Why Does My Desktop PC Randomly Restart? Causes I See on the Bench in New Jersey
Your tower reboots or goes black in the middle of a game or a render, with no blue screen to explain it. The cause is almost always hardware. Here's what I look for, and what you can check at home before you bring it to my desktop repair shop in Somerville, NJ. Drop-off, no appointment needed.
Why There's No Blue Screen
A crash and a power cut look different in the logs.
When Windows itself falls over, you get a blue screen with a stop code, then a reboot. When the screen goes black and the fans spin down, or the tower reboots instantly like someone pressed reset, Windows never had a chance to crash. Something cut the power, and that's what this page is about.
One setting muddies this. Windows has "Automatically restart" under Startup and Recovery, on by default, and with it on a real blue screen flashes for a fraction of a second before the reboot. Turning it off is step two below.
Either way, Windows leaves a note: Event ID 41, Kernel-Power in the System log. Microsoft's troubleshooting page says it means "the system has rebooted without cleanly shutting down first," and lists a faulty or underpowered supply, overclocking, memory misconfiguration and overheating as causes. It tells you the lights went out, not who flipped the switch.
If the log also shows WHEA-Logger entries or a stop code of 0x124, the processor reported a fault it couldn't correct; Microsoft calls that bug check "typically related to physical hardware failures." I cover that branch on my WHEA uncorrectable error page.
The Causes, What You See, and How I Confirm Each One
In the order I usually find them.
| What you see | Likely part | How I confirm it on the bench |
|---|---|---|
| Instant black screen and reboot when a game or render starts | PSU overcurrent or overpower protection tripping; aging supply out of margin | Load test at rated wattage with a scope on the 12V rail; swap in a known-good ATX 3.x unit |
| Hard shutdown after 10-30 minutes of heavy use; case hot | CPU thermal shutdown: dried paste, clogged cooler, dead pump or fan | Stress test with per-core temperature logging; check paste and mount |
| Dies only in demanding 3D scenes | GPU transient spikes overwhelming a marginal PSU; GPU thermal limit | Log GPU power and temperature under load; compare to PSU rating and ATX version |
| Restarts at idle, worse after a BIOS update or new RAM | RAM instability under an XMP or EXPO profile | BIOS defaults, then overnight memory test at stock and at profile speed |
| Cuts out under GPU load; connector discolored or warm | Failing 12V-2x6 or 8-pin PCIe connection: bent pins, partial seat, worn cable | Inspect pins under magnification; measure voltage drop across the cable under load |
| Started in summer or after the tower moved to carpet; fans loud first | Dust choking the CPU cooler and GPU heatsink | Pull heatsinks, clean, repaste, rerun temperature logging |
| Restarts when the AC compressor kicks on or lights flicker | Power strip, worn outlet, or dirty wall power | Run the tower on a known-good circuit through a UPS |
These stack: a dusty cooler raises CPU temperature, which raises power draw, which pushes an aging PSU over the edge.
What to Check at Home in 15 Minutes
None of these need the side panel off. Do them in order.
Read Event Viewer for Kernel-Power 41
Right-click Start, open Event Viewer, then Windows Logs, then System. Filter Current Log for event ID 41. Each entry is one unexpected restart. Open the newest, click Details, and note the BugcheckCode.
Turn off Automatically Restart
Press Windows key + R, type sysdm.cpl, press Enter. On the Advanced tab, under Startup and Recovery, click Settings and uncheck "Automatically restart." If the tower still goes black with no blue screen after this, you've confirmed a power or thermal cut.
Watch temperatures with the vendor's software
Your motherboard and GPU makers each publish a monitoring utility. Run something heavy for ten minutes and watch the CPU package and GPU readings. Intel's support page explains the processor throttles as it nears its limit and "will automatically shut down to prevent permanent damage" if throttling can't hold it. Mid-90s and climbing means the cooler is the problem.
Turn XMP or EXPO off
Restart, tap Delete or F2 at the splash screen, load Optimized Defaults, save and exit. This drops DDR5 or DDR4 to its base JEDEC speed. If the restarts vanish over a couple of days, the memory profile wasn't stable on that board.
Try a different outlet, with no power strip
Plug the cord straight into a wall outlet on a different circuit, away from a space heater, window AC or laser printer. If the restarts stop, read my page on power surge damage before plugging back into the old one.
Unplug every USB device except keyboard and mouse
Hubs, external drives and RGB controllers can short internally without looking any different. Pull everything, run for a day, then add devices back one at a time.
What I Do at the Bench
The home checks narrow it down. The bench proves it.
PSU load test. I pull the supply and put it on a load tester that draws a set wattage from each rail while I watch the 12V line on a scope. A tired unit sags, ripples or drops out, and its protection circuit shuts it off. Tom's Hardware's power supply guide describes those protections: overcurrent protection "kicks in when the current in the rails surpasses a certain limit," overpower protection trips when total draw exceeds rated capacity, and overtemperature protection shuts it down when it runs hot.
Stress test with temperature logging. With the supply cleared or replaced, I run the CPU and GPU flat out together for at least an hour and log every sensor once a second.
Memory test. If the restarts happen at idle or Event 41 shows a bug check code, I run a full memory test at JEDEC speed and then at the XMP or EXPO profile.
Swap in known-good parts. I swap one part at a time and rerun the same test until the restarts stop.
If the tower has simply outgrown its parts, a new GPU on an old supply for instance, I'll say so at pickup and we can talk upgrade instead of repair; my custom PC repair and upgrade page covers that. A machine that hangs without rebooting is a different problem; see my page on a PC that freezes every few seconds.
When a Restart Means a Dying Power Supply
This is the cause that costs the most if you ignore it.
A power supply doesn't fail all at once. Its capacitors dry out over years and the current it can deliver in a burst shrinks, while the graphics card you installed two years ago asks for bigger bursts than the supply was designed around. That collision is behind most of the load-triggered restarts I see from Somerset County.
The numbers matter. Intel's ATX 3.0 design guide, which Tom's Hardware broke down when it was published, requires a supply over 450W to deliver 200% of its rated power for 100 microseconds, 180% for a millisecond and 160% for ten milliseconds. An older ATX 2.x unit was never asked to do that, and a tired one trips overpower protection and goes dark. A 750W supply from 2018 and a 750W supply built to ATX 3.1 are not the same thing under a current graphics card.
The connector is its own failure. A 12V-2x6 or older 8-pin plug that isn't seated all the way carries the current through fewer pins, heats up, and loses contact under load. If the plastic is brown or the plug is warm after gaming, read my page on a melted GPU power connector before running it again.
For the full symptom list, see signs of a failing power supply. When the test confirms it, I supply and install the replacement with the service; details are on the power supply replacement page. If dust was the trigger, a proper desktop cleaning with repaste is often all a tower needs.
What I See on the Bench in New Jersey
Composite examples, built from the kinds of jobs that come across my Somerville counter.
A Bridgewater customer's three-year-old gaming tower with a new graphics card. Dead within two minutes of any current game, Event 41 with a bug check code of 0 every time. The original 650W supply dropped out on the load tester the moment I hit it with a transient burst. An ATX 3.1 unit ran the same game for two hours without a restart.
A Hillsborough customer's home-office desktop that rebooted a few times a week at idle. Automatic restart was on, so nobody had seen the blue screen underneath. With it off, the stop code pointed at memory; the kit passed at JEDEC speed and failed within twenty minutes with XMP enabled. A stable manual setting ended it.
A Piscataway customer's prebuilt that shut off after twenty minutes of video editing. Four years on carpet; the cooler fins were a felt mat. Logging showed the package hitting its limit and cutting out exactly as Intel describes. Full cleaning and fresh paste fixed it.
Bring the tower in. The $75 diagnostic ($149 for custom-built and boutique PCs) is credited toward the repair when you approve the work.
📞 Call (908) 428-9558Drop-off shop at 75 N Bridge St, Somerville NJ · Mon-Fri 10-5, Sat 9-2
Directions to Dave's Computers in Somerville, NJ
Downtown Somerville, one block off Main Street, behind Bank of America.
From Bridgewater
Route 22 or Route 202/206 to the Somerville Circle, then Route 28 east (West Main Street) into downtown. Left onto N Bridge St.
Open in Google MapsFrom Hillsborough
Route 206 north into Somerville, right onto West Main Street, then left onto N Bridge St.
Open in Google MapsFrom Bound Brook
Route 28 west straight into downtown Somerville. Right onto N Bridge St.
Open in Google MapsFrom Piscataway
I-287 north to Exit 17 (Route 202/206 south), then Route 28 east into downtown Somerville.
Open in Google MapsFrom Flemington
Route 202 north to the Somerville Circle, then Route 28 east into downtown. Left onto N Bridge St.
Open in Google MapsFrom Princeton
Route 206 north through Montgomery and Hillsborough into Somerville. Right onto West Main Street, then left onto N Bridge St.
Open in Google MapsTimes are typical drives outside rush hour. Call 908-428-9558 before a long drive if you want to confirm I'm at the counter.
Random Restart Questions From NJ Customers
Asked at the counter.
Why does my PC randomly restart with no blue screen?
A restart with no blue screen usually means power was cut before Windows could crash: the power supply tripped a protection, the CPU or GPU hit its thermal limit, or a connector lost contact. Windows logs it afterward as Kernel-Power Event ID 41. Also confirm Windows isn't set to restart automatically.
Why does my desktop turn off while gaming but not on the desktop?
Games push the GPU and CPU to full power at once, which is when a marginal power supply, a hot cooler or a weak connector gives up. Idle draw is a fraction of gaming draw. If it only dies under load, I check the PSU and temperatures first.
Can a power supply cause random restarts even if it turns on fine?
Yes. A supply can hold voltage at idle and still fold when the graphics card demands a spike. Its protection circuit shuts it off in a fraction of a second and the tower restarts. Older units do this more as capacitors age. A load test on the bench shows it.
Is Kernel-Power Event ID 41 a hardware problem?
Usually. Microsoft lists power loss, an underpowered or faulty supply, overclocking, memory trouble and overheating as causes. A bug check code of 0 inside the event means Windows never crashed and power was simply lost, which points at hardware. A non-zero code means a real blue screen happened first.
Should I turn off XMP or EXPO if my PC keeps restarting?
Yes, as a test. XMP and EXPO run memory above its default JEDEC speed, and a kit that isn't stable at that profile can crash or restart, especially with four sticks. Load BIOS defaults and run the PC for a few days. If the restarts stop, the profile or the sticks need attention.
What does it cost to find out why my PC keeps restarting in NJ?
The diagnostic is $75 for a Dell, HP, Lenovo or other major-brand desktop or $149 for a custom-built or boutique PC, credited toward the repair when you approve the work. Repairs are quoted after the diagnostic as a flat rate, and most are done in 24-48 hours. Bring the tower and its power cord to 75 N Bridge St in Somerville, no appointment needed.
Sources and further reading
- Microsoft Learn, Advanced troubleshooting for Event ID 41: "The system has rebooted without cleanly shutting down first": what the Kernel-Power event means and its listed causes.
- Microsoft Learn, Bug Check 0x124: WHEA_UNCORRECTABLE_ERROR: the fatal hardware error and its typical causes.
- Intel, Processor temperature, Tjunction and thermal throttling: throttling near the junction limit and automatic shutdown.
- Intel, ATX Version 3.0 Desktop Power Supply Design Guide, PSU Power Excursion: transient requirements for supplies over 450W.
- Tom's Hardware, Intel's ATX v3.0 PSU Standard Has More Power for GPUs: Power Spike Tolerance: the ATX 3.0 power excursion rules explained.
- Tom's Hardware, Power Supplies 101: PSU Protections: what OCP, OPP, OTP and the other protections do.
Bring the Restarting Tower to Somerville, New Jersey
Bring the desktop and its power cord to 75 N Bridge St, no appointment needed, and I'll find the part that's failing. The $75 diagnostic ($149 for custom-built and boutique PCs) is credited toward the repair when you approve it; the full rundown is on my desktop computer repair in New Jersey page.
📞 (908) 428-9558