PMG5 1kW AcBel PSU, Blown Fuse

[tedjar]

I just got (for free ) a Late 2005 Power Mac G5 Dual. This is NOT the one with watercooling..... (yipee)

Bad news is the PSU is totally dead, and after taking apart just about every component in the thing to get the PSU out (think different, indeed), I have discovered the fuse is blown, but amazingly everything looks fine inside the PSU. All Japanese caps, nice clean layout, and just a few dust bunnies. This looks like a lower-hours machine (it did come from an office setting though, so maybe lots of time on standby?)

I've done a bunch of recapping projects and other electronics so I have no problem poking and prodding and soldering, but I have no idea where to start since everything looks pristine in here, just the fuse is blown. (And there doesn't seem to be an obvious dead short either)

I couldn't find much info on this model as it seems to be the "reliable" one as opposed to the LiteOn version, so anything the great minds of Badcaps can come up with would be great.

On the sticker:
AcBel
API4FS13
ID: 291G
Apple P/N: 614-0373
1000w Max

On the board inside:
API4FS13-291G PCB1 R03Q65-0002I

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Can you do some side shots please? Tough to determine what's where under those heatsinks..

Unfamiliar with these as they're always $100+ on eBay and I never see a "parts or repair" one.

With ohmmeter, check resistance from output side of fuse and the other side of line. A low resistance may indicate a shorted coil, cap, or blown bridge rectifier. From what I can see, the BR appears mounted to the heatsink near the white relay(?).

If that doesn't show low ohms, then check the BR with the diode-check function and see if it's okay.

From there, I'd move to the main switcher MOSFETs and check them for shorts.

Anything further at this point is a mere guess.

Toast

PS: Where in NJ?

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Yeah, this seems to be the one that never causes trouble......

From the output side of fuse to other side of line is open. No change on the meter when connected.

I tried testing the bridge rectifier using various tests i found by Googling "test bridge rectifier" and I couldn't come up with a clear result. The values seemed to keep changing as I was measuring them, I don't know if something in-circuit is affecting it or if I'm just doing it wrong. Maybe you can fill me in on the correct procedure... and I will grab my Fluke DVM over the weekend so I have a better tool for the job.

I'm based in Bergen County, NJ, but I do travel around the area quite a bit.

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Great photos!

Bergen is waaay up there from me. I'm in Gloucester. Exits 3 & 4.

Google bridge rectifier and you'll see the 4 diodes you're checking and the polarity.
The center 2 pins are the AC in and the outer are DC +/- out. Black probe to cathode (-|) positive to anode (<-). Reverse the leads and it should show open/infinity.

The switching FETs all seem to be right in behind the bunch of 6 UCC, KMG, brown sleeve filter caps. What you're looking for there is an obvious short with diode test on those FETs across their 3 legs. Looks like 3 pairs of them.

Was the fuse body glass or ceramic?

Toast

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

So close but yet so far...

I grabbed my Fluke 117 this weekend and started poking around (so nice having a real meter). I found a good diagram on Google but nothing was making sense compared what I thought I should be seeing. I took an old, dead ATX PSU and unsoldered the bridge rectifier from that, and it responds as I would expect. So I'm not sure if something is funky here or if the other in-circuit components are messing with my readings.

I can desolder the BR but I will have to remove the white relay and two (I think) film caps which obstruct access to the screw which holds it to the heatsink.

The readings I am getting look like this... (using diode check on the meter)

For the pulled BR from the junk psu (assumed good)
With the - probe on the + terminal, and the + probe on either of the ~ terminals, I get about .500v
With the + probe on the + terminal, and the - probe on either of the ~ terminals, I get OL (no reading)

With the - probe on the - terminal, and the + probe on either of the ~ terminals, I get OL (no reading)
With the + probe on the - terminal, and the - probe on either of the ~ terminals, I get about .500v

For the BR in the G5 PSU (in circuit)
With the - probe on the + terminal, and the + probe on either of the ~ terminals, I get about .500v
With the + probe on the + terminal, and the - probe on either of the ~ terminals, I get about .500v

With the - probe on the - terminal, and the + probe on either of the ~ terminals, I get about .500v
With the + probe on the - terminal, and the - probe on either of the ~ terminals, I get about .500v

Now obviously, the results from the G5's PSU do not look right, it's almost like there are diodes going every which way...

The components behind the filter caps
The row looks like this, from left to right:
[20N60C3][20N60C3][STTH8R06D][20N60C3][20N60C3][20N60C3]

The two left-side 20N60C3's have some shorts and some opens depending how you mix the leads around. The STTH8R06D is a two lead component which acts like a diode for the most part.

The right side 3 20N60C3's all read short in all directions. This would seem to be a problem....

Fuse body is ceramic.

I'm guessing I will be desoldering things shortly to test them out of circuit...

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

>>Now obviously, the results from the G5's PSU do not look right, it's almost like there are diodes going every which way...<<

It could be shorted or it's indicating a short farther along in the circuit. Desoldering 2 legs could tell you better.

>>The STTH8R06D is a two lead component which acts like a diode for the most part. <<

That's because it is. Left leg (facing) is the cathode, right the anode. {see attached datasheet}

>>The right side 3 20N60C3's all read short in all directions. This would seem to be a problem....<<

It could be, but one could be bad making all 3 appear bad. The Gates and Drains are tied together on all 3. The Source legs could be also (looks like it's a topside trace), making them operate as one large 60A FET.

Pinout 20N60C3 (facing): Gate - Drain - Source {see attached datasheet}

You could try desoldering the same leg on each one and see how they read then. If still no good go for a second leg. That would isolate them enough and you wouldn't have to remove them to test. The caveat here would be if they're all soldered to bottom -and- top traces.

The reason I asked about the fuse: A ceramic fuse won't show -how- it blew. A glass one would. A glass one that is blackened or appears 'spattered' with metal would indicate a large current failure, such as a short. One that is just a parted wire indicates an overload, but not necessarily a short. Most of these are time-delay fuses, so momentary overloads don't pop them

Now, from a strictly experimental point of view...
I would solder a single strand of wire from a piece of stranded wire such as a lamp cord or PSU cabling across the fuse. Something no larger in diameter than the period at the end of this sentence.

Then, using a long extension cord and NO LOAD on the PSU, stand -waaay- back and plug it in. If the wire flashes *poof*, it's a short, and you need to start checking that bridge and those FETs. If it doesn't, then perhaps it was just an overload and the fuse could be replaced.

If it does not fail, then try jumpering it on between the PSU ON wire {trace wires back to the circuit board, they're all marked out} and a ground (output side negative). Remember that at least one fan needs to be plugged in for these to operate.

Toast

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

I decided to try the single strand of wire in place of the fuse trick before unsoldering parts, figuring that once I did that I'd have to put it all back to try the single strand of wire test if I did not get conclusive results. I don't know if this ended up being a good idea or not, but I may have made an interesting discovery in the process.

After unsoldering the fuse, I tacked on some wire from some Christmas lights to the bottom of the PCB near the fuse holder. I then ran this over to the aptly-named 3rd hand and soldered one strand across the two wires. Plugging in the resulting contraption resulted in small pop as the wire melted.

Partially against my better judgement, and partially thinking the wire was so fine that maybe 1 strand was not enough for the inrush current on this monster, I tried two strands. Powering this up did much the same thing to the wire, but I saw smoke rising from the PSU as well as from by the wire.

Pic of the resulting carnage is #1 in the attached photos.... If you look closely you can see the 2nd one is developing a crack as well, it's not just goo sprayed on it by the explosion of the first.

Googling the part number (SCK 206L 541) on the surviving one, I discovered this interesting thread which is pretty much the same PSU and problem, but unfortunately they go off into retrofitting an ATX instead of repairing the original PSU. However there must be something in common with these failures...

I also found a datasheet which I think is the right one and attached that as well.

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse



"If it don't work, force it. If it breaks, it needed fixin' anyway..."

AKA: The larger hammer syndrome

Replacements:

http://search.digikey.com/us/en/prod...1042-ND/749882

Might as well pull the Bridge and the FETs and test them.

Check the diode in picture #2, Post 3. In front of the transformer.

Toast

[Th3_uN1Qu3]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Next time use a lightbulb in place of the fuse. Keeps the smoke inside the components. http://antiqueradio.org/dimbulb.htm Of course, it needs to be a regular incandescent type, a CCFL won't work.

60 or 100W does it for most PSUs.

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Well, I've been soldering for the last three hours. This board is one of the toughest things I've ever worked on. There is so much thermal mass that I could not get the solder to pull out of the board fully so the parts would release. (everything is plated through). I finally resorted to pre-heating the board with a heat gun (I used an IR thermometer to make sure I wasn't going nuts) and then I was able to successfully remove parts.

I cleared the relay and film caps out of the way, and removed the bridge rectifier and the thermistors I exploded . Series light bulb next time, good idea. This is a learning experience for me, and I learned something. I'll try to suppress the Tim the Tool Man section of my brain for the remainder of the project...

Then I moved onto the FETs, the middle legs all connect to a large trace on the top of the board which was especially troublesome to get hot enough. I unsoldered them from left to right, testing as I went...

Now guess which component was bad......... yep, the last FET I pulled. That one is shorted every which way.

The other two FETs seem OK.

The bridge rectifier tests good.

The diode next to the transformer tests good.

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

>>I finally resorted to pre-heating the board with a heat gun...<<
+1

>>Series light bulb next time, good idea. <<

Following instructions and not wandering off into Toolman territory would not require the light bulb... I'm just sayin'

The light bulb circuit is on here (mine): https://www.badcaps.net/forum/attach...chmentid=10462

Best to replace all 3 FETs. Same manufacturing run and likely to be a match. Especially since these run together as one.

DK: http://search.digikey.com/us/en/prod...C3IN-ND/593226

Toast

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

I started putting together my Digi-Key order (might as well get some parts for those other projects at the same time....) and I got 3x SPP20N60C3IN-ND FETs and 2x 570-1042-ND to replace the ones I blew up.

Last thing I need is the fuse. It needs to be 5x20mm axial sized, and 15a 250v. The only ones I see in that size and amperage on DigiKey are:

http://search.digikey.com/us/en/prod...336-ND/1023997

Which is normal/fast acting, or

http://search.digikey.com/us/en/prod...1839-ND/777611

which is time delay/slow blow.

Neither of these are ceramic as the original was. Will either of these work as a substitute or do I need to look somewhere else to find the right fuse?

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Markings on the original fuse?

You can stand the fuse up as there is enough vertical clearance. You'll need to insulate one leg and cover the whole thing with appropriate heat shrink tubing. MAINS end goes down on the board, the FUSED end feeds the supply. Ceramic is recommended and there are both available in larger sizes.

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

On closer inspection, I realized the lead ends are attached with little cylindrical metal things pressed onto the ends of the fuse, thereby covering up the other markings on it. I used a Dremel to slit the sides of them so I could take them off and now I can see everything on the fuse.

It says 15a 250v and GBM (<- was completely hidden) and then what looks like a bunch of regulatory approvals with little logos that have JET, UL, SA, CO in them.

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Fast Acting.

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

I picked out DK P/N F3330-ND

http://search.digikey.com/us/en/prod...3330-ND/778519

as the replacement. It fits the bill, 15a, 250v, fast acting, and ceramic. I'll just have to stand it up which should be easy enough.

Now for the 2nd to worst part of every project.... waiting for the parts to arrive

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

Link does not work. Search on part number does.

Looks fine.

Now, build the series lamp circuit while you wait.

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

So, the 2nd worst part is over. Parts have arrived. Not only that, but I had time to install them too

I didn't have any major difficulties installing the parts, the worst was trying to get those nuts back on in the tiny space without ripping any of the tiny SMT parts off on the board standing up next to the heatsink. The pictures pretty much speak for themselves. It might not look like it, but I did apply thermal paste to the BR before I installed it. I verified as much as possible that I had solder all the way through the board. The misc solder spatter was also cleaned up...

I got the lightbulb tester ready in the meantime waiting for the parts, so I gave it a go. With a 75w bulb in there (where DID all the 60's go anyway ) it does the following:

Plugging it in gives one little bump on the bulb and the relay clicks. The +25v standby and +5.1v standby come up and work fine.
Jumping the PS ON (brown) wire to ground results in a few quick flashes from the bulb and then the relay clicks out and it shuts down. Unplugging it and plugging it back in will let you repeat the process.

So the short seems to be gone now.... The suspense in knowing if this thing will work is driving me nuts...

[Toasty]

Re: PMG5 1kW AcBel PSU, Blown Fuse

That is perfectly normal. Try a higher wattage bulb or go without the bulb in series.

[tedjar]

Re: PMG5 1kW AcBel PSU, Blown Fuse

So, I went for it.



Magic click, fans, voltages all look good. Ah the magic feeling of the resurrection of a pile of dead electronics

Now I just need to make the random piles of G5 parts and a box of screws back into a G5, and I might be able to claim success...


Huge thanks to Toasty for all the help. No way I would have been able to get this working without you. Dead on advice every step of the way.