Reading Time: 4 minutes

This is a continuation of “Recapping an Xserve G5 power supply – Part 1 – Preparations” and “Recapping an Xserve G5 power supply – Part 2 – Recap and test” so read that first if you haven’t already

I didn’t want to squeeze this into the previous articles and clutter them up, so here are some additional notes.

First off, here is the pinout for the January 2005 G5 Xserve and G5 Xserve Cluster Node PSU:

This is looking at the connector when the PSU is extracted from the machine and on a desk facing up, like this:

When in doubt, just look closely at the board and you’ll see pins 1, 8 and 14 are numbered.

If you’re not comfortable testing your recapped PSU inside the machine, you can use a voltmeter as well.
Connect the power supply to power, then connect the black lead to pin 8 (Ground) and the red lead to pin 1 (+12V DC).


If you get 12V or a little higher, your power supply is likely OK.

If you get less than 12V, disconnect the power cord, wait a minute and then check if the pins are clear of dust and debris. Reconnect the power cord and test again. If you still get less than 12V, your power supply is not OK.

++++++++

About the revision 01 power supplies

The revision 01 power supplies have an aluminum electrolytic capacitor, labeled C903, that the revision 02 doesn’t have. The revision 02 instead has a ceramic capacitor (or maybe a tantalum electrolytic capacitor but visually I’m pretty darn sure it’s a ceramic capacitor) (which I did not replace in my recap). Apple maybe determined that this capacitor was a common point of failure so if you have a revision 01 power supply, it may be worth replacing that one first. Now in my revision 02 power supply C903 (ceramic) did not fail but one or multiple of the other electrolytic caps did. So while C903 may have been a failure prone cap in revision 01 PSU’s, it’s likely other caps are bad as well.

As far as I can tell, the different capacitor types in C903 is the only difference between revision 01 and 02. All other capacitors are the same and have the same values. This makes the list for the rev 01 PSU’s as follows (list changed to a table format and updated last in December 2022):

Component ID	Capacitance	Voltage	Lead Spacing	Diameter	Length/Height	Temp Rating	Hour Rating	Ripple Current	Link
C10	47 µF	50V	2.5mm	6.3mm	11mm	105C	1000	115mA	Mouser
C103 / C104	2200 µF	16V	5mm	12.5mm	25mm	105C	10,000	2.23A	Mouser (get 2)
C111 / C112	1000 µF	16V	5mm	10mm	20mm	105C	7000	840mA	Mouser (get 2)
C350	1 µF	50V (or 100V)	2.5mm	6.3mm	11mm	105C	1000	24mA	Mouser
C900 / C353	100 µF	25V	3.5mm	8mm	7mm	105C	5000	112mA	Mouser (get 2)
C901 / C907	100 µF	35V	2.5mm	6.3mm	11mm	105C	6000	540mA	Mouser (get 2)
C903	47 µF	16V	2mm	5mm	11mm	105C	2000	85mA	Mouser
C905	10 µF	25V	2mm	5mm	11mm	105C	2000	53mA	Mouser
C909	22 µF	25V	2mm	5mm	11mm	105C	4000	144mA	Mouser

Note, this list still links to some of the caps that are not quite the right size. They work but may not fit snug. Both this list and the list in Part 1 will be updated at some point in the future. The C903 I linked to is a radial aluminum electrolytic capacitor, just like what’s already in the rev 01. When I find a good ceramic capacitor to replace it with, I will update the link. First I have to make sure it is identical in values etc.

Adding this extra capacitor to the graphics (C903):

I’ll add notes worth mentioning to this article as I work on the other PSU’s.

After recapping, you may notice a high-pitched whine sound coming from the PSU when it is connected to power (but not on). This has been a very common find and while the cause is unknown, it does not affect the functionality of the PSU.

April 2020:
I found a Rev 2 PSU with 13 capacitors. This is the first out of dozens I’ve opened up so far. While rare, it does happen.

You have everything you need to get your Xserve G5 PSU’s recapped but if you’d still prefer someone else to do it, I can help out. For $89 + shipping I will do the recapping and testing for you. This includes the purchasing of the new capacitors.