Failing audio PA transistor biasing measurements...

[Radio Wrangler]

Finest guessing:

The big capacitors you replaced will be the output DC blocker en-route to the speakers.

To the left of each of those is a pair of long dark resistors, and you can see they're wirewound. These will be the emitter resistors for the power transistors. The things jammed in the heatsink are probably the bias control transistors. Controlled by the pot with the red paint, and stuck there as Mark says to compensate for the temperature of the power devices.

The smaller capacitor you replaced is probably the bootstrap for the collector load resistors of the voltage amplifier stage.

I'm not sure that there are enough transistors for a diff pair on the input and then the voltage amp stage. I wonder if it's a single transistor input and the voltage amp?

Fairly standard stuff for its era.. late sixties into early/middle 70s.

David

[Al (astral highway)]

Thanks again, Mark, for this very detailed post. Really helpful, I appreciate it.

Originally Posted by mhennessy

Straight away, it appears that the blue and white wires could be where you measure the quiescent current of each channel.

Well spotted!

The white lead is wired to the collector of power transistor 1, Left channel. Quiescent current is 5.4mA.

The blue lead is wired to the collector of power transistor 3, Right channel. It instantaneously blows the fuse of the meter on the 200mA setting, and I'm not tempted to test it on the 10A range!

Originally Posted by mhennessy

As a starting point, could you give us the voltages at the positive end of the output capacitors on both channels? Should be 22.5V in theory.

Left channel output cap volts =16.4
Rt channel output cap volts =22.4

Originally Posted by mhennessy

Also, see those yellow wires that emerge from the other part of the PCB? The two that are joined at one end (left of picture 5)... I reckon that' a PSU of some sort, and it's possible that they are regulated to a different supply voltage and are supplying the earlier stages of the amplifier. It'll be worth checking the voltage there.

33.2V intriguing, as this is approx twice the voltage on the left channel output cap.

The test points on the board read approximately correct, except test point 45V on the Left channel. It reads only 16.4V, exactly as across the output cap, instead.

So there we have some results.

Since last posting, I have cleaned up the balance control internally and there is, crucially, now an output on both L and R channels. I am only using the headphone jack and phones at present, to avoid overloading the power supply or output transistors since a fault condition obviously exists.

I take on board your observation about all those electrolytics. It will be a hell of a job to fix these, fiddly and tedious, as the values aren't visible until I remove them one by one.

I'm putting this on hold for the moment in case we get to an 'ahah moment' based on the voltages and quiescent current findings.

Thanks again, looking forward to hearing what you think!

[Al (astral highway)]

Originally Posted by Radio Wrangler

Finest guessing...

Intuiting, David! Thanks for adding these insights, really interesting and helpful.

What do you think the transistor with the clip on heatsink is doing, up near the power supply? Is it some kind of pass device? It runs very hot, but is thermally stable after a point (as far as I can see/ feel)

I can't account for the low voltage on one half of the board... 16.4V instead of 45V on the Left channel. I had expected maybe that the quiescent current on that side would be far too high and maybe pulling the voltage down, but it's the other way round. Quiescent current is low on the Left.

[mhennessy]

Originally Posted by astral highway

The white lead is wired to the collector of power transistor 1, Left channel. Quiescent current is 5.4mA.
The blue lead is wired to the collector of power transistor 3, Right channel. It instantaneously blows the fuse of the meter on the 200mA setting, and I'm not tempted to test it on the 10A range!

OK - left seems reasonable, but clearly the right channel is unwell. We need to work out how to get the current down.

As a first strike, I would join together the bases of the two TO5 driver transistors and see if this takes it down. There will be crossover distortion, but that's not a worry at this stage. The objective is to work out if the quiescent current is high because of a fault in the bias generating circuitry (including whatever that device on the heat sink is), or a fault in one of the output transistors that is causing them to conduct, despite being fed with a reasonable bias voltage.

Originally Posted by astral highway

Left channel output cap volts =16.4
Rt channel output cap volts =22.4

This is interesting, because it suggests that the right channel is actually the more healthy of the two. This conflicts with the above, but it is entirely possible to have 2 faults at once...

As the left channel is drawing a harmlessly low Iq, I suggest that we fix the right channel first. At least we're then in a position where we can power the unit up for extended periods to take measurements, etc.

Originally Posted by astral highway

33.2V intriguing, as this is approx twice the voltage on the left channel output cap.

Almost certainly a coincidence. But yes, I reckon that'll be a regulated (or just filtered) rail for the input stages of the power amp.

Originally Posted by astral highway

The test points on the board read approximately correct, except test point 45V on the Left channel. It reads only 16.4V, exactly as across the output cap, instead.

OK - I'm guessing that the 45V test points will be straight off the collectors of the upper output devices. For that to have fallen to 16-odd volts suggests some sort of open-circuit somewhere. A lack of "raw" HT, but the continued presence of the "regulated" HT could easily explain a lot...

It'll be worth checking for cracked tracks, etc. Are the bolts on the output devices nice and tight?

Cheers,

Mark

[DangerMan]

Just a couple of observations from here... all guesswork but maybe worth checking.

Those two rectangular blocks on the heatsinks look like diodes, possibly twin diodes within the same package connected in series (these were commonly used at around the time of this amplifier).
I remember some of these used to go open-circuit. I assume their lead out wires are insulated as they must not come into contact with the heatsink itself, and their "twist" makes it look like they're close to doing so. (although the body of the package should be in good thermal contact)

The two pcb-mounted pots are probably the quiescent current adjuster pots and, if so, will be connected in series with those diodes between the bases of the pnp and npn drivers. These pots have a nasty tendency to developing dead spots and going open circuit. They did so back in the day and they are a lot older now! If they go open, the quiescent current is uncontrolled and will blow fuses and/or transistors.

A DVM on diode check range will soon identify whether the diodes are singles (~0.6/0.7V) or doubles (~1.2/1.4V) or kaput.
You will need to remove or disconnect them to check else you will get a reading of both transistors b-e junctions anyway, which will cause confusion.

Pete

[Al (astral highway)]

Originally Posted by DangerMan

Just a couple of observations from here... all guesswork but maybe worth checking.

Thanks, Pete... interesting and really helpful. I will check the thermal compensation diodes as you suggest, and thanks for the advice on taking them out of circuit first.

Ah, I was just beginning to wonder about those pots bias going o/c, curiously enough, so fascinating to hear that it was a relevant thing, back in the day!!

One way or another, it looks like we could be closing-in on what's wrong. It definitely looks like more than one little thing at once, now.

[Diabolical Artificer]

Dodgy diodes and faulty pots killed a few amps prematurely back in the day, notably the Sony Vfet amps.

Andy.

[Radio Wrangler]

The thermal compensation bias scheme will either be a few diodes (2 to 4) in a string, or maybe a couple plus a variable resistance or it may be a transistor with a couple of feedback resistors (one variable) simulating a variable diode. It's a very elegant circuit with few components and works a lot better than a diode string, so the transistor 'Vbe multiplier' became the most common way of doing it. In soma cases the accountants thought a few diodes were cheaper than one transistor. In other firms the accountants took into account the loading time and the cost of the solder So your circuit could use either approach. Look at a number of transistor amplifier circuits from the 70s and 80s and play 'spot the Vbe multiplier'. It's a common feature and a useful landmark.

What's the small transistor with the big temperature and the tubular heat dissipator clipped on? Well, around it is a number of small/medium size electrolytics and what look like rectifier diodes. I think it's a voltage regulator on a lower voltage supply for the preamps and microphone input circuits, tone controls and all that jazz. It too is probably in a very simple circuit... It may be just a zener diode and its associated bias resistor driving an emitter follower to boost the current capability.

A PA amp is expected to travel a bit and sometimes to be run from portable generators (Yamaha brand of course...) So the mains input voltage can be a bit iffy and regulating supplies to spread the pain a bit can help keep things working.

Oh, the diode string and variable resistor approach tries to go to huge quiescent current in the power transistors if the resistor slider goes open or even only skips a little while being adjusted. The transistor Vbe multiplier version can have the trimmer in either of two places. One exhibits exactly the same problem and risk of damage, the other makes the quiescent go low if the pot skips and risks no damage, you just get no quiescent, and a still-working amplifier but with bad distortion. You'd think designers would always get it the right way round? Ha! The wrong way saves one resistor that the right way needs to stop the quiescent going almost infinite at one end of the pot travel.... and resistors cost a penny.

David

[Al (astral highway)]

Originally Posted by Radio Wrangler

A PA amp is expected to travel a bit and sometimes to be run from portable generators (Yamaha brand of course...) So the mains input voltage can be a bit iffy and regulating supplies to spread the pain a bit can help keep things working.

Hi David, my bad, the thread title refers to PA transistors but I didn't mean a portable PA style amp - I meant the transistors in the power output stage, which is actually an old-school 'music centre', as they were quaintly called at the time.

Your analysis of the transistor/ diode temperature compensation adjustments is illuminating and helpful, as ever. Thank you.

I have cleaned up those presets and will test the heatsink-clamped diodes out of circuit. Of course, if they are faulty it will be a triumph of sorts -but also a headache as I'm unlikely to be able to source replacements with the same clampability!

Cheers

[Al (astral highway)]

Originally Posted by mhennessy

Because it's meaningless. It tells you nothing about transistor operation - when the transistor is doing its thing in a real circuit*, the BC "diode" is reverse-biased. The output of a transistor is a current source, just like a valve...

Update:

1) Mark, you were completely right, kudos to you I took device one (channel L, the one passing a high quiescent current) out of circuit. Once I realised that the 1A fuse had blown in my digital voltmeter, while testing quiescent current on the 200mA setting, I was worried that the device had blown, so I desoldered it (with difficulty, I might add!) out to test it.

The B-E junction reads completely normal when forward biased and O/L when reverse-biased.

So that can go back in now.

2) The thermal feedback diodes (the little black devices) don't look to be easily removed. I measured them in circuit and they read as if they are a pair of anti-parallel diodes. If this is true, they are healthy.


Back to resolder device one and then to look for the origin of the under-voltage problem on the Left channel. I'm checking all tracks and potential dry joints on the silk screen.

See you later, folks

[Al (astral highway)]

Originally Posted by Dr Wobble

Dodgy diodes and faulty pots killed a few amps prematurely back in the day, notably the Sony Vfet amps.

Interesting, Andy, cheers!

[Lucien Nunes]

Can't see any mention of lamp limiters above, if you are poking around suspect bias diodes and presets it can be worth leaving one in circuit. I disturbed an unsuspected intermittent in the bias circuit of an apparently working Pioneer SX-1980 and immediately consumed £100's of special rare output transistors, having just installed a set in another amp where IIRC they had been destroyed by transistor failure upstream. Bias double-diode failure permanently silenced a TA-5650 V-fet amp while I looked on helplessly. Neither unit was thought to have a problem in the power amp but might have been saved by the limiter.

[julie_m]

Building in circuit protection requires one extra resistor, probably adding 0.1 p to the cost of an amplifier. Building a circuit with the potential to self-destruct actually increases the number of units sold, as long as it doesn't happen often enough to persuade customers to look elsewhere for its replacement.

[Al (astral highway)]

Originally Posted by julie_m

... Building a circuit with the potential to self-destruct actually increases the number of units sold, as long as it doesn't happen often enough to persuade customers to look elsewhere for its replacement.

Hi Julie, what do you mean, explicitly? I'm not sure if I follow??!!

[julie_m]

If a piece of equipment lasts long enough before it breaks, people are likely to buy another one the same to replace it, as opposed to having to learn the ways of a competing product.

[Al (astral highway)]

Ahah, yes, good point...without going OT too much, I guess this was a lot more applicable in the past, when consumer electronics were expected to last longer anyway. And this one clearly did have a fault way back in the past - the DC blocking output cap fudge proves it (one 1000uF, 16V working (fudge), one 470uf, 35 working (correct). We've already seen that the voltage across each cap is designed to be 23V, near enough, so this was never going to end well. Bizarrely, someone took care to replace the wrong value cap with the same brand as the set original, however - both in situ were from ELNA.

[Al (astral highway)]

Hmmm. Still at least two faults to troubleshoot, no joy after another hour or so of poking around systematically.

I'm trying to visualise the relationship between the power supply and the output transistors. Since I can't find (nor can anyone else!) the circuit diagram, I've hunted for a conceptually similar one that may help us to visualise this. I know we don't know exactly, but I think (comment, please?) the one below might be a similar topology, from Mark and Davids' comments, except that the biasing arrangements are far more intricate in the set we're working on... - Anyhow I'm focussing more on the power devices, for the time being. NB: this is not the actual circuit, just an attempt at conceptualising the kind of design!

I'm trying to account for the 22.5V across one DC blocking cap and only 16 odd across the other, you see. We're agreed on a common supply rail of it seems 45VDC.

It really looks virtually impossible to remove the anti-parallel diodes to test them out of circuit. I tried desoldering them, very patiently but the tracks are thin and the leads were pressed down sharply in manufacture. I could easily damage the silk screen. Alternatively, I could cut them out from the top, but then they'll need some kind of fix before I can re-attach them, assuming they test ok. How difficult would it be to obtain replacements?

[Lucien Nunes]

Have you tried Mark's suggestion of shorting out the bias components yet? This will prove whether faulty power devices are causing the excess current, in which case the bias diodes might be removed from suspicion. Alternatively, can you isolate one end by removing one driver and the VAS or current source?

[Herald1360]

I can't see any antiparallel diodes in that circuit, only series connected ones.

[Al (astral highway)]

Originally Posted by Lucien Nunes

Have you tried Mark's suggestion of shorting out the bias components yet? This will prove whether faulty power devices are causing the excess current...Alternatively, can you isolate one end by removing one driver and the VAS or current source?

Hi Lucien, no, I haven't done this yet. I'm going to recap the electrolytics and then head in that direction.