GUIDE

Signs your power supply is failing, and how to test it

Published October 11, 2026 · Who writes this

The classic sign is a PC that shuts off or restarts under load, like in the middle of a game. Before you blame the power supply, rule out the simple stuff, then test it.

Common symptoms

• Shuts off or restarts under load, like in the middle of a game or a render, but runs fine at the desktop.
• Won't turn on at all, or the fans twitch for a second and stop.
• Clicking or buzzing from the power supply, or coil whine that keeps getting louder.
• A burning or hot-plastic smell. Shut it down and unplug it right away.
• The PSU fan never spins, even when the PC is working hard. (Many units stop the fan at low load on purpose. That's normal; it should start once things warm up.)
• Random crashes, USB devices dropping out, or drives disappearing and coming back.

Every one of these can have another cause. Overheating, a bad RAM stick, an unstable overclock or a failing graphics card can look exactly the same, so check before you buy a new PSU.

First, the easy checks

1. Rule out the outlet. Plug the PC straight into a different wall outlet, skipping the power bar.
2. Reseat every power cable: the 24-pin, the CPU cable and the graphics card plugs. Look for scorched or melted pins while you're there.
3. Undo any overclocks and check CPU and GPU temperatures under load.
4. Check Windows Event Viewer. A "Kernel-Power" event with ID 41 means the PC lost power without shutting down properly. It confirms a sudden power loss, not what caused it.

Test 1: Swap in a known-good PSU

This is the most reliable test. Borrow a power supply that has enough watts and the right cables for your build, and run the PC the way that made it crash. If the problem goes away, you've found it.

Test 2: Check the voltages with a multimeter

With the PC running, measure at a spare Molex or SATA plug, or back-probe the 24-pin connector. Check at idle, then again under a heavy load like a game or a stress test. Healthy readings stay inside the ATX spec's range:

Newer ATX 3.0 and 3.1 units are allowed to dip to about 11.2V on the 12V rail during heavy load spikes. A reading that sags well below 11.4V under steady load, or wanders around, is a bad sign.

Software readings from HWiNFO or the BIOS come from motherboard sensors and are often off by a few percent. Use them to spot a trend, not to judge the PSU.

Test 3: A PSU tester or the paperclip test

A cheap plug-in PSU tester checks that each rail is present and roughly in range. It's quick, but it barely loads the power supply, so a unit that passes can still fail under a gaming load.

The paperclip test only checks that the PSU turns on at all:

1. Switch the PSU off and unplug it from the wall. Disconnect it from every part in the PC.
2. On the 24-pin plug, bridge the green wire (PS_ON) to any black wire (ground) with a bent paperclip.
3. Plug it back in and flip the switch on. If the fan spins, the PSU powers on. Units with a fan-stop mode may stay silent with no load.

Passing the paperclip test tells you nothing about how the PSU behaves under load. Switch it off and unplug it before removing the paperclip.

Never open the power supply

The capacitors inside can hold a dangerous charge long after the unit is unplugged. A suspect PSU gets replaced, not repaired. If it's still under warranty, contact the maker; good units often carry 7 to 12 years.

When to replace it

• The swap test or the voltages point to the PSU.
• There's any burning smell, or scorched or melted connectors.
• It's old, out of warranty and you're about to put in a more power-hungry graphics card.

If your old unit is modular, don't reuse its cables with the new one. Here's why.

Check your exact build in the calculator