Tranny failure

[Diabolical Artificer]

I've been working on a couple of vintage transistor amplifiers recently, that have presented odd issue's. A Marantz 1070M ( PM400 in all but name) works fine with music but stick a sine wave in and the OP power goes down and there's other weirdness. A Sony 1630 has an output but with the top of the sine cut off, it's not a faulty emmiter resistor or leaky cap, power supply issue or any of the usual issues.

After exhaustive testing of the discreet components, checking diodes and checking of voltages which didn't reveal anything much, that leaves the transistors. So how can a transistor fail? Sure it can get zapped and go SC between EBC but are there more subtle failure modes?

I've read they can go leaky but can they just get a bit old and tired and not quite up to the job? I intend to build a tranny tester to test for leakage and hfe, but would be interested in how others go about diagnosing and testing potentially dodgy trannys.

Andy.

[Mike Watterson]

but stick a sine wave in and the OP power goes down

Unless it's a regulated PSU, that's normal.

A Sony 1630 has an output but with the top of the sine cut off

Bias?

[Murphyv310]

but stick a sine wave in and the OP power goes down

Unless it's a regulated PSU, that's normal.

What??? That's an open statement Michael. If the OP is checking the amp in the correct way, (power meter, scope, rated sine wave at the frequencies stated in the manual, dummy load etc) the Power should be as per the specification laid down by the manufacturer.
If the power sags below the rated sine wave output then there is a fault no arguments.
Actually testing an amp with music doesn't give a proper RMS figure, it will more likely be a PMPO figure so any accurate testing can only be done with the above equipment and a sine wave.

I have found many transistorised amps to be down from spec, usually caused by small changes in resistor values, transistors that exhibit small leaks, power supply caps with poor ESR and other maladies, transistor amps are rarely better than their valve cousins, both suffer from cap and resistor troubles, at least on a valve amp the replacement of valves takes seconds rather than a full dismantling.

[Mark Hennessy]

It's true that measuring audio power is a complex matter.

The main advantage of sine wave testing is that it's easy to do and standardised. But it doesn't tell you much about how the amp will behave with real audio. A sine wave has a low PMR (peak to mean ratio) compared to music (which might be 20dB or more - a factor of 100 in terms of power), so with the usual unregulated PSU, the instantaneous peaks might well reach higher levels than the "RMS" power output suggests. That's an advantage in reality.

I put "RMS" in quotes, as "RMS power" is a meaningless marketing term that came about because "continuous average sine wave power" is a bit of a mouthful

Instantaneous or short-term peak power is a figure of interest, and there are ways to measure it (IHF, etc). By comparing that with the continuous power, that gives an indication of the "stiffness" of the power supply. It can make a subjective difference to the sound quality, so it's something good designers pay attention to.

In this case, I'm not 100% clear of what Andy is seeing, but I'd start by checking out the PSU capacitors, the fuses and the bridge (always start with the power supply!). There's no output capacitors, but others might cause problems (e.g. the bootstrap caps - C709/710).

Silicon transistors are usually pretty good when not abused, and failures tend to be CE shorts, or a loss of Hfe. Leakage can happen as well. Some plastic packages can suffer from moisture ingress...

Those sub-£20 eBay kits can be pretty good for basic testing (and they do just about every other type of component as well). I rarely feel the need for more exhaustive testers because nothing tests a transistor like the circuit it's working in, so usually, you can tell from the surrounding conditions if the transistors is doing the right thing or not. Not always, but usually. Occasionally I might build a basic "jig" for testing more esoteric things, but one of the aims of good circuit design is to get away from depending on certain parameters of the transistors - especially if it's going into production

[Mike Watterson]

but stick a sine wave in and the OP power goes down

Unless it's a regulated PSU, that's normal.

What??? That's an open statement Michael. If the OP is checking the amp in the correct way, (power meter, scope, rated sine wave at the frequencies stated in the manual, dummy load etc) the Power should be as per the specification laid down by the manufacturer.
If the power sags below the rated sine wave output then there is a fault no arguments.

It was a confusing statement: "but stick a sine wave in and the OP power goes down"
Down from what? How was it measured without sticking a sinewave in? I made the probably unwarranted assumption he really meant the theoretical power dropped, due to PSU voltage dropping.
Many stereo amps could give more power on one channel alone, power dropping if both were loaded.

Yes, absolutely all you say is true.

[Diabolical Artificer]

Right chaps, to clear things up the Marantz in question led me a merry dance by presenting odd issues every time I turned it on, after a fault finding/testing spree, like buzzing and a bright bulb on the lamp limiter, OP trannies shorted? No, they're fine. This fault disappeared after I unsoldered the OP trannies for testing. Then I had oscillation of the OP, IE instability and a VHF signal sitting on top, had the top of the sine chopped off and other intermittent faults. All I found was two or three off spec resistors and dry joint's.

I tested the PSU, which for the OP section is a bridge and two big caps, it does have quite a big ripple current 300mV P-P. The caps are fine as are the individual diodes that make up the bridge, that is they test ok on a diode checker, but a leaky diode could be possible.

There is a regulated PSU for the graphic EQ and other sections, tests ok.

So eventually I got two good sinewaves out. This is with a 1khz sine at the voltage as stated in the manual, 8 ohm dummy load. Music in, all well, power meters indicate a power near to the stated OP. Sign back in to test for rated power OP, IE check RMS voltage at OP, squared divided by 8. the meter briefly reads around 25v, which is about right for a stated power of 70w, then it crashes to 2v ish and won't go back up. Music back on, all fine. WTF???

You make two good points Mark, IE look at the bootstapped Q's and their caps and look at the bridge. I've pulled and tested all the trannies (Q's)and in some cases temporarally popped in a new substitute. Found no issues, all Q's test ok.

The Sony amp is very simple, the power amp consists of a LTP/differential input which is a Sony 2SA884 DIL package ( two PNP Q's), a class A stage, and an OP stage. Sine in, all ok up to power amp, then the sine shows X over distortion. All ok up to the power Q's, but now were getting a voltage across the right hand side of the combined E resistor, non to speak of on the left. The e R is one of those 3 legged jobbies. Bias pot does nothing to speak of. One possible suspect, there is a diode that is mounted on the HS, for thermal coupling that sits between two Q's, but this tests ok.

These old amp's arn't well designed, because the input signal gets sent from PCB to PCB back and forth all over the shop, through switches, etc before it gets to the power amplifier.

So, hence my question, when you've tested and affirmed as good all resistors, caps, diodes etc your left with the transistors. I'd read that some Q's in certain amps go "bad" and need replacing, probably from moisture ingress so wondered if there were other ways a Q can fail apart from the usual.

I wanted to build a semiconductor tester for three reasons, A) I hate buying new stuff especially from China, yes I know everything is made in China, but I try and limit the ammount i buy. B) As you know I'm not that experienced, havn't been at it as long as most of you, so I learn by building stuff, and C) Its more fun building than buying.

Andy.

[Murphyv310]

Hi Andy.
Sounds like you have covered more or less everything.
A few small things do come to mind.
You say you have checked the smoothing capacitors, is that for value or for ESR? Why I ask is that you can often get an accurate value reading but a poor ESR value. It could be that.
Soldering...... Now I have also had issues with soldering especially with Sony, not the familiar "dry joint" but where the joint looks perfect but the copper track of the print has not been tinned and there is only a friction connection to the joint, these need a good old prod to show up.
Transistors......... Yes they can fail under stress, go O/C until the power has been removed and will then work seemingly fine until you crank up the volume, usually driver or outputs do this but often accompanied by a noticeable crack in the speaker.
Good Luck

[Diabolical Artificer]

I hardly ever check a cap for value Trevor, just use an ESR meter.

As you say, sometimes the best diagnostic procedure is poking a circuit with a stick. I'll go over the Sony and redo the solder joints. As one OP tranny had blown on one channel, I'll order a couple of pairs and see what happens.

As the Marantz works ok with music, its probably ok, but I'll re test and listen for the crack you mention.

Thanks, Andy.

[Mark Hennessy]

As the Marantz works ok with music, its probably ok

No!

Something is wrong, and you need to do diagnosis. What you've done so far has been very "general", and that's fine and understandable, but that approach doesn't deal with the difficult ones.

As hinted at earlier, the clue to this is the average level of sine waves compared to music. Ignoring modern hyper-compressed rubbish, music has a PMR of 20dB or more. 20dB. In power terms, that's a factor of 100.

To put that into context, a 100W amp that is on the cusp of clipping the peaks is only delivering an average power of 1 or 2 watts. Yes, incredible. But true.

Compare that to a sine wave, where the PMR is very much smaller. Just 3dB.

So as soon as you do sine wave testing, you're asking the amplifier to deliver much more power than music. That means that much more is demanded from the power supply, and the power transistors are going to get a lot hotter. Something is falling over, and you must determine what. Because if you don't, you'll find out the hard way that the owner like to listen to hyper-compressed modern rubbish through very inefficient loudspeakers while wrapping the amplifier in a duvet, having first placed it on top of a radiator! Well, not exactly, but you get the point.

Your fault reports are getting better, but we need to be a lot more specific.

When the output falls to almost nothing, what exactly happens? Does the sine wave become severely clipped (and equally so on both peaks)? Do the power supply rails stay where they should be, or does one (or both) fall accordingly? That's why - BTW - we need several multimeters

Does the input to the power amp itself stay the same?

Does it happen equally to both channels? That's a really powerful question, BTW, and one that is unclear to me so far...

What temperature are the output devices when this happens? (IR thermometers are good for this, and only ~£10 these days).

Is the current limiting causing the clipping? And if so, is that because of a dry joint on the output emitter resistors (reasonably common).

This amplifier has DC offset detection. Assuming this works properly, that'll fire in the event of many faults.

BTW, what's a "ripple current of 300mV"? That's a ripple voltage, and a perfectly reasonable one for an unregulated PSU. That ripple voltage will reach several volts at full output, but the amplifier will be designed to reject that. Not a problem. What's more interesting, however, is the frequency of the ripple. It should be 100Hz on both rails, but it it's 50Hz, that points to an o/c diode (or dry joint or similar). It happens.

Can we just check, BTW, that we're talking about the right amp. I got the 1070 manual from Hi-Fi Engine. It's only 38 watts per channel, and that's into 4 ohms. That's a long way from 70 watts. The rails are only +/-36V when idling, so it would struggle to reach 50 watts on a good day. If the "M" version differs significantly, have you got the manual? A link or similar would be good...

As for the Sony, you need to measure the DC voltage across the bias circuit as a starting point. There's little point dabbing the 'scope probe around the circuit as the negative feedback loop will confuse everything. If the adjustment is doing nothing, that points to a pretty specific part of the circuit. So let's focus on it, rather than everything else. (NB: I haven't yet looked up the schematic for the Sony).

Finally, the above was dashed out in a hurry, so apologies for the highly unstructured brain-dump. I'll have more time later this evening.

Don't give up!

Mark

[Diabolical Artificer]

Thanks for the pointers Mark, I'll get back with more info.

Andy.