crwdns2933423:0crwdne2933423:0

Crash every few minutes with flickering patterns on screen

I have an old Mid-2012 MacBook Air which some years ago started developing a problem that it would freeze, with a flickering blocky pattern on screen for a while, and then shut down.

At first this was only occasional, but has grown more frequent with time. Now it won’t stay running for more than a few minutes. It seems to stay on a bit longer when started for the first time with everything cold.

This doesn’t seem to be related to system load though, and also happens at an idle recovery screen with no actual OS running. I have tried resetting nvram/smc. Incidentally, the same has happened a few times on a windows laptop in the family, and I’ve seen a picture of the exact same problem on stackexchange (unfortunately did not bookmark the question).

I think the most obvious common factor would be the cpu (or, possibly, a flaw in support circuitry following the same reference design) – or the integrated gpu more specifically. It seems unlikely that e.g. memory corruption could cause the exact same symptoms on this many different systems.

Has anyone encountered and been able to fix this issue? How likely do you think this could be heat related? Should I try to reapply thermal paste? Or anything else I could do?

At this point I’m mainly interested to make it stable enough for a local trade-in campaign, and ideally long enough to be able to wipe the disk and reinstall MacOS first.

Block Image

-----

UPDATE:

I changed the thermal paste. There was plenty of it, completely caked, which still did not cover the entire cpu die. It can't ever have made very good thermal contact.

Unfortunately, the problem was not solved (I would say not even improved).

The heat pipe seems ok, or at least I could not see any sign of damage or leaking. The fan works. The cpu end of the heat pipe is only mildly warm when the crash happens, so this does not seem to be a cpu heat issue after all.

I believe the SSD can be excluded as a potential cause, as the laptop will also crash when booted into recovery mode. At that point everything is in ram, the drive can even be erased entirely. There cannot really be any such disk activity happening that would crash the system.

This is starting to look like a logic board issue, whether that be solder joints, ram, cpu power circuitry, or something else. That is beyond what's sensible to repair in a 2012 laptop, so it seems I'm out of options for any more things to try.

crwdns2934089:0crwdne2934089:0 crwdns2934093:0crwdne2934093:0

crwdns2934109:0crwdne2934109:0

crwdns2889612:0crwdne2889612:0 0
crwdns2934285:0crwdne2934285:0

crwdns2933313:01crwdne2933313:0

crwdns2934057:0crwdne2934057:0

This sounds like a thermal stress kind of problem. The simplest would be what we call a cold solder joint. Here it could be the RAM chips or the flash chips within the SSD drive, next would be the CPU module either a bad heat sink or poor thermal bonding between the chip and heat sink due to old thermal paste.

It you have access to a hot air rework station you could try reheating the RAM chips using some good flux and fresh solder. If you have access to a spare SSD drive see if that works, and lastly carefully inspect the heat sink heat transfer tube for a hole. Look for staining on the tube and bottom plate where the tube sits. Then clean and reapply new thermal paste.

crwdns2934105:0crwdne2934105:0

crwdns2889612:0crwdne2889612:0 1

crwdns2944067:05crwdne2944067:0:

Thanks! My reasoning was that since the exact same (or extremely similar) issue has been seen on three different laptops, with different logic boards and running both MacOS and Windows, it would almost have to be something to do with the CPU/GPU itself, not the system around it.

There is nothing out of the ordinary that I could see on either the logic board (e.g. blown capacitors) or the heat transfer tube, though the tube I haven't been able to inspect on its bottom side yet. The SSD is removable but uses an Apple proprietary connector, so not possible to swap or to run tests on without buying adapters.

As thermal paste seems to be the only thing I'm able to do (or willing to spend money on for such an old laptop), I'll probably have to try that.

crwdns2934271:0crwdnd2934271:0crwdne2934271:0

@jay82379 - What are you running on this system (and the others)? Maybe you are over stressing them.

These older systems can easily be driven into the ground like a horse. While horses are built for running, they can be pushed to their limits, potentially leading to death from exhaustion, dehydration, or heart problems if not properly cared for and rested. Running heavy CPU and/or GPU based games or other programs can be the root cause here.

I don’t think that’s your issue at this point but after repairing the cold solder joints I think you’ll find the problem will be gone.

As far as the SSD, you can use an external drive to test that. It just needs to be setup as a bootable drive and setting it as the boot drive.

crwdns2934271:0crwdnd2934271:0crwdne2934271:0

Well the strange thing is, I pretty much don't need to run anything at all to make it crash. Happens just as well if I boot into recovery mode (cmd-R) and just let it sit there idling.

Good point about using an external drive. Unlikely to be the cause though, as the crashes can happen regardless of disk activity – which anyway could likely not cause GPU or kernel level issues unless through corrupting swapped-out memory, and that can't really be happening in idle state either – but probably still good to test after ruling out other possibilities.

Today I tested letting it run in cold weather outdoors for closer to an hour (no crashes). So looking more and more like either a thermal paste or heat pipe issue.

crwdns2934271:0crwdnd2934271:0crwdne2934271:0

@jay82379 - Don’t discount a cold solder joint!

crwdns2934271:0crwdnd2934271:0crwdne2934271:0

Certainly not! But in any case not the first thing to try (and not worth paying for on a laptop this old). The gradual worsening has my hopes on a simple thermal paste issue.

crwdns2934271:0crwdnd2934271:0crwdne2934271:0

crwdns2934285:0crwdne2934285:0

crwdns2934229:0crwdne2934229:0

jay crwdns2934231:0crwdne2934231:0
crwdns2936625:0crwdne2936625:0:

crwdns2936751:024crwdne2936751:0 0

crwdns2936753:07crwdne2936753:0 0

crwdns2936753:030crwdne2936753:0 1

crwdns2942667:0crwdne2942667:0 71