My brother's Windows 11 PC powers off at completely unpredictable times. The screen goes black, while the fans and lighting continue briefly before slowly stopping, and there are no useful entries in Event Viewer or Reliability Monitor. It can happen while gaming or sitting idle, with anywhere from an hour to two weeks between incidents. Temperatures appear normal, and the latest shutdown happened about two weeks after replacing the power supply and its cables. We have already reinstalled Windows, monitored the hardware with HWiNFO, removed overclocks and XMP, and replaced the PSU. The system has an ASUS ROG B550-F motherboard, Ryzen-based platform, 32GB Corsair memory, an RTX 2080 Ti Founders Edition, a Corsair RM750e PSU, a 1TB SSD, and several case fans with a radiator. What should we test next before replacing parts at random?
4 Answers
There are several hardware variables worth isolating: test with one memory stick at a time and try different slots, remove the graphics card temporarily if the processor has integrated graphics, and test a different wall outlet. You could also temporarily disable CPU C-states, since some systems become unstable during low-power idle states. If those tests don’t help, the motherboard’s VRM or another board component may be failing.
Disable Windows Fast Startup and see whether the next failure produces more useful power or hibernation events. Fast Startup uses a partial hibernation process, so the shutdown may not be logged in the way you expect. Also check for hibernation-related events rather than looking only for a normal shutdown event.
Check for a mechanical short or grounding problem inside the case. A misplaced motherboard standoff, loose screw, or contact with the case can cause unpredictable power loss. If possible, test the board outside the case on a nonconductive surface with only the essential components connected.
That’s a good lead. Someone else with hardware experience suggested checking the standoffs too, so I’ll inspect them carefully.
A motherboard fault is a strong possibility, especially with completely random timing and no thermal or software evidence. Still, test the memory properly with MemTest86, preferably overnight. Windows’ built-in memory check can miss intermittent RAM problems.
I’ll run MemTest86 overnight. The previous built-in test found nothing, but I don’t fully trust it for intermittent errors.
The long gaps between failures make testing difficult, but this gives me a useful order for narrowing things down instead of swapping parts blindly.