Table of Contents:
This remains not the most stunning of subjects, but it is equally as fraught as the last post about it; fraught with trauma and money-burning stupidity.
This is the second thrilling instalment of the saga that is the new PC (and it hasn’t ended yet, by the way). This has got all sorts of tangents, by-the-ways and trouserly bifurcations in it, so buckle up, strap yourself in.
Where it was left. Some time ago.
At the end of the first post on the subject at the end of July, I’d got it working as a PC after a fashion. At the very least I’d got to the point at which I could use it for a couple of hours before it turned itself off.
Over the next few months – up until December 2025 – I spent an awful lot of time attempting to troubleshoot why the PC was turning itself off.
Power issues?
The PC was just turning off, as if someone had turned the power off at the wall socket. Thinking it may be a power supply problem (as in the actual electrical supply to the PC), I duly swapped cables, changed power sockets, eliminated (and then reinstated) the UPS. Checked the UPS out to make sure it was doing what it was supposed to be (it was). None of this made any difference: it still turned itself off.
I bought a new power supply for it and swapped it out. It was a lesser capacity power supply than the one I’d originally bought, but I’m unable to run any PCIe cards in the PC anyway, so there’s no extra power draw. I did have to re-cable it with the supplied cables though, as they were slightly different to that of the original power supply (which I thought odd, I assumed all that was standard). Anyway, none of that made any difference: it still turned itself off.
So that was actual power supply issues near enough ruled out. I left the new PSU in the PC, as it was doing the job perfectly well.
Moving on…
I asked the dreaded AI
I recently scribed a post about AI: where I waffle briefly about asking AI about the PC issues. And I have to say, it was sort of helpful, but not as much as I thought it would be. It didn’t instantly diagnose the issues, but over a period of several weeks going back and forth between me and ChatGPT (and sometimes CoPilot, if I was feeling frivolous), I got enough clues to narrow things down a bit more.
But do not labour under any misconception, dear reader; it took quite a few weeks of asking, interpreting, re-asking, re-interpreting and so on and so forth before I could get any closer to what I thought the issue was. When I say quite a few weeks, I mean 18 weeks. And when I say asking, I mean attempting to get a sensible answer instead of a load of contradictory bollocks from AI.
I trawled the forums
Ai was not the only route for research. It couldn’t be – it’s terrible. I also trawled various forums, manufacturers support pages and various other “sources of information”. As I point out in this post, 99% of the information that’s out there is mainly crap. Much in the same vein as AI responses, there were clues that had potential to narrow the fault down a bit.
This also takes a good deal of time. I ended up chatting to AI about an issue, then verifying it using web searches and forums. Then I’d do it again and see what it came up with this time. Sometimes, it would suggest something – no more than a mere breadcrumb – that would lead to quite some useful pages, describing how things (i.e. modern motherboards) worked.
Why not just send it back?
This is a very good question. It’s still in warranty, so I could have just uninstalled it and sent it back to Overclockers UK. But rather than just send the motherboard back to Overclockers UK straight away, I just wanted to make sure that I hadn’t done something stupid to cause the fault to happen. I was after all, running extra disks and whatnot, I just wanted to make sure that I wasn’t the cause of the fault, whereupon OcUK would test it, determine it was my fault, then send it back and charge me for the privilege. And then I’d look a twat. To some people, that wouldn’t have mattered and they wouldn’t have cared. But not me. Call me old-fashioned.
Moving on…
What did I do?
What I did over the course of 18 weeks was split into two parts: dealing with the turning off issues and dealing with BSOD’s, or Blue Screen of Death’s (or Black Screen of Death as it is now in Windows 11).
A Blue (or Black) Screen of Death – shortened to BSOD – is when a Microsoft Windows PC suffers from some sort of issue and crashes. It could be a driver issue, it could be a hardware issue, but the computer will display a blue (or black) screen with some technical text before rebooting. Linux and other operating systems suffer from them too, but are called something slightly different: a kernel panic, an abend etc. The end result is the same – there’s been an operating system problem, the PC reboots.
I’m pretty sure at this point that the underlying power issue was contributing to the BSOD issue(s) as well – but at the time of attempting diagnosis, this wasn’t evident, or indeed obvious.
The turning off problem
I removed the peripherals: as much as I could, anyway. I had a PCIe fibre network card in it to start with. Once that was removed, it started turning off less frequently.
Turned off anything that really wasn’t required: things like the LED RGB lighting in the case and on the fans. Some forums mentioned that particularly on ASUS motherboards, the RGB controllers could cause issues.
Performed driver and BIOS updates: over the course of a few weeks, ASUS released three BIOS updates for the motherboard. The first two didn’t make a good deal of improvement to anything much. I’d saved the overclocked profiles and any tweaks that I’d done, which did improve things for a while. Following the second BIOS update, I could go for days before it turned itself off.
Removed the M.2 NVMe SSD drives: some forums (and AI) suggested removing all of the disk drives apart from the system disk. There was talk of PCIe lane speeds and power drain with PCIe 5.0 drives. The suggestion was to throttle the PCIe lane speed to 4.0 for all M.2 drives in the BIOS, which I duly did. Running with one drive only seemed to be pretty stable and lasted a good 14 days without shutting down. The other disks were re-installed one-by-one over a period of three weeks, leaving a few days in between installations to see if it stayed up. It did. For a while. All the time noting the BIOS tweaks to throttle the speed down to 4.0. It worked for a while. Until – of course – it didn’t. sigh
Ran Memtestx86. A lot. It wasn’t clear at the outset whether this was a motherboard or a memory, or even a CPU issue. So in an attempt to eliminate something, memory integrity was the easiest to do, Memtestx86 was run several times and for several passes per time. It was run with the overclocked settings both on and off. The results were mixed: but it didn’t seem to like the overclocked settings much (as they are mainly memory speed settings anyway)… but sometimes it did! The best results were with the default BIOS settings for memory. The memory was ruled out quite early on in the diagnostic process – sticking with standard BIOS settings gave the best and most consistent results. By the end of the many series of passes, it passed the memtestx86 tests consistently and many times. That pretty much ruled out memory anyway.
BIOS tweaks: several BIOS tweaks were performed to try and improve things. The BIOS on this particular motherboard has a lot of options for overclocking and advanced tweaking. I made sure I backed up the BIOS setting profiles that were there (mainly mine and the OcUK overclocked profile settings) so I could restore them if necessary. Knowing that when a BIOS is updated, all settings are returned to factory settings. From there, various profiles were loaded – with varying success. CPU power consumption tweaks, RGB tweaks, memory speed tweaks. The default factory settings were tried for a while, before loading the overclocked profile. It didn’t seem to make a great deal of difference to the PC turning off. It still randomly turned itself off whatever the BIOS settings.
It boiled down to this: no matter what tweaks, adjustments or fine tuning was made; it still persisted in turning itself off for no readily (or otherwise) apparent reason. It just did. Sometimes it pranced off down a garden path of uncertainty, skipping merrily away for a few days before coming back and laughing as it just shut down.
BSOD Issues
Well, it is Windows 11 I suppose. And it is fraught with Event Log warnings and application crashes and whatnot, even from a fresh install. Hours spent chasing DCOM Errors in the event log all led to the same thing: it still shut down.
I had some periods of time when I couldn’t boot to Windows due to BSOD’s. I couldn’t play any videos without it crashing and it wouldn’t stay on for very long without BSOD and rebooting. There was a dalliance with both page and minidump files. WinDBG and event log tracking. This was all mostly driver errors of some kind, whether it was display drivers or VirtualBox executables (I had to discontinue VirtualBox, was I was considerably displeased about), they were solvable by updating or rolling back drivers.
Apart from the Intel Management Engine drivers, which we’ll come to in a sentence or two.
Several fresh reinstallations were executed. Some with ASUS drivers, some just with Windows updates supplying the drivers. And some with the drivers supplied with the motherboard, contained on a handy USB stick.
The Intel Management Engine Drivers earned a special place in hell, all of their own. The Intel Management Engine (IME for short) appears to be linked to the version of BIOS that’s on the motherboard. A BIOS update will require a different set of IME drivers (in Windows, anyway), otherwise it won’t run. It took a little while to work that one out. And quite a bit of swearing.
But still, it shut down. Of course it did.
The unconvincing evidence
By the time a few weeks had passed, several rebuild and BIOS updates, the PC was still shutting down on its own. There was no solid evidence to indicate what the hell was the matter with it, other than there may be a fault on the motherboard causing the power to shut off. Because it was a sudden power off, the indication was that something on the board, or even the CPU was triggering it because of [condition] and it shut itself down as a safety precaution.
It’s definitely not the software that was causing anything to happen, as several reinstalls of Windows and a couple of Linux ones proved not to make an iota of difference; effectively ruling out operating systems.
Having said that, the two faults (power shut offs and BSOD’s) weren’t directly linked – as in the operating system isn’t causing or triggering the power shut offs – but it is suspected that the power shut off problem may have had an effect of the operating system installed. I.e. the power issue was causing driver upsets and therefore BSOD’s.
Monitoring
Several monitoring tools were employed to attempt a diagnosis. One of them (HWINFO64) even caused an unexpected shutdown, which did end up being relatively useful in narrowing down the general area of where things may be awry on the motherboard. Days and Gb’s of logs later, the conclusion was, er… inconclusive.
There was no indication of anything going awry before the PC shutdown. Nothing. All of the temperatures (CPU cores, memory, M.2 disks), all of the voltages, all of the averages for all of the readings – all remained totally stable up until the point at which the PC shut down. There were no spikes, no dips. Not even a burp.
The final straw
The event that triggered the return of the motherboard from whence it came (Overclockers UK) happened around mid-December 2025. A couple of BIOS updates had been released by ASUS to address different things, along with their respective IME updates. The latest one was downloaded and duly applied in the manner that was accustomed to for the last two BIOS updates: backup the BIOS profiles, run the update, see what happens.
This time, things went awry very, very quickly. The PC wouldn’t turn on for quite a while. When it eventuality got going (it goes through some kind of training stage), it wouldn’t stay powered on for very long. Minutes, sometimes.
A few hours later. I’d reset the BIOS (several times). Hard reset it, removed the BIOS battery. Still no joy whatsoever. The longest time I managed to have the PC running was 42 minutes. Although incredibly infuriating, being a Hitchhiker’s fan, it did bring a wry smile to my face. Only very briefly, mind you.
At that point I uttered the immortal words “fuck it” and logged a web ticket with Overclockers UK on December 22nd. The PC was shut down, unplugged and left there until I had instruction from OcUK.
Contingency measures
Given that it was the Christmas period, I wasn’t really expecting anything from OcUK until January, so the Microsoft Surface Pro 8 was pressed into service as the main PC for a while, along with a couple of M.2 NVMe disk caddies to handle the storage that would normally be in the main PC.
The Surface Pro actually does do a good job at playing “main PC”. Hooked up to the main monitor via the Thunderbolt dock and with a Bluetooth keyboard, you would barely notice that it was a tablet pretending to be a grown up.
It served me well for the next couple of weeks, whilst the holiday period was being enjoyed by one and all.
The OcUK verdict
It didn’t take OcUK very long to determine the board was crap. They sent several emails along the way with progress, but it boiled down to a replacement board, which they configured and tested with my CPU and memory.
They sent the replacement board to me – and that is, dear reader, where we are now. The new board is in, Windows is built, drives are in and work is ensuing.
So far so good.
TL;DR
The PC’s still broken and AI isn’t much help, if any. Managed to narrow the fault down a bit, determined it wasn’t me at fault and sent it back under warranty. New board arrived and is in and running. Everyone for the moment is happy. Will it stay that way?
