fixed amp, get heat in one channel

[jol50]

Trying to fix a sony xm2002gtr, a 2ch or 500wrmsx1 auto amp made in '06. I found a shorted power mosfet on each bank of 3 that blew a trace. After some searching found a replacement for the sanken $15 sony part and installed all 6 for 15, as you can buy this amp for roughly $100+ship. Looked very close but new ones a little faster timing, can post if needed. Resistors looked good, tested all good, solder looked good. Started on limited power, eventually ran it on a car HU and to 8ohm house speakers and it played fine at low volumes I ran. Then I noticed one channel of the outputs was getting warm so I killed everything.

I tested all of the outputs for shorts and the resistors...everything looks ok. Don't have a manual for it yet but it has bias pots it looks like, they do affect voltage to speaker leads. I turned that one down and it still has more voltage (2.6 think it was, compared to 1.8 on other channel and .8 on other leads). Nothing is grounding to the heatsinks, no other visible damage I can find. Rails show + and - 48v, only thing I found is near a surface mount in the OP amp there was a black ink looking spot. Could be ink or something puked a little not sure. Black one with three leads. Ran the amp again for a while with no load or input, the one channel gets maybe 100F on the 1" block of aluminum the chips are mounted to. That block mounts to case for more heat transfer when assembled. Can't really tell by feel if one or more chips do/don't get hot or all, it is warm to hand but can keep hand on no problem.

Not really sure what to look at now, everything else in amp and power supply is cold and it plays music. At this point my direction is getting fuzzy so I thought I would ask for some suggestions. Here is a pic of amp, surface mounts are on back side mostly. Two power fets that were blown are marked, the sink at other end is the warm one.

[Perry Babin]

If the bias control won't reduce the current to a level that allows the outputs to run cool, you either have a drive problem or a leaking output.

The bias controls shouldn't have an effect on the DC offset on the speaker terminals. Were you measuring the voltage across the speaker terminals or did you use the chassis ground as the reference?

[jol50]

That was to ground for whole amp, just wanted to see if they were the same or close. I did lead to lead but don't remember exactly what it was. Also that was the leads from the outputs, one side is reversed for bridging I believe so the output is the negative there. Right, the pot lowers voltage to a point and no more, and it is still higher on that channel.

I also get higher voltage (about same, 1/3 more) at the pot for that channel but let me double check that if important. My next tentative step was to pull the outputs one by one, can't really test for leaks can you? Looked at the surface mount stuff for a while and can't find any visible clues in the driver/output areas. Checked the drivers for shorts but figure they would not play music if they were.

I got it broke, have no idea what happened to it but it looks fairly nice.

[Perry Babin]

If you haven't yet pulled the output transistors, measure the DC voltage from pin 1 to pin 3 on all 6 outputs of the defective channel.

Each one of the 3 in parallel should read the same but you need to check all of them to confirm this.

If the voltage is the same for each transistor in each group of 3 parallel transistors, post the voltage for the N-channel group and for the P-channel group. If you have one that's not the same as the others, also post its voltage and circuit board dseignation (Qxxx).

Measure the DC voltage across the speaker terminals again.

[jol50]

All three were the same for both sets. Rails were at 43.4v on charged but weak battery no load or input. 3 on N side had 3.6v, P side had 3.3v. Two outside pins.

Checked speaker terminals to each other with meter at 200Mv and for + to - hot side showed 2.5 and good side was 1.4. + to + was 1.1 and - to - was 0. That was treating the outputs as + even though one is reversed, the good one is.

BTW, thanks for the help and what a great site you have

[Perry Babin]

It looks like the outputs are probably OK and simply biased too high. If you get run out of things to check, you could pull them and check them but I don't think you'll find that any are defective.

If the bias pot can't get the bias current down to an acceptable level, look at the solder connections on the bias transistor and the surface mount resistors in the area of the outputs and the bias pot. You may need to use a high power, lighted magnifying glass to see any defects. Desolder (very important) and resolder any suspect solder joints.

If you don't have an amp meter on your power supply, you can monitor the bias current by measuring the DC voltage across the emitter (source) resistors (the large white resistors). Low voltage means low current. Use the voltage across the source resistors of the good channel as a reference.

If you measured 200mv DC across the outputs, the amplifier may have other problems. 200mv isn't likely to damage any but the most fragile speakers but it's excessive by most standards.

[jol50]

Yes I was thinking of a magnifying lamp on an arm.

The meter was at 200mv setting, for some reason it has 200 and 2000 as lowest settings instead of 10 base? So I guess those readings would be in mv. The speaker leads show 0 at 200v setting I checked the transistors with, the regular volts reading.

[jol50]

So it should be 2.5mv and 1.4mv at the speaker terminals. I have good light and a big 4" mag glass, but those things are tiny what a pain. I will check best I can next pass at it.

The one with the spot next to it might be suspect, its in that area. It may be a transistor with the three leads but I can't see a number on it. I will test it and see if the other channel has the same, once I trace it more and see where it divides on the board better.

[Perry Babin]

If you suspect that you have a defective transistor, that's where you should begin. It should be marked but will probably have only 2-3 letters/numbers on it. You'd need to look up the actual part number. Google 'smd codebook'. If there are several different part numbers for the code on the transistor and the case styles are very similar, you'll have to carefully measure the transistor with dial calipers to determine which transistor you have.

Most of the lighted magnifying glasses on an arm are useless for this type of work. The distortion and chromatic aberration are so bad that you can't see much more than you can with the naked eye. If you intend to continue doing this type of work, buy a lighted magnifier like the one in the attached image. I've tried many different lighted magnifiers and these are the best I've found. They are rated at 10x mag.

[shagone]

Perry,
where would one find that specific 10x magnifier in the picture? do you happen to know the maker or part number? i was thinking of getting a cheap lighted one on an arm but since you say these are better i'm willing to save some bench space and give this small one a shot.