Some faults walk through the door with a perfectly reasonable explanation. A dropped laptop with a cracked screen, a flooded keyboard, a machine so clogged with dust it could stuff a pillow. You look at it and you know. Then there are the ones like Margaret's.

Margaret — not her real name, but the kind of name that fits — ran a small bookkeeping practice from home. She'd had her Dell laptop for about four years, used it every day, and had never had a moment's bother with it. Until, one Monday morning in February, it crashed before she'd even finished her first cup of tea.

She restarted it. It crashed again. She waited, tried once more, and it crashed a third time — this time with a blue screen she didn't catch fast enough to read. By half nine it was running normally. She wrote the whole episode off as "one of those Windows things" and got on with her day.

Except it happened again on Tuesday. And Wednesday. Always before nine in the morning. Always fine by mid-morning. Always completely normal for the rest of the day.

What She Thought Was Wrong

By the time she brought it in, Margaret had a theory: a virus. She'd read somewhere that malware could be triggered by scheduled tasks, and this certainly felt scheduled. She'd also wondered whether a Windows Update had broken something, because — in her words — "they always seem to break something."

Both are reasonable guesses. If you described the symptoms to most people, they'd probably land in the same place. A fault that only shows up at a specific time of day, then vanishes? That does sound like software.

It wasn't software.

What Was Actually Going On

The blue screen error, once we caught it properly by connecting an external monitor and running the machine through a cold start, pointed squarely at a RAM fault. Specifically: a memory error under low-temperature conditions.

Here's the thing about RAM that most people — quite reasonably — don't know. Solid-state components, including memory chips, can behave differently when they're cold. In most laptops this is never a problem; the components are well within tolerance and warm up in seconds. But occasionally, particularly in older machines where the RAM has been through thousands of thermal cycles of heating and cooling, a memory module starts to develop marginal stability. It works fine when warm. When cold — say, first thing on a February morning in a home office that's been unheated overnight — it throws errors.

Margaret's laptop was sitting in a room that dropped to around 10–12°C overnight. By the time she sat down to work, the machine was cold to the touch. That first twenty or thirty minutes of use, while everything warmed up to normal operating temperature, was exactly the window in which her RAM was unreliable. Once it was warm, it tested fine. Every single time.

This is why the fault was so consistent and so baffling. It wasn't a virus. It wasn't Windows. It was physics.

How We Diagnosed It

Diagnosis started, as it almost always does with laptop repairs like this, with the basics: running a full memory test using MemTest86, which boots independently of Windows and hammers the RAM with read/write tests. The first run — done at room temperature in the workshop — came back clean. That was actually a useful result, not a frustrating one, because it told us the fault was intermittent.

The next step was deliberate: we left the laptop powered off in the coolest part of the workshop overnight, then ran MemTest86 again first thing in the morning before the room had warmed up. This time, errors appeared within the first few minutes — dozens of them, clustered in a specific address range that pointed to one of the two RAM modules.

We pulled that module, ran the test again on the single remaining stick. Clean. We reseated the suspect module, ran it again cold. Errors straight away. The culprit was confirmed.

The Fix — and the Cost

The failed module was an 8GB DDR4 stick, a completely standard size. A compatible replacement cost under £20. Fitting it took about ten minutes once we had the right part. Total bill including diagnosis: modest enough that Margaret admitted she'd been quietly bracing herself for far worse.

We also ran a full health check on the remaining original stick and the SSD while the machine was open, because it would have been daft not to. Both were fine. The laptop left in better shape than it had arrived, and Margaret reported the following week that her mornings had become considerably less eventful.

The Practical Lesson

There are a few things worth taking away from this one.

  • Intermittent faults are the hardest to diagnose remotely — if your computer only misbehaves under specific conditions, it's worth trying to describe those conditions as precisely as possible. "It crashes" is hard to work with. "It crashes every morning before half nine, then runs fine" is a proper clue.
  • Temperature matters more than most people realise — computers in cold rooms, garages, or outbuildings can behave in ways that look like software problems but are actually hardware ones. If your machine is noticeably cold when you start it, that's worth mentioning.
  • Blue screen errors are not all the same — they look alarming, but they usually contain a specific error code that points toward the cause. If you can photograph the screen before the machine restarts, that information is genuinely useful.
  • RAM faults don't always show up on the first test — a clean memory test result doesn't always mean the RAM is healthy. A good technician knows to reproduce the conditions of the fault before ruling things out.
  • A repair isn't always expensive — the assumption that a persistent fault means a big bill puts people off getting things looked at. Sometimes the actual fix is a £20 part. The diagnostic work is where the skill is, not necessarily the component cost.

Margaret's case is a good reminder that computers are physical objects, not just software running in mid-air. They get cold. They age. Their components develop quirks. And sometimes the most baffling fault has the most straightforward explanation — once you know where to look.

If your machine is doing something odd and you can't work out why, a good description of when it happens is often more useful than a description of what it does. The "when" is where the diagnosis usually starts.