Data Recovery Case File · Desktop Externals & Aging Drives · When the Driver Gives Up
External Drive That Crashes Windows When Accessed
Her enquiry described a symptom that people usually blame on the wrong machine. A Toshiba external hard drive: "it has stopped working, causing the laptop to have a Windows error and to restart when I try to access the hard drive. Believe the hard drive is mechanically broken. Please could you provide a quote for the worst case scenario on retrieving the data." Her instinct is right and worth confirming: the crash is the drive's symptom, not the computer's fault. A stop error naming a system thread is Windows reporting that something it depends on stopped responding — and when that happens the moment a particular drive is accessed, the drive is what stopped responding. This page explains the mechanism, why it makes the situation more urgent rather than less, and gives her the honest worst case she asked for.
| Media | Toshiba external hard drive — host operating system halts with a stop error and restarts whenever the drive is accessed; contents required |
| Reported situation | Drive non-functional · host crashes on access attempts rather than reporting an error · owner suspecting mechanical failure · worst-case quotation requested |
| Fault class | Drive stalling on reads long enough to hang the storage subsystem — indicative of failing media or mechanism; repeated crash cycles applying uncontrolled load |
| Equipment used | Assessment on diagnostic instruments rather than a host computer · Atola Insight Forensic head and surface evaluation · PC-3000 Express with Data Extractor prioritised imaging under per-sector timeouts · losses reported per file |
The decode: why a drive crashes a computer, and what that implies
What is actually happening: when software asks for a file, the request passes to a storage driver, which asks the drive and waits. On a healthy drive the answer arrives in milliseconds. On a failing one the drive retries internally and may never answer at all — and the driver, waiting inside the operating system's core, blocks. Windows monitors for exactly that condition, because a core component that stops responding cannot safely be left, and it halts the system with a stop error rather than continuing in an unknown state. So the crash is not Windows being fragile. It is Windows detecting that the storage subsystem has stalled, and the stall is the drive.
Why this points at the media or the mechanism: a purely logical fault — a damaged filesystem over a healthy drive — does not do this. Reads still complete; the data returned may be nonsense, but it arrives, and the operating system reports an ordinary error instead of halting. A drive that hangs the host is one that fails to answer, which means failing sectors, a struggling head, or a mechanism that cannot complete its reads. Her instinct about mechanical trouble is well founded.
Why the crashes make it more urgent: each cycle is worse than an ordinary failed access. The drive spends the whole stall period retrying the same unreadable region — seconds or tens of seconds of a head passing repeatedly over the weakest area — and then the machine restarts, which power-cycles the drive abruptly. Repeat that a few times and it is a meaningful amount of punishment concentrated on exactly the ground that is failing. The instruction is therefore firm: stop connecting it to a computer at all. There is nothing further to learn from another crash.
The worst case, honestly: she asked for it, so it should be given rather than softened. The worst case for a drive that hangs its host is extensive media damage or a failing head, where imaging returns a substantial part of the disk and some regions never yield — and where the honest deliverable is a per-file map of what came back complete, partial and not at all rather than a percentage. The realistic case, which is more common, is bounded weak regions where patient re-reading under timeout control recovers most of it. What can be promised in advance is the method and the reporting, not the proportion.
On the bench
The drive was never connected to a general-purpose computer. Assessment ran on the Atola Insight Forensic, where a drive that fails to answer stalls nothing and damages nothing further — head condition, surface behaviour and error patterns read directly, turning her crashes into a measured map of where the drive still reads. Imaging followed on the PC-3000 Express under Data Extractor with per-sector timeouts enforced, which is the decisive difference from what her laptop was doing: a failing sector is given a strict budget and deferred rather than retried until something breaks. The healthy expanse was captured first and completely, weak neighbourhoods revisited on later passes, and every unreadable region resolved to the files it affected.
The outcome
The drive assessed and imaged on instruments that tolerate what a computer cannot, with the losses reported per file. Free assessment, one fixed written figure including VAT; where a drive has to be opened, 50% of parts and labour is payable upfront with the balance only on success — otherwise no recovery, no fee. The decode, for anyone whose drive crashes their computer: the stop error is the operating system detecting that its storage driver has stalled waiting for a drive that will not answer — so the crash is the drive's symptom rather than the computer's fault, and it points at failing media or mechanism rather than a damaged filesystem, because logical faults return errors instead of hanging; each crash concentrates punishment on the weakest region and then power-cycles the drive abruptly, so stop connecting it; and ask for the result as a per-file map rather than a percentage.
Drive that blue-screens your computer
Stop plugging it in — that is the whole of the advice and it matters more than it sounds. The crash is your computer detecting that its storage driver has stalled waiting for a drive that never answered, so the fault is the drive rather than Windows, and it points at failing media or a struggling mechanism rather than a damaged filesystem, because logical problems return errors instead of hanging the machine. Each crash is worse than an ordinary failed read: the drive spends the entire stall retrying the same unreadable area, and then the restart power-cycles it abruptly. A few of those concentrate real punishment on exactly the ground that is already failing. Don't try it on a second computer to see whether that one copes, don't run repair utilities, and don't leave it connected hoping it settles. Ask for the recovery to be reported per file rather than as a percentage.
Don't connect it again — call Glasgow Data Recovery on 0141 404 0294; assessed on instruments that tolerate a stalled drive, imaged under per-sector timeouts, losses mapped file by file.
Request a quote online →
Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.