How to Check Your PC's Power Supply Unit (PSU)

Your power supply is one of the most overlooked components in a PC — until something goes wrong. Whether you're diagnosing a system that won't boot, investigating random shutdowns, or planning an upgrade, knowing how to check your PSU can save you time, money, and frustration.

Why Checking Your Power Supply Matters

The PSU converts AC power from the wall into the DC voltages your components need. When it underperforms or fails, symptoms can look like almost anything — crashes, blue screens, freezes, or a system that simply won't turn on. Before replacing expensive components like your GPU or motherboard, it's worth ruling out the power supply first.

Method 1: Check PSU Output Using Software

The easiest starting point requires no tools at all.

Hardware monitoring software can read voltage data reported by your motherboard's sensors. Tools commonly used for this include HWMonitor, HWiNFO, and Open Hardware Monitor. These applications pull data from your motherboard's embedded controller and display readings for the +3.3V, +5V, and +12V rails — the three primary voltage lines a PSU delivers.

What healthy readings generally look like:

These tolerances follow the ATX specification, which allows ±5% variance on each rail. Readings consistently outside these ranges suggest a struggling or failing PSU.

One important caveat: motherboard sensor accuracy varies. These software readings are useful indicators, not lab-grade measurements. A reading slightly outside range could reflect a sensor issue rather than the PSU itself.

Method 2: Use a Dedicated PSU Tester

A PSU tester is an inexpensive device — typically a small unit with an LED display — that plugs directly into your PSU's connectors and reads each rail's voltage without a motherboard in the loop. This removes the sensor accuracy variable entirely.

To use one:

1. Power down and unplug your PC completely
2. Disconnect the PSU from all components
3. Connect the PSU tester to the 24-pin ATX connector (and optionally the EPS CPU connector)
4. Plug the PSU back into the wall and switch it on
5. Read the voltage values displayed on the tester

More advanced testers include a load resistor that draws actual current while testing, which gives a more realistic picture of PSU behavior under stress. Basic testers that apply no load can miss problems that only appear when the PSU is actually powering components. ⚡

Method 3: The Paperclip Test (Power-On Without a System)

If you want to confirm whether a PSU powers on at all — without any other components connected — you can use the paperclip test. This works by shorting the PS_ON pin (pin 16) to any ground pin on the 24-pin ATX connector, which tricks the PSU into turning on without receiving the signal from a motherboard.

Steps:

1. Unplug the PSU from all components and the wall
2. Locate pin 16 (green wire) and any black ground wire on the 24-pin connector
3. Bend a paperclip into a U-shape and insert both ends into those two pins
4. Plug the PSU into the wall and switch it on
5. If the fan spins, the PSU is at least functional enough to power on

This test only confirms basic operation. A PSU can pass this test while still delivering unstable or out-of-spec voltages under load.

Method 4: Multimeter Testing 🔧

For the most direct measurement, a digital multimeter lets you measure voltage directly at the PSU's connectors while the system is running. This is more involved but gives you real-time, hardware-level data.

Common test points:

• Molex connector: +12V (yellow to black) and +5V (red to black)
• 24-pin connector: Multiple rails accessible with pin reference charts
• SATA power connector: +3.3V, +5V, and +12V lines

Set your multimeter to DC voltage, connect the black probe to a ground (black wire), and touch the red probe to the voltage wire you're measuring. Compare readings against the ATX tolerance table above.

This method is most useful for spotting voltage sag under load — checking readings both at idle and while running a demanding task like a stress test or gaming session.

Method 5: Observe System Behavior for Warning Signs

Not every PSU check requires tools. Certain behavioral patterns are strongly associated with power supply problems:

• Random shutdowns or restarts, especially under load (gaming, rendering, compiling)
• System won't POST despite all other components appearing functional
• Burning smell or visible scorch marks near the PSU
• Coil whine or buzzing from the PSU area under load
• USB devices behaving erratically due to unstable +5V delivery
• GPU artifacts or crashes that disappear when the GPU is swapped but return with the same PSU

These symptoms overlap with other hardware issues, which is why isolating the PSU through testing — rather than relying on symptoms alone — gives you more confidence in your diagnosis.

Variables That Affect What You're Actually Testing

How you check your PSU matters less than understanding what you're checking for. Several factors shape the picture:

System load: A PSU might test clean at idle but sag badly under a full CPU + GPU load. Testing while running a stress tool like Prime95 alongside FurMark gives a more honest picture than testing at desktop idle.

Age and wear: Capacitors degrade over time. A PSU that tested fine three years ago may no longer hold stable voltages under the same workload.

Wattage headroom: A PSU running near its rated maximum continuously stresses components and runs hotter. How much headroom exists between your system's actual power draw and the PSU's rated output changes what "normal" looks like.

Rail configuration: Some PSUs use a single +12V rail; others split it across multiple rails. This affects how you interpret multimeter readings and which connectors matter most to test.

Connector and cable condition: Loose connectors or damaged cables can cause voltage drop that looks like a failing PSU but isn't. Reseating all power connections is worth doing before drawing conclusions from any test.

Whether your system needs a full diagnostic workup or just a quick voltage sanity check depends entirely on the symptoms you're seeing, the age of your hardware, and how much load your components actually demand — factors only visible from your side of the setup.