9906 Views 10 Replies Latest reply: Jul 24, 2010 4:40 PM by iGary
Before you begin, open the computer, lay it on its side with the access side facing up, and remove
• Hard drives and hard drive carriers in drive bays 3 and 4
• Optical drive carrier and optical drives
• Any PCI Express cards that block access to the power supply mounting screws
1. Using a short-handled, magnetized 2.5 mm hex screwdriver, remove the four power supply
mounting screws from the bottom of the media shelf.
2. Depress the upper right corner of the power supply and slide the power supply toward the
front of the computer.
3. Lift the power supply a short distance and rest it on the edge of the enclosure.
4. Starting with the top connector and working down, disconnect the four connectors on the
power supply cable from the four connectors on the power harness cable. Note: You must release the locking latch on each power supply cable connector before detaching the connector.
5. Remove the power supply from the enclosure.
Replacing the Power Supply:
1. Rest the power supply on the edge of the enclosure so that its cable can reach the media
shelf divider. Note: The following four steps explain how to reconnect the power supply cable connectors with the power harness connectors. When the power harness connectors are in their original position in the media shelf divider, it is not possible to see where they mate with the power
supply cable connectors. Therefore, the next step shows how to temporarily remove the power harness connectors from the divider for a better line of sight.
2. Starting with the top connector and working down, release the four power harness cable
connectors from their openings in the media shelf divider. Note: Each connector has two locking latches that hold it to the divider. To release the connector, depress the bottom latch first and then the top latch. You may find a flat-blade screwdriver useful in depressing the latches.
3. Placing a hand on either side of the media shelf divider, thread the J1 connector on the power supply cable through the top connector opening in the media shelf divider and connect it to the PS1 connector on the power harness. Repeat for the other three power supply cable connectors.
4. Important: The first three connectors on the power supply cable are marked J1, J2, and J3; the fourth connector is not marked. The four connectors on the power harness cable are marked PS1, PS2, PS3, and PS4. You must connect J1 to PS1, J2 to PS2, etc.
5. Starting at the bottom connector and working up, depress the locking latches on each power harness connector and reinsert it into its opening in the media shelf divider. Important: Test that all power supply cable connectors and power harness connectors are fully seated by tugging on the cables on both sides of each of the four paired connectors.
6. Fold the power supply cable on the bottom of the power supply bay.
7.Lift the power supply and holding the power supply at an angle, llower it into the enclosure until its
upper right corner slides under the corner lip of the enclosure.
8. Slide the power supply toward the back of the computer as far as possible.
9. Make sure the power receptacle aligns with the opening in the enclosure’s back panel and
the four screw holes in the power supply align with the screw holes in the media shelf.
10. Replace the four power supply mounting screws.
The reason for the power supply removal was because I have the spontaneous reboot problem I've read about in other threads.
I bought 3 Mac Pro towers at the same time. One was to use as a spare and as a test environment.
So I exchanged power supplies. The spontaneous reboot Mac remains a spontaneous reboot Mac.
I'm going to try to exchange RAM boards. And a third video card. But I'm thinking the worst.
2008's had to have EFI/SMC firmware updates, the early models would freeze on wake from sleep. The fix should be applied and solve, though obviously better with later models.
There was also a bug with ATI 2600s that got a firmware fix that was much less acceptable fan speed than the next revision board.
The new SMC Reset method is to unplug everything, no power, and hit power on for 10 seconds. Old method was unplug for couple minutes... and do nothing (didn't work 80% of the time, and unplug over night would),
And that is one draw back of buying three at same time - or upside, that you have identical parts and systems so you can swap parts and narrow down.
I'd already tried every SMC reset method I could find!
Turns out to be bad RAM (or perhaps a bad riser): When I exchanged video cards and RAM, the Mac Pro that was running well suddenly became the spontaneous reboot Mac. And much more so than the other one had been with this same equipment. And after a few hours of use the original one that was spontaneous-reboot-happy has yet to encounter any error.
I exchanged the video cards back to the way they were, and the problem didn't follow the video card.
Now I'm running tests on these 8 chips trying to narrow down the turkey.
A few months ago I had tested this same RAM with both AHT and memtest. Results with both were for good RAM.
My experiences with RAM is I used to never ever see RAM go bad after months or years of use. When I found bad RAM it was either in a new computer or the RAM itself was just installed. A couple of years ago I started finding RAM that had worked well at first start to fail after some use.
In fact for all my own computers, and for key client computers like servers, I always run an overnight AHT right out of the box and before deployment. I had also done that with this RAM when it was brand new. So this is really the third go-around with tests on these boards.
What an odd change in behavior.
Message was edited by: iGary
In the last two years I guess I have seen more of this marginal RAM where the most common symptom would be reporting half the capacity of a DIMM, sometimes a pair.
But yes RAM issues can have profound side effects. And the older Fully-Buffered had to have those heat sinks because they can run too hot without. Apple even advises to wait 15 minutes before opening the side door, and that just the sudden change in temperature can be a shock, and harmful.
Heat can stress and age RAM and we have never had or seen RAM this hot. I replaced my Apple RAM when it was about 2 yrs old.
Even PCs have their own problems and DDR3 can be fussy, or so it seems. Mostly, I try to stay with Crucial as much as possible, but even there they have had their own issues, sometimes a change in batch or construction design.
A chip plant having to shut down or hit by storm can take weeks or longer to get back online and back to reliable batches, too. Which happened in fall of 2000 I think.
Mixing different batches, even from the same vendor or factory can even have problems and not get along. You could have two or three sets that are fine until used together. Which is why the ideal is to buy and replace in full sets, or take a chance.
But if Riser, Apple has been known to advance exchange if taking it in isn't easy. And easier to take or send in a Riser than the MP!
Riser for 1st gen
The US Military did some great work on predicting Mean Time Between Failures MTBF for a whole lot of different kinds of equipment. It is not Classified. For Electronics, the manual is Mil Handbook 217c.
There are a number of complex factors involved in why electronic stuff dies. But if you set aside the shock of launching things into space and all the other extraneous factors, it boils down to two predominant factors:
Number of "gates" (more memory cells means more gates) and Junction \[on-chip] temperature.
Our memories are getting MUCH denser than only a few years ago, and they are running hotter. We WILL see more failures.