Resolved 2009 8-Core 4,1 - 5,1 suddenly stopped booting

[Bob-K]

About 9 months ago I flashed my 2009 8-Core 2.66 GHz Mac Pro to 5,1 firmware and upgraded the processors to x5680s which I bought on eBay and de-lidded.

I've been running Sierra 10.12.6. Things were fine until a few weeks ago, when the following symptoms started happening erratically:

1. Bluetooth disconnecting and reconnecting (through a new GMYLE Bluetooth 4.0 USB adapter; internal Bluetooth is disabled).

2. Computer would seemingly-randomly put itself to sleep.

3. IIRC, the wired USB keyboard sometimes would stop working. It's connected to a front-panel USB port with an active extension cable. I don't recall for sure, but sometimes un-plugging and re-plugging helped.

Then, about a week ago, the computer shut itself off. It would power up, the white LED was on (not flashing), and fans were on, but there no startup chime, and it wouldn't boot.

I ran through some troubleshooting steps in the Apple service manual, including checking the PSU by pressing a button on the backplane board. The expected LED lit up.

Then I replaced the 5680s with the original processors, and it booted up without issue.

Today I did further tests and found:

1. The computer will boot with either of the 5680s in slot A and nothing in slot B.

2. It will NOT boot with both 5680s - with one exception, described below. I tried both combinations: CPU 1 in slot A, CPU 2 in slot B and vice versa.

I doubt it's matter of insufficient thermal paste or bolt tightening. I tried tightening the screws tighter than the service manual directs, and it didn't help. (Apple says "finger tight plus 1/4 turn".)

The exception is interesting: the first time I tried both 5680s, I forgot to put on the plastic retaining frame on the processor in slot B and it booted without issue.

I shut down, put the retaining frame back, and now it wouldn't boot. I tightened the bolts what I considered a good bit tighter than Apple's recommendation, but that didn't help. Then I tried loosening them a bit, and that didn't help either.

I didn't try removing the frame additional times to see if this is repeatable, but if it boots OK without the frame, does anybody what the risks might be of leaving the frame off?

Any other ideas? I'm wondering if maybe the power supply is starting to go, but I don't know if they gradually fail, with intermittent issues, or if it's a binary thing.

Any suggestions much appreciated.

Thanks!

Bob

 

[pdafan]

About 9 months ago I flashed my 2009 8-Core 2.66 GHz Mac Pro to 5,1 firmware and upgraded the processors to x5680s which I bought on eBay and de-lidded.


I doubt it's matter of insufficient thermal paste or bolt tightening. I tried tightening the screws tighter than the service manual directs, and it didn't help. (Apple says "finger tight plus 1/4 turn".)

The exception is interesting: the first time I tried both 5680s, I forgot to put on the plastic retaining frame on the processor in slot B and it booted without issue.

I shut down, put the retaining frame back, and now it wouldn't boot. I tightened the bolts what I considered a good bit tighter than Apple's recommendation, but that didn't help. Then I tried loosening them a bit, and that didn't help either.

I didn't try removing the frame additional times to see if this is repeatable, but if it boots OK without the frame, does anybody what the risks might be of leaving the frame off?

Bob

You mean you used to BLACK retaining CPU frame and it wont boot?
Of all the Mac Pros I upgraded to X5680 (lidded), I never use the BLACK plastic frame and they work.
Just sharing.

 

[Bubalight]

The black plastic frame distributes the force from the heat sink evenly across the entire delidded CPU so that all the pins make good contact, and so that your CPU doesn't flex which can cause it to crack internally. (which sounds like what happened)
It can work without the black clip but it isn't safe.
The screws should be tightened all the way. I have worked on several of these macs and it isn't always the same amount of effort to tighten but when you have it all the way screwed in it stops. Of course don't use a ratchet or put all your strength into it, you would break the bolt off.

 

[Bob-K]

You mean you used to BLACK retaining CPU frame and it wont boot?

Yes.

When I originally installed the (de-lidded) x5680s, I put the black plastic frames back in, because they were used for the original, de-lidded processors.

Today, I put the frame in for CPU A but accidentally left it out for CPU B, and the computer booted fine (recognized both processors and all RAM). When I put the frame back in, it powered up but wouldn't boot.

Of all the Mac Pros I upgraded to X5680 (lidded), I never use the BLACK plastic frame and they work.

In these cases, you replaced de-lidded processors with lidded ones - correct? In that case, it would be impossible to use the frames.
[doublepost=1533256194][/doublepost]

so that your CPU doesn't flex which can cause it to crack internally. (which sounds like what happened)

There's no evidence of cracking - at least not yet. I only had the CPU in without the frame for a minute or two, and at that point I'd tightened the screws what I consider a "normal" amount, which might be a little on the light end.

Also, when the non-booting issue arose, both frames had been in use since the installation about 9 months ago.

The screws should be tightened all the way. I have worked on several of these macs and it isn't always the same amount of effort to tighten but when you have it all the way screwed in it stops. Of course don't use a ratchet or put all your strength into it, you would break the bolt off.

I don't think I've ever tightened the screws until they stop, so I might try tightening more.

I thought I read somewhere that if they were too tight, the computer wouldn't boot - is that correct?

(I know that if they're not tight enough, there are problems, like not all the RAM is recognized.)

Thanks, Bubalight!

 

[Bob-K]

Today I tried a few more things, none of which resolved the issue.

1. Removed the power supply and opened its case. It didn't seem excessively dusty, but I cleaned out what I could with a vacuum cleaner (with small attachment intended for this sort of thing) and a rocket air blaster. Didn't see any obvious signs of fried components or bad solder joints.

2. Reinstalled the x5680s and made sure the bolts were good and tight.

3. Removed the black plastic frame from CPU B to see if it would boot, like it did yesterday. It didn't.

However, inspecting the CPUs, I remembered that when de-lidding last year, I accidentally sliced off the top of one of the small rectangular protrusions, which I've recently read are tiny capacitors.

Is that enough to cause the problem, and if so why did it work fine for 9 months? And why does it still work if installed alone in slot A?

 

[alphaod]

Sounds like a problem with a socket than the CPU. When you delidded, did you remove all the solder?

 

[Bob-K]

Sounds like a problem with a socket than the CPU.

The original processors boot just fine, so I doubt the socket is the issue.

When you delidded, did you remove all the solder?

It sure looks that way. Otherwise, I would have expected the problem to arise earlier. It's been working fine for 9 months.

 

[Bob-K]

RESOLVED:

One of my x5680s was bad.

The surprising thing is that it's NOT the one with the partially-shaved-off capacitor.

I bought another 5680 on eBay and de-lidded via the clothes iron method. Installed it in slot A (with slot B empty) and it booted without issue.

When I added the "good" old 5680 (without the shaved capacitor) in slot B, the original problem returned - it powered up but wouldn't boot.

I put the old 5680 with the shaved capacitor into slot B, and it booted.

All RAM is recognized, everything seems fine. Fingers crossed.

 

[dangerweenie]

About 9 months ago I flashed my 2009 8-Core 2.66 GHz Mac Pro to 5,1 firmware and upgraded the processors to x5680s which I bought on eBay and de-lidded.

I've been running Sierra 10.12.6. Things were fine until a few weeks ago, when the following symptoms started happening erratically:

1. Bluetooth disconnecting and reconnecting (through a new GMYLE Bluetooth 4.0 USB adapter; internal Bluetooth is disabled).

2. Computer would seemingly-randomly put itself to sleep.

3. IIRC, the wired USB keyboard sometimes would stop working. It's connected to a front-panel USB port with an active extension cable. I don't recall for sure, but sometimes un-plugging and re-plugging helped.

Then, about a week ago, the computer shut itself off. It would power up, the white LED was on (not flashing), and fans were on, but there no startup chime, and it wouldn't boot.

I ran through some troubleshooting steps in the Apple service manual, including checking the PSU by pressing a button on the backplane board. The expected LED lit up.

Then I replaced the 5680s with the original processors, and it booted up without issue.

Today I did further tests and found:

1. The computer will boot with either of the 5680s in slot A and nothing in slot B.

2. It will NOT boot with both 5680s - with one exception, described below. I tried both combinations: CPU 1 in slot A, CPU 2 in slot B and vice versa.

I doubt it's matter of insufficient thermal paste or bolt tightening. I tried tightening the screws tighter than the service manual directs, and it didn't help. (Apple says "finger tight plus 1/4 turn".)

The exception is interesting: the first time I tried both 5680s, I forgot to put on the plastic retaining frame on the processor in slot B and it booted without issue.

I shut down, put the retaining frame back, and now it wouldn't boot. I tightened the bolts what I considered a good bit tighter than Apple's recommendation, but that didn't help. Then I tried loosening them a bit, and that didn't help either.

I didn't try removing the frame additional times to see if this is repeatable, but if it boots OK without the frame, does anybody what the risks might be of leaving the frame off?

Any other ideas? I'm wondering if maybe the power supply is starting to go, but I don't know if they gradually fail, with intermittent issues, or if it's a binary thing.

Any suggestions much appreciated.

Thanks!

Bob

I am having a very very similar issue in my 2009 / 4,1 Mac Pro - except it is with the stock 2.26 ghz quad-core processors, not anything upgraded.

Currently I can only boot with something in socket A - if I put procs in socket A and socket B, no chime. Either processor just in socket A, and I get a chime.

No CPU error lights when theres something in socket B, just a solid power light in the front panel, and no boot.

Leaving CPU B out, the CPU-B error light next to the socket is illuminated (red LED).

Will keep you posted if I can figure anything out - the DIAG LED and technicians manual doesnt seem to help this particular issue (have examined it thoroughly).

good luck!

 

[Bob-K]

good luck!

Thanks, but did you read my last post? The problem was one of the CPUs.