Repair of the M-Audio M3-8 speaker for what may be a common fault

The short story

• Strange noises that start coming from the bass driver when there is no input signal may be caused by faulty electrlytic capacitors on the class D amplifier board.
• Distortion coming from the bass driver even at low volume, but which gets worse as the volume is increased may be fixed by replacing the 220uF bipolar cap on the main amp board next to the class D amplifier board.

The long story

A friend came to me with a tale of woe. His beloved M-Audio M3-8 active loudspeakers had developed a fault. One of them would produce a strange swishing sound on occasion, even when there was no input to the speaker. He had taken the speaker to the offical M-audio repair centre but from what I hear, after a long while they had declared it unfixable. He had looked around on the Net for a schematic diagram but none could be found.

Taking the box apart was quite straightforward. A rear metal panel with a sizeable heatsink was removed and all the electronics were found to be mounted to this panel. A toroidal transformer, a board with what looked like active crossovers, and another board with what appeared to be the three amplifiers. The previous repairer had not fully re-assembled the speaker so there was also a bag of parts including a power resistor, a small PCB, as well as a print out of a class D driver chip data sheet. The small PCB appeared to be some kind of class D driver board. This had been desoldered from the main PCB of the speaker. Looking at the electronics there appeared to be a chip amp for the tweeter, another chip amp for the mid-range, both running off one power supply. Off a second power supply there appeared to be the bass amp comprising the class D driver board and a couple of output transistors. The data sheet for the driver chip included a typical application schematic but little more. I said to my friend that I had no experience with class D amplifiers so I had my doubts as to whether I could fix it.

I looked the bits over. The board where the resistor (a fairly large 3W or so type) had been removed had been quite hot and was discoloured. The resistor itself didn't appear heat damaged. Markings looked like brown-black-brown with black for tolerance (100 Ohm) but the resistor measured 1900 Ohm. One has to be careful with resistor colours however as what looks like brown is sometimes actually red and so maybe it was red-black-red (2k Ohm). 2k is not a standard value however, so this cast yet further doubt over its intended value. I tried to match the resistor up with any part on the typical application schematic but found nothing. After some pondering I wondered if there was no fault with the resistor but that the circuit had actually been designed this way. Perhaps the designer has simply not allowed for the amount of heat being produced.

My friend had done a lot of searching on the Net and found many reports of faulty capacitors in relation to M-Audio gear. This would turn out to be useful information. I looked at the caps on the amp boards and the unknown brand name of the electrolytic caps did not inspire confidence.

After a few days of pondering what to do I decided to replace the hand full of electrolytic capacitors on the class-D driver board for the bass driver amp, and to simply put the big resistor back in. So the following caps came out; 4.7uF 50V C90, 10uF 50V C78 and a bunch of 22uF 50V including C79, C93, C95 and C96. The caps were replaced and measurement on my LCR meter showed the old ones were down noticably in capacity, particularly the 22uF caps. One of which read about 3uF. So definitely suspect.

With the parts replaced and the class-D driver board soldered back in, a misplaced mylar insulation pad from the previous repair was found and it all went back together without too much hassle. Turning the speaker on resulted in no strange sounds and it played music fine. Success!

Things were not great for long however as Murphy was watching and turning on the second speaker - the one that didn't have a fault and had been turned off while the first speaker was away being repaired for an age - showed it was now making a noise without any input at all! So I set about taking that one apart too and replacing the same capacitors. I also measured the big resistor in circuit and found it was certainly not a 100 Ohm, but more like 2k. The stripe colours were not at all clear but it was likely red-black-red with a black tolerance stripe or 2k 20% resistor. Unusual. The PCB was also suffering from heat damage under the resistor on this speaker. With the second speaker fixed I powered it up for a test. The swishing noise was gone, but anything above low volume caused the bass to sound like it was bottoming out or limiting in some way. Now it just so happened that when I had it apart I had noticed a bulging electro on the main PCB. I now replaced this part - marked as NP 220uF 25V. I didn't have a non-polarised cap but I made one from two 470uF 25V caps back to back. The replacement was quite a lot larger than the original which had me wondering. Also the PCB was labelled with a + for one of the leads. Things that make you go hmm. Anyway, cap replaced the speaker now sounded great! So number two speaker went back to its owner.

A few days later the owner reported that the first speaker didn't sound quite as good as the first. I suspected the NP electrolytic cap so he brought the speaker back. In the meantime I had ordered some replacement parts from Digikey. I decided to replace the 2k approx 3W resistors with a series pair of 1k 3W metal oxide film resistors that will be good for high temperatures. I also ordered some decent audio grade 220uF NP caps. Taking the speaker apart I expected to find a 220uF NP cap perhaps bulging, but instead I found a cap of a different manufacture and not NP at all but polarised! Maybe this had been replaced by the first repairer? Anyway, that cap came out and the new caps from Digikey went in. All good except they were physically considerably larger. Hmm.

So now both speakers sound great! Of course in the ideal world one should go through and replace all of the electrolytic caps with new better quality ones but with three amplifiers per speaker there are quite a number of parts. Let's just hope the caps on the class-D driver board were being stressed more than others elsewhere in the design and that the speakers will keep working for a good while yet with their new lease on life!