I'll leave the specs and whatnot for inspection at your own leisure, on the manufacturers' product pages. The good(?) stuff is on the inside - isn't that pretty much always the case?
Speaking of the inside - there's not all that much that meets the eye, right off the bat. The left side has space and connectors for the various optional extras of the range (analog or DSP filtering, four-channel models, etc). The power supply takes up the rear-right side of the space, while the amplifier section occupies roughly the front half of the enclosure. The cooling's handled by a pair of 80mm (temperature controlled) fans in the bowels of the thing, and the two cardboard shrouds over the amplifier heatsinks channel the fans' output while also providing some measure of electrical isolation from the top casing ("just in case", i would think).
The six fat input capacitors (for the rectified mains supply) are Samxon 3300u / 200V - "good enough"; at least they're not CapXon or other utterly infamous brands. The ones on the output of the SMPS are a whole bunch of SamYoung 680u / 63V. Not the greatest either, but at least there's quite a few of them to spread the load, and they've survived this far.
Quite a chunky transformer, too - then again, this thing is supposed to be capable of putting out near to 6KW in total, so you'd pretty much expect some "beef", some heft in there. The label indicates this might be from week 31 in 2011.
Since we're on the power supply, the control circuitry for that is conveniently located on a vertical daughterboard, plugged into a standard 0.1" dual row socket header. Somebody was a practical thinker when they designed this thing.
After unplugging the fans (a pair of Sunon 24V-rated PMD2408PTB1-A), the separating wall they're mounted onto slips right out, providing a(n even) better look at the insides.
Among other things, the reason for the shroud over the heatsinks (apart from aerodynamics) - the output devices are not isolated from the heatsinks. That does of course maximize thermal transfer, but it also means you'll want to ensure they don't come into electrical contact with anything else. Otherwise, things go "poof", or even "bang", and you're gonna have a bad time.
The construction is admirably modular, too - the supply voltages, as well as the amplifier outputs, go through an individual bundle of cables and connector per each channel.
Apparently this had been serviced before - seemingly with a similar issue, if not identical. Fortunately, the service manual was easy enough to get a hold of. With the aid of that and some preliminary probing, at the very least one of the four IGBT's (forming a full-bridge converter) was a dead-short, as well as at least one of the IR2110S driving them.
And yes, the obvious non-factory soldering was indeed a dead (no pun intended) giveaway. See if you can notice any of the telltale signs...
[sigh] Yep, "of course" it just HAD to happen, didn't it? The last screw holding in the power supply board 1) is tighter than a nun's... unmentionables, and 2) has its head damn near totally stripped. Oh joy... And since Murphy never sleeps, it's also in the midst of a group of tall capacitors on the secondary side of the supply.
[1-2 hours later, split between an evening and an afternoon]
<insert long and thorough list of expletives here> It's out!!! But not without using several flathead screwdrivers as chisels, to try to create a slot wide and deep enough to use to drive it out.
The back side greets us with some... interesting details. Nothing terribly interesting on the secondary (left side), but rather on the primary.
The slot by which the copper trace in the bottom-right corner was cut, looks like a factory solution to a small board layout boo-boo. The bigger solder blob near the cut is the connection for the mains live; the one connected to the screw hole "halo" is the earth, and the two through-hole pins visible right in the corner are from C1, connected between mains live and neutral. The pin that's part of the cut trace (also) connects to the mains neutral (bigger solder blob a bit further "north"), while the pin going to the live is right in the corner of the board. So at some point, C1 got turned around, and its connections overlapped, or... something.
Also, that small blue capacitor and the blue mod-wire are... intriguing, as well. The capacitor seems to be between neutral and the negative of the rectified-mains DC bus, while the mod-wire goes between Q2's emitter and D2's anode (see service manual linked above, pages 42-43). Not sure why it's been left quite that long; i'd imagine one would want that connection to be as short and beefy as feasible.
Speaking of the service manual - i was thoroughly impressed by what i saw on page 5. Service instructions as detailed as recommending how to remove the blown IGBT's! As well as listing the most likely collateral victims, in such events (like the gate drivers and such).
Right, out come the IGBT's, and surprise-surprise - the origin of the mod-wire is revealed. One of the previous servicings managed to rip the Q2 emitter via upon removal. Nice going, guys...
Actually no, i take that back - even with the wire disconnected, both the via and the trace are fine. So what gives? Well, just to be on the safe side, i WILL solder that wire back, albeit shortened.
After removing the four IGBTs, further on-board probing revealed the freewheeling diodes to be fine. Speaking of the IGBTs, turns out only Q2 had given up the ghost; the other three are not acting as wires. Either way, they're all getting swapped, just to be on the safe side. No point in risking the need to fix this sooner, rather than later.
PS: I don't know about you, but i'm ever so slightly suspicious about these IGBTs i'm replacing. Maybe i'm just paranoid, but the little "window" in the plastic moulding, where the markings reside, looks like it's been milled out, and the markings are really faint and tricky to distinguish unless you catch the light at the right angle. Very much unlike the replacements i got from Mouser...
Either way, swap done. I even took the liberty of cleaning up all the stock white thermal goop from the heatsink and the IGBT mica thermal pads, and replacing it. Now then, on to the control board...
Swapping out the pair of IR2110S drivers went as smoothly and uneventfully as could've been hoped. The couple of missing pads were missing beforehand, so that's none of my doing. That being said, upon plugging the control board back into its socket, as opposed to the initial state (where the little LED near the top edge merely blinked), the power indicator LED stayed on. Unfortunately though, a faint "tick-tick-tick" sound could be heard from within the power supply. That usually indicates an output-overload-induced "hiccup mode".
After unplugging the two multipin power-and-output connectors between the PSU and the audio board, sure enough, when flicking the power switch, the fans ramped right up. A quick probing across the emitter and collector of one of the accessible transistors from channel A confirmed it - dead short. Oh joy...
After removing the amp board from the casing, we finally get a look at the bottom side. I wanna say it's all still as it came out of the factory, but they sure did a kinda-sorta messy job. The soldering itself is ok, but hot dang, that's a lot of flux spatter, all over the place.
But what do we see here, near the front edge of the board, on channel A? Skidmarks coming out of a small transistor that's cracked in half, and a tiny pinhole "crater" in one next to it (Q78 and Q81, respectively). A "promising" start...?
Moving on - since each "half" of each amplifier has six beefy (MJW21195 and MJW22196) output devices in parallel, they need to get pulled out and individually tested, to make totally sure of the facts. A bit of "manhandling", desoldering and probing later, resulted in all six 21195's of channel A testing as good. The smaller MJE15033 driving them, on the other hand, not so much. Still, that's good news - it's just ONE (medium) power device, and i have a few identical ones in my parts stock.
As a sanity-check, before removing any transistors, i probed across the collector-emitters of the negative side of channel A - around 130mV, which is a bit low. Measured about 120mV across the collector-emitter of Q15 (MJE15032), the negative side driver; granted, that had all the base-collector junctions of the six output devices in parallel, but that's still quite low. Pulling Q15 from the board cleared the air - it's toast as well, and now i'm measuring a much healthier 430mV across the collector-base junction of the six negative side output transistors. But that "only" means i'll need to double-check all the silicon on channel A. It would be a major shame to install new driver transistors just to have them blow up due to remaining "upstream" problems.
[Some time later...]
Well, good news - all the rest of the silicon components test out fine, and no other "skidmarks" are visible anywhere so... replacing the two blown transistors it is. Slight problem though - i have no SOT-23 packaged transistors anywhere near the voltage ratings of the originals (160V for the MMBT5401, Q78, and 300V for the MMBTA42, Q81). Rummaging through my parts collection produced a KSP44 (500V NPN to replace Q81) and a 2SA1018 (200V to replace Q78). Even the gains match up more than well enough, especially considering that Q81, as far as i can figure, acts as a switch, as part of the muting circuitry, and Q78 seems to act somewhat as a so-called "Vbe multiplier".
No, it ain't pretty, but you gotta do what you gotta do, to get the job done. Especially since, as the owner just told me, he'd like the amp back ASAP because he has a gig to leave for tomorrow before noon. Anywhosits, that's about that. Due to the pinout and mechanical factors, i "knitted" that strip of Kapton tape in-between the new Q78's legs, to ensure the base won't make erroneous contact with the collector.
Right, once i shoehorned the amp board back, i took it to the bathroom to plug it in. The main reason being that the power socket in there is on a separate circuit breaker than my living room / workbench here, and if the worse happens, everything here wouldn't suddenly turn off. Work smarter, not harder...
After plugging in, i flicked the power switch on the front panel and backed up a bit (just in case). About half a second later, the blue power LED came on, along with the fans, and so did the "PMS" LED indicator (cue jokes...), but that went off 3-4 seconds later. I think we've got a live one here!
Sadly, i don't think i have any Speakon connectors i could use to hook up a test speaker or dummy-load for signal tests, but still - it's a helluva lot more and better than before...
In the mean time, it's been handed back to its owner, and will get a real-world workout on tomorrow's gig, so... Here's to hoping for no bad news, eh?
No comments:
Post a Comment