Randall RG200

[J M Fahey]

Hi Lowell.
Yes, for me series (with the 110 or 220V) lamp is the same as bulb current limiter.
Yes, voltages in general will be off, lower, but anyway proportional to the original ones.
The idea is to be able to measure something, which a continuously blowing mains fuse or new parts dying won't let me do.
I neither like those 8.5V Vbe in Q1 and Q2, but since they are muted, that doesn't mean much to me, yet.
That's why I ask you to do certain things and take some measurements, which, by the way, is what I would do if I had that amp on my bench.
Please unmute Q1 and Q2 to get some meaningful voltage measurements.
All those I asked before are a "catalog" of what I would measure.
Of course I am "hunting" for abnormal ones.

[lowell]

ok here we go:

DC output:47mv

Q1
E:368mv
B:198mv
C:70v

Q2
E:368mv
B:208mv
C:70v

Q3
Vbe:568mv
Vce:61v

HT
+73v/-61.8v

[J M Fahey]

Hi Lowell.
Now we're *much* better. Q1, Q2 and Q3's voltages are much more reasonable , and you have perfect 47mV DC at the output.
Remember, until I tell tou the contrary, we will work *with* the amp plugged into the serial/lamp bulb/current limiter.
Since output DC voltage looks normal, turn the amp on, check that you still have less than 100mV DC in the output and hook the speaker, in that order.
It should at most make a very faint click or nothing, and the series lamp , which should be glowing barely orange should stay so, or at most increase *very* little its brightness.
If that happens, inject some low level music and listen, it should sound good, if low level, and the series lamp should pulse brighter, following the music.
*IF* the lamp becomes suddenly *much* brighter, the speaker clicks loudly or hums and the cone jumps either forward or backwards, we still have a big problem around.
Hopefully not.

[lowell]

Ok I'll try it. Do you think the HT voltages being off could be an issue?

[lowell]

Ok did it and it worked great. Good signal volume and series lamp was dimly lit until I really cranked it up. Even cranked it wasn't glowing too brightly. Where to next JM?

[J M Fahey]

Goooooooooooooddddddddddd
Now we have a working amplifier, already at least 90% original.
Let's reattach bells and whistles one by one.
*still using the bulb limiter*
1) Re-check that with a speaker connected and music in, we have a few watts available.
2) For now, somewhat lower or somewhat unsymmetric +/- Hv are not important, we *are* current limiting the amp for safety.
3) Turn the amp off, wait for capacitor discharge (if way too slow, solder a 4K7 2W in parallel with each big power supply filter capacitor and leave it there forever)
4) re-solder the free ends of D3 and D4 into their original pads.
Turn the power on and check that everything works fine as before.
If any of the ugly symptoms reapears, we have a protection circuit problem.
If everything is fine, post it and we continue with the next suspect, the mute circuitry.
Good luck.

[lowell]

JM, whoops, I never disconnected D3 and D4. So the protection circuitry is ok..?

[lowell]

Where to next JM?

[J M Fahey]

Hi Lowell.
Now we get into the complex muting circuit.
I guess they avoided using an "expensive" relay yet tried to mimic all that could be made with it. Maybe they saved 1$ per amplifier, and hey!!, in a million amplifiers that means a million dollars !!!
The bean counters must be real happy on this one.
Oh well !!
For the amp to work, Q3 must supply current to Q1 and Q2.
It was not happening, but wiring R44 to ground directly unmuted the amp . so .....
There are many "muters", so let's separate them for now.
Now I'll name "clusters" of components that are joined together by copper traces or pads on the pcb.
When I say "D7/Q21/R42/C15/Q22" see that one pin from each of them is connected to one pin of each of the others, I call that a "cluster".
That one is cinnected to another cluster formed by Q20/R39. Both clusters are joined by a track , which on the schematic is a long horizontal line.
*On the actual board it might be anything*, short, long, thin, fat, a piece of wire, whatever.
What I want to emphasize here is "what-is-connected-to-what" is the important thing.
Now, identify and follow that track, the one that joins both "clusters" and cut it on any convenient place. Later on you'll be able to scratch the etch resist and re-join it with a short piece of wire and a dab of solder. You may even trace it with a Sharpie thin point marker.
Now we've removed the "right side" from the equation, and which I don't fully understand yet , we´ll study the left side without interference.
Reconnect R44 to Q22C(ollector).
Measure Voltage across R7, Q22 Vce and Vbe, V across ZD1, *very specially* Voltage across PTC Thermistor RT1 pads (if it's open, even for cracked pins of whatever, it'll mute the amp, thinking it's overheated), Q21's Vce and Vbe , actual voltage across ZD3.
As you see, I rely heavily on Amp-ON, voltage measurements, for which I only need a multimeter.
It's more boring and definitively longer *writing* what to measure, than actually measuring them, going through the "normal" ones until "a-ha !!!" , some impossible value appears.
Only the it pays to measure component by component in a very small area, and in the worst case, replacing 3 or 4 heavily suspect components.
OK friend, post what you measure, although I distrust highly the PTC or its actual connection, it must be glued to the heatsink and joined to the board or the rest of its cluster with some wire or long, thin tracks.

[lowell]

Ok JM,

R7:9.6v across it

Q22
Vce:195mv
Vbe:653mv

ZD1:10v across it

RT1:22mv across it

Q21
Vce:10.8v
Vbe:34mv

ZD3:15.3v across it

From the little I know of transistors it seems Q21 is off. ZD3 seems fine. Q22 may just barely be on. ZD1 seems good. RT1 does not seem open, I believe it'd have -15v across it if it was.

[J M Fahey]

Hi Lowell.
For what I see, your amp could be pronounced healthy.
Of course I think it still works as before , when the top end of R44 was grounded with a piece of wire.
Your amp is alive since you replaced driver and output transistors, but it was muted by a complex protection circuit.
It was confirmed by the experiment with R44, the amp worked properly.
Just disabling the mute circuitry would have left a "normal", run of the mill power amp, very usable as is, and protected from shorts by Q10 and Q11, conventional overcurrent clamps which limit peak current per device to a conservative 5.5 Ampere peak. (1800/1000)x.65V/.33ohm=5.51A.
That's usually enough, or better said, neither better nor worse than most others use.
Yet they added the clamping circuit.
I fully understand the left (thermal) part of it, which is triggered by the PTC jumping its value , triggering Q21 which cuts off Q22 and its "slave" Q3, muting the amp.
I do not fully understand the right part, which looks like some SOAR limiter or something like that, and which is muting the left part.
Personally, *I* would stop here and consider the amp repaired, because I find thermal protection very important, and that kludge on the right not that much.
If you find that *very* important, we go on.
Anyway, it was *not* that good to begin with, or the amp would not have landed on your bench !!
Try it with a speaker and some guitar, still with the bulb limiter.
If it works, just for kigks try it with an even larger bulb, say 150W or 200W.
If it works, hook it straight into 120V and play loud for sometime, feeling the dissipators with your hands every 5 minutes.
If it works .... I'd call it a day.
ADD-ON: *IF* you want, just replace Q18, Q19, and Q20, re-join the cut track, maybe you don't need do more than that.
Problem is, it's not that I can't analyze it, but I see it kludgy, poor design, and it scratches me the wrong way.

[Steve Conner]

I think the right hand part of the mystery circuit is designed to mute the amp if the current limit triggers too often or for too long. When the current limiter activates it'll also turn on Q18, and all the rest is just a level shifter to allow the current from Q18 to turn off the input tail source. Looks a bit kludgy here too...

I can only see current limiting on it. SOAR protection would sense the voltage across the output transistors too.

The daddy of all of these circuits is Peavey's DDT, with the protection system hooked up to a compressor. But they have the patent on it, everyone else has to kludge their way round.

[lowell]

Ok so leave the R39/Q20 disconnected? The amp will still work it's just protection circuitry? Finally, what did we fix? We broke THAT connection, but you say if I want I can reconnect it? How does this fix/unmute the amp? I'm not sure how this muting circuitry works.

[J M Fahey]

Hi lowell and steve.
I'll answer you both, not getting too deeply into the water.

the right hand part of the mystery circuit is designed to mute the amp if the current limit triggers too often or for too long.

Agree, *basically* Q18 gets triggered by voltage drop across R29+R30, we can safely assume that current across R25 and R26 will be practically the same, and some time delay is provided by the time constant of R31+R34, which are in series, and C11. Q18 is connected properly: it's base will always be more positive than its emitter, and its collector is tied to the most positive point in that amp, so no matter where it sits it works (it floats attached to what we call "the output bar", the high current net that drives the hot speaker out pin).
So far, so good, ...
*but* ... there are a couple extra elements: R32+D5 and R35+D6, *very* suspiciously similar to the elements usually added to a basic fixed-level current clamp to twist its curve to follow the SOAR curve.
Remember the classic "RCA protection"?, very much used in big PA power amps and derivatives such as Hartkes and avoided if possible in music instrument amps for its harsh clipping sound (although Laneys often use them).
They are sensing the output voltage and modifying Q18's bias (thus its "trigger level") depending on the actual voltage difference between the output bar and ground.
Anyway my head hurts because I'm minimalistic, I *hate* kludgy design and this is a good example.
Please tell me what you think, I may be very mistaken or failing to see an important element, I hate this design, it could be so much simpler.
To boot, it's unsymettrical , one branch 1k2>27K>diode, the other one 270r>2K7>diode. Ugh !! Not even the ratio is the same.
I get the designer hat some great idea, it didn't work as expected, and he added and modified things until it worked, sort of.

Finally, what did we fix?

We?, nothing, *you* did.
The real problem was burnt/shorted output devices and you fixed that at once, congratulations.
We found that the left side of the clamp trigger works properly (really you should use your hot air gun and check that it actually triggers at, say, 60ºC to 85ºC) Sorry, no Farenheit for me.
We also found that the right side is defective. I suggested not to waste time measuring and checking it, just replace the transistors because the resistors are very high value and unlikely to be damaged. Ah, just in case, also replace D5 and D6. I'm all against shotgun tactics, but in this case ....
After that, reconnect the right side. If everything works, good; if it mutes again, just separate the right half and that's it for today.
All I want to read now is you saying that it went through your door together with a happy client.
The following problem was this ill-conceived protection circuit, which only "protects" it from working properly.

[lowell]

Ok thanks JM for all your in-depth help. How did you come to the conclusion that the amp was muted?