Power Amp Repair - Please Help!

[Aniol1349]

Hi everyone,

I have a problem and I hope someone can point me in the right direction.

I have recapped a big power amp and during disassembly, I misplaced one mica but I decided to power the amp to see if everything is ok. I have bent the Out device off the heat sink and when I powered it up it went through the cycle, relays engaged etc, all good, and suddenly smoke appeared - I cooked the out transistor. ( I know bloody stupid ;/) I marked it on the schematics. Along with that transistor one of the fuses went, oddly it was one of the negative rail - also marked on the schematics.

So anyway, I have re-tested every transistor in the amp taking it out of the circuit and all apart from the one that cooked are fine (faulty one is shorted).

I’m putting the amp back together and somehow the collectors of all the out devices are reading very high resistance to the chassis, starts quite high (hundreds k)and then charges up to O.L, this is with the ground lift switch in OPEN, when ground is on CHASSIS then resistance is low and also slowly rising.

The bipolar h voltage rails are also connected together with the resistance between them rising when probed. (I guess going through a cap)

I’m slowly losing the plot on this one

Could someone give me a hint where I could start looking for that short?

I’m attaching schematics to the channel and PSU.

 

[Aniol1349]

PSU Attached

 

[scott2000]

I'm not really following what's going on beyond the blown transistor.....

is the resistance abnormal ?

Maybe make sure you didn't miss a fault on one of the transistors? I chased my tail for an entire day only to find out I skipped a measurement.....  and I even had it out of circuit.

As soon as I put it back, I noticed a funky reading  that didn't jive ...

Surprised you cooked it that fast......Was there a load on it when you powered it up??? Cool looking amp.... Bias switch??

 

[Bo Deadly]

First of all, be careful. That thing looks like it could singe your eyebrows.

So anyway, I have re-tested every transistor in the amp taking it out of the circuit and all apart from the one that cooked are fine (faulty one is shorted).

I’m putting the amp back together and somehow the collectors of all the out devices are reading very high resistance to the chassis, starts quite high (hundreds k)and then charges up to O.L, this is with the ground lift switch in OPEN, when ground is on CHASSIS then resistance is low and also slowly rising.

The bipolar h voltage rails are also connected together with the resistance between them rising when probed. (I guess going through a cap)

Collector resistance to chasis? Never heard of that. Is that a thing? I didn't think you could do that.

Which collectors? The positive side collectors are connected to 86V and the negative side are connected to the output.

At any rate ... maybe the servo circuit is horked. Is the output stuck high or low?

One thing known to break a servo is if something is only getting one half of the a bipolar supply.

Then again I probably shouldn't be answering this question because I know nothing about power amps. I honestly didn't understand the FB at first. The p channel JFETs through me off.

 

[Aniol1349]

thanks for all the replies,

Basically I'm measuring resistance between collector ( metal casing of the output device which is normally shielded by the mica) and the chassis.
The amp has a ground lift switch, when in OPEN position I'm getting now no continuity but when I switch it to CHASSIS I get resistance random resistance readings which are dropping or rising.

Bottom line is - I'm not 100% sure what the reading should be but in my understanding there shouldnt be any continuity between collector and the chassis at any point, right?

There was no speakers attached when I powered the amp after the recap, it was on for maybe a minute when I noticed smoke, F4 blew and the 5A plug fuse blew. I have tested every transistor, all testing good apart from the one that wasnt attached to the heatsink - hence my understanding that I've cooked it.
I also tested all the big caps which are testing good.

When testing between D2 and R16 to chassis I get very low resistnace when the ground switch is on CHASSIS, I get a brieft resistance reading in Mohm and then OL when the ground switch is on OPEN.

 

[JohnRoberts]

That is a typical qusai-complementary output stage, so the collectors of the top power devices connect to the + rail, and the collectors of the bottom power devices connect to the output .

That bias switch on wiper of the relay is something I have never seen before (perhaps to make it heat up faster? ) and not typical of such designs. No current limiting protection, but an added bias tweak??  Maybe some engineer had a left over switch contact on the relay he wanted to use. 

With 80+V rails the power devices will definitely need solid mechanical bonding to the heatsink for heat transfer.

Try not to blow it up worse.

JR

 

[Bo Deadly]

Basically I'm measuring resistance between collector ( metal casing of the output device which is normally shielded by the mica) and the chassis.
The amp has a ground lift switch, when in OPEN position I'm getting now no continuity but when I switch it to CHASSIS I get resistance random resistance readings which are dropping or rising.

Bottom line is - I'm not 100% sure what the reading should be but in my understanding there shouldnt be any continuity between collector and the chassis at any point, right?

You cannot measure resistance through transistors / diode junctions AFAIK. The meter will jump about. I don't think that is abnormal. If there were a short, the meter would read a solid 0 ohms. So I think you might be ok. But you cannot leave the mica insulator out. Those transistors have to be thermally controlled or something will be imbalanced and runaway. It sounds like that's what happend. If you don't have a replacement transistor it *might* be ok to remove that transistor AND remove the complementary transistor from the other rail to balance things out. But that is potentially a dubious thing too because the supply voltage may exceed the SOA of the remaining transistors combined. But with 5A fuses I would think it should be ok.

I say this all cautiously because, again, I know very little about power amps. But I appreciate that that amp looks like it's really throwing a lot of current around. If it's not working perfectly I would not be entirely confident about using it. If I were working on this I would be wondering two things:

1) will it burn out my speakers because there's no protection  and the output offset drifts a little

2) what am I going to say when the fire marshal pulls out what remains of it and says "well this is your problem right here"

When testing between D2 and R16 to chassis I get very low resistnace when the ground switch is on CHASSIS, I get a brieft resistance reading in Mohm and then OL when the ground switch is on OPEN.

I'm not sure you can measure resistance through diode junctions. And the resistance will jump around depending on the voltage across it. Maybe because the meter puts out a small voltage and it builds up a charge across it but then it exceeds the diode voltage and starts to discharge. And so on. So it's not a reliable measurement. But if there were a short it would read near 0 ohms.

 

[scott2000]

Can you compare readings from the other channel???

 

[PRR]

> a typical qusai-complementary output

Except the AuriCaps and MOSFETs suggest a high aspiration. It may be a last-gasp quasi-comp, done that way because 200V PNPs at this level lagged the designer's aim. 

> That bias switch on wiper of the relay is something I have never seen before (perhaps to make it heat up faster?

I have seen a similar recently. Also with two trimmers. The amp had cool class B mode and a hot "class A" mode, user-selected. That also had two supply windings, and wasn't true A up to the A mode's clipping. This one does not have two main power voltages (73V is not much drop and the '317 can't feed the power amp), so I suspect door two is a "rich AB" mode which may sound sweeter but will run-up your electric bill much quicker than "plenty good AB" mode.

 

[JohnRoberts]

> a typical qusai-complementary output

Except the AuriCaps and MOSFETs suggest a high aspiration. It may be a last-gasp quasi-comp, done that way because 200V PNPs at this level lagged the designer's aim. 

> That bias switch on wiper of the relay is something I have never seen before (perhaps to make it heat up faster?

I have seen a similar recently. Also with two trimmers. The amp had cool class B mode and a hot "class A" mode, user-selected. That also had two supply windings, and wasn't true A up to the A mode's clipping. This one does not have two main power voltages (73V is not much drop and the '317 can't feed the power amp), so I suspect door two is a "rich AB" mode which may sound sweeter but will run-up your electric bill much quicker than "plenty good AB" mode.

More like class aB vs class Ab, both will be class AB at higher power.

The only quasi-comp output stage designs I've seen in recent decades were literal copies (repackaging of thru hole to surface mt) done late last century by Crown... no actual design engineers were required for that new SKU  : .

It seems odd to add presumably higher quality features while ignoring basic protection like current limiting.  Current limiting can be blamed for spurious seeming audible artifacts in combination with poorly engineered passive crossovers that present deep mid band impedance dips, but an amp that doesn't blow up is also a useful feature (IMO).

JR

 

[Khron]

Given that rare and unusual dual P-ch jfet input stage, the relay-switchable bias, those 1.5ohm(!!!) emitter resistors, the Elna Cerafine electrolytics and perhaps the DC servo, it kinda "smells" like one of those snooty 'audiophool' amps (or borderline that).

I'd be surprised if any cost-conscious designer decided to use near-unobtanium parts ;D Not to mention the unjustified complexity... And obviously, the lack of safety features / protections and the like.

It seems odd to add presumably higher quality features while ignoring basic protection like current limiting.  Current limiting can be blamed for spurious seeming audible artifacts in combination with poorly engineered passive crossovers that present deep mid band impedance dips, but an amp that doesn't blow up is also a useful feature (IMO).

JR

Whic mica cap DID you "misplace", by the way?

I have recapped a big power amp and during disassembly, I misplaced one mica but I decided to power the amp to see if everything is ok.

 

[scott2000]

I think he was talking about the mica pad for the transistor.......???

 

[Aniol1349]

thanks for all your input guys! The amp is up and running again!

In the end I just had to replace the damaged NPN transistor and the rest was just fine.

my lack of understanding of the design shown and I misinterpreted the readings, I was looking for problems which weren't there!

 

[scott2000]

That's great!

The end result is all that matters and I know I've taken much longer roads to nowhere to get there ...

.....even recently! :-[