Solid state amp repair - distortion problem

[Doris The Diode]

Hi everyone,

I hope I’m in the right place for general repair etc.... I usually reside in the radio repair section.

I have taken it upon myself to repair a friend’s solid state 100W amp.
Upon a cold start, with a sine wave going into it and a dumy load and scope at the other end I get a nice clean 20V rms so 100W into a 4ohm load.

Once the amp has been on for about a minute or two the sine wave on the outpu distorts, the top and bottom clip and go square wave.
If I turn it down to 4V rms, 4W it is only at this level or bellow that the sine wave is not clipped any more.

The amp is a clasic long tale pair followed by a pre driver, driver and PA transistors, class AB, +43V -43V rail with negative feed back.

I would like to ask you for your thoughts on how to fault find the problem once the signal goes past the long tail pare and into the complexities of solid state with negative feedback.
The signal is good from input through a few op amps up to the long tail pair.

I’m good with valve amps and fault finding and repair but this is the first time repairing an amp of this classic transistor nature.

Any tips would be appreciated.

Cheers,
DD

[paulsherwin]

I would guess you have a transistor fault in the driver or output stages.

[Doris The Diode]

Paul,

Thank you for the reply.

I’m hoping that too.
The problem with these darn things is the negative feedback as a fault in let’s say the PA will still be shown on the scope on the input.
I didsconected the feedback loop, well that gave me a surprise and quickly turned off the amp!
Feedback loop now resoldered.

I’d like to disconect the feedback loop in some clever way that I can then probe with my scope every component but at the moment I have little experience with this sort of beast appart from start replacing transistors going from right to left on the diagram, not a very technical approach.

DD

[1100 man]

Hi DD,
If it was me, I would be reaching for my can of freezer and gently cooling things down whilst monitoring the output on the scope. That's after checking the power rails of course! With everything DC coupled, faultfinding is made rather difficult- a fault round one transistor will alter the DC conditions on all the other transistors.
Freezer spray, a small hot air gun and the handle of a screwdriver (for finding dry joints) were some of my most frequently used diagnostic aids!
Don't blast everything down to sub- zero: that can be misleading and can cause more problems.
Of course, it may not be temperature related at all: that might be a red herring
Good luck
Cheers
Nick

[Nuvistor]

There are quite a few Zener diodes, is it worth measuring the voltage drop across them when the amp is faulty, or could that damage other items?

[ms660]

First thing I'd do would be quoting the manufacturer and the model and the specs. for that model so it's known that the rating is 100 watts and that it's a continuous rating.

Lawrence.

[Doris The Diode]

Thank you for all the ideas.

Freeze spray, I had forgotten about that one, will give it a go.

You’ve given me another idea about zener diodes and measurements in general, I’ll jot down when first switched on as many voltages as I can around the circuit, probably an attempt over a few days I think.
Then jot down the voltages after it has been running for at least 10 minutes and then compare.
I’m hoping the freeze spray will be quicker.

Not sure about continueous 100W, it will happily put out 100W for some of the time, at least
It’s a Fender FM212r guitar amp, nasty little thing, the valve amps are so much nicer in sound as well as to work on.
Wether it’s a hifi amp or a solid state guitar amp I’m more interested in the repair technique in these amps with large transistors and lots of negative feedback and how one goes about fault finding these things in general.

DD

[paulsherwin]

Measuring lots of circuit voltages is usually a good place to start.

[Doris The Diode]

DC voltages on all semiconductors have already been measured and I compaired one side to the other symetrical side and all are the same on each side, the other side having negative voltages.

Although this may not mean much as there are no DC voltages on the circuit diagram, at least there doesn’t seem to be anything that stands out.

Just ordered the freeze spray

[Skywave]

Looking at the cct. diagram, I see what looks like a thermistor. I also see that the entire cct. is d.c.-coupled. So the d.c. operating conditions of all the transistors in the main signal path are dependent on the correct operating conditions of that thermistor and its associated cct. Perhaps that part of the whole cct. is where the problem lies. Is that thermistor in contact with the heat-sink for the O/P transistors? Could it be defective?

You say: "I’m good with valve amps and fault-finding and repair but this is the first time repairing an amp. of this classic transistor nature."

The really big difference in comparing the repair of valve amps. to most transistor amps. is that valve amps. tend to be a.c. coupled; transistor amps tend to be d.c. coupled. (Apart, that is, from the difference in their HT voltages). Therefore, the repair philosophy is not readily transferable from one type to the other. Unlike an a.c. coupled amp., with a d.c. coupled amp., a fault in the d.c. operating conditions in an early stage of such a design will propagate through through the entire amp., often causing the destruction of the O/P transistors and their drivers. In the event of discovering such a fault, it is pointless measuring d.c. conditions. The only effective method is to remove every transistor and check it for faults. If you miss one transistor and if it is faulty, all the other transistors that you have then replaced will probably be destroyed at the next switch-on.

Al.

[Ambientnoise]

If the fault only appears after say a minute from cold and fairly immediately from "hot" then I suggest something heating up or a leakage. Some careful feeling around straight after switch off may show something up.
One useful technique during troubleshooting power amps is to put resistors in the main power lines. The amps prob idle at maybe 100mA but draw several amps at full power so resistors can drop little at idle but protect against major faults.

Ken

[McMurdo]

Does it clip symmetrically when it's distorting?

[Hartley118]

As Al has said above, there's so much interaction between the stages in a DC-coupled amp like this that a cascade of semiconductor failures is all too familiar. Faults can be a nightmare to diagnose, so much so that some professional repair shops simply replace all the semiconductors as a job lot. Not that I find that very satisfying, but it does illustrate the diagnostic problem.

One test approach that I've found useful is to test the transistors and any diodes without removing them simply by measuring the forward voltage across each junction with a DVM on its 'diode test' setting. A healthy transistor will normally read between 0.6V and 0.8V across each forward-biased junction. It's not wholly foolproof but will point the way to likely short or open circuit junctions and saves a lot of unsoldering.

Martin

[Silicon]

Can I assume that you are injecting the test tone into the power amp and you are bypassing the preamplifier stages?

The electrolytic capacitor in the feedback circuit might be worth looking at.

Is there any muting or anti-thump circuitry?

[Silicon]

I've just found an exploded image of the poweramp on the internet.

Its on the vesselyn.com site.

It shows a mute circuit with a ptc thermistor and there are some voltage levels added to the image.

[Skywave]

Originally Posted by McMurdo

Does it clip symmetrically when it's distorting?

In the opening post - about 4th. para. down - Doris the Diode says that it does.
With that in mind and looking at the cct., sounds like a biassing problem to me.

Al.

[Radio Wrangler]

With a dead amplifier, it's relatively easy. With power off, test all the components, out of circuit if necessary, and reassemble. Don't apply power until you are sure they are all OK and you will avoid the endless circle of replacement trap.

With distortion, it's a harder job. Something in the loop creates distortion, and the loop carries it around. It is therefore very difficult to determine where the source is. The loop should be trying to cancel it, so your best clue may be seeing a stage with a distorted drive and a distorted output, but the distortion on the input is the reverse of what you'd expect would cause the output.

Matched probes and a dual channel scope can help.

The warmup sensitivity suggests something is partially breaking down. The voltage amplifier stage can be susceptible to damage especially if it's running a bit close to its second-breakdown limits.

Starting the amplifier from cold where it's OK and applying a bit of localised heat from a soldering iron may give some clues, but I must admit I've had most success from swapping things and hoping

David

[Doris The Diode]

Thanks to the replies!

The distortion first starts on the positive peak and starts by slightly deforming the peak and then starts going into a square wave.
The bottom trough starts to distort a little after the peak does and forms a square wave with very clean sharp corners.

I will also try by putting in a few low ohm resitors in the pa and driver and measure.

Some other good thoughts too with the thermistor although I'm not sure how that would distort the signal.

I'll measure again a look for 'reverse distion' as suggested.

I'll report back soon.

DD

[Marconi_MPT4]

Originally Posted by Doris The Diode

I will also try by putting in a few low ohm resitors in the pa and driver and measure.

Be careful adding extra resistance to circuit nodes, it may introduce additional problems or mask the fault.

Is the power output rated at 100W? According to the Fender schematic text, the amplifier input is "adjusted to produce 90W RMS at 5% THD into 4Ω resistive load at rated line voltage". It will be interesting to see if this condition can be met.

One measurement that usually can be checked without de-soldering any components is the output stage quiescent current. This can be achieved by measuring voltage drop across 0.22Ω resistors R109 and R112. Since most of these amplifiers tend to be class B, some perhaps class AB, quiescent current expected to be 50mA or less giving a voltage measurement of <= 0.011V neglecting resistor tolerance. Might be worth checking before and after to see how much Q-current has changed.

Cheers
Rich

[Skywave]

Originally Posted by Doris The Diode

Some other good thoughts too with the thermistor although I'm not sure how that would distort the signal.

The D.C. biassing on every transistor must be as per the design. In that design, some transistors will be designed to operate in Class-A, so the collector voltage will be (approx.) half of its collector supply voltage; some - typically the O/P transistor pair - will be biassed so that each transistor of that pair conducts on alternate +ve and -ve going half-cycles of the signal waveform: that's Class-B operation (with slight variations whereby there is some overlap between the conduction times of that pair - that reduces distortion). If the bias on that O/P pair is wrong, so that each transistor only conducts for part of the incoming waveform, distortion will result.

In a d.c.-coupled amp., the d.c. operating conditions are predominately determined by the d.c. operating conditions of the first transistor in the entire chain. Since the Vbe of all those transistors is a function of temperature, that d.c. operating conditions must be correct and stable and, ideally, immune from any temperature changes. That's where the thermistor (and other regulating devices, such as Zener diodes and Vbe multipliers) come in. The items that will undergo the largest temperature change in that entire d.c.-coupled amp. are the O/P transistors. That why it is common practice to find a thermistor that is mechanically-coupled to the heat-sink of those transistors. In effect, producing a temperature-related -ve feedback loop to stabilize the entire d.c. operating conditions.

Al.