Suspect Bad PS Apple AcBel API 1PC36

[Kansas Sparks]

Re: Suspect Bad PS Apple AcBel API 1PC36

CORRECTION: C7 Capacitor noted on the above MAP.jpg shows incorrect info. Should read : C7 - 220 uF /35V

[Toasty]

Re: Suspect Bad PS Apple AcBel API 1PC36

While you're at it, would you change the Yellow dots to something that contrasts better? Orange, purple, polka-dot...

[Kansas Sparks]

Re: Suspect Bad PS Apple AcBel API 1PC36

Purple it is.

[Toasty]

Re: Suspect Bad PS Apple AcBel API 1PC36

[chris_peak]

Re: Suspect Bad PS Apple AcBel API 1PC36

Thanks to folks on this forum (Toasty especially) for everything on this and related threads which enabled me to repair the ACI1PC32 unit in my G4 MDD. This is my experience for the record…

Initial symptoms: power light flashed but no boot. On removing the PSU, I measured good standby voltages but no 3.3/5/12 on grounding power-on. The fans twitched but that was all. I pulled the guts out and could measure good B+ (370v). There was no evidence of bulging caps or other distress.

First, I replaced 10 of the 12 shotgun caps using this fine list (https://www.badcaps.net/forum/showpo...&postcount=112) from Digikey. I was able test and eliminate C7 in circuit and C8 by removal. None of the others was bad but a couple were leaky. I used this excellent ESR meter: https://www.digikey.com/product-deta...004-ND/9687220 Sadly, this didn't do the trick, so I ordered the 7 of the 10 “other” caps (the big guys C5/C4/C37 tested healthy in-circuit and certainly seemed to work well).

From various discussions, it seemed like C45, C32 and C30 might be the next suspects. So I pulled and tested those. C45 and C32 had a high ESR but they weren't completely fried .. and C30 was too small to check. I replaced these, retested and .. praise be! .. all outputs fired up. So I applied new silicone caulk, partially re-assembled and re-tested. Nothing. Back to standby only. Rats.

At this point, I could only guess that other caps were marginal. So I pulled C53/C55/C56/C57. I could only test the 4.7uF guys and they both checked good. But replacing all of them indeed got it working properly. So I can only conclude that one or both of the small 0.47uF caps were the culprit.

Note that the best way to deal with the original adhesive caulk for me was to cut between caps with a modeling knife so that each cap could be easily desoldered individually.

Also: after a test, to safely discharge and measure the high voltage on the primary caps, I'd hook a multimeter between B+ and R4, the PFC current sensing resistor, and used a 10K+ resistor to bleed between meter terminals. Otherwise, it would take at least 20 minutes to discharge adequately. This was after It'd been tickled once!

[useanavo]

Re: Suspect Bad PS Apple AcBel API 1PC36

Hi there. I have a Power Mac G4 with an Acbel APi1PC36 PSU. The fault symptoms sound similar to those reported here - pressing the power on button produces a twitch on the fans and a red LED (D518) on the motherboard flashes once then nothing else happens. I have removed the PSU and, using a 22R resistor to bridge pin 6 (3.3v sense) to ground and a jumper between pin 11 (green) and ground, it starts and runs delivering the expected voltages at the other pins. I have since replaced the following caps: C8, C7, C10, C16, C13, C21, C29, C59, C60, C35, C41, C57. I have an analog Ohm meter to test leakage and all of the removals seem to charge up and hold charge as expected. Only C60 was slightly domed at the top but still seemed to be OK. Obviously, I have left all of the new replacements in and the PSU still works on the bench, but not when connected to the Mac. Is there any point in my replacing any more caps, or should I be looking for a motherboard fault, in which case some help and advice is sorely needed!
P.S. I have just attempted another re-start and the Mac powered up with a start-up tone, ran for maybe 5 seconds, then powered off.
P.P.S. It now seems to keep running...... Advice please - change the rest of the electrolytics or no?

[chris_peak]

Re: Suspect Bad PS Apple AcBel API 1PC36

I'd replace all "shotgun" and "other" caps before looking elsewhere.

[TheJeffChen]

Re: Suspect Bad PS Apple AcBel API 1PC36

This is someone from the future... *ahem* Many thanks to all of you in this discussion.

I managed to fix my AcBel by replacing the bulging big ones (2x2200uF, 1x1000uF) and the small ones (C30 C32 C45, but not sure which one was the cause). Before it was fixed. it was turning on and off instantly and standing by with 25V and 5V.

I did notice that even they went to town with the glue, some parts managed to lift the traces and became loose. I had to put more solder on them so they could bridge to something next to them to stay in place.

[sam_sam_sam]

Re: Suspect Bad PS Apple AcBel API 1PC36

Lifting board traces is either to much heat or staying one the trace to long with to much heat
or not enough heat and you are having to stay to long on the trace

One solution is add fresh solder to the joint with the right amount of heat

To determine if you are using enough heat is to the iron on the joint and as long it is not a double sided board with large ground planes you should be able to melt the the soldering joint in less than 3 second if not increase the heat setting

Now with my soldering station I set it for 675*F but I get on a soldering joint and once the parts pin is loose either suck the solder off the joint or move the pin out of the hole

Now I also have a Desoldering Station I set it for 775*F so it quick to melt the solder and suck it out of the joint hole then take the tip off the board as soon as possible to keep from burning the board now on double sided board or ground planes you have to leave it a little bit longer for the solder to melt

I hope this helps you

[TheJeffChen]

Re: Suspect Bad PS Apple AcBel API 1PC36

Thanks for all the soldering tips. I know how to solder pretty well. What I meant was the parts started lifting traces from the board on their own from the other side. That surprised me since they went nuts with the white stuff, everywhere, all over.

[sam_sam_sam]

Re: Suspect Bad PS Apple AcBel API 1PC36

What a crap-e board

[betaporter]

Re: Suspect Bad PS Apple AcBel API 1PC36

I've got this same PSU. Initial symptoms were the same as described elsewhere in this thread. I replaced all the "shotgun" caps, and found three that were far out of spec. None were bulging particularly.

Since replaced, the PSU when jumpered on spins the fans briefly. I attempted to force the supply on by soldering a resistor to Pin2 of the M8 optocoupler, but this causes the fans to spin briefly and causes a rapid clicking noise from the supply.

Would anyone happen to have any diagnostic suggestions for fixing this supply? When not jumpered on, I get the expected voltages (25v white, 5v purple).

[adderthorn]

Re: Suspect Bad PS Apple AcBel API 1PC36

I've got this same PSU. Initial symptoms were the same as described elsewhere in this thread. I replaced all the "shotgun" caps, and found three that were far out of spec. None were bulging particularly.

Since replaced, the PSU when jumpered on spins the fans briefly. I attempted to force the supply on by soldering a resistor to Pin2 of the M8 optocoupler, but this causes the fans to spin briefly and causes a rapid clicking noise from the supply.

Would anyone happen to have any diagnostic suggestions for fixing this supply? When not jumpered on, I get the expected voltages (25v white, 5v purple).

If it makes you feel any better, I am getting similar issues. I've replaced all the secondary caps mentioned in the thread (C7,C8,C10,C13,C16,C21,C29,C35,C41,C43,C59,C60) and am getting a few clicks from what seems like maybe one of the transformers? before it shuts down. Standby voltages look fine. I'll be trying the resistor test this week and see if I can measure any other voltages.

[adderthorn]

Re: Suspect Bad PS Apple AcBel API 1PC36

Finally had some time to troubleshoot some more on this.

Bridged the M8 pin with ground and the PSU mostly turned on. The fan was spinning normally and something was clicking at a good clip (sounded like it was coming from one of the transformers maybe?).

I measured all the voltages and they were all just a bit short, 12v was closer to 10v, 5v was closer to 4v and 3.3v was about where it should be, the numbers were fluctuating a lot, so I'm guessing it was cycling on/off with the clicking.

Would this indicate one of the ICs?

[BruceC]

Hi everyone! Does anyone know where a schematic for this AcBel API1PC36 power converter can be found?

I'm currently troubleshooting three of these for personally owned Macintosh Mirror Door Drive systems but have no new significant insights to contribute toward the most common specific electrolytic capacitor component failures at this time. But I'll add my results when they arise.

I replace any relatively low voltage low capacity electrolytic capacitors with tantalum capacitors which of course provide the great advantage of being solid state and not inherently life limited. I view this option as limited to about 100 µF at 16 V or equivalent capacitance times voltage product, and almost never available at high voltages such as above 50 V. But I eagerly parallel them to avoid electrolytics too - I view five 100 µF 16 V tantalum capacitors as far preferable to a single 470 µF 16 V electrolytic capacitor for 12 V power rails for example. For capacities and voltages outside the range of tantalum capacitors I highly prefer hermetically sealed electrolytic capacitors when available, because:

----------------
Bruce: ChatGPT, polymer sealed electrolytic capacitors are life limited components which cause serious failure problems in long life electronics products. Are hermetically sealed, such as with glass or epoxy, electrolytic capacitors still available? If so about how much greater longevity to they generally provide, and very roughly what extra cost do they typically impose compared to similar polymer sealed capacitors?

ChatGPT: Yes, hermetically sealed electrolytic capacitors are still available in the market, particularly for applications where longevity and reliability are critical. These capacitors are sealed using glass or epoxy to create a hermetic seal that protects the internal components from environmental factors such as moisture and humidity. The hermetic seal ensures that the electrolyte inside the capacitor remains stable over time, reducing the risk of electrolyte drying out or leaking, which are common failure modes in traditional electrolytic capacitors.

The longevity of hermetically sealed electrolytic capacitors can be significantly greater compared to polymer-sealed capacitors, depending on factors such as operating conditions, voltage stress, and temperature. In general, hermetically sealed capacitors are designed to have a longer operational life, often ranging from tens of thousands to hundreds of thousands of hours, or even longer in some cases.

As for cost, hermetically sealed electrolytic capacitors tend to be more expensive than polymer-sealed capacitors due to the additional manufacturing processes involved in creating the hermetic seal and the use of high-quality materials. The exact cost difference can vary depending on factors such as capacitance, voltage rating, size, and manufacturer, but in general, you can expect hermetically sealed capacitors to be several times more expensive than polymer-sealed ones.

Despite the higher cost, hermetically sealed electrolytic capacitors are preferred in applications where reliability and longevity are paramount, such as aerospace, defense, medical devices, and industrial equipment, where the consequences of capacitor failure can be significant.
----------------

In my view the considerable labor required to remove this power converter from the Macintosh MDD, plus the additional considerable labor required to replace capacitors well warrants upgrades to tantalum and hermetically sealed electrolytic capacitors. Otherwise we trap ourselves in an eternal rather short cycle time and effort intensive maintenance task. This particular power converter seems reliable except for capacitor failures, as does the Macintosh MDD. So for those who wish to preserve this computer tantalum and hermetically sealed electrolytic capacitor upgrades, where available, seem like a worthy investment.

(The RTC (real time clock) battery is a corrosion time bomb and the only other inherently life limited component. I replace them with super capacitors as described here.)

All just in my personal view of course... Cheers!

[BruceC]

I repaired one of my five MDD power converters, crudely but sufficiently, in the process developing the following approach which replaces many life limited electrolytic capacitors with unlimited life MLC (Multi Layer Ceramic) or tantalum capacitors, or longer life PAE (Polymer Aluminum Electrolytic) capacitors, and minimizes new component ordering complexity by substantially consolidating the range of values.

The consolidation is practical because none of these capacitors appear to be utilized in frequency or timing related circuits but rather only filtering roles, where more capacitance is essentially always better (and where if capacitance degradation occurs the power converter will nonetheless remain functional for a longer time). I've yet to prove success in all consolidation cases, but many were successfully implemented in my first crude repair and I'll test the full array later.

I couldn't find suitable hermetically sealed electrolytic capacitors so the PAE option seems to be the best practical alternative to ordinary aluminum electrolytic capacitors in these times.

I purchased from the vendors linked. Values must be selected from the arrays provided on the linked pages.

This completely ignores all values above 1 mF. I judge those as much less troublesome but of course they could be replaced too, such as from this vendor (the same as the 1 mF PAE vendor below), for a fully thorough capacitor upgrade.

Everything below "Original Capacitors, 28 total:" is compliments "Toasty", FdB@MacOS9Lives.com, and perhaps others, with my very modest refinements, thanks tons all! To wit:


--- Macintosh Mirror Door Drive Computer's API1PC36 REV:A Power Converter Capacitor Upgrades ---

Replacement of 1 mF or lower values, 20 total of 10 values:

[X] Indicates quantity of the particular value in the original power converter.
Higher values seem functional in all tests thus far. Some cases not tested yet but dysfunctions seem unlikely.
Ideally Tantalum or MLC, otherwise PAE.

470 nF, 50 V: [4] Replace with 1 µF MLC.
4.7 µF, 50 V: [2] Replace with 10 µF MLC.
10 µF, 50 V: [4] MLC.
47 µF, 25 V: [3] Replace with 100 µF tantalum.
100 µF, 25 V: [1] Tantalum.
220 µF, 35 V: [1] Replace with 1 mF 35 V PAE.
470 µF, 16 V: [1] Replace with 1 mF 35 V PAE.
470 µF, 35 V: [1] Replace with 1 mF 35 V PAE.
680 µF, 35 V: [1] Replace with 1 mF 35 V PAE.
1 mF, 10 V: [2] Replace with 1 mF 35 V PAE.


Components and generous quantities ordered for upgrade of 5 power converters, consolidated to 4 values:

1 µF, 50 V, MLC: 100
10 µF, 50 V, MLC: 200
100 µF, 25 V, Tantalum: 30
1 mF, 10 V, PAE: 40

No attempt to match lead spacing was made so lead bending is required in some or all cases. The last (1 mF) component is 1 x 1.7 cm and thus at least 2 mm larger in diameter than the original capacitors. I suspect sufficient space is available but have not confirmed this yet.


Original Capacitors, 28 total:


Fan Control:

C401: 47 µF, 25 V
C402: 470 nF, 50 V
C403: 10 µF, 50 V
C405: 10 µF, 50 V
C407: 10 µF, 50 V
C416: 470 µF, 016 V


Generally Primary Adjacent:

C4: 100 µF, 450 V
C5: 100 µF, 450 V
C30: 470 nF, 50 V
C32: 47 µF, 25 V
C37: 100 µF, 450 V
C45: 10 µF, 50 V
C53: 4.7 µF, 50 V
C55: 470 nF, 50 V
C56: 470 nF, 50 V
C57: 4.7 µF, 50 V


Generally Secondary:

C7: 220 µF, 35 V
C8: 47 µF, 25 V
C10: 2.2 mF, 16 V
C13: 2.7 mF, 6.3 V
C16: 2.2 mF, 6.3 V
C21: 3.3 mF, 10 V
C29: 2.2 mF, 6.3 V
C35: 470 µF, 35 V
C41: 680 µF, 35 V
C43: 100 µF, 25 V
C59: 1 mF, 10 V
C60: 1 mF, 10 V

Cheers! End.

[BruceC]

Please ignore my previous post, or better yet moderator please delete it. Here's a more refined version, with previous mistakes corrected, for a more thorough restoration:

I repaired one of my four AcBel API1PC36 MDD power converters, crudely but sufficiently, and in the process developed the following approach which replaces all but three life limited electrolytic capacitors with unlimited life MLC (Multi Layer Ceramic) or tantalum capacitors, or longer life PAE (Polymer Aluminum Electrolytic) capacitors, and minimizes new component ordering complexity by substantially consolidating the range of values.

The consolidation is practical because none of these capacitors appear to be utilized in frequency or timing related circuits but rather only filtering roles, where more capacitance is essentially always better (and where if capacitance degradation occurs the power converter will nonetheless remain functional for a longer time). I've yet to prove success in all consolidation cases, but many were successfully implemented in my first crude repair and I plan to test the full array in April 2024.

I couldn't find suitable hermetically sealed electrolytic capacitors so the PAE option seems to be the best practical alternative to ordinary aluminum electrolytic capacitors in these times.

I purchased from the vendors linked. Values must be selected from the arrays provided on the linked pages.

This ignores the three 100 µF 450 V primary side rectified line storage capacitors. I judge those as much less troublesome but of course they could be replaced too for a complete electrolytic capacitor upgrade.

An illustration of the board's capacitors and everything below "Original Capacitors, 28 total:", in some cases with my very modest refinements, is compliments "Toasty", FdB 'a.t' MacOS9Lives.com, and perhaps others, thanks tons all! To wit:


--- Macintosh Mirror Door Drive Computer's AcBel API1PC36 REV:A Power Converter Capacitor Upgrades ---

Replacement of all but the three 100 µF 450 V capacitors, 25 total of 14 original values:

[X] Indicates quantity of the particular value in the original power converter.
Higher values seem functional in all tests thus far. Some cases not tested yet but dysfunctions seem unlikely.
Indefinite life tantalum or MLC used wherever practical, otherwise PAE.

Original component [Original quantity]: Replacement

470 nF, 50 V [4]: 1 µF, 50 V MLC
4.7 µF, 50 V [2]: 10 µF, 50 V MLC
10 µF, 50 V [4]: 10 µF, 50 V MLC
47 µF, 25 V [3]: 100 µF, 25 V tantalum
100 µF, 25 V [1]: 100 µF, 25 V tantalum
220 µF, 35 V [1]: 1 mF, 35 V PAE
470 µF, 16 V [1]: 1 mF, 35 V PAE
470 µF, 35 V [1]: 1 mF, 35 V PAE
680 µF, 35 V [1]: 1 mF, 35 V PAE
1 mF, 10 V [2]: 1 mF, 35 V PAE
2.2 mF, 6.3 V [2]: 4.7 mF, 6.3 V PAE
2.7 mF, 6.3 V [1]: 4.7 mF, 6.3 V PAE
2.2 mF, 16 V [1]: 3.3 mF, 16 V PAE
3.3 mF, 10 V [1]: 3.3 mF, 16 V PAE


Quantity required for each power converter, consolidated to 6 values, example vendors linked:

1 µF, 50 V, MLC: 4
10 µF, 50 V, MLC: 6
100 µF, 25 V, Tantalum: 4
1 mF, 35 V, PAE: 6
3.3 mF, 16 V PAE: 3
4.7 mF, 6.3 V PAE: 2


No attempt to match lead spacing was made so lead bending is required in some or all cases. And some are larger than the original component. For example the replacement 1 mF capacitor is 1 x 1.7 cm and thus at least 2 mm larger in diameter than the original components. I suspect sufficient space is available but have not confirmed this yet.


Original Capacitors, 28 total:

Fan Control:

C401: 47 µF, 25 V
C402: 470 nF, 50 V
C403: 10 µF, 50 V
C405: 10 µF, 50 V
C407: 10 µF, 50 V
C416: 470 µF, 016 V

Generally Primary Adjacent:

C4: 100 µF, 450 V
C5: 100 µF, 450 V
C30: 470 nF, 50 V
C32: 47 µF, 25 V
C37: 100 µF, 450 V
C45: 10 µF, 50 V
C53: 4.7 µF, 50 V
C55: 470 nF, 50 V
C56: 470 nF, 50 V
C57: 4.7 µF, 50 V

Generally Secondary:

C7: 220 µF, 35 V
C8: 47 µF, 25 V
C10: 2.2 mF, 16 V
C13: 2.7 mF, 6.3 V
C16: 2.2 mF, 6.3 V
C21: 3.3 mF, 10 V
C29: 2.2 mF, 6.3 V
C35: 470 µF, 35 V
C41: 680 µF, 35 V
C43: 100 µF, 25 V
C59: 1 mF, 10 V
C60: 1 mF, 10 V

Cheers! End.

[PeteS in CA]

Replacing electrolytics with high value MLCCs is probably not a great idea. They have very different impedance vs. frequency characteristics, and high value MLCCs usually (almost always?) have tempcos such as Z5U or Y5V, in which the capacitance at operating temperatures could be 20%-40% or the nominal value.

[Sorc]

Time to add my voice to the chorus here. (textually speaking that is). I too have one of these PSU that have failed. Knew what it was before my friend gave it to me. and sure enough 2 of the shotgun 12 electro's were swollen so replaced them all with good quality equivalents from RS and mouser.
Symptom: PSU will not start.. pushing power button at attempts then immediate shutdown.

Replaced ALL electolytics on main board. PWM daughter board, and the Fan controller daughter board.
still doing the attempt to start immediate shutdown.

So pulled the primary mosfets and tested them out of circuit. one of the 20n6055's was measuring short. (0.5v D-S one way and 0.14v D-S the other way) so replaced both as a pair with modern equivalents.

PSU now starts and runs Unloaded. but as soon as you plug any load on either the 5v or 12v lines it shuts down (OCP/OV/UV protection possibly) or possibly issues with the regulation in the 2ndary side silicon. So i replaced all schotky, and regulators on the secondary heatsink as well as the single mosfet.
did not solve the issue.

last ditch effort, replace the other two primary Mosfets (K2011's), with modern equivalents (this slightly improved things. The PSU will now run for a second or two with a 5w load on the 12v rail (break light from a car).

but it still will refuse to cold start with any load on either the 12v or 5 v rail.
No load it will happily run forever.

when the PSU trips with a load attached both the 12v, 5v and 3.3 v rails collapse. (suggesting something on the primary side, or the WT7515 is telling the PSU to shut down.

measured voiltages running no load.
3.3v rail = 3.31 VDC
5v rail = 5.17v DC
12v rail = 11.87v DC (i am thinking this may be a little low)

12v Reference to VCC on WT7515 is 12 v.

Unfortunately my test equipment is not good enough to measure milli ohms. so testing the current sense resistors. statically is no go (they both measure 1 ohm as that is as low as my DMM will go. ) could use ohms law and measure voltage drop across them and determine the value that way but this is hampered by their position (hard to get to when PSU assembled) I refuse to run the PSU out of is cage. I am overly cautious about the Live primary Heatsink (connected directly to the Bulk capacitor negative so this is sitting at -400v DC) I have a Digital pocket Oscilloscope but this is not sufficient to capture the trip can only measure as high as 20miliseconds scan rate so it never triggers when the PSU shuts down.

So not sure where to go from here. Without a circuit diagram. envisioning the circuit diagram for the Main PSU board is not too difficult but the PSU protection is split between the 2 daughter boards (Supervisor on fan control board (WT7515 and TL 431), and the PWM controller on the Primary side Daughter board (ML4800) ) both the daughter boards have SMD and are double sided in the least (possibly multi layered)) so envisioning a circuit for these 2 boards is significantly more difficult. Also some of the protection is on the main board as well (LM339 quad comparitor and Opto's)

So it seems i am looking at some OV/UV or OC protection issue being triggered. Possibly if this PSU is cross regulated loading 1 rail cause the other rail to trip OV as the PWM ramps up to deal with the current draw ?? Have not tested this yet was my next test.

Any tips or suggestions welcome at this point

Traced the VCC of the ML4800 to the main board it goes across a wire link to a 100ohm resister then to the base of a PN2222a and PN2907a transistor pair. Possibly one of these transistors may be faulty under load especially with a suspect Mosfet short may have damaged them?? but if that is the case it likely would not start at all.