Solid state troubleshooting

[faloon]

Hi all,

I have an old Japanese Companion A-200G 20W solid state amp (also sold under different brand names; Shin-Ei, Univox U-150R, Teisco) from the late 70s/early 80s that stopped working after my friend yelled a bit too hard through it with a normal voice microphone.

After switching it on it works (clean tone, full volume) for a few seconds and then fades down to a low output which is heavily distorted. Switching it off and back on does not help, ie. it is distorted from the second I turn it on. However, if I leave it unpowered for a longer time, it works for a few seconds after switching on (cold start).

I have some experience with electronics but I'd really appreciate if you could share some ideas where and how to start troubleshooting. Here's a schematic I found on the internet for a newer model with different power transistors and some other minor differences

http://www.aijaa.com/v.php?i=6727635.jpg
http://www.aijaa.com/v.php?i=6727633.jpg


The amp has a single clean channel meaning that this has nothing to do with channel switching problems.

[Roly]

Ah, a circuit, ripper!

Yeah, one look at the insides shows the driver tranny so this is an older style tranny driven totem-pole.

As always, disconnect the speaker and connect a limiting lamp in series with its mains supply.

It's a single rail supply with output coupling cap.  Looks like the supply from the mains filter cap to the power output board should be easy to lift "(A)" at the cap, so do that and confirm healthy supply voltage on the cap.

Restore the supply and measure the output half rail to ground i.e. the transistor side of the output coupling cap.  My guess is that this will rapidly drift down to ground or up to supply rail.  If it does then you need to track down what is doing that in the output section (and it's a fair bet it will be one of the output transistors, but keep an open mind).

If the half rail is settling okay then you can reconnect the speaker and need to work back stage by stage injecting a signal and confirming each is clean until you find the one that isn't.  Initially you need to determine if the problem is in the power amp section, the part powered by "(A)", or the preamp powered by "(D)" etc.

The one good thing about this style of amp is that they are not DC coupled, which makes them a lot easier to service stage by stage.

[faloon]

Thanks Roly!

While waiting for a reply I studied the amp in more detail, just in order to understand the circuit and to find out the differences compared to the schematic I have.

Here are some images

http://aijaa.com/ILCPZg
http://aijaa.com/P4mXXW
http://aijaa.com/cByEEq
http://aijaa.com/mcgdSU
http://aijaa.com/ZdjIDy

As you can see, the output stage's 25V/1000uF cap before the speaker is gone and so is the smaller ceramic capacitor (?) beside it. I had a hard time identifying this component on the schematic. All I know is that it's connected from the other side to the the broken 25/1000 which means that it's obviously the "1" or "25/11". However the sizing might have changed compared to the schematic. Luckily there's a similar one on the other side of the output transformer (also in the photos) that shows the color marking. Any ideas?

[Roly]

Uh huh.  {Don'tcha think the socket set is a bit premature?  :lmao: }


I think from these pix it's pretty obvious that your problems are in the output stage after the driver transformer.  It also looks like the circuit is a pretty fair representation of what you've got ('tho some values and transistor types might be different).

In http://aijaa.com/ZdjIDy - 11396885.jpg the small grey disk with two red dots and a black dot is a thermistor, a temperature dependent resistor.  In pic http://aijaa.com/cByEEq - 11396880.jpg you have what looks like the remains of the thermistor for the other power transistor.

You're lucky in that one seems to have survived, so you need to carefully extract it and measure what its resistance is at 25 deg C (room temp).  Both of these will be the "resistors" shown in parallel with the diodes in the output stage.  They are to prevent thermal runaway.  (and this may explain the okay-then-not as the good one heats up).  If they test the same you could be lucky (the heat from your fingers should/will change their value), but don't be surprised if the overheated one falls apart as you take it out.

The square ceramic resistor is another sign of a previous blow up, and I suspect that the thermistors are supposed to rest against these 0.5 ohm emitter resistors to monitor the idle current by their temperature.

Just for the moment, if you need to, you can replace the burnt thermistor with a fixed resistor of half the value of the good one at room temperature.  This won't sound great, but the amp should then work consistently without fading out.  Finding a suitable replacement thermistor could be a bit interesting, might require a bit of shopping around.

Both of the electrolytics in pic http://aijaa.com/mcgdSU - 11396883.jpg look like they have been given a good cook by the overheating component(s) in-between, so I'd replace both of those.  The components in-between look like they are the Zobel network (0.1uF & 5 ohm in series across the output) but they don't look damaged, so I'm guessing that they have been replaced after a blow-up, possibly due to high frequency instability (ono).  Really the electros should have been replaced at the same time.  Check the Zobel cap, but if it looks okay and isn't shorted it can stay.

Just from the look, I'd guess that the output pair of transistors aren't original either, but no matter at this stage.

{Good pix BTW, tell the story}

HTH

[faloon]

You're probably right in that the transistors (mine are TIP33A and BD245A, original 2 x C840) and some other parts have been replaced, the board had some home-made soldering traces on the other side.

Thanks again for taking the time to reply..! I'll look into the business more next weekend and report back then

[faloon]

I finally managed to desolder the other leg of the intact thermistor and it gave a value of 100ohms. The first measurement gave 93ohms but it could have been that the thermistor was warmer than ambient after being in contact with heat?

So my question is that is this value (100ohm) in line with what it should be and would it be more likely to be a PTC than a NTC? Or is a total waste of time looking for a replacement, meaning I should go with the normal resistor at half the value?

[Roly]

In that position it should go lower in resistance as it gets hotter.  First check the dodgy looking one and see if it's the same.  If it is we need to look elsewhere, but my guess is that it will be much higher or open circuit.  If so you can try replacing it with a 47 ohm (ono) for the moment and see if the amp then gives consistent results, that is, operates but perhaps with some distortion.  Again, if it continues to do this fade and die number then we have to look elsewhere.

If you aren't getting anywhere with the thermistor, you could post a copy of the output stage circuit annotated with the voltage you measure.  That should provide a strong clue as to where the trouble is.

[faloon]

I finally got the parts, replaced the two electrolytes and thermistor,  and now I have a working amp!

http://aijaa.com/iLIxBb
http://aijaa.com/Qp3yud

If somebody needs a 100ohm NTC thermistor, you can get one here

https://www.elfa.se/elfa3~eu_en/elfa/init.do?item=60-305-89&toc=19509

I also bought replacement transistors (10 euro on Ebay) but obviously I didn't need them. Yet

Once more, thanks for the help Roly!

[Roly]

Yo!  Well done!   :dbtu:

My pleasure.   :cheesy:

I was a bit worried that you were going to be up the creek with that thermistor since it's a pretty old-style circuit and termistors were never very standardised at the best of times; so thanks for the link.

A word of caution; I would find a place to monitor the idle current and thermally cycle the output stage, say by driving it hard into a dummy load until it's cookin', then idle it and check that the idle current is reasonable and trending back to its cool idle value - no sign of runaway.  {I once had a Gibson repair that was thermally oscillating over a several minute cycle until I thermally coupled the compensation diodes to the output devices - easy to overlook.}

[faloon]

Is it accurate enough to measure the current by using a current clamp on the mains cable, or should I connect a multimeter in series somewhere within the the amp circuit? And what would be the easiest way to  generate a suitable dummy load? What comes to my mind first is pulling the speaker plug out for a while but I guess it's not what you had in mind

[Roly]

Is it accurate enough to measure the current by using a current clamp on the mains cable, or should I connect a multimeter in series somewhere within the the amp circuit?

No, not really.  Rather than in series, which is always a bit of a pest to break the path and connect your meter, you can measure the current through the output pair as the voltage across either of the 0.5 ohm emitter resistors, the lower one with one end connected to chassis would be the obvious choice, then apply Ohm's Law;

I = E/R

I(mA) = E(mV) / R (0.5)

I'd expect somewhere between 20 and 50mA, but we are interested in its stability and how it trends over time and temperature - back home (good) or out to lunch (not good).

What comes to my mind first is pulling the speaker plug out for a while but I guess it's not what you had in mind

Well no; this is like revving an engine without load.  What we want to do is give it a realistic thrash under normal load.  You could just do this into its speaker but this generally quickly drives you nuts and causes complaints, hence using a dummy load for the sake of sanity.

Dummy load?

Basically you need a resistor of the speaker nominal impedance, say 8 ohms, and with a power rating in excess of the amp (but if you are going to knock one up you may as well build something that will cover all the amps you are likely to come across, say 100 watts).  Since X-ohm/100 watt resistors are a bit thin on the ground we normally use a number of power resistors to make up the wattage.  The self-powered fan is optional but blown resistors will handle a fair bit more than their free air rating (for a shortish time).

An important point, if you don't want to melt stuff on your bench, is that these resistors will get quite hot, so some sort of metal enclosure (here an old computer PSU case) is a good idea.

{here's one I prepared before the show  :cheesy: }














This circuit is for a different load but I have included it to show the general scheme of how to make the fan self-powered.



For an 8 ohm/100 watt load you can connect just 10 x 82ohm/10 watt resistors in parallel ( =8.2 ohms).

If you want to power the fan then one of the resistors needs to be split so that the voltage across the lower one produces no more than the fan voltage (12 volts) at full rated power in.  This split resistor should add up to about the value of the others, i.e. 82 ohms.

The PCB was just a scrap cut off something out of the junk bin.  This is only one of many ways of doing this, so look around your scrap pile/junk bin and see what you can press into service.

HTH

[faloon]

Wow, that was detailed, thanks. I'll probably have to build myself a load soon enough as it seems that there's more to do. And I have another amp project on the way

Anyhow it was a bit early to pop the champagne for the amp. I had the chance to play with it a bit longer today and it seems that once it's warmed up (by playing, not by idling) it begins to rattle (scared the hell out of my 4-year old son) and distort stronger signals (low E strummed hard). After cooling it works again for a while.

Am I correct if I suspect that this comes from the new thermistor?

[Roly]

Test gear conceptually comes first.  Thorts.

A sharp eye, a wet finger, and paying attention will take you a long way in electronic fault-finding, but at some point you need to confirm this voltage, that current, or the other resistance (not to mention the occasional watt), and you need a multimeter of some sort, any sort.

A meter of any sort, digital and analogue.  Test signal source (Sig-gen, PC, MP3, CD, tape &c) and dummy load/wattmeter of any sort.  An oscilloscope of any sort is highly desirable because it provides a much better view of what is going on in audio circuits.  Failing an oscilloscope then a simple monitor amp as a signal tracer is important as a substitute {e.g. powered 'puter speakers, DIY mono amp, ono}

Without at least these minimal "eyes" and "ears" you are playing a blind man's buff with your problems. {just watch out for the limited specs on some cheap e-Bay "DSO" Digital Storage Oscilloscopes.  While these are fine as high capacity low voltage data loggers they are not "oscilloscopes" in the sense of audio work}

If you're seriously into it a millivoltmeter/dB meter/THD meter/spot SigGen, frequency counter.

There is a huge range of software, a lot of it free, that will do some useful things via a 'puter/soundcard both as signal source as recorder/analyser.

But a physical front end is required.  This can range from anything ranging from an old portable radio/cassette machine with lots of controls, big stereo meters, and DIN Line In and Out; up to domestic stereo amps like old Sanui's again with lots of controls, but more in the area of monitor amp/signal routing.  Or DIY.

While this covers audio reasonably well, it isn't down to DC, and you still don't get ground isolation, which is not only important for safety, it also causes hum/instability problems in test rigs due to ground loops and common ground with digital circuitry {but a laptop on battery power would get around}.
8|

As I said, you need to thermally cycle this amp on the bench under controlled conditions.  It's a fair bet that the thermal coupling between the sensor/thermistor and the controlled loud/the output stage won't be good, and anything done to close that loop will help.

A potential problem is that your replacement thermistor may not have the same t:R relationship/curve/log gradient.


While you are concentrating on the output stage, don't forget you could have, say, a leaky coupling cap between preamp stages, or any other fault for that matter.  The main volume control normally represents a point between pre and main amps, and a logical point to extract pre output and inject main input for testing.

At this vintage, be sure to give all the sockets and pots a good clean too.

[arskas]

I looked at this amp recently and I found that one of the diodes at the (pi/output stage) transformer secondary was open. Is it possible that it caused the thermal runoff (?) reaction?

[Roly]

I looked at this amp recently and I found that one of the diodes at the (pi/output stage) transformer secondary was open. Is it possible that it caused the thermal runoff (?) reaction?

Thermal runaway.

Yes.