Troubleshooting M-Audio BX8 powered monitor

[dissonance]

I'm trying to troubleshoot a problem with one of my M-Audio BX8 powered monitors. At higher volumes, the low frequency driver begins to cut out periodically- It plays for about eight seconds, cuts out for two, etc. After ruling out a driver problem, I started looking at the amplifier itself. It's using a pair of LM3886TF's, and scope / VOM checking pins at the low frequency amp has revealed the following:
Pins 1 & 5 (V+): ~32V
Pin 3 (output): Loses signal when cut-out occurs
Pin 8 (mute): ~3V (never goes open or to 0V, so I assume that is not an issue, but I don't really understand how this works)
Pin 9 (Vin-): Loses signal when cut-out occurs
Pin 10 (Vin+): Maintains signal when cut-out occurs

I haven't been able to locate schematics for the BX8 (a tech at M-Audio said he has a difficult time getting them himself, for what that's worth), but I've been referring to the following datasheet for the amplifiers:

http://cache.national.com/ds/LM/LM3886.pdf

I'm at a loss as to where to go next. If anyone has suggestions, or knows where I can find schematics, I'd appreciate any assistance you can offer.

[Jazz P Bass]

Pin #9 is the audio signal input pin in this case.
It sounds like the problem is before this.
You are going to have to trace it back to the failure.

[J M Fahey]

I'm sorry for seeing it from another angle.
For me (and I can be wrong), *signal input* is on Pin 10 and it never dissappears, so I can't blame it on an earlier stage; Pin 9 is the feedback one, where we have a sample of output signal; more precisely 1/20th of it, if following datasheet application.
It's normal for it to dissappear if output signal (pin3 ) dissappears too.
V Mute not changing on pin 8 , only means that there's not an "external order" to mute the amp; it says nohing of internals.
I guess that the thermal or Spike protection is acting up, specially because this is an LM3886TF, the plastic insulated one.
Many use it for convenience, but forget that it dissipates half as much as the metallic flange one.
Touch it to see if it's way too hot.(the body, not the heatsink)
Cool it with a small fan and see what happens.
Of course there may be *other* motives too.
Good luck.

[Jazz P Bass]

Thankyou for the correction J H.
I failed to read the datasheet.
Good call.

[dissonance]

Thanks very much for the input, folks. When I initially noticed the problem, my first thought was that it behaved as though a protection circuit was causing the cut out. I thought this because it works normally for a few minutes after being powered on, and once it does begin to act up, it's very consistent (works for a few seconds, cuts out, works, etc.) Tomorrow I'll check to see how hot the actual housing is getting, and try cooling it. I know the heatsink was definitely getting a bit warm.
I'll report back with my findings.

[dissonance]

So the body of the amp is getting very hot, and I was able to temporarily stop the cutting out by chilling it with compressed air. Is it safe to assume the internal protection circuit is the problem, and if so, should I replace it with a standard LM3886 (without the plastic insulation)? Also, do you think it's prudent to replace mid/high frequency amp as well while I have it apart?

[J M Fahey]

An LM3886 with traditional mica and grease will work definitely cooler.
You should also check it for oscillations (which would definitely overheat any SS stuff).
If available use a scope, otherwise an easy to build RF probe.
HF amplifiers work cooler .
I'd change any low frequency plastic insulated chip, maybe not the HF ones, although it's your time and money.

[doctor]

An LM3886 with traditional mica and grease will work definitely cooler.
You should also check it for oscillations (which would definitely overheat any SS stuff).
If available use a scope, otherwise an easy to build RF probe.
HF amplifiers work cooler .
I'd change any low frequency plastic insulated chip, maybe not the HF ones, although it's your time and money.

I hope that the schematic may help you

[dissonance]

An LM3886 with traditional mica and grease will work definitely cooler.
You should also check it for oscillations (which would definitely overheat any SS stuff).
If available use a scope, otherwise an easy to build RF probe.

Thanks, I do have access to a scope. Can you elaborate on what it is that I should be looking for?

[dissonance]

I hope that the schematic may help you

Thanks very much, although that's for the newer, "A" model. Looks like those use a different amplifier (TDA7294). Mine are the original BX8 model.

[J M Fahey]

Inject audio of the appropriate frequency and drive it until you get around 5 to 10 volts on the corresponding speaker (half that on the tweeter) and manually sweep it up and down.
You should see just the clean thin trace drawing a sinewave on screen.
If you see "grass on the floor" with no signal or superposed over the clean sinewave that's an unwanted oscillation which will heat your chip a lot.
You may also see a sharp kink in the sinewave or "the tit grows a nipple".
Sorry for the example but that's what it looks like.
Regular flat-top clipping, even with some ripple modulating it, is normal.
Today's Saturday, lots of work, but in my free time I'll draw and scan the typical waveforms.
Note: sometimes the oscillation frequency is so much higher than your audio frequency, that you don't see small sinewaves superimposed over the audio one, you just see the trace going appreciably thicker.

[Steve Conner]

Just another couple of things you might want to consider:

Are you putting too much bass energy through it? (Try swapping the feeds to the left and right speakers, and see if the other one starts to cut out.)

Is it mounted in some place, like stuffed in a bookshelf or against a curtain, where it's not getting free air circulation around the back panel?

[km6xz]

If you are unsure of the stability of the amp, use a dummy load, or better, no load if you have a scope to verify if it is stable before connecting the woofer again. If the chip still get hot with no load it is probably at fault. If it stays cool when running without a load, it is probably not an oscillation(which usually have audible beat notes in the output) and a closer look at the load might be in order.
I assume you made sure the matting surface between the chip and heat sink is flat, clean and tight.

[dissonance]

@ J M Fahey: I just tried connecting a signal generator to the input, and drove it with a sine wave. The scope pattern to my eyes looked great- Nice and crisp, and nothing superimposed, even when sweeping frequency and amplitude. I then let it sit there for five minutes or so while driving it with a steady 100 Hz signal at a moderately high volume, and the chip got very hot and started cutting out.

@ Steve Conner: The other speaker is stable when presented with the same source material. Also, I can get the problem unit to overheat when it's apart and on my bench, with plenty of air circulation.

@ km6xz: I tried running it without any load as you suggested, and the chip did not get hot. However, I had previously swapped the woofer from the other monitor that is working normally, and the problem still persisted, so I don't think that the load is at fault.

What Steve mentioned regarding too much bass energy got me thinking- Since I assume that the crossover takes place before the amps, is it possible that the problem could lie there?

Thanks again to everyone for the input.

[J M Fahey]

Well, after your tests, specially after having swapped the woofers , it only remains the heat dissipation itself. Try a metalflange/mica/grease option .
It has also been suggested to consider bad thermal contact.
Sometimes happens, it's easy for a clipped component leg, a bit of plastic insulation, etc. to get stuck to one of the already greased mating surfaces, before mounting.
After all, they fly all over the place.
I once repaired a guitar amp where the PCB's edge was not properly cut and the ChipAmp barely reached the heatsink.
Top edge mated properly, but you could easily slip some *thick* paper between the lower edge and the heatsink.
Owner was fed up with the amp cutting off in the middle of his best solos, yet *never* doing so during rehearsals.
I almost believed the amp had ears, because his playing was terrible.[]