Mac Pro 2009 Yosemite: freezes, restarts, kernel panic, no boot...wait a day or two, and it's fine?

[WhizGeezer]

Photos 1, 2 & 3 make me think a spill of some sort but they're to "geometric." There may have been some sort of adhesive insulating pad there at some point. I'd suggest cleaning with a toothbrush and 99% isopropyl alcohol.

The marked capacitors uncertain... But the tolerance on cheap capacitors is +/- 20% a manufacturer may then test the capacitors and then mark and bin them to a tighter tolerance. I see no evidence of leakage.

The oily residue around the connector is a very good question. Potentially someone used some variety of contact cleaner in the past? Are the connectors themselves clean? Toothbrush & alcohol here as well.

The situation with the processor tray is more or less normal on older systems. Over time the silicone based oil in the thermal paste separates out. I would recommend removing the heat sink, cleaning off the old thermal paste and replacing it with new paste. This is a distinct possibility as root cause.

Those resistors are effectively nothing more than jumpers. The "blistering" that you see is known as a cold solder joint. This again is a possibility but they really don't seem that bad. You would have to give all 16 connections a near microscopic examination to tell for sure that you have a bad connection. Resoldering would be the fix, but given the scale of things, I wouldn't try that unless I were a pro.

 

[BradHatter]

You might also want to try another OS version (like Mavericks) in case there's some type of software bug. Swapping out components for testing if available is OK but I wouldn't go around buying and swapping out parts in the hope you'll get luck.

Let us know how this turns out. I was considering getting a similar system myself.

 

[Jolio]

I've been pretty busy the past few days, and only just now got the chance to put everything put back together. I'm not even going to power in on tonight, so if history is any guide, it will be a few days after I first do that this problem will recur, and I'll have a chance to resume troubleshooting.

If the supply faults while in use it's unlikely any monitoring software will catch it because the fault will reset the logic board and in doing so terminate that program.

Which makes isolating the problem, if it is in fact the power supply, exceptionally difficult, as such failures leave little definitive evidence.

I'd try to find one of the service manuals...

I've got the Apple Technician Guide for this model, and I've working through it. I've got one thing left to try, Minimum Configuration Testing; it's tedious and time-consuming, but I will perform it when I have the chance to. However, there are very few avenues which don't involve replacing an expensive part, either the processor board, processor, or backplane board. Oddly, none of the sequences lead to replacing the PSU. The power supply is, in fact, listed in most scenarios as an unlikely cause.

I would replace the power supply.

I tend to agree with you, despite what the ATG says. However, the price of a new PSU (about $250) is a bit steep to test this theory, without exhausting all possible options first, in the hopes of finding the problem definitively. The price of the processor board or backplane board are about the same, too...and I'm certainly not going to spend more than I paid for this machine just trying to track down the problem.

Does anyone in the Washington, D.C. area have spare parts I could use for testing? PSU, processor tray, or backplane board?

 

[TWB]

The symptoms described here fit a heat-related hardware component failure. It's easy to go on wild goose chases with sporadic failures like this, so you should start with the cheap and easy things.

1. Oily deposits and flux on circuit boards are always a bad sign. Generally, flux and oily deposits are conductive, and can cause failures. It may not be causing this failure, but it should be cleaned off first. Contrary to what another poster said, the stuff in the pictures is not production-related. A circuit board would never pass QA with that on it.

2. There are two ways to test for heat-related failures: freeze spray and heat guns. A can of freeze spray costs $10, and will probably temporarily fix the problem if you spray the offending component after it has failed. You could also try to trigger the problem with a heat gun (sort of like a hair dryer, but hotter and more focused), but that is difficult to do, and you may damage components in the process. (Many years ago, I delaminated a $10k circuit board with a heat gun. Lesson learned.)

If, after cleaning and freeze-spraying, you haven't identified the problem, it's probably not worth fixing.

 

[BradHatter]

Out of curiosity, has a different OS version been tried? Yosemite isn't exactly known for being bug free.

 

[Jolio]

Hey, sorry for the staggered replies, I drafted my most recent post early, then and touched up and posted late it without checking for new replies.

Photos 1, 2 & 3 make me think a spill of some sort but they're to "geometric." There may have been some sort of adhesive insulating pad there at some point.

I noticed the unusually regular shape, too. If you look very closely at the photos, you'll see some strange patterns: lines between solder points, circles around posts not directly connected to the PCB...to me, these looked suspiciously like electric field lines.

Then I came across this video of wave soldering; have a look, near the beginning of the video, at the tray the PCBs are placed into before being soldered. Shaped cutouts such as those would explain perfectly what I'm seeing on my board. The white residue seems most likely to be from the soldering process; perhaps it's from flux, or a cleaning agent, or something else.

My first thought was that the fine patterns may have been formed by electric fields acting upon conductive particles in the contaminant. Alternatively, they have been deposited as their solvent evaporated. I'm leaning towards the latter, because some of the patterns occur where there are no nearby contacts (see the bottom of this image), and the lines seem to generally form concentric patterns from the perimeters of the affected areas.

I'm still not sure whether or not this white residue, or the oily residue, is a problem. I hear what you're saying, TWB, but after some searching, there seems to be no clear consensus on whether, in general, solder flux or cleaning agent residue is harmful. I suppose they almost certainly aren't doing anything helpful, so it should be safe to clean them off. I will do this if I have to, but I'm still hoping to find access to spare parts to rule out other components before I disassemble the machine again.

It may be worth noting that this machine has Boot ROM Version MP41.0081.B08, which, I understand, is only available in refurbished Macs. Perhaps the three sections that appear to be wave soldered on the backplane board are what was repaired.

The situation with the processor tray is more or less normal on older systems. Over time the silicone based oil in the thermal paste separates out. I would recommend removing the heat sink, cleaning off the old thermal paste and replacing it with new paste. This is a distinct possibility as root cause.

I've not before heard of the issue you're referring to, thermal paste separating over time. The oil spot I pointed out is a bit of a distance from the CPU; it's right next to the thermal pad that contacts the VRMs. If the oil is actually coming from the CPU thermal paste, there must be quite a lot of it seeping out to have spread that far. If something doesn't turn up soon, I will indeed be having a look under that heatsink, although I was hoping to avoid that delicate and tedious procedure.

You might also want to try another OS version

BSODs in Windows 7 under bootcamp, Machine Check Exception, similarly after some period of usage, apparently without any noticeable abnormalities in temperature, voltage, or any other sensor readings. I think it's unlikely the OS is at fault here, particularly given that the problem persists even with the HDD removed.

freeze spray

Brilliant! I had imagined a similar procedure for isolating failures, but I had no idea such a thing existed! One question, though: how do you prevent condensation from forming on frozen components?

Let us know how this turns out. I was considering getting a similar system myself.

If...it's probably not worth fixing.

Oh, I intend to get to the bottom of this, at this point as a matter of principle if nothing else. I'll most certainly share my experience. For the time being, I'm stalling (not to mention, busy), hoping to find access to known good parts (PSU, processor tray). Also, if I can find a way to more consistently produce this issue, I'd considered having it diagnosed at an Apple Store.

 

[BradHatter]

The Apple store will just try and sell you a new computer.

 

[Jolio]

It had gotten to the point where the machine wouldn't start up at all, even after a rest.

The CPU being the cheapest part ($10 used, eBay), I decided to just go ahead and try a replacement.

Booted right up after the procedure. Seems to be working fine, time will tell.

 

[Jolio]

Same problem after about an hour of uptime.

 

[Jolio]

It's looking like the processor board is at fault.

I picked up another 09 MP, which seems to work fine, and swapped the processor boards for testing. The problem follows the old board to the new machine, while the old machine with the new board seems to work fine.