PSU Failure Symptoms: How to Tell If Your Power Supply Is Dying (2026)

by

By Richard Anderson — PC Repair Veteran (16 years), Owner of Cincinnati PC Repair LLC

In 16 years of fixing desktops professionally, I’ve replaced more power supplies than I can count — and the sneakiest part of the job is that a dying PSU almost never announces itself cleanly. It pretends to be a bad stick of RAM, a corrupted OS, a failing GPU, a flaky motherboard. I’ve seen technicians replace three or four components chasing ghosts, when the real villain was a $40 power supply going bad six months after its capacitors started drying out.

That’s why PSU failure symptoms are the first thing I check now, not the last. Once you train your eye to recognize the signs, you’ll save yourself hours of unnecessary troubleshooting. This guide covers everything: the eight warning signs I look for, how to test a PSU with a paperclip and a multimeter, how to use software monitoring, and — critically — what not to do when you find a dead unit.

Whether you’re chasing random shutdowns, mysterious BSODs, or a PC that just won’t post, understanding PSU failure symptoms on desktop systems is an essential skill. Let’s get into it.

Safety first

Never open a power supply. A PSU can hold dangerous charge even after it is unplugged. You can inspect cables, test voltages carefully, and swap the unit, but do not open the metal PSU housing or try to repair internal components.

PSU failure symptoms quick table

SymptomWhat it usually meansWhat to check first
Random shutdowns under load12V rail may be unstableGPU/CPU load, PSU wattage, cables
Clicks, then no powerPSU protection may be tripping24-pin, CPU power, shorts
Random rebootsVoltage instabilityEvent Viewer, load testing
Different BSODs each timeHardware instabilityRAM, PSU, storage
Burning smellPossible electrical failureShut down and replace
PSU fan never spins under loadFan failure or semi-passive modePSU model behavior
Coil whine getting worseElectrical stress or aging componentsLoad changes, PSU age
USB devices disconnect5V rail instabilityUSB ports, motherboard, PSU
PC turns on but no displayGPU/RAM/PSU power issueGPU power, RAM, monitor
PC shuts off while gamingLoad spike or weak PSUGPU load, PSU wattage
System only boots after unpluggingPSU protection resetPower cycling behavior
Burning or hot cable/connectorBad connector or overloadStop using immediately

Four more PSU warning signs to watch for

9. PC Turns On But No Display

A no-display problem is not always a graphics card issue. If the GPU is not getting stable power, the fans and lights may come on while the system never completes POST. Before replacing the GPU, check PCIe power cables, reseat the card, test one RAM stick, and consider the PSU if the issue appears after upgrades or under load.

10. PC Shuts Off While Gaming

Gaming loads can expose a weak PSU faster than normal desktop use. A system may browse the web all day but shut off the moment the GPU pulls real power. If shutdowns happen during gaming, rendering, or stress testing, the PSU belongs high on the suspect list.

11. System Only Boots After Unplugging It

If the PC refuses to start until you unplug the power cord, wait, and plug it back in, the PSU may be tripping protection and resetting after discharge. That can also point to a shorted component, so test carefully, but do not ignore the power supply.

12. Hot, Melted, or Discolored Power Connectors

Any connector that looks browned, melted, loose, or unusually hot is a stop-now warning. Shut the PC down and inspect the cable path. Do not keep testing under load with a suspect connector.

The 8 Warning Signs of PSU Failure

These are the symptoms I see most often in my shop. None of them alone guarantees a bad PSU — but two or more together, especially if they appear under load, and I’m pulling the power supply first.

1. Random Shutdowns Under Load

The PC runs fine at idle or during light tasks, but the moment you fire up a game, start a render, or run a stress test, it powers off without warning. This is the most telltale sign of a PSU that can no longer deliver its rated wattage under sustained demand. The unit is brownoutting internally and triggering its over-current protection to kill the system rather than damage components. The critical clue here is reproducibility under load — if it happens every time during the same heavy tasks, look at the PSU before anything else.

2. PC Won’t Turn On — Clicks Then Nothing

You press the power button. You hear a faint click (sometimes multiple rapid clicks), and then absolutely nothing happens. No fans spin, no POST beep, no motherboard LED. This usually means the PSU’s internal protection circuit is tripping immediately on startup because the unit can’t hold its rails under even minimal load. I’ve had customers bring me machines where they’d already replaced the motherboard and the CPU before they called me. In nearly every case, it was the PSU.

3. Random Reboots Without Warning

Unexpected reboots — especially without a BSOD or error code — often point to voltage instability on the 12V rail. The CPU and GPU draw heavily from the 12V line, and when the PSU can’t sustain it, the system resets rather than crashes. If Windows is configured to restart automatically on system errors and you’re not seeing stop codes, check Event Viewer, but also strongly consider the PSU if the reboots coincide with any kind of computational load.

4. Blue Screens of Death That Don’t Match Software Issues

Random BSODs with stop codes like WHEA_UNCORRECTABLE_ERROR, KERNEL_POWER, MEMORY_MANAGEMENT, or CRITICAL_PROCESS_DIED can all stem from unstable power delivery. When a component like the CPU or RAM receives voltage outside its specification, even briefly, it throws errors that Windows interprets as software or hardware faults. If your stop codes are inconsistent — different errors each time, no single driver implicated — the PSU deserves serious suspicion. A faulty PSU can mimic bad RAM almost perfectly, which is how RAM gets replaced unnecessarily on a regular basis.

5. Burning Smell From the Case

A burning or acrid electrical smell is always an emergency. Shut the machine down immediately and unplug it from the wall. Don’t restart it. A failing PSU can arc internally, cause a capacitor to vent, or develop a partial short that generates real heat. In rare cases, this escalates to a fire risk. I’ve opened PSUs (safely, after discharge) and found blown capacitors, scorched PCBs, and melted plastic connectors. If you smell burning, the PSU is done — don’t test it further, just replace it.

6. PSU Fan Not Spinning

A PSU fan that has completely stopped can mean one of two things: the fan has failed mechanically (and the unit is overheating silently), or you have a modern PSU with a fanless/semi-fanless mode that’s working correctly at low load. I cover this in the myths section below. The diagnostic question is: does the fan spin up under heavy load? If you’re running a stress test and the PSU is hot but the fan is dead, that’s a problem. A genuinely failed fan will cause thermal shutdowns and accelerate internal component degradation.

7. Coil Whine Getting Worse Over Time

Some coil whine from a PSU is normal — it’s the electromagnetic vibration of inductors under load. But coil whine that’s progressively getting louder, changes pitch unpredictably, or starts appearing at loads where it previously didn’t, is a warning sign. It usually indicates that the filtering components inside the PSU are degrading and voltage regulation is becoming less stable. It’s not a “replace immediately” alarm on its own, but pair it with any other symptom on this list and I’d start shopping for a replacement.

8. USB Devices Disconnecting Randomly

This one surprises people. USB controllers on the motherboard are powered by the PSU’s 5V rail. When that rail becomes unstable — even subtly — USB devices lose power momentarily and disconnect. If your mouse, keyboard, or external drives are dropping out randomly without any Windows error or driver issue, and you’ve already ruled out the cables and USB ports themselves, test the PSU. I see this frequently in systems with aging budget power supplies.

The Paperclip Test: What It Tests and What It Doesn’t

The paperclip test — also called the jumper test — is a quick way to determine if a PSU can power itself independently of the motherboard. It isolates the power supply so you can rule out whether a dead system is caused by the PSU or by something else causing a short downstream. Here’s what you need to know before you try it.

What the Paperclip Test Actually Tells You

If the PSU fan spins when you perform the test, the unit is minimally functional — it can energize itself under no load. That does not mean it’s delivering correct voltages or that it can handle system load. A PSU can pass the paperclip test and still be delivering a sagging 12V rail under load. Think of it as a first-pass check, not a verdict.

If the fan does not spin, the PSU has almost certainly failed. You’ve confirmed it as the problem.

How to Do the Paperclip Test Safely

  1. Power down completely. Shut down Windows, flip the PSU’s rear power switch to OFF (O), and unplug the power cord from the wall. Wait 10 seconds.
  2. Disconnect the 24-pin ATX connector from the motherboard. This is the large flat connector on the right side of the board.
  3. Identify pins 15 and 16 on the 24-pin connector. Pin 15 is the green wire (PS_ON). Pin 16 is a black wire (ground). These are the two adjacent pins you need to bridge. If you have a PSU with a different pin layout, consult your manual — but on virtually all modern ATX PSUs, it’s the green wire to an adjacent black wire.
  4. Bend a standard paperclip into a U shape so both ends can insert into those two pin sockets simultaneously. The paperclip must be bare metal — no coated or plastic clips.
  5. Insert the paperclip into pins 15 (green) and 16 (black) of the 24-pin connector. Do not let it contact any other pins.
  6. Plug the power cord back in and flip the PSU switch to ON (I).
  7. Observe. The PSU fan should spin up within a second or two. If it does, the PSU is at least minimally alive. If nothing happens, the PSU has failed.
  8. Power down immediately after testing. Flip the switch off and unplug the cord. Running a PSU under no load for extended periods is not ideal for the unit.

Important: Do not touch any exposed pins or connectors while the PSU is powered on. The voltage levels inside a running PSU are not what kills you — but the unit contains capacitors that can store dangerous charge. Handle with care.

Interpreting the Results

  • Fan spins: PSU is minimally functional. Proceed to multimeter testing to check actual voltage output.
  • Fan doesn’t spin, no activity: PSU has failed. Replace it.
  • Fan spins but system still won’t boot after reconnecting: The PSU may be failing under load, or another component (GPU, hard drive) is causing a short. Disconnect peripherals one at a time and retest.

Testing With a Multimeter: Checking the Voltage Rails

A multimeter gives you actual numbers. If you want to know whether your PSU is delivering what it claims, this is how you check it. You’ll need a basic digital multimeter — I use a Klein Tools MM400 Digital Multimeter, which has been my go-to for shop work for years, though any decent DMM will do the job.

What You’re Measuring

A standard ATX PSU has three primary DC output rails:

  • +12V rail — Powers the CPU, GPU, and most drives. The most critical rail for modern systems. Acceptable range: 11.4V to 12.6V (±5% ATX spec).
  • +5V rail — Powers USB controllers, some chipset functions, and older drives. Acceptable range: 4.75V to 5.25V (±5%).
  • +3.3V rail — Powers memory and certain chipset functions. Acceptable range: 3.135V to 3.465V (±5%).

Voltages within these ranges under load are healthy. Voltages outside these ranges, or significant drops under load, indicate a failing PSU.

How to Test: Step by Step

  1. Set your multimeter to DC Voltage (V⎓) mode. Select a range above 15V if your meter isn’t auto-ranging.
  2. With the system running, locate a free Molex connector or use the back of the 24-pin connector while it’s still plugged into the motherboard. You can also use a PSU power supply tester adapter that breaks out pins for easy probing.
  3. Insert the black (negative) probe into any black wire pin on a Molex connector. This is your ground reference.
  4. Test the 12V rail: Touch the red (positive) probe to a yellow wire on the same Molex connector. You should read between 11.4V and 12.6V.
  5. Test the 5V rail: Move the red probe to a red wire on the Molex connector. You should read between 4.75V and 5.25V.
  6. Test the 3.3V rail: For this, you need to probe the orange wire on the 24-pin ATX connector (while it’s connected to the board). Insert the black probe into a black pin on the Molex, then probe the orange wire at the back of the 24-pin connector. You should read between 3.135V and 3.465V.
  7. Test under load. The critical step — run a stress test (Prime95 + FurMark simultaneously) while watching the multimeter. Voltages that sag significantly below the minimums during load confirm PSU weakness.

What the Numbers Mean

Steady voltages within spec at idle and under load: PSU is healthy. Voltages within spec at idle but dropping outside spec under load: PSU is failing and needs replacement. Voltages outside spec at idle: PSU has failed — replace immediately.

Pay particular attention to the 12V rail. On modern gaming PCs, the GPU and CPU together can pull the vast majority of system power from that single rail. A PSU that can’t hold 12V under combined CPU/GPU load is the number-one cause of the gaming shutdown symptoms I described in Section 1.

Software Monitoring With HWiNFO64

If you don’t have a multimeter handy — or you want a continuous view of voltages while running under load — HWiNFO64 is the best free tool for the job. Download it at hwinfo.com. It’s free, lightweight, and actively maintained.

Setting It Up for PSU Monitoring

  1. Install HWiNFO64 and launch it in Sensors-only mode.
  2. Scroll down to find your motherboard’s voltage section. Look for entries labeled +12V, +5V, and +3.3V.
  3. Right-click on a voltage row and select “Show Graph” to track changes over time.
  4. Run a demanding application — a game, a benchmark, or a stress test — and watch the voltage graphs.

What Voltage Fluctuations Mean

Small variations of a few millivolts under load are completely normal — no PSU is perfectly flat. What you’re looking for is:

  • 12V rail dropping below 11.4V under load: The PSU is sagging. This is a confirmed problem.
  • Rapid oscillations (the value bouncing around erratically): Voltage regulation has degraded significantly, likely due to failing capacitors.
  • 5V or 3.3V outside their ±5% range: Less common with modern PSUs, but if you’re seeing it, the unit needs replacement.

One important caveat: HWiNFO64 reads voltages as reported by the motherboard’s sensors, not directly from the PSU. The motherboard’s sensor resolution can vary, and some boards report more accurately than others. Software monitoring is a useful first look, but multimeter testing at the connector is the definitive check.

PSU Myths Debunked

I hear these constantly — from customers, on forums, and from people who should know better. Let me clear them up.

Myth 1: “My PSU Is 750W, So It’s Definitely Fine”

Wattage is not quality. A no-name 750W unit from an unknown brand might deliver only 600W at its actual rated temperature before the rails start sagging — or it might deliver 750W total but with terrible voltage regulation that creates ripple and instability. I’ve tested $30 “750W” power supplies that couldn’t sustain 500W without the 12V rail dropping below 11V. What matters is the brand, the tier (check the Tier List maintained by the Linus Tech Tips community), the 80 Plus certification, and whether it’s been reviewed by reputable testers. A 650W unit from Seasonic or Corsair will outperform a 1000W unit from a shelf-brand manufacturer every single time.

Myth 2: “The PSU Fan Stopped, So It’s Dead”

Many modern PSUs use what’s called a semi-fanless or hybrid mode. Below a certain load threshold — typically 20-30% of rated capacity — the fan doesn’t spin at all. This is a feature, not a fault. It extends fan life and reduces noise during light use. The problem only exists if the fan fails to spin up when the system is under heavy load and the unit is getting warm. If you’re idling at the desktop and the fan isn’t running, that’s probably fine. If you’re gaming for an hour and the PSU is hot and the fan is silent, that’s not.

Myth 3: “My PSU Is Only 2 Years Old, So Age Doesn’t Matter”

Age matters in two ways. First, the electrolytic capacitors inside a PSU have a finite lifespan measured in hours of operation and temperature cycles. A cheap PSU running hot in a poorly ventilated case can degrade its capacitors in 2-3 years. Second, even a well-made PSU that’s been running hard for five or more years will start showing rail sag under load as capacitors lose their capacitance. I’ve replaced PSUs that were “only” 4 years old but had been running a 24/7 workstation in a dusty environment. Always factor in hours of use and operating temperature, not just calendar age.

When to Replace vs. Repair — And Why You Should Always Replace

I’ll make this simple: never attempt to repair a PSU yourself. This is not me being overly cautious. The large capacitors inside a PSU — especially the main filtering caps on the primary side — can store lethal voltages of 300-400V DC even after the unit has been unplugged for hours. Professionals with proper equipment and training discharge these before working on the internals. Without that step, probing around inside a PSU is genuinely dangerous. There is no repair worth your life, and for what a replacement PSU costs, it doesn’t make financial sense either.

If a PSU is under warranty (most reputable units come with 5-10 year warranties), file an RMA claim with the manufacturer. Otherwise, buy a replacement.

How to Choose a Replacement PSU

  • 80 Plus Gold minimum. I won’t recommend anything less for a primary desktop. Gold efficiency means the unit is at least 87-92% efficient at typical loads, which means less heat generated inside the PSU, longer capacitor life, and more stable regulation. Platinum and Titanium are better still, but Gold is the minimum I’ll put in a customer’s machine.
  • Size the wattage correctly. Use a PSU wattage calculator — the one at be quiet!’s website is solid — then add 20-25% headroom. Running a PSU at 70-80% of rated capacity is the sweet spot for both efficiency and longevity. A PSU running at 95% capacity is stressed and runs hot.
  • Stick to reputable brands. Seasonic, Corsair (their RMx and HX lines), EVGA (SuperNova G series), be quiet!, and Fractal Design are names I trust in my shop. Avoid unknown brands, regardless of how impressive the spec sheet looks.
  • Check the tier list. The community-maintained PSU tier lists are genuinely useful for sorting out which specific models are worth buying.

If your system is also experiencing no-display issues after a PSU-related failure, check out our guide on diagnosing desktop PC no-display problems — a bad PSU can corrupt display output in ways that look like a dead GPU or monitor.

Leave a Comment