Various notes on recapping a G4 MDD Power Supply (AcBel API1PC36)

UPDATE: This project is now finished. I ended having to order the missing caps, and thankfully that was enough to get the MDD to start up and work. I’ve since started pushing back information to the CapsWiki page, so if you look there now, the missing info is filled in along with replacement parts for the caps that are no longer stocked. I hope this info helps someone. I’ll also likely be updating the scope of this post to cover the Samsung PSU, as I’ve got one of those on hand to mess with. Original post follows:

Boy, this was an adventure and a half. I’m not even done yet as of this writing but most of the really hard stuff is out of the way and all I have left to deal with is making another DigiKey/Mouser order for the caps that were the wrong value.

Given how involved this is, I just rolled with the parts list from Caps Wiki. I didn’t have the brain bandwidth to pull every single cap, measure everything, and try to navigate a vast list of caps. Thus far, the replacement caps list has been spot on with one notable exception: The replacement caps listed for C8, C32, and C401 appear to be wrong. Rather than being 47µF/25v capacitors, the replacement link goes to a capacitor that fits, but is 1000µF/10v. Now, I’m not sure if somehow, some way, the 1000µF cap will work just fine in place of the 47µF one, but general googling suggests that yes, you can go up in µF, but you cannot go up in voltage. So I just cursed under my breath and decided to carry on with the rest of the recapping, and maybe I’ll just place an order for the proper caps later.

This job has also tested damn near every tool I have at my disposal, and my patience. If you are looking at attempting this yourself, I’m going to say 100% a desoldering gun is required. This job sucks major, even moreso than the IIsi power supply that I recapped that was flooded with cap juice.

Various tips ‘n things

There is lots of adhesive all over the damn place. Bring lots of isopropyl alcohol. This has been the bane of my existence. Bunch of caps are stuck together, some of them stuck to nearby components. You’re going to have to break the caps free to get them out and it’s a really delicate process. I’ve been told IPA will make short work of the adhesive but when I was going in I just opted to do things the hard way because I’m low on IPA. So it goes.

Speaking of that adhesive, it’ll make a filter cap near the AC input seem like it’s soldered to the PSU’s main PCB. It is not. This puzzled me for a good bit, as there’s this filter cap that feels for all intents and purposes like it’s soldered to the main PCB, but the leads going from it to the AC input are so short that you have no room to desolder anything. Good news, though, that filter cap is just adhered to the board, and once you break it free of the adhesive it’ll just hang out behind the AC input. This will give you enough slack to get the board out, turn it over, and desolder the wires from the IEC connector to free the board from the metal housing. Photo here if you want to see what I’m talking about.

The little daughterboard needs to be removed, even if you’re not recapping it. It’s that little sub-board that’s toward the bottom center if the board’s heatsinks are facing up and the molex cables/etc are coming out toward you. This is where the desoldering gun will come in clutch: there are a bunch of little joints that need to be cleanly desoldered for this board to come out. Worse yet, it’s also adhered to the main board. Shouldn’t be too hard to break it free, though.

I’m not 100% on if the capacitors on this board (it’s the fan controller board, for the curious) need to be replaced, but they’re listed in the Caps Wiki parts list (save one, which is listed incorrectly, which is one of the aforementioned 47µF/25v capacitors that got switched) and thus I had already ordered them, so why not just get it done while we’re in here.

Even if this board were entirely skippable, we want it out anyway because it makes accessing everything behind it that much easier, and space is at a premium in here. There are a bunch of caps by the output lines/coils/etc, and they’re really tightly packed. You’ll need all the room you can get.

You don’t have to desolder the wires going to the fan controller, though! Once you’re done with it I find I can just route the wires up and around the nearest heatsink, and tuck the board in between the two heatsinks where it’ll stay out of your way for the most part.

Sometimes you just…gotta pull. A few capacitors didn’t want to come out even though they had all the solder sucked away, and the legs were very clearly loose and not attached with the tiniest amount of solder. The stock capacitors have a little bend in the legs that cause them to sort of clip into the holes, and that can cause some resistance when trying to pull the capacitor. There have been more than a few where I just had to grip the damn thing and give it a nice tug before it’d come free. So if it seems like you’re getting punked in this regard, well…just try pulling harder.

Oh yeah, you’ll probably need a damn good soldering iron for this. I use one of those KSGER T12/T15 knockoff irons, and up until now the thing has been totally great in soldering shenanigans! But the absolutely beefy ground planes of more modern PSUs have this poor thing struggling a fair bit. It’s getting the job done but I’m noticing I’m having to boost the temps a fair bit higher than usual. Just be forewarned that if you have a weaker soldering iron you may well be testing its limits here. (Heh, maybe I should consider stepping up to one of those fancy C245 irons…)

(This is a work in progress, I will update more as I finish. Also, if you want to see photos, I threw them in my Repairs album on Flickr.)