SWR 750 blowing output transistors

[windsurf22]

I'm repairing an SWR 750 watt bass amp. The power amp board will blow ( short) all 6 of the 2sc3264 Tanken transistors,
This only happens under load. without load, I can crank up the gain and volume to above 100v peak to peak. As soon
as i load the output ( resistive 5 ohm) it wil shortly blow all the expensive transistors. I've ohmed out or tested all
the transistors and resistors ( .1 ohm) in the power circuit. I can't find anything wrong. My next step was to leave out
the final four 2sc3264s and try with just the first two, and measure the collector current. Any other ideas? thanks

[nevetslab]

Do you have a scope, and perhaps a distortion analzyer? If the driver stage is heavy into oscillation, it can take out either top or bottom half, and it sounds like the short circuit protection, which is only on the first pair of devices in the schematic, it's not clamping them down.

If there is oscillation and severe enough, it can be seen without the aid of an analyzer.....with one, you get to look deeper and see it as you advance the current into the load.

What's the bias set for? how much current is flowing in your output stage (looking across the 100mOhm resistors at idle? And, is there any significant DC level on the output. Usually, it would be under 50mV in either direction...less than 10mV even better.

Are the power transistors fully insulated? If you're using silicon rubber pads of some sort, and NOT greased Mica, beware....the grease-less insulator pads break down from thermoset on the original parts, and aren't to be re-used. They can kill your new devices. Not normally a problem with greased mica, unless they're fractured.

[J M Fahey]

1) Fully agree, even if transistors appear to work well static or with no load , unseen/unheard oscillation can cook them .

But ... wouldn't the oscillation overheat them even without a load?
Not necessarily, the amp may be marginally stable and oscillate only with a load.
No need to oscillate all the time either, sometimes you get oscillation only during part of the cycle.
Similar to:



Unfortunately, for some problems a scope is necessary.

2) it might also be overbiased.
The heatsinks might cope with the extra idle dissipation, but at higher power it might runaway.

3) that said, my main suspects are FAKE power transistors.

Buy them only from a large and serious distributor, such as Mouser, Digikey or Farnell ... not many more trusty ones.

In what Country do you live?

EDIT: just look at the chilling fake transistor pictures here:

Transistores Falsificados.

no need to say that: FORGET EBAY .

[windsurf22]

Thanks for the help! I have a scope. The waveform looked clean, with no load, and load. I did however hear , what sounded like
and oscillation, when I cranked it up , with no load. Very subtle, but I could hear it.
The trasistors are mounted with mica, and heat sink compound. They are from China, so that could be the issue.
What quiecent current should i see across the 100 ohm resistor, on the emitter of Q7. What direction should I turn the bias adjust,
to decrease the foward bias on Q4 ?

[Jazz P Bass]

SWR 750 Schematic

Trying the output with one set of transistors is o/k, as long as you don't ask too much of the output.

Usually when both sides blow, it is because they both turned on at the same time.

For starters, I would remove the limit transistors.
And replace the drivers for sure.

[nevetslab]

The schematic for the 750X power amp section (and for that matter, other SWR amps I quickly looked at) doesn't give you any technical support details like that. If you were to measure across the emitter resistors of the driver stage...R22 & R25, from which the bases of the output transistors are connected, setting a voltage of 375mV to 425mV would probably be a safe bet. It's a pre-bias, not enough to turn the power transistors on. Some mfgrs will give you a voltage value across the power stage emitter resistors, of say 2.5mV (which would be 25mA per xstr, if all measure the same).1mV would be 10mA. I'd look at both, and see where your are at idle. Don't need to have load connected.

As to which direction to crank the bias pot? CW, CCW? which ever direction sends you in the right direction....just noticed the CW note (on the schematic) on the bias pot, so I'd guess CW will increase bias.

With greased mica, you're probably in good shape. But, with the parts mounted, it wouldn't be a bad idea to see if there is any resistance path from the collectors on the top half, and emitters on the bottom half to the heat sink...which should be at ground.

The output stage will run without the 3 pair of power transistors...the pair of complimentary drivers completes the circuit with feedback, so one could operate without the power xstrs. Loading it would NOT be a normal speaker impedance. You are sitting with +/- 100V supply rails, so that approach requires a higher value power resistor to load it.....to see under current drive what's causing your failure. You've already found no problem under no load...only when you're driving current...which in part is being sourced from the drivers to the 3 pair of bases of the outputs. There are 560pF miller integrator caps around the base/collector of each driver. No base resistors on either the driver nor outputs so if your replacement transistors are in fact much faster and gets the drivers into oscillation at some point in the waveform, that needs to be corrected. What JM Fahey mentioned regarding re-marked parts bought at bargain prices could be an issue.

I just looked at the Sanken spec sheet on the 2SC3264...MT-200 package.....ft is 60MHz typical. Very fast parts. But, SWR has been using those and similar parts. You could try slower and more rugged parts, such as OnSemi MJL21194, which is also a 200W rated part, much slower but very rugged, though different package, so mounting issues are involved.

[Jazz P Bass]

[QUOTE=windsurf22;370094] They are from China, so that could be the issue.QUOTE]

2sc3264 Stock and Price by Distributor | FindChips.com

Newark or Digikey has them.

[windsurf22]

Nevetslab..If i run it with just the first two 2sc3264's and leave out the last four, what ohm load should I
use ?

Jazz P bass.. I'm assuming the limit transistors are Q5 and Q6, But why would removing them help?

I will replace the driver transistors, and get all replacements from a reliable source.

[Jazz P Bass]

I do have one question: What was the condition of the original transistors?
Did the original fault take out all six transistors?

If you are concerned about shorting out more transistors, you can run one pair.
It is not only the resistance of the load that matters.
It's the overall power that you are expecting to run through the one pair. (P= VxV/R)


I suggested removing (and testing) Q5 & Q6 to remove then from the equation.
There is obviously something seriously wrong, so you need to limit the fault conditions.

Purchasing from a known reputable supplier is a good step.

(if you care to, pop open the China transitors & compare them to the original ones.)

[windsurf22]

I tossed out the original transistors, so I can't compare them.
The original fault did not take out all six, just two.

I know how most of the circuit works, but i don't understand the reason for
Q15 and Q16, where they bring in an ac source and rectify it. ??

[Jazz P Bass]

How about posting the schematic?

[windsurf22]

swr%u002520550x%20750x%20amp.pdf

[Jazz P Bass]

SWR 550X- 750X Schematic

So it's a 750X?

That AC/-100V/+100V circuit appears to be a power on mute.
Possibly so the preamp tube can have a bit of time to warm up.

[windsurf22]

It's a 750 Bass amp. The schematic matches the board exactly.
I was thinking that the circuit had something to do with monitoring the power.
My next step is to extend the board outside the chasis, replace q3,4,5,6,7,8 ( they are cheap)
install the first pair of power transistors, and measure the emitter currents of q7 and q8,
if that looks good, put on a 16 ohm 200 watt load, and draw a small load ( 50 watt).
If that works and two more power transistors, then tthe final two.
Any other ideas welcomed.. I've tested the miller caps and most of the caps and resistors in the cicuit.
Would testing with a square wave help, to look for oscillations? I have a square wave generator.

[Enzo]

I haven't turned on a square waver in YEARS.

Q15,16 serve as mutes. they control power to the diffy pair. AC from the transformer comes through D7 to make a small supply by charging C16. Takes a brief moment to charge. Pulling th base of Q16 down makes it turn on, and current can flow into the amp, turning it on. We pull down that base by turning on Q15, with a positive current at its base. That positive charge on C16 would do that except for that -100v trying to counter it coming through R37. The -100 byy itself would hold Q15 off, but the charge of C16 to some positive voltage averages out to enough positive to turn on the q15. The difference between R36 and R37 handles the fact the C16 voltage is not nearly as high as th -100.