Apple MDD G4 PSU (AcBel API1PC36) - Recap question

[Agent24]

Got around to pulling the PSU from my old MDD G4 Mac

It's an AcBel API1PC36 Rev. B which is nice because it doesn't have the CapXon HP problem (or the heatsink in the way, luckily!)


It seems mine has gone similarly down the same path as the one severach posted about here: http://www.badcaps.net/forum/showthread.php?t=13985

That is, that C60 is bulged but nothing else (yet). And the Mac still starts most of the time. Not sure if the brands\values match his Rev. A model but I guess we will find out.



In any case, looking in that and other threads, it seems that the smaller capacitors might give issues too, so I'm thinking it sounds a good idea to just replace most everything, including them?

Also, what about the ones on the little daughter-boards?

[Toasty]

Yup. Daughterboards are okay though. Just be sure to get that smaller one right at the end of the output heatsink near the opto's. C64....? Lemme grab the sheet.

Nope, C43. 100µF / 25v

Stay with PW or around that class of caps for the outputs. FM's and such are a waste and really don't lend themselves to these PSU's.

Toast

[Agent24]

Originally Posted by Toasty

Stay with PW or around that class of caps for the outputs. FM's and such are a waste and really don't lend themselves to these PSU's

Ahh yes, I think I saw that you wrote that in another thread - FM has too low ESR for this, and could increase ripple?

(I usually try to match ESR as close to the originals though, anyway)

[PeteS in CA]

FM has too low ESR for this, and could increase ripple?

Toasty means that you don't need super low impedance caps for such P/Ss. In a P/S output, the ripple current the cap sees is determined by the AC line voltage and the load voltage, not the cap impedance. For a given line-load condition, the ripple voltage (E = I * R) and power dissipated by the cap (P = I * I * R) will decrease with a lower impedance cap. If a PW or LXZ cap has the ripple voltage low enough, little is gained by using a lower impedance cap - i.e. the improvement in ripple voltage is not useful, and the reduction in dissipated power isn't very significant. OTOH, the regulator circuit has components that compensate the regulator for the effects of the cap ESR. Significantly decreasing the ESR can affect the regulator loop stability.

[Toasty]

No, what I meant was it is useless to expend money on "super" caps when less expensive, but still good grade caps will do just fine. OTOH, I've bought FMs for less or at lower grade caps prices...?

There was a problem encountered by (IIRC) JonnyGuru when replacing caps in a PSU that actually -increased- ripple.

I've also found that not enough ESR causes other problems in some PSUs that will look great under a standard load test, but fail or shutdown when under "real world" conditions. Changing the caps to a lesser grade than FM corrected the problem. Yes, their ripple handling was a bit less, but I can guarantee it was superior to the OEM junk.

~Sometimes~ the engineers surprise me and "get it right".

Toast

[Scenic]

Originally Posted by Toasty

I've also found that not enough ESR causes other problems in some PSUs that will look great under a standard load test, but fail or shutdown when under "real world" conditions.

I've had something like this happen in a LCD TV PSU. Worked fine with relatively low loads after recapping, but back in the TV, the whole thing squealed and hissed like crazy and ran super hot. Replaced the caps again with higher ESR ones and everything's back to normal (no squealing or hissing, running cooler). Those caps were apparently "too good" (in terms of grading them) for this circuit..
I think it was Panasonic FM (hissing) vs. FR (working fine)

[Agent24]

Originally Posted by PeteS in CA

If a PW or LXZ cap has the ripple voltage low enough, little is gained by using a lower impedance cap - i.e. the improvement in ripple voltage is not useful, and the reduction in dissipated power isn't very significant.

Don't you mean "If a PW or LXZ cap has the ESR low enough" ?

And it's ripple current, not voltage, right?

[Toasty]

He's saying that if using a PW cap provides a low enough ripple voltage, then nothing is gained by using a lower ESR cap.

Ripple current is what the cap can handle and not overheat or be damaged.

Ripple voltage is what we measure and see at the output and as the end result of the filtering. It's what's seen on a scope as an AC voltage that effectively "rides on top" of the DC voltage produced. Typically measured in millivolts.

e.g., - A PW costing 20¢ is used and the measured ripple voltage is 15mV. An FM costing 45¢ is used and the ripple voltage is 14mV. There is really no improvement in ripple at a cost of 225% more. All that for a decrease of <7% that the motherboard is going to (easily) filter even further anyway.

As JG, Scenic, I, and several others have found is that the "better" cap is not necessarily the proper way to go.

Toast

[Agent24]

Sorry about that, now I got it!

When PeteS said Ripple Voltage I thought he was talking about a property of the capacitor itself. Something mustn't be working properly in my head .. good grief!

[Agent24]

Started on the removal today, removed all the big capacitors and tested them so far.
All had low ESR except the bulged C60.

The brands lineup is a bit different to the Rev A. version in the other thread.

C8 = 	47uF	25V LTEC 	TK SERIES 	4101D			1.00 ESR
C7 = 	220uF	35V LTEC 	LZG SERIES 	3O01D			0.08 ESR
C10 =	2200uF	16V TEAPO	SC SERIES 	12/03 A3		0.02 ESR
C13 =	2700uF	6.3V RUBYCON	YXG SERIES 	S0326 			0.02 ESR
C16 =	2200uF	6.3V TEAPO	SC SERIES 	10/03 A3		0.02 ESR
C29 =	2200uF	6.3V TEAPO	SC SERIES 	10/03 A3  (RED MARK)	0.02 ESR
C21 =	3300uF	10V TEAPO	SC SERIES 	10/03 A3		0.01 ESR
C60 =	1000uF	10V TEAPO	SC SERIES 	12/03 A3		2.70 ESR
C59 =	1000uF	10V RUBYCON	YXG SERIES 	T0341			0.04 ESR
C35 =	470uF	35V UCC 	KY SERIES 	3(7) NC			0.03 ESR
C41 =	680uF	35V LTEC	LZG SERIES	3D03D			0.04 ESR
C43 =	100uF	25V TEAPO	SEK SERIES	12/03 A3		0.51 ESR

I'm guessing the Rubycons and the UCC should be good to keep.

Teapo is probably useless but what about the Ltec?

Haven't looked at the smaller capacitors yet but after taking another look at severach's post I notice he did do some replacements on the daughterboards so perhaps it would be a good idea in fact to do that too?


Also, the smaller ones seem to be mostly Ltec TK. There are no ESR values in the datasheet. I assume I can use any General Purpose series from a decent brand?

[Toasty]

Any update on this?

Originally Posted by Scenic

I've had something like this happen in a LCD TV PSU. Worked fine with relatively low loads after recapping, but back in the TV, the whole thing squealed and hissed like crazy and ran super hot. Replaced the caps again with higher ESR ones and everything's back to normal (no squealing or hissing, running cooler). Those caps were apparently "too good" (in terms of grading them) for this circuit..
I think it was Panasonic FM (hissing) vs. FR (working fine)

Odd, because FR is aimed at replacing the FM series...

In this case it's best to follow Per Hanson's sig: "The one who says it cannot be done should never interrupt the one who is doing it."

[PeteS in CA]

I've also found that not enough ESR causes other problems in some PSUs that will look great under a standard load test, but fail or shutdown when under "real world" conditions. Changing the caps to a lesser grade than FM corrected the problem. Yes, their ripple handling was a bit less, but I can guarantee it was superior to the OEM junk.

~Sometimes~ the engineers surprise me and "get it right".

There are compensation components in the feedback loop that compensate for the RC of the capacitance and the ESR of a capacitor. change the ESR too much and stability can be affected. Also, response to load steps can be affected, possibly causing ringing, which could look like noise.

The occasion when I was doing some work in Delta's Engineering lab let me see that they were well equipped, and though I didn't see it, I'm sure they had the equipment necessary to measure and optimize loop response. I doubt El-Cheapo Inc would have the equipment, and possibly not the expertise. El-Cheapo Inc.'s engineers probably use/adapt app note and cookbook circuits and/or recycle previous designs.

[Agent24]

Originally Posted by Toasty

Any update on this?

Not yet.. I'll try to get some more done on it soon - I guess you're keen to see what happens!

[Toasty]

They're rather resilient supplies, sans the crap caps. Haven't seen any silicon failures, even with bulged & leaking caps. Primary filtering is solid with 3 UCC KMG caps.

Even the Samsung of the same genre is well made. It's primary failure(s) is(are) due to a bad mains cap (Capxon HP series defect), and 4 small startup caps. The secondary filtering is fine and I've not encountered a bad cap out there yet (Capxon GL series).

Toast

[Agent24]

In my AcBel the primaries are a trio of Rubycon AXW.

Pulled the rest of the capacitors this afternoon.
The following list is a continuation of the first, the capacitors on the main board:

C57 = 	4.7uf	50V LTEC	TK SERIES 3D01D			2.10 ESR
C56 = 	0.47uf	50V LTEC	TK SERIES 3D01D			7.00 ESR
C55 = 	0.47uf	50V LTEC	TK SERIES 3D01D			8.10 ESR
C53 = 	4.7uf	50V LTEC	TK SERIES 3D01D			2.10 ESR
C30 = 	0.47uf	50V LTEC	TK SERIES 3D01D			6.90 ESR
C32 = 	47uF	25V LTEC 	TK SERIES 4101D  	 	1.60 ESR
C45 = 	10uF	50V TEAPO 	SEK SERIES A3 12/03		3.20 ESR

Next list is capacitors from the daughterboards:

PCB4  (FAN CONTROLLER?)

C403 = 	10uf	50v LTEC	TK SERIES 3D15D			1.90 ESR
C407 = 	10uf	50v LTEC	TK SERIES 3N15D			1.20 ESR
C402 =  0.47uf	50v LTEC	TK SERIES 3915D			6.40 ESR
C416 =	470uf	16v TEAPO	SC SERIES 11/03 A3		0.04 ESR
C405 = 	10uf	50v LTEC	TK SERIES 3D15D   		2.20 ESR
C401 =	47uf	50v LTEC	TK SERIES 3D15D			1.10 ESR


PCB2 (PWM CONTROLLER?)

C20 = 	3.3uF 	50V UCC?  ?? SERIES	15.0 ESR

Not sure of the brand of capacitor C20 on PCB2. Logo looks similar to UCC but doesn't appear to have the points on the top, bottom and top corners. There is no apparent series name marked.

I'll post a photo if anyone wants one (or might post one anyway!)

Now for the fun task of finding replacements... looks like Farnell will be cheaper, but I'll probably have to wait until I get some other stuff lined up to fill the order to the free shipping value.

[Toasty]

Nothing to do on the fan controller or PWM boards.

Just the 47/25 & 10/50. C32 & C45.

[Agent24]

Got an order sent away today. Most of the capacitors are coming from the UK warehouse so I expect it in about a week

[Agent24]

Did the replacement a couple of days ago. All seems good now - haven't had boot failure with a blank screen and roaring fan since.

Some photos attached

Will try to get around to listing what I used for replacements if anyone wants to know... (Mostly Nichicon, and a couple Rubycon and Panasonic) Used what I had left over from other things and then ordered what I didn't have. Left the good brand capacitors in.

[Scruffy]

Alright... here goes, my apologies for bringing this thread back to life, but I believe I've got a capacitor problem in my Rev. B AcBel API1PC36 PSU from my G4 MDD.

I've spent the past few days reading over the relevant threads, but this one obviously deals with the Rev. B.

So after reading over this thread and inspecting my unit, there do not seem to be any visibly damaged caps, but the symptoms are all too familiar; power light comes on while button is pressed and fans twitch. Both +5V and +25V standby voltages are present.

I've already noticed a few minor differences in my unit from Agent24's unit, such as C45, where mine has an LTEC.

So my question to Agent24 would be this; do you happen to have the order list you made from Farnell for this project?

Thanks in advance.

[Agent24]

Not sure that you do have the same symptoms as me... for me the fans did not twitch, they all started, and went top speed while the machine had no video output, and no start-up chime. But your problem could still be caused by bad capacitors...

I can give you what brands\series I used. I chose what was probably cheapest at the time that fit the specifications for ESR and RCR of original capacitors.

Note I kept 3 caps as they were already decent brands:

C13 = Rubycon YXG
C59 = Rubycon YXG
C35 = United Chemi-Con KY

The rest I replaced (from the main board) as follows:

C8, C32 = Rubycon YXF
C7 = Nichicon HE
C10 = Nichicon PW
C16, C29 = Nichicon HE
C21 = Nichicon HE
C60 = Panasonic FC
C41 = Nichicon HE
C43 = Nichicon PS
C57, C53 = Rubycon YXF
C56, C55, C30 = Nichicon VZ
C45 = Nichicon PW

Hope this helps