe Q5040 current sink intermittent failing would nail this thing positive and kill Q6050 (top output).

Check across D5000 for 5.6 V
Something shorting there would make this to go all positive.

[Yamin]

A problem with the diode idea is that the drivers will still be damaged if there is a bias problem. 100 ohm resistors would be safer.
I usually just put a jumper across the B-E pads of the NPN output BJT. Both output BJTs removed.

I also wanted to ask - you recommended using 100 ohm resistor instead of of the didoes (in the same place)?
Could you please explain about the jumper across on Base to emitter scenario as well, its on both the output transistors right? In that case I assume there is not 100 ohm resistor right?


Follow up on my last post - could the DC offset blow up the output transistors, maybe something in the crossover circuit of the speakers shunting to ground and putting a short across the terminals?

Thanks

Yes, the 100 ohm resistors would go in the same place as the diodes.
It doesn't matter witch B-E pads are jumpered but the NPN like I said is best for the readings making more sense.
Using voltage  measurements, calculate the current through Q5000 and Q5010.

Sorry @xavier60 I am a little bit confused. It was suggested to use two diodes for the two transistor’s B-E (Q6050 & Q6060). So I should use two 100 ohms instead as you suggest right?

With the B-E jumper option- you suggest connecting only one transistor’s B-E via jumper right (the NPN)? So the PNP transistor’s B-E pad should be left open or should I connect a 100 ohm resistor or diode there?

Would I be able to measure across the test points to see whether the bias is holding up with the above mentioned procedure?

Sorry again for the noob question, really do appreciate the time.

[Yamin]

We've kind of been going on the presumption that it's a bias failure.
Since only the top output transistor blew (is blowing), might it be a DC balance failure going positive and dumping current into the speaker?
When it blew did it just go silent or was there a "whump" or bad hum for a second?

How about checking leakage in C5040?
When you remove it, you can check your output DC offset without it.
(With that cap removed you have reduced your amplifier for both DC and audio to a gain of 1.)

Still, even if C5040 was shorted it would only amplify the DC offset of the input pair by 28 times.

Maybe Q5040 current sink intermittent failing would nail this thing positive and kill Q6050 (top output).

Check across D5000 for 5.6 V
Something shorting there would make this to go all positive.

Yes I’ve been trying to cross-out the possible reasons for this failure. I had a feeling that it might be something to do with the biasing as I have checked all the other factors to my best knowledge initially.

The first time the NPN was shorted, but the second time the two transistors had shorted out.

The borrowed output transistors are holding up so far (like last time). This time I want to run it on my bench long enough till I am confident that there are no issues with the amp. I will obviously return the amp with the new components soldered in and after running it again.

I wasn’t there when the amp failed but the owner did mention that there was a ‘pop’ sound when it went out.

I have checked C5040 with an ESR meter and it seems to be in order. I could remove it to make sure.  If there was an issue with the capacitor being leaky what should I expect to see when I power up without it like you suggested?

I did check across D5000 and it seems to show proper forward voltage  drop. I’ll check it after powering up for the 5.6V drop across it.

Any suggestion on how to suss out the possible intermittent failing Q5040?

Also please note that I measured the 50mV at the right channel output the left channel output measured around 30mV - the left channel was what had failed before.  What would be the acceptable DC on the outputs?

Thanks for the help.

[Audiorepair]

At this point I would be thinking this:

It's possible one transistor fails over time.
There are 9 transistors in the pre-amp stage before the main power transistors.
Replace all 9, probably half an hours work and a couple of dollars in parts.

Way, way less than you have spent on this already.


If, for some reason, this doesn't fix the problem, you will KNOW that it is not due to any of the transistors, and you can use this fact to test the rest of the circuitry that might be bad.



Currently you know nothing other than it is not the output transistors, you need to establish some facts.

[xavier60]

Sorry @xavier60 I am a little bit confused. It was suggested to use two diodes for the two transistor’s B-E (Q6050 & Q6060). So I should use two 100 ohms instead as you suggest right?

With the B-E jumper option- you suggest connecting only one transistor’s B-E via jumper right (the NPN)? So the PNP transistor’s B-E pad should be left open or should I connect a 100 ohm resistor or diode there?

Would I be able to measure across the test points to see whether the bias is holding up with the above mentioned procedure?

Sorry again for the noob question, really do appreciate the time.

Using two 100 ohms instead of diodes will give better protection to the driver transistors if the bias circuit develops a fault while testing.
Don't bother with the B-E jumper idea, it makes things too confusing.

[Yamin]

At this point I would be thinking this:

It's possible one transistor fails over time.
There are 9 transistors in the pre-amp stage before the main power transistors.
Replace all 9, probably half an hours work and a couple of dollars in parts.

Way, way less than you have spent on this already.


If, for some reason, this doesn't fix the problem, you will KNOW that it is not due to any of the transistors, and you can use this fact to test the rest of the circuitry that might be bad.



Currently you know nothing other than it is not the output transistors, you need to establish some facts.

Thanks @Audiorepair this was my plan at one point too, I tried sourcing the parts but it turns out the exact parts are now obsolete and the only ones available are from not so reputable shops so I was afraid of putting in counterfeits. I tried looking for equivalent parts but it always turns out it's not an exact match if one parameter is superior or matched the other parameter would be inferior to the original. But that's an excellent suggestion

[Yamin]

Sorry @xavier60 I am a little bit confused. It was suggested to use two diodes for the two transistor’s B-E (Q6050 & Q6060). So I should use two 100 ohms instead as you suggest right?

With the B-E jumper option- you suggest connecting only one transistor’s B-E via jumper right (the NPN)? So the PNP transistor’s B-E pad should be left open or should I connect a 100 ohm resistor or diode there?

Would I be able to measure across the test points to see whether the bias is holding up with the above mentioned procedure?

Sorry again for the noob question, really do appreciate the time.

Using two 100 ohms instead of diodes will give better protection to the driver transistors if the bias circuit develops a fault while testing.
Don't bother with the B-E jumper idea, it makes things too confusing.

Ah thanks so much for clearing it up

[Yamin]

Guys so the mystery of this amplifier's plots thickens! just realised something very interesting -
When I checked the amplifier last time I used all the analog inputs to input my signal and even this time I have been using the analog inputs to test the amp.
The owner however uses the HDMI ports. He has got his DVD player, TV and some other devices connected to the HDMI ports and the amp has got some kind of automatic function of detecting which device is currently being used - so I guess its switching from one to another. The owner asked me even at that time could that have caused the amp to blow up and I thought that its highly unlikely, since the fault is on the output side of things.
Reading online about Onkyo amps many have had issues with the HDMI board, but none have reported of the outputs blowing. Do you guys think that its a possibility? the input is saturating the outputs during switching? I was under the impression that these amps would be designed in such a way that there would be some sort of 'limiting' if the input is driven hard.
Thanks so much for the help guys.

[Audiorepair]

It looks like there are 3 protection lines on the speaker output.

IPRO, providing protection against excessive output current.
FLVPRO providing protection against escessive output voltage, I suspect.
FLVOLH providing some kind of I don't know what signal.
(looks like high DC, but why no low DC?)

The amp is well protected.


Replace the transistors.  One of them probably fails randomly.

And if in doing so your DC offset falls considerably below 50mV, then that might be a clue that one of them was actually duff.


You could have done that instead of thinking about and composing this post.

[Audiorepair]

One other thing,

the fact that the working side also has a high DC offset, may not actually be an indicator of normality, but might just mean this channel is also on the verge of self destruct, due to the same component failure.

[dietert1]

Have seen quite some amplifiers and when i look at the schematic, i am wondering how this amplifier is meant to be RF stable. This must be a very old design and it appears strange. Amplifiers can be killed by RF oscillation and dynamic instability will cause horrible sound quality.

There should be a 1 nF cap over R5120. There should be a miller cap on Q5030 (maybe 100 pF+3K3) and there should be a capacitive voltage divider in the feedback path (e.g. 2.7 nF parallel to R5040 and 100 pF parallel to R5030). C5080 can go. Next thing to check is the presence of a snubber parallel to the output. It may appear outside of the schematic shown. Many amplifiers also include an inductor in the output path.

Recommended reading: apexanalog.com/resources/appnotes/an19u.pdf

Regards, Dieter

[Audiorepair]

What about the grounding?

The schematic as posted shows the power amp ground and the pre-amp ground separate and dissapearing stage left, and no speaker ground indicated.

If there is a problem here where they don't all three connect properly, then, er, er, I dunno what might happen to the power amp.
Maybe DC offset or something.

Its ground might be floating.

[Renate]

There should be a miller cap on Q5030...

Yeah, that would be more typical.
Still, there is C5080 to cut the high frequency.

[dietert1]

Yes, that capacitor may have the intention to limit high frequency gain, but in my understanding that "solution" is invalid. It's exactly what you can't do: Feedback over the full loop gain without bandwidth limit. A perfect way to make it oscillate. I mean there is always a frequency where negative feedback becomes positive feedback due to delays.

Just imagine some clever ONKYO engineer determine a capacitance value experimentally such that the input capacitance of Q5010 makes it a capacitive divider and it remains roughly stable, but that is a bad approach. Though it appeared in many amplifiers, typically with a 10 pf to 20 pF capacitor.

Regards, Dieter

[Yamin]

Have seen quite some amplifiers and when i look at the schematic, i am wondering how this amplifier is meant to be RF stable. This must be a very old design and it appears strange. Amplifiers can be killed by RF oscillation and dynamic instability will cause horrible sound quality.

There should be a 1 nF cap over R5120. There should be a miller cap on Q5030 (maybe 100 pF+3K3) and there should be a capacitive voltage divider in the feedback path (e.g. 2.7 nF parallel to R5040 and 100 pF parallel to R5030). C5080 can go. Next thing to check is the presence of a snubber parallel to the output. It may appear outside of the schematic shown. Many amplifiers also include an inductor in the output path.

Recommended reading: apexanalog.com/resources/appnotes/an19u.pdf

Regards, Dieter

Thanks so much @dietert1, very informative.
There is a snubber like circuitry on the outputs.