A particular sort of job comes in from the light-industrial units around Bonnington: an office PC that restarts by itself, with no blue screen, no warning and no pattern anybody can pin down. It has usually been happening for months. Somebody has already replaced the power supply, somebody else has reinstalled Windows, and it still does it. The machine tests perfectly on our bench and does not misbehave once in three days of burn-in.
That result is the diagnosis. If a PC that reboots daily in a workshop unit behaves impeccably in a quiet workshop across town, the fault is not in the PC — it is in the supply the PC is plugged into. This is a common pattern in the trade units off Bonnington Road and Newhaven Road, where an office corner shares a consumer unit with compressors, extraction, kilns, presses and three-phase machinery. If the reboots are already eating into your working days, same-day computer repair in Bonnington is the quickest way to get it looked at properly.
What Business Owners Actually Describe
The description is nearly identical every time. The PC goes off and comes straight back to the boot logo, as though somebody had pulled the plug and pushed it back in. No blue screen, no error, no memory dump. The monitor may or may not flicker at the same moment.
Then comes the sentence that gives it away: “it mostly happens in the morning”, or “it happens when the lads are on the saw”, or “it never does it at the weekend”. Those aren’t coincidences. They are the shape of the electrical load in the building, and they are the single most useful piece of evidence you can give a technician.
Occasionally the tell is smaller: the till or card terminal beeps, a fluorescent tube in the workshop dips, or the extraction fan changes note a half-second before the screen goes black.
Why the PC Isn’t Usually at Fault
Big motors draw a very large current for a fraction of a second when they start. A compressor, a dust extractor, a large fridge or a welder can pull several times its running current on start-up. On a healthy, generously wired circuit you never notice. On an older unit — long cable runs, a shared distribution board, a bit of borrowed capacity from when the unit was subdivided — the voltage on the whole board dips for a moment while that motor gets going.
That dip is called a sag, or a brownout. It does not damage anything dramatically. It just takes the mains below what the PC’s power supply needs for a few tens of milliseconds, the supply drops its “power good” signal, and the motherboard does exactly what it is designed to do: it shuts down immediately and restarts. From the PC’s point of view the behaviour is correct. From yours, the machine rebooted for no reason.
This is why swapping the power supply so often doesn’t help. A brand-new supply behaves the same way, because it is responding correctly to a supply problem outside the case. Genuine supply failure looks different — our guide to the signs a PC power supply is failing covers the symptoms that really do point at the unit inside the case, and those are worth ruling out first.
Proving It: Event Viewer and the Timing Test
Two checks separate a building problem from a machine problem, and both are quick.
The Event Viewer check. Open Event Viewer, go to Windows Logs → System, and filter for Event ID 41, source Kernel-Power. That event means Windows was not shut down cleanly — power was removed while it was running. Note the exact timestamps of the last dozen of them.
What matters is not the presence of Kernel-Power 41 — every dirty shutdown logs it. It is the pattern. A dying component tends to produce reboots that cluster under load, or grow steadily more frequent, or come with a Machine Check Exception or a WHEA warning nearby in the log. A supply problem produces reboots at working times only, often close to round-the-hour starts, always with a completely clean log either side of them — no thermal warning, no disk error, no driver crash. Just a normal machine that suddenly stopped existing.
The timing test. Write those timestamps down and take them to whoever runs the workshop floor. Ask what starts up at those times. In our experience this conversation solves it more often than any hardware swap: the first coffee-time compressor cycle, the extraction going on at eight, the delivery shutter motor. If the timestamps line up with machinery, you have your answer without touching the PC again.
If they don’t line up with anything, the fault is more likely internal after all — the general causes are covered in our guide to a PC that keeps restarting.
A Surge Protector Is Not Power Protection
This is the part most guides skip, and it is the reason so many businesses believe they are already protected.
A surge-protected extension strip clamps voltage that goes too high — a spike, typically from lightning or a switching transient. That is a real hazard and the strips do their job. But a workshop reboot is the opposite problem: voltage going too low for a moment. A surge strip does nothing whatsoever about a sag, because there is nothing for it to clamp. You can have every socket in the unit on surge strips and still lose the office PC every time the compressor kicks in.
What answers a sag is a line-interactive UPS — one with AVR (automatic voltage regulation). It holds the output steady through small dips using its transformer taps, and switches to battery for the deeper ones. The battery runtime is almost beside the point here; you are buying the AVR and the ride-through, not an hour of working in the dark. A basic standby UPS without AVR will also cover most of it, but it switches to battery for every little dip, which wears the battery out faster in a noisy unit.
The Damage You Don’t See Until Later
The reboot is the visible symptom. The part that really hurts is what it does to your data over time, and it usually shows up weeks after the event that caused it.
Every drive holds recently written data in a small volatile cache before committing it. Databases — accounting packages, EPOS back-ends, job-management systems, stock files — also keep in-flight transactions and index structures in memory. Pull the power mid-write and you can lose a small amount of that in-flight metadata. Modern SSDs and journalling filesystems handle this well most of the time. The damage is cumulative and probabilistic: dozens of dirty shutdowns, and eventually one lands at exactly the wrong moment.
What you see later is a company file that won’t open, an EPOS system reporting a corrupt index, or a folder full of documents that open as blank pages. By then nobody connects it to a reboot from a month ago. This is exactly the situation that turns into a data recovery job, and it is entirely preventable.
Two habits blunt it considerably. First, make sure the backup is actually running and actually restorable — a shocking number are not, as our guide to silent backup failure sets out. Second, if the business genuinely cannot trade while that PC is down, plan for the outage in advance rather than during it, the way we describe in the till PC failure guide.
A Recent Job Off Bonnington Road
A small fabrication and finishing business in one of the Bonnington Road units called us out after eighteen months of intermittent reboots on the machine that ran their job sheets and invoicing. They had already fitted two power supplies and a new surge strip, and had been told by a previous visitor that the motherboard was on its way out.
Event Viewer had forty-one Kernel-Power 41 entries over six months. Nothing else in the log — no thermal events, no disk errors, no driver faults. The timestamps clustered between 07:40 and 09:20 and again just after lunch, and there was not one single entry on a Sunday. When we read the times out to the owner he identified them immediately: the compressor and the extraction rig, both on the same board as the office socket, both cycling hardest first thing and after the lunch break.
We moved the PC and monitor onto a line-interactive UPS with AVR, on its own socket, and left the laser printer and the little fan heater off it entirely. Their inrush would have tripped the UPS into overload every time somebody printed. The reboots stopped that day and have not come back.
The part worth noting is the second half of the job. Their accounts file already had index corruption from all those dirty shutdowns — it opened, but reports were quietly wrong. That took longer to sort out than the electrical side did, and it was the reason the business had been struggling to close its year. We also spec’d a replacement machine to sit better in that environment — a filtered, positive-pressure case rather than an open mesh one, because grinding dust had been packing the heatsinks solid. That side of it is a custom PC build question more than a repair one.
When It Really Has Damaged the Hardware
Repeated hard power loss usually just corrupts data. Occasionally it does physical damage — most often to the board rather than the drive. We see failed VRM components, blown MOSFETs near the CPU socket, and swollen capacitors on machines that lived on a rough supply for years. Those are board-level faults, and depending on the machine they can be a microsoldering repair rather than a replacement.
If it comes to that, get the data off first and worry about the board second. That order matters more than most people expect, and it is the same rule we apply on every dirty-shutdown job that lands on the bench.
What To Do This Week
- Pull the Kernel-Power 41 timestamps and match them against what runs on the floor. Ten minutes, no tools, and it usually settles the question.
- Get the critical PCs onto a line-interactive UPS with AVR — the till or invoicing machine, the server if there is one, and the network switch and router. A PC on a UPS behind a dead switch is not much use.
- Nothing with a heating element or a motor goes on the UPS. No laser printers, no heaters, no kettles. Their inrush will overload it and you will end up with the UPS beeping and everyone ignoring it.
- Verify the backup by restoring something, not by looking at a green tick. If it has been silently failing, a run of dirty shutdowns is exactly when you find out.
- Get an electrician to look at the board if the sags are severe. A UPS protects the PC; it does not fix an under-specified circuit that is also affecting your machinery.
- Check the database, not just the PC. If the machine has been rebooting for months, assume the accounts or EPOS data needs verifying.
We look after this sort of thing for trade units and small offices across north Edinburgh — Bonnington, Leith, Granton and out towards Meadowbank — as part of ongoing business IT support, and we cover the Lothians and Livingston for larger site visits. If a machine has already gone down and the day’s work is stuck behind it, same-day repair in Leith covers the neighbouring units on the same run.
The short version: if a PC reboots at work and behaves perfectly anywhere else, stop buying parts for it and start looking at the socket it’s plugged into.
Last updated: 6 September 2026