#2 Amphion Impact 500 (DB Series DB5004) subwoofer amplifier autopsy (and repair, hopefully)

Right, so i've had this amplifier brought in last week (sans the subwoofer enclosure itself). When i noticed the brand, i remarked to the owner that, to the best of my knowledge, Amphion stuff is anything BUT cheap. He confirmed, this whole thing having costed him something over a grand(!!!), a couple years ago.

It had allegedly been serviced before, and he pointed at an electrolytic cap. A ChemiCon SXE - totally out of place on a board otherwise full of CapXons. Yep... Oh joy...



First noticeable thing - this crock stinks like an absolute mother...lover. Fine splatters of some oily-looking residue nearly all over the place. But sweet heavens, the SMELL...

Okie-dokes, on to the preliminary visual inspection:

• Great big honkin' toroidal transformer - this may well be the one and only thing the OEM got right in this whole debacle. You didn't imagine Amphion (Finnish speaker manufacturer) would have designed and made the electronics, did you?

• Four TO-247-cased devices, two reasonably sized toroidal inductors - pretty much a dead (no pun intended) giveaway this is a full-bridge class-D amplifier. Power devices are IRFP250 MOSFETs.
• Interestingly enough, they went with a 4th older low-pass on the outputs (two LC filters, cascaded).

• Two DIP-8-cased chips with (most of) the markings scraped off. WHY?!?!? Fortunately they did a pretty lousy job on one of them, and by "pixel-peeping" at a zoomed-in photo and comparing the pinout to some datasheets, i managed to determine these are IR2101 gate drivers.

• Speaking of gate driving, instead of the more usual (and arguably cheaper) resistor (for limiting the rate of charge for the MOSFET gate) and reverse-parallel diode (for quick gate discharge), they went with a resistor and diode + PNP arrangement.

• Oddly enough, they didn't bother to un-mark the TL072 and TL074 opamps on the bottom side of the same board. Plus points for consistency, guys! <end of sarcasm... for now>
• Two big fat snap-in bulk filter caps; one domed to all hell (and likely vented), the other one not, but likely not far behind. Gonna both get replaced as a matter of course.

• Electrolytic(!) caps in the output filter. That's a death-sentence / guarantee of eventual failure if i've ever seen one. Bloated on one side, non-bloated on the other. Non-polarised, but still...
• One wide-ish output trace blown - i mean BLOWN - off the board. One almost as bad on the negative output of the bridge rectifier. Same side where the crapped-out snap-in cap is. Not a coincidence, rest assured...
• Nicely "tanned" (actually, more like "toasted") part of the board in the linear regulator area. Cheap OEMs always skimp on thermals, don't they?

• Weird mix of adhesives, though. In most places, they squirted gobs of this gummy translucent-ish silicone caulk / RTV, which is just as elastic as you'd expect, even in the hotter areas of the board. But for whatever reason, they chose to also employ a few dabs of the "cancer for electronics" yellow-glue-that-turns-brittle-brown-and-corrosive-with-time-and-temperature, aka "brown glue". One of those dabs is (or rather, was) right where the two wires going to the heatsink-mounted thermistor are soldered into the board. Sure enough, one of the wires is just flappin' around in the breeze.
• Cute touch, though - they actually bothered to use a uPC1237HA protection chip, and an output relay and all. Well, we can see just how well THAT worked out... Then again, odds are good the speaker driver itself survived, so it might actually have done its job(?).
• On the bright side, the mounting screws of the output devices are easily-enough accessible (nothing directly in the way).
• I'm... actually mildly impressed - they used what looks to me like aluminium oxide insulation / thermal-transfer pads between the MOSFETs and heatsink. This must be the first time i've seen these things used in some equipment, if memory serves. They're not exactly what you'd call cheap either - in the region of 50cents in bulk quantities. I was wondering what's with the metric ton of white thermal goop there; turns out 1-1.5mm of the thickness of the white stuff is the pad. I would've expected some plain old cheapo mica sheets(a mere 10-15 cents, if not less, in quantities).

Couldn't find anything (relevant) about this plate amplifier, save for the user guide of the exact same thing (the electronics) as employed in another active subwoofer - the Bolzano Villetri DB 5004 BP-AB. There's also no real information about the OEM of this thing, save for the model numbers this board (set) is used for / sold under:

DB Series
DB2504
DB5004
DB7504

P/N: 015-170900-001

Next step, measurements:

• Two of the four output MOSFETs shorted. Surprise-surprise, on the same side where the blown trace is - colour me shocked (nooot...).
• The gate drivers would seem to have survived - couldn't find any of the supply / output pins shorted together, so... I'd dare call them ok.
• The electrolytic that had been replaced in service before, seems to be on the supply rail for the two gate drivers. 
• The bridge rectifier seems to be just fine. The only issue was measuring a dead short between the + and - terminals, but that was before removing the (dead) MOSFETs. After that, all is well in that area.

Preliminary plans:

• Replace the two bulk filter caps - one's long gone, and it's not worth taking a chance on the other one. It IS a CapXon after all, and thus, it WILL fail. Sooner, rather than later.
• Replace all four output devices - if i gotta replace two of them, might as well go whole hog and swap'em all out.
• Assuming i can shoehorn them in, replace the output filter caps with "plastic" ones (PET or PP, depending what i can find in my parts bins; worst case, i'll order a few, along with the new caps and MOSFETs).
• Obviously, reconstruct the blown / questionable traces, but not before "amputating" all the darkened areas.
• <sigh> Wouldn't you know it? "Of course" one of the ceramic thermal pads just HAD to crack in two...

I need a few other items from Mouser, so i might as well add the stuff i need for this bastard, to the order. I'm looking at a pair of Nichicon LGU 5600u/63v (stock ones are 4700u/63v) for about 3.50e a pop, four Infineon IRFP250M's for 1.70e each, a couple of those TO-247-sized alumina (aluminium oxide) thermal transfer pads, and at least for now, that should do it.

First step = removing the big caps (in order to facilitate the removal of the apparently-good pair of MOSFETs). The bloated / blown one is exactly that, and then some - it must've went literally ballistic!

Also feels / sounds like whatever inside is ever so slightly rattling around in there. Right now i'm starting to wonder whether this cap crapped out and caused one half-bridge to blow, or whether that half-bridge caused this cap to kick the bucket. For some reason, the former "feels" a bit more plausible to me. I've seen shorted-out caps "bake" connectors and/or rectifiers before, so i wouldn't be surprised if this was the case here, as well.

I was mistaken!!! This is NOT a full CapXon rig. The electrolytics in the output filters are, of all things, "Elytone". I... I've got nothing...

Aaaaaanyway, without spare parts, there's not much else i can do right now. Maybe reconstruct the output trace, but that might just get in the way of soldering in the new MOSFETs, so... I think that can wait. Now, off to ordering parts.