SWR LA15 Bass Amp - Blowing "outputs" - Class-A amp mod question.

[rdtsc]

Hello, I was given this 100W bass amp "cube" from a fella whom repeatedly "blew" the 2SC2837, TO-3P output transistors. After the third time servicing it, the shop told him "sorry, replacements are no longer available..."

In any case, he is moving so gave it to me. I've located some similar domestic replacement output transistors, however the issue is, if I'm gonna repair this thing, I'd like to eliminate the possibility of it happening again.

Here is the schematic of the original power amp board. The assembly itself looks similar to this. (I have the new version with the integrated soft-start.) Notice that "driver" Q5 is middle-left black, and Q4 is middle-right black and they are both not heat-sinked. On my board, the three transistors on the heatsink, Q1-3, have melted solder joints I assume from overheating (before the fuse blew.) The board is still in good shape and can be reused.

Now my question is this: rather than repair it and risk another "meltdown" later, I'm wondering if I could parallel a second 2SC2837 with Q1 and Q3 and mount them underneath, and add heatsinks to Q4 and Q5? Or would that skew the biasing? I realize I'm not gonna get 200W out of this thing, just want to avoid another repair later.

Or, is the issue purely heat? Meaning, if I machined vent holes in the metal mounting plate and installed an air circulation fan to cool things off, would it probably never fail again? The design is pretty poor... solid aluminum plate, to which a horizontally-oriented heatsink is mounted. (Against convection flow.) On the exterior, it's just a solid, smooth plate. Here is a photo of the newer assembly removed from the plate. "Up" is to the right when mounted. The complete exterior looks like this. One person did a thermal mod of this already: Modded my SWR LA-15...ro - TalkBass Forums

Thanks for your time.

[Jazz P Bass]

I would not advise paralleling the output transistors.Not knowing the repair history is not helping with the repair.
If the amp failed multiple times you would have to question "what did the tech miss".
A tried & true method is replacing the drivers if the outputs shorted.
The very small ohm ballast resistors must be checked.
Often the current transistors(Q6 & 10)take a hit but still function.
I like to highlight the failed components when repairing a failed SS output section.
This can give you a grasp of the current path at failure.
From this data, an educated guess can be made as to which components likely got strained.
These are the parts that will bite you in the tush.
They static check fine.
They dynamic check fine.
Then they blow.
Yeah, solid state repairs are a real challenge.

[R.G.]

Hello, I was given this 100W bass amp "cube" from a fella whom repeatedly "blew" the 2SC2837, TO-3P output transistors. After the third time servicing it, the shop told him "sorry, replacements are no longer available..."

Clearly the tech decided the amp was haunted. It's really not difficult to find transistors that will work there, as the application is not terribly stressful.

Now my question is this: rather than repair it and risk another "meltdown" later, I'm wondering if I could parallel a second 2SC2837 with Q1 and Q3 and mount them underneath, and add heatsinks to Q4 and Q5? Or would that skew the biasing? I realize I'm not gonna get 200W out of this thing, just want to avoid another repair later.

I'm with Jazz P on a couple of issues. Let's reserve the paralleling issue til later. Right now, the problem is finding out what happened. This one can't be normal for the type, as there exist other amps with this design that work, so this one's bad somehow.

Meltdown deaths in SS amps similar to this come from three causes:
1) Bias voltage gets too high (or the bias source opens) and the output trannies cross conduct themselves into ash and smoke
2) Inadequate heatsinking lets the outputs get into thermal runaway even if the bias is right.
3) high frequency oscillation runs the outputs so fast that they can't shut off fast enough, again leading to cross conduction and heat death.
Sometimes an inductive spike will kill one output, and the other dies trying to hold the output in the middle of the supply. But there is almost no way to find this out, other than only one of the outputs has melted its solder joints, not both.

My best guess is that the bias components are hosed on this amp. If there is an undiagnosed intermittent there, it would eat output transistors. So would a marginal oscillation.

If it were mine, I would remanufacture the bias string components: Q2 and everthing that touches it, as well as the wires to the heat sink from the main board for it. I'd then replace the output devices, and the drivers, and bring it up slowly with Q2 shorted, living with the crossover distortion until I'd verified it's full power operation with no bias. Once you find it's working OK with the new output stage, then remove the short on Q2 and bring it up with no signal, adjust bias, and then try load tests again.

Once you get there, you can start thinking about improving it.

I would put catch diodes from the output before the choke to the power supplies to prevent inductive spikes from puncturing an output.

Adding more heatsinking never hurts.

And now for paralleling. I'm not as concerned about parallelling more outputs as JazzP. It's fine, but do it right. Any power amp with "pf" caps in it anywhere other than the single dominant pole cap on the voltage amplifier transistor (C7 in this case) has had oscillation problems in design. There are those on the output transistors, the feedback path, and the input pair. You have to assume that there is at least the potential for it to sing again.

If I added more output devices, I would give each one its own 0.1R/5W emitter resistor, and I would put a 47R 1/2W resistor in series with each output transistor base. This helps substantially with the tendency to cross-excite each other.

Finally, I notice that this design has no output protection circuits at all. A short circuit will kill it dead, semi-instantly. What does the speaker wiring on this amp look like? Extension jack?

[Ronsonic]

This may just be hazy memory, but wasn't there a factory change that replaced the entire power amp in those?

I'd call and check.