For the past three days, my desktop has been shutting off completely while gaming and sometimes even during light browsing. The lights and fans turn off, but pressing the power button initially lets it boot again. It would then run for anywhere from 10–30 minutes, or up to about an hour after I replaced the power supply, before shutting down again.
I tested a different PSU, reseated the RAM and SSD, and temporarily disconnected a second SSD and several case fans. After the latest shutdown, the PC only stays on for two or three seconds before powering off again, without reaching Windows. Event Viewer only shows a Kernel-Power critical entry indicating an improper shutdown. I also monitored temperatures with HWMonitor, and the CPU and GPU stayed below 90°C.
What could cause this behavior, and what troubleshooting steps can I safely try at home? I can't currently afford to take the computer to a repair shop.
4 Answers
You can also temporarily test power-management settings, such as disabling PCIe Link State Power Management or setting the graphics driver to prefer maximum performance. However, because the machine now turns off after only two or three seconds, I would prioritize hardware isolation and checking cables over Windows settings. If it shuts down in the BIOS with the same minimal setup, the operating system is unlikely to be the cause.
The Kernel-Power entry usually only confirms that Windows lost power unexpectedly; it doesn’t identify the failed component. Try running separate CPU and GPU stress tests while watching temperatures and voltages. If it powers off almost immediately even with minimal hardware connected, the motherboard, CPU, graphics card, or a power connection may be involved rather than Windows itself.
This pattern can point to overheating, especially if the shutdowns happen sooner each time. Check both CPU and GPU temperatures while idle and under load, and inspect the cooler, thermal paste, fans, and airflow. HWiNFO is useful for recording maximum temperatures. A photo of the inside of the case and the complete system specifications would also help narrow it down.
Please list the full specifications, including the motherboard, CPU, graphics card, RAM, storage, and both PSU models. Also verify that the replacement PSU’s modular cables are the ones supplied with that PSU—mixing modular cables between brands or models can damage components. Reseat the 24-pin motherboard connector, CPU power connector, and GPU power cables, then test with only one RAM stick, the boot drive, and the minimum hardware needed to start.
I checked the temperatures and neither the CPU nor GPU went above roughly 90°C. I haven’t run a dedicated stress test yet, only tested with a demanding game and normal browsing, and both eventually caused shutdowns.