Output transistor on - side keeps blowing

What can cause the final Transistor of the - side of a class B to blow but not the Driver for it or the + side? Can it be bias or perhaps some kind of weird half wave from the pre amp?

I already checked the resistors and they are high tolerance and right on. There is no load atm. This keeps reoccurring. Last 2 times it blew the driver too but not this time.

Q22 is the one that keeps blowing. Please refer to page 13.

Originally, Q22 and it's driver Q20, as well as Q3 were all blown. I replaced Q22 and Q20 because I did not know about Q3 yet. Found no adjust-ability out of R21 so then I discovered Q3 was blown. Replaced that and Q20,22 again. Adjust-ability returned but found base of 1.004 +vdc with R21 all the way counter clockwise. Q20 was only about 1.004 -vdc. Tried to adjust R21 and blew Q20 and 22 again. Replaced em again. Pot all the way CCW again and this time Q22 just blew right away.

Is it safe to just pull out Q20 and Q21 and run it to try to see what is coming in at the base of those?

Been struggling with this one for a while now.
Thank You

Page 13.pdf (350 KB)

I don't see a reason why it would be unsafe to remove Q21/Q22 and measure the voltages at the bases. You have a symmetrical situation where with zero volts across the Q2 differential inputs you should have complementary voltages on the push and pull sections of the circuit (i.e., the base of Q21 is approximately the negative of Q22, the base of Q20 is the negative of Q17, etc.). You can check for symmetry at all points you can get a probe on and hopefully get some clues.

Hi -

You should be careful adjusting the bias setting on the pot.
If you get it wrong, you will saturate the drivers Q17/Q20 and then rapidly the output stage Q212/Q22.
Before you even notice then, the weakest end of the pairs will burn out in a rapid thermal run-away.

I suspect that you have a problem in the bias stage Q3/Q18, which are set up as an adjustable Zener diode circuit.
If the pot is aged and worn out, the wiper may loose connection during adjustment, causing the circuit to be biased too high.
Also if capacitor C12 is a dried out electrolytic one, it will no longer dampen rapid changes during adjustment.

Notice the diagram voltages written at Q8/Q19 emitters.
These voltages should be spot on and replicated in the high end emitters Q1/Q16.
If these voltages are not spot on, you should investigate the differentail amplifier Q2 and its emitter circuit.
The two collectors should have equal voltage rops over R12/R13

have you done a basic check on the power supply ?
If one side is out of tolerance, the whole amplifier circuit becomes unbalanced.

charliesixpack:
I don't see a reason why it would be unsafe to remove Q21/Q22 and measure the voltages at the bases.

You can't just remove them in THIS amp. Then the whole amp has no DC feedback anymore.
You could remove them, and temp use a ~100-220ohm resistor between B and E.
NO (speaker) load ofcourse... Just a scope on the output.

Anders53:
If the pot is aged and worn out, the wiper may loose connection during adjustment, causing the circuit to be biased too high.

In this amp, as in most amps, bias will be at minimum when the wiper looses connection.

Just short the whole bias circuit. e.g with a jumper/solder blob over C12.
That way you can rule out the bias circuit while fixing the amp.
Without bias, the amp just has a lot of crossover distortion.

Did you use branded transistors. There seem to be a lot of "fakes" around.
Does the amp also blow without a load. With or without input signal.
Do you use 8ohm speakers. IMHO, this amp can't handle 4ohm loads!
120volt rail (2x60) and just a single output pair is typical "home theatre only" stuff.
Leo..

Lot's of Great help here. I have been doing a lot of reading and YouTube to solve this. What you folks are asking is related to some of that. Let me address your questions as best I can.

You should be careful adjusting the bias setting on the pot.
If you get it wrong, you will saturate the drivers Q17/Q20 and then rapidly the output stage Q212/Q22.
Before you even notice then, the weakest end of the pairs will burn out in a rapid thermal run-away.

I suspect this is what happened when I tried to adjust the bias the first time. The Positive side (Q17 base) was very close but the negative side (Q20 Base) was way off (1v). After I replaced 21/22 AGAIN they just blew right away, even with the pot all the way CCW like the drawing instructs.

I suspect that you have a problem in the bias stage Q3/Q18

Me Too. I found Q3 to be bad but Q18 I had no replacement for at the time. I never tested it but I have a replacement coming now anyways. I got adjustablity at Q17/20 back after replacing Q3.

Also if capacitor C12 is a dried out electrolytic one, it will no longer dampen rapid changes during adjustment.

Had replaced all caps with brand new ones just in case.

Notice the diagram voltages written at Q8/Q19 emitters.
These voltages should be spot on and replicated in the high end emitters Q1/Q16.

Will do. I am not 100% clear on the purpose of these transistors except that they seem to be somehow related to the feedback loop?

If these voltages are not spot on, you should investigate the differential amplifier Q2

I have recently become very suspicious of this section and doing a lot of research on "Long Tail Pairs". I actually only half understood it before but I think I have the whole Idea now. It's a phase inverter also and not just a diff amp. Seems like I have had this go bad in a few amps now. One was spitting out DC on the output jack. I did not check Q2 originally because it was a hard to get 2 in 1 package but I have GK recommended substitutes in the mail now.

have you done a basic check on the power supply ?

I checked the plus and minus 60 but I forgot the + and -15. Embarrassing. Will check it after all the transistors get replaced. I already paid for them. Going to put them in.

You can't just remove them in THIS amp. Then the whole amp has no DC feedback anymore.
You could remove them, and temp use a ~100-220ohm resistor between B and E.

I realized about the feedback issue after a while. Good tip about the resistors.

Just short the whole bias circuit. e.g with a jumper/solder blob over C12.
That way you can rule out the bias circuit while fixing the amp.
Without bias, the amp just has a lot of crossover distortion.

Are you sure this is ok??

Did you use branded transistors. There seem to be a lot of "fakes" around.

I got the first batch from Digikey. I'm getting this batch right from GK themselves.

Does the amp also blow without a load.

Yes :frowning:
It also blew with an 8 ohm speaker but not for a much longer time and I think that time I turned it up to high for too long.

Thank You All.

Hi,
A long long long time ago I built an Amp, Electronics Australia, Twin 25, to be exact.

They had fuses in the two 40Vdc busses that fed the final transistors.
To set it up from scratch, you replaced the fuses with 100R 2W ,I think resistors.

This limited any serious current, but did not impose a great drop on the busses when the bias/quiescent current was set.

This saved lots of blown transistors, as in my case the supplier had supplied four transistors with alternate pin outs, but did not tell the constructor.

You may be worth cutting the 60V bsses and fitting a resistor to limit your current.

Just a suggestion.

Tom...... :slight_smile:

TomGeorge:
To set it up from scratch, you replaced the fuses with 100R 2W ,I think resistors.

I think you mean 100R 2W resistive fuses. :grinning:

(25V-a-side supply? 6 watts at 100 ohms.)

I R to dumb to understand this advice :smiley: I should swap the fuses in the circuit with resistors?

Hi,
I said it was a long time ago, found the article. 1976.

https://dl.dropboxusercontent.com/u/44941643/Playmaster-Twin-25-v1.pdf

Its 8Meg in size, scanned three part article.
You also had to cut a couple of links to set it up, yes 100R resistors, read, and + and - 30Vdc buss.

Tom... :slight_smile:

jarrod0987:
I R to dumb to understand this advice :smiley: I should swap the fuses in the circuit with resistors?

There are fuses on the plus and minus side of the output transistors?

Yes, put 47 or 100 ohm resistors there instead whilst you "tune it up". Cheaper than transistors for starters.


Couldn't find a diagram on Google to show it, but that "Playmaster" design has the fuses only in the collectors of the power transistors, which is preferable for the purpose of calibration.

jarrod0987:
Are you sure this is ok??

Yes.
I have designed/build many power amps, starting with a Quad 303 clone in the early 70s.
That was after several valve amps, and a 600watt guitar amp that blew five from the eight speakers connected to it within 10 minutes :frowning:
And I have fixed a truckload of amps, working as a consumer audio tech for 40+ years.

The advice about replacing the fuses with 100ohm resistors is an old trick to set the bias without blowing up the output stage if you get it wrong. It also is a perfect current sense resistor to measure the bias.
But it won't stop the amp from blowing up if something else is wrong. Transistors die faster than fuses.
It also depends on WHERE the fuses are in the schematic.

Blowing the amp without a load (and input) can mean two things:
The transistors are able to withstand the 60volts idle, but not the 120volt output swing (fake transistors).
Or the bias goes up with voltage swing (bias circuit).
Remember that these transistors have a 100-110watt SOA (safe operating area). That's only 1A at 120volt.
Advice for a non-tech: carefully inspect/measure EVERY part.
And replace every electrolytic cap under 47uF.
Don't ever party with amps like this. This supply voltage needs two or three pairs of output transistors per channel, and a massive heatsink.
Leo..

I have re attached the print for convenience. The only fuses I see are in the power rail before the driver and power transistors. This circuit also has some kind of fault protection which frankly doesn't seem to do diddly except on start up.

I bought my replacement transistors from Digi Key and all the caps as well, replaced all the caps.
I don't understand the part about the bias going up etc. Do you mean it is contaminated with AC and rising above it's set level when the positive alternation comes? How can I check for this without blowing my transistors again? it doesn't look like fuses are an option in this case?

I am no audio tech by any means but I would like to be. I have a solid understanding of old school linear power supplies and NPN's set up as class A and that is about it. I'm trying to retrain in switch modes and AB audio amps etc. Learning a lot but still feel like a total newb. Sometimes it's hard to find instructional info on the web because I am so new I don't know the right words to search for etc.

Also...I have been struggling to understand how the bias works in this amp. I don't have a clear procedure in my mind. I'm just trying to make the dc voltages on the drawing match reality and it isn't going well. Maybe someone can point me to a reliable method for this type of biasing somewhere?

Thank You.

Page 13.pdf (350 KB)

The bias circuit is probably the easiest part of this amp.
It consists of C12 and every part to the right of it, R18, R22, R23 and pot R21 Q18 and Q3.
It has the function of a "super" zener diode. Current path is Q16 > bias "zener" > Q19.

The zener voltage is the same as the four driver/power transistors's B/E "diodes" need, to just start conducting.
The bias circuit is designed to follow the changing bias voltage of the power stage due to temperature fluctuations (power) as close as possible.

Bias is adjusted by measuring the voltage over the string of emitter resistors from the power stage.
J2-1/J2-2 is the connector for that.
20-40mA (~8-16mV over 0.4ohm) is common. Or follow the designer's recommendation.
Or measure the still working channel, and copy.

A better way is using a scope on the amp's output.
Feed a tiny 1Khz sinewave into the amp, and adjust for just disappearing crossover distortion.

Q11 measures the current in the power stage, and generates a fault condition if this gets too high.
Hope this helps a bit.
Leo..

P.S. Did you measure the 0.1ohm emitter resistors. The sometimes suffer after a shorted power stage.

So I get that it is supposed to set up a DC bias for Q17/20. That is correct isn't it? The bases of those Q's should be +1v and -1.2v respectivly if the bias is all proper?

Originaly I was getting no adjustability there and then discovered Q3 was cooked. Replaced it and with R21 all the way CCW Q17 base was at 1.004vdc (with no freq injection btw) and also Q20 was -1.0x vdc. I adjusted R21 until Q20 was -1.2 vdc then a strange buzzing noise and pow, no more output Q's. Replaced em. Put R21 all the way CCW again. Turned it on....Pow...no more output transistors again. This time my plan is to change out every transistor from exact replacements obtained from the manufacturer at twice the price.

I did just verify Q2 is good on both sides as per DCA75 tester. If you have any idea what this means please do tell :frowning:

Thank You

One of your problems is, that you do a destructive testing, which means you are set back, everytime you are trying to do something, by blown transistors.

Solve that problem first.
Fuses and resistors have been mentioned as current limiters.

Why not try a bulb ?
It wont blow up and it will limit the current, and have a nice side effect of giving you a faint visual alarm light.
This is my preferred method of fault finding 12V electric circuits in cars.

Donno what your local power is, if you are in a US based location, you would have easy acces to 120V bulbs.
But 240V bulbs can be used as well.
A 120V 40W bulb in each power side would limit the current to something around 100mA.
240V 60W bulbs will give you an even lower maximum current.

I fear, that you might have the infameous problem of a self oscillating amplifier, which would exactly explain a buzzing sound.
Such a problem is normally due to a broken feedback loop, even a hairthin crack in a pcb copper track or a bad solder joint.

Wawa:
You can't just remove them in THIS amp. Then the whole amp has no DC feedback anymore.
You could remove them, and temp use a ~100-220ohm resistor between B and E.

You still have bias for the input through R17 so I don't see your point. There is evidently a bias problem and if the DC feedback is causing the problem you will learn something.

Yes I understand about the whole resistor thing but there is not one in the right place for that trick apparently. I too suspect the feedback loop may be broken but not sure where. I just verified Q2 (both of them) are good. I don't know why the Output stage keeps blowing. I don't have any shorts across the rails at this time. I am going to try to pull all the transistors and replace the ones in the power section first and see if the rails are right. Then put in all the ones for the amp section and try again to set the bias.

What about replacing the .1 ohm emitter resistors in the final stage for some 100 ohm ones? I don't know if that would screw up setting the bias at all?

charliesixpack:
You still have bias for the input through R17 so I don't see your point. There is evidently a bias problem and if the DC feedback is causing the problem you will learn something.

Bias is the C12/Q18/Q3 block.
DC feedback is voltage divider R17/R20.
AC (and DC) feedback is R17/R20/R19/C11.
HF (oscillating) feedback is C10.
Don't confuse things.
With the power transistors removed, there is NO link anymore between R42 (drivers out) and R17 (DC feedback in). Without that, the output voltage of the drivers will be all over the place.
Leo..

jarrod0987.
Try to get the amp running with the power transistors replaced by 220ohm resistors.
You can connect a small speaker THROUGH A 1K RESISTOR to the amp, to hear what's going on.
Low volume, and keep an eye on the temp of the drivers.
Carefully measure everything. A scope makes things a lot easier.
DC (0.0volt) on the out rail is important.
Measure the voltage on the driver bases (+ and - ~1volt).
Before you put the new power transistors in, set the bias to minimum!!!!
Minimum voltage between driver bases, minimum over R42 (maximum pot resistance).
Leave it there untill you're 110% sure the amp is fixed. THEN turn bias up SLOWLY.
An amp without bias just has a small amount of crossover distortion at very low volumes.
Leo..

Wawa:
With the power transistors removed, there is NO link anymore between R42 (drivers out) and R17 (DC feedback in). Without that, the output voltage of the drivers will be all over the place.

The OP asked "is it safe." What is unsafe? The input stage Q2 is obviously a differential input with the common mode voltage at GND. Without the DC feedback you will still maintain symmetry between the differential outputs of Q2 (assuming, of course, you have 0 volts differential input). You should have equal currents through R12 and R13, Q8 and Q1, Q16 and Q19, R11 and R28, etc. The bias will not be "all over the place." The bias will be very well defined. The bias voltages may be uniformly high or low but if they are all over the place you have a much greater problem.

The question I have is what are the voltages across R12 and R13. If the bias of the input stage is off nothing will be right.