Sunn Concert Bass PA crossover distortion

[Nyqusit5]

Hi,

I am repairing a Sunn Concert Bass that had a catastrophic power amp failure. All 24Ohm resistors were burnt and the power transistors shorted. After replacing these resistors and the 2n3055 transistors the amplifier powers up and puts out its rated output into a load or speaker. The problem is that there is a noticeable crossover notch in the sinewave (clean otherwise) even when signal is fed directly to the power amp input, and it is audible as a slight distrotion when the amp is played. I have changed Q1, Q101 and Q102, and tried different settings of the "bias" preset pot but it has no effect on the problem.

R113, 115 , 121 and 119 were replaced with 22Ohm as I couldn't get 24Ohm ones, but I doubt this is the reason of the problem. All diodes check ok.
Can anyone suggest a possible reason for this crossover distortion?

[J M Fahey]

Learn to live with it, itīs inherent to all transformer driven designs.
Bias is fixed, unadjustable and does not thermally track , so itīs set at factory so amp does not thermally runaway after 3 or 4 hours of playing *loud* , so when cold itīs underbiased.

IF you increase it , at high power it will suicide.

"Bias" pot there handles only the *driver* transistor Q1 and is set to get symmetrical clipping.

In any case nobody played in a bedroom way back then, where itīs annoying, but live, besides a drummer, which absolutely covers any crossover distortion and then some.

It was a LOUD amplifier, I bet the "Concert" part of the name hints at that.
It was often used with this loud/big/heavy 4 x 12" cabinet as standard, or a similarly large 2 x 15" :

[Nyqusit5]

Hi JM, thank you for your input.

I hate to contradict you but I personaly own a Concert Lead (same PA), my bassist plays one in front of me every week, and I have repaired quite a few Concert Bass and Concert Lead. None of them had this corssover notch, so I truly believe that this one has some kind of fault...

[J M Fahey]

Mmmmmhhh, I stick to my observations , but wonīt argue with you on your Sunn

Now going to the generic case, resistive uncompensated bias amplifiers, of which transformer driven ones are just but a particular case, by definition are either underbiased and safe, or properly biased (at ambient temperature) and potential time bombs when hot, power transistor BE junctures drop around 2mV per Deg C so drop is , jus to use easy numbers, 200mV when going from 25C ambient and amp just turned on and on the bench to 125C after a couple hours shaking walls.

125C too much?
Not that much, remember thatīs actual junction temperature; heatsink might be at hot but almost bearable 70/75C , remember we have junction to case Rth plus interposed mica/goop plus internal heatsink drop, a fin will be a few degrees lower than main body and so on.

In any case 200 mV variation in bias is *a lot* .

And consider it as twice as much as we usually see in modern amps, which apply bias from base to base of the output pair, be it Darlington or Quasi complementary so each sees hlf of total bias, while here both the lower and theupper rail receive each a 200mV variation.

Way more than what is neded to go from underbiased to overbiased.
Unfortunately you can NOT add a diode there to compensate because of the high currents involved in direct driving the mainpower transistors, it does cause a current inversion through the diode which of course is impossible by definition.

Th old way of compensating such amps is adding Thermistors to the biasing network; of course thermistors are bidirectional; hereīs a properly compensated example, the Philips/Fapesa 100W amplifier:


In any case the thread amp seems to be even more underbiased than others , so canīt strictly be compared to them.

And biasing: notch and voltge drop through emitter resistors plus scoping waveform should be checked on a cold amp to be certain it does not distort at bedroom levels, then current at idle / voltage drop through emitter resistors should be checked again with amp real hot, at live playing conditions, or simply driving it to just below clipping on a nominal load resistor for at least 30 minutes.

I would be surprised if idle current did not change a lot from cold to hot.

Using relatively large emitter resistors (0.47 ohm instead of 0.33 / 0.22 or 0.1 ohm) will diinish the effect but not neutralize it.

FWIW even today, with a biasing transistor epoxied to the heat sink, I still bias my amps quite cold ..... same as peavey and others did for ages.

PS: so if you wish, please measure voltage drop across emitter resistors both in this amp, in the others, and compare.
Also scope them if you wish.

FWIW IF rails are at rated +/- 34V , bias should be (24/1524)*34=0.535V=535mV which is just whatīs needed for a 2N3055, a similar value was used in other amps same technology.

But it wonīt track, so it will quickly become overbiased.

22 ohm resistors will offer: (22/1522)*34= 490 mV.

That 45mV difference, which is a lot if we are talking bias, might explain your problem.

Another difference might be that currently available 2N3055 are faster but weaker, higher fidelity Epitaxials, while way back then, and specially for abused MI amplifiers, way more robust (but slower) Hometaxials were preferred, the reference to which all others compared being the famous RCA 2N3055H

[Nyqusit5]

Thanks JM. As you point out, the 22Ohm resistor are probably the origin of the problem. I'll go on a quest for 24Ohms then.
Thanks again!

[J M Fahey]

I suggest a Plan B which allows fine tuning of bias voltage:

1) leave 22 ohms (R113/115/119/121) as is, itīs a standard easy to find value, hope you used straight 22 ohm 2W in each case and not a kludge of various values in parallel .

2) measure Vbe at Q2/3/4/5 to check actual voltages, I calculated what *should* be there but parts have tolerance, wall voltage too, remember this is a not compensated voltage, either against temperature or rail voltage.

3) to have a clue on the actual result of biasing, measure voltage across R116/117/123/125 .
I expect 0mV , meaning 0 bias current, but who knows?

4) it would be great if you could repeat same measurements on one of your properly working Concerts, that will give us a clue on what we "should" aim at; again I already calculated those values but measuring beats anything else.

5) IF we confirm underbiasing is the source of the problem, we can try ***slightly*** increasing it; easiest and relatively safest way is to tack solder some quite higher resistor in parallel with th 1k5 5W ones; I guess 33k 1/4W in each case would be about whatīs needed because it would increase bias voltage by about 5% ; try it; but always measure before/after and compare with "good" Concert voltages, also scope output at a couple Watts into a load.
If not enough but a step in the good direction, add an extra 33k resistor in parallel with the 1k5 ones.

Different to other cases, you can not bias while using a lamp limiter, Vbe (bias) is a straight percentage of rail voltage since it comes from a simple voltage divider.

Ok, somewhat messy, but since there is no way to adjust with a pot, and to boot thereīs 4 values to adjust at the same time, we must take this with lots of patience.

Good luck.

[g1]

5) IF we confirm underbiasing is the source of the problem,

This is important. As JM mentioned earlier, you should double check that your other units do not have the crossover notch.
Otherwise it is still possible that the audible distortion is something other than the crossover notch.
If you have verified this already then sorry I missed it.