Why a power amp catastrophically fail (sparks), and how can I protect it (me)?

elnaldo

Lunatic Member
Hello. I have a powerful chinese power amplifier and mixer, very useful in my bench, that has failed 2 times catastrophically, sparks and explosions inside, burned resistors and outputs, but fuses not blown.

I don't have the schematic, but each channel output has 2 x 2SC5200 and 2 x 2SA1943 (original toshiba bought from mouser). It has 2 fuses at the AC secondary going to the rectifier bridge, 8 A each (PCB mark is 10A) (tried smaller, 6A, but it burns after several power-on cycles, it gets red with the surge current and fails later)

The "incidents" were produced by some loud crackling noise or thump coming from preamplifiers under test, a loud BANG and the disaster started, 2 times already. I want to fix it and protect it from another event.

My 2 questions: Why a power amp could fail that way? some output transistor that can't hold some high current peak, starting a cascade fail? If so, how that high current peak is produced? The PS is +-60V

And then, how could I fuse this, to avoid the disaster again? Fuses or polyswitches at each V+ and V- rails before the output transistors, each channel separately? I don't need the full power, the amp will never play loud PA levels, it's in a small room, so I can "under" rate the fuses to be on the safe side...

Any ideas will be welcome.

Thanks!
 
Assuming you just have it on the bench to have audible means of testing.

This "sounds like" you might be oversteering its inputs so much, it makes the amps settings a mess, it even may destroy or deteriorate its input opamp or transistors.
Make sure, hard limit the amp for input voltage and make sure it is capacitor coupled.

If on the bench not needing any loud volume and if it is a class ab amp, you also can add appropriate resistors in the driver and power transistors collectors. This really messes up anything at any higher power but at low volumes such is not audible and usually the amp will behave. It just slows down the drivers and output transistors a lot.
You may choose the power resistors for the output transistors to be around 50 ohms or so. It will drop 5 Volts if current is 100 milliamps. Assuming the bias is not really high. You still can get a few watts having enough voltage left over the transistors. Any resistance making sure the transistor stay within their safe operating area will do. 10 Watts inflammable resistors will do I guess.

(this all assuming you are not using the amp for playing loud music)

Note this is just ideas, although I DO use blown fuses having power resistors soldered over them if having sideways accessible fuses, testing seventies vintage amps. Small amps like some Sansui have fuse holder clips making such safe limiting possible.
 
It is possible the amp is unstable and can break into oscillation. If it has an opamp in the front end, when it is overdriven it may be "sticking" with its output stuck to the rail, putting a large DC voltage on the output.
 
Not to be a smartass, but a dumpster might sort that out.

Failing that, try slow-blow fuses of a lower current rating. they will stand the inrush but open at lower current. Fast blow on the rails though since they shouldn't have inrush issues.
 
Last edited:
Might consider adding a pair of reasonably fast diodes somewhere near the input to act as input voltage limiters.
 
Wow! I've only ever ran across one D.C. amp that had a full on cascade failure from front to back.
 
Amigo!
agreed to power supply area .. 220v down there... A Variac would be ideal but you could make a DBT array? if one bulb isn't avail to cut down the current. like 1-2-3 15w bubs? Other thing is use it when needed and shut off? Does sound like the rectifier area or above noted so maybe cut the rush down..
 
I'm temped to trash the unit (actually this was my idea before starting this thread). It was given to me burned, and it burned 2 more times. But it's so heavy and has a mixer and a very heavy toroidal transformer inside. Current sensors to activate the fan cooling, so it's silent at low levels. If I could protect it from a next event, I could use it for some more years before trashing it.

I was thinking is some zeners at the input, or at the power amp input, to limit the max. input signal, and fuse the DC rails (they are not fused, just the AC side)

I didn't see any oscillation when fixed, but this 2 events were started by a loud noise or peak signal getting into the unit from malfunctioning preamps, and all was very fast.

Now that you are mentioning oscillations, perhaps I could add some network to avoid that.

Anyway, trashing the unit is the 1st idea, fixing it again is plan B by now.

The thing that gives me a little bit of light, is that this second time the channel burned was not the channel I fixed 1st time (with all new transistors). This blown channel had original small signal transistors, original drivers, just new outputs. Perhaps it's just coincidence.
 
Im not really an expert on this stuff but my AU-X1 is infamous for oscillation occurring in the pre-stage that makes its way to the output stage to cause a mini-Chernobyl.

It has a triac between the pre and power stages to apparently stop DC(?) going between the stages.

Could this work?
 
I'd like to add some protection to avoid oscillations, to limit the input signal (if that can cause an amp to blow), and to limit the current when some event like this happen again, as said, fusing the rails to the output transistors. As mentioned, under this event, the fuses didn't blow. I could fit smaller slow blow fuses, as suggested, or some NTC thermistor to limit the current during startup.

I think the amplifier has some coupling capacitors between stages, and at the input. I could replace them with smaller uF and higher Voltage ratings.

Anyway I don't understand very well how the problem starts. A loud peak at the input signal, and output transistors starting to explode.
 
Last edited:
One other possibility -- you don't have any kind of load across it that puts a large capacitance (say 0.5uF or more) right across the amp's output terminals by chance? That will make a lot of amps go into oscillation, and bipolar class AB amps don't oscillate "gracefully" - each output transistor turns ON fast, but turns off a little slower so that current blasts through both causing heat and.... fire. Maybe check what kind of circuit the amp is hooked up to on its output. Shunt capacitors are a no-no.
 
couple thoughts from me

add a Zobel at the very output.

put heatsinks on your drivers, TO-3 heatsinks are somewhat rare and/or pricey so use a
terminal (with a split round female insert for the wire and widen or narrow as necessary)
with a fork - and bolt with a small heatsink if needed.

measure current at the B+ supply to the output stage and add fast blow fuses. think
about adding a series fuse to the outputs.

if you have a schematic or can read the original parts - get more robust parts:
to-220 versions for example.

up the ratings for any power-assisted resistor.

I suspect what happened is a failure in preamp supplying this amp, perhaps
a transistor failing, send a rail-voltage sized square wave into your amp.

others have mentioned zeners to prevent a large spike - good idea
(find one with FAST response time - these square waves are the tsunamis
of the SS world.), but not sure of what a great solution to this would be.

if this spike came at power startup then you could build a muting circuit
on both units.
 
Thanks, I appreciate the answers and the ideas.

The preamp section was not refurbished but the power amp has all fresh caps and output transistors (one channel fully restored 100% new transistors, other channel just the outputs). I didn't double check the Zobel network at the output, I'll check it and replace the capacitor just in case. The unit has a speaker protection relay, I don't know how efficient, since under this 2 events, I've heard loud noises at the speakers.

Drivers are insulated TO-220 without heat-sinks. I'll add them. And outputs are 2SC5200 and 2SA1943, TO-3P (TO-247) package.

Regarding the Zeners to limit the input, I suspect it will produce square waves easily (smaller though)

I think with all this ideas I'll give the amp another chance... Fusing the +- rails for sure. The transistors are 15A each, 2 in parallel make 30A. I'll add something like 5 or 6 A fuses at each channel, I think that will cover the sound levels that I need and keep the transistors safe.
 
Fuses only prevent disastrous fires - they're not fast enough to prevent transistors from frying. The old engineer's joke is to "fit some transistors to protect the fuses". :D

Sounds like some circuit modifications to make the amp less fragile are in order, but that isn't a trivial task. One common way to protect the output transistors is to use the voltage across the emitter resistors (on the output transistors) to turn on the base of a small signal transistor which then limits the voltage drive to the output stage.
If the problem is the amp suddenly going into oscillation you'll never be able to monitor it on a 'scope because it will all happen too quickly (unless perhaps you run the amp into a high impedance load of maybe 100 ohms which will limit the power dissipation but maybe allow the oscillation to be studied).
 
I know he fuses are not fast enough to protect a transistor, but I was thinking in small fuses to limit the output, so it never reachs the max. 15A transistor rating.

I think the amp already has a current sensor circuit. The circuit seems quite complex, even the cooling fans are "current controlled", they go faster when the music is louder, and they stop at low levels.
 
Back
Top Bottom