Random crashes to a black screen, and four reasonable troubleshooting steps taken in a sensible order. Each was more invasive than the last, and the fourth turned a machine that still booted into one that did not.
A laptop crashing to a black screen without warning. The timing was inconsistent — sometimes minutes after starting, sometimes only after hours — which made it hard to attribute to anything in particular.
The owner then did what most people do, in an order that is entirely sensible if the fault is software.
A malware scan. Reasonable and non-destructive. It found nothing. The cost was several hours of sustained reading across the entire disk, which on a failing drive is not free — but it is not damaging either.
chkdsk. Also reasonable, and it reported no problems. That result is where the diagnosis went wrong, and it deserves its own section below.
Uninstalling suspect software. Harmless, and it changed nothing — which by that point was informative in itself.
A system restore. This is the one. A restore rewrites system files, registry hives and boot configuration across a substantial portion of the disk. Performed on a drive already struggling to complete reads reliably, it is a large sustained write with a meaningful chance of not finishing.
It did not finish. The laptop froze part-way through and afterwards would not boot at all.
Worth being exact, because this is the single most useful thing on the page.
chkdsk verifies that the file system is internally consistent: that the master file table agrees with itself, that the allocation bitmap matches what the records claim, that no two files believe they own the same clusters, that the directory tree resolves. All structural questions, all answered from metadata.
It does not ask whether the drive is healthy. A disk with hundreds of weak sectors, taking many seconds of internal retrying to complete each read, will frequently pass chkdsk with a clean report — because everything chkdsk needed did eventually come back, and eventually is good enough for the question it was asking.
The tool answered correctly. The question was the wrong one.
What nobody had looked at was the drive’s own health data. A minute in any free SMART utility would have shown attribute 5 (Reallocated_Sector_Ct) elevated, attribute 197 (Current_Pending_Sector) climbing, and attribute 187 (Reported_Uncorrect) non-zero — and reframed the whole exercise, because the correct response to a failing drive is to copy the data off it, not to repair the operating system sitting on top of it.
The pattern that made this so hard to diagnose falls straight out of the mechanism.
A drive stalling on a read does not crash a machine directly. It stops responding for as long as its internal retries take, and if that exceeds the storage driver’s patience the system either resets the device or gives up. Whether that happens after five minutes or five hours depends entirely on when the system happens to touch a region containing weak sectors — which is a function of what you were doing, not of how long the machine had been on.
So the timing looked random because it was effectively random, while having a completely deterministic cause.
The drive was removed and imaged read-only. SMART showed exactly what nobody had checked: substantial reallocated sectors and a pending count still rising.
Imaging ran in graduated passes, banking the readable majority first at large block sizes and short timeouts, then returning to the weak regions with longer timeouts and reads in both directions.
From the image the NTFS structures were rebuilt, including the boot records and partition table the interrupted restore had left inconsistent — and this is where the fourth step really cost something. The restore had been writing when it stopped, so the damage was not merely the original bad sectors but half-written system structures over the top of them, in regions the file system needs to describe itself.
Everything the client needed came back, on fresh media. The laptop was written off, correctly — the drive was failing and the machine was old enough that replacing it made no sense.
The useful order is the reverse of the one followed here. Establish whether the hardware is sound. If it is not, copy the data off before anything else. Only then start repairing software.
And if a machine is crashing unpredictably, spend the minute on SMART before you spend the afternoon on a restore. If reallocated or pending counts are climbing, every repair operation you run is a large write to a disk that may not survive it. Single drives are from £300 +VAT after a free 48-hour diagnostic.
Post it or bring it in. Either way the diagnostic costs nothing and commits you to nothing, and you get one figure in writing before a single screw is turned.
Nothing can begin until the device is on the bench, which makes this the only step that needs anything from you. Pack it properly, put your details in with it, and send it over. What follows is a free diagnostic and a written figure, in that order.
Posting it? Use something tracked and insured — whatever is on the drive is worth considerably more than the postage. Bringing it in? Weekdays, 9am to 5:30pm, and it still wants packing as above for the journey.
Tell us what the device is, what it is doing, and anything already tried — an engineer will come back with a realistic view of the odds and a price band, before you commit to posting anything.
We’ll get back to you soon. Anything pressing, call 0161 871 0788.
Where the timing is inconsistent, a failing drive is a strong candidate — the system stalls on reads it cannot complete, and which regions get touched varies from session to session. Check SMART before assuming it is software.
No, and this is a common misreading. chkdsk verifies that the file system is internally consistent; it does not assess the health of the disk. A drive with a rising reallocated sector count will frequently pass while being clearly unwell.
Not if the drive might be failing. A restore rewrites system files, registry hives and boot configuration — a large write operation with a real chance of not completing on unreliable hardware. An interrupted restore commonly leaves a machine unbootable.
Check SMART first. If the counts are healthy, software troubleshooting is reasonable. If they are climbing, copy your files off to another drive, most important first, and stop using the machine. Repairing the operating system on a dying disk is the wrong order.
Almost never. An interrupted restore damages system files and boot structures, not your documents. The drive is imaged and the file system rebuilt from the copy, and personal data usually recovers completely.
From £300 plus VAT for a single laptop drive, fixed in writing after a free 48-hour diagnostic, with most jobs no fix, no fee.
Kick off with an instant online quote, or ring us and talk it through first. Either way you’ll know a clear, fixed price before any work starts.