The drive "is not readable", and before it came to us an IT department had "tried it, briefly, with Linux, which suggested a hardware fault". That is a more useful starting point than it might sound, because it eliminates the entire category of problem that software can address. Linux is a good instrument for this: it will attach a drive that Windows has given up on, and its logs report read errors, resets and timeouts in plain language rather than shrugging. When it too reports the drive misbehaving at the hardware level, the argument is essentially over. What remains is to work out which piece of hardware, and there are only three candidates worth considering.
| Media | A WD Elements 5TB desktop external drive that will not read, previously tested briefly under Linux by an IT department whose conclusion was a hardware fault. |
| Reported situation | WD Elements 5TB · not readable · briefly tested with Linux by an IT department · testing indicated a hardware fault rather than a software one · recovery sought. |
| Fault class | Hardware, to be narrowed — mechanical, firmware or board-level. The recorded data is generally intact and is reached with equipment that addresses the drive beneath its normal interface, once the specific fault has been identified. |
| Equipment used | Read on a PC-3000 UDMA through a hardware write-blocker and the hardware fault narrowed — firmware and service area, head stack and motor, and the board each assessed in turn · the specific fault then addressed: firmware repair, clean-air head work under laminar flow, or a matched donor board carrying this drive's own ROM · imaged using DeepSpar, the filesystem read out in R-Studio and the data validated. |
Give credit where it is due: an IT department that reaches for Linux before reaching for the phone has done the diagnostic step most people skip. Windows tends to hide a struggling drive behind a spinning cursor and a vague message, while a Linux system will attach whatever it can and log what actually happened — read errors at specific offsets, bus resets, timeouts, a device dropping off and reappearing. Those are hardware symptoms, and they are not curable by any amount of software.
So the field narrows to three. It may be firmware: the drive's own management tables or service area damaged, so it cannot describe itself properly or cannot translate addresses. It may be the head stack: reading unreliably or not at all, in which case there is usually a noise to go with it. Or it may be the board, including the bridge electronics inside the enclosure, which on a five-terabyte desktop unit is a real possibility in its own right.
Separating them takes equipment that speaks to the drive at its own level rather than through the operating system, which is exactly what the previous testing could not do. That is the first hour of the job. It is also the whole of the free diagnostic, which is finished inside 2 working days of the drive arriving and ends in one fixed written figure — because the price of a firmware repair and the price of clean-air head work are not remotely the same number, and quoting before knowing which is guesswork with somebody else's money.
It was read on PC-3000 UDMA with a write-blocker wired in line, which allows the drive to be interrogated at firmware level even when it will not present itself normally. Service area, head stack and board were each assessed in turn rather than by elimination, and the specific fault addressed on its own terms. The drive was then imaged out to DeepSpar, the filesystem worked through in R-Studio from that image, and the recovered data checked before it was written out to fresh media.
Hardware fault identified, addressed and the drive imaged, with the data written out to fresh media. The diagnostic cost nothing; the figure that followed was fixed, written and inclusive of VAT. Where the fault resolves at firmware level the work runs on no fix, no fee; where it requires opening the drive or rebuilding the board, half the quote is payable in advance against parts and bench time with the balance due only on success. Which of those applied was established and stated before any money was discussed.
If someone technical has already tested it and concluded the fault is hardware, believe them and stop testing. Further attempts on a drive that is throwing read errors and bus resets do not add information, and on a mechanical fault they actively subtract data. Two things worth passing on to whoever looks at it next: what the logs actually said, because specific errors at specific offsets narrow the field usefully, and whether the drive makes any noise it did not make before. And do not let anybody run a filesystem check or a repair pass in the meantime, however tempting — that writes to a drive that is already struggling to read, which is the least useful direction to be pushing in.
Every case file here is written up from a real enquiry handled for households and businesses across Norwich and Norfolk, anonymised so that nobody is identifiable. Each sets out the reasoning behind the diagnosis and the method used on that particular fault, with the equipment named.
A general repair shop has general tools; below a certain fault they run out, and the fee should have run out with them
Three clicks and silence is a drive that has counted its attempts — and heads wear out without ever being dropped
A glued shell is a five-minute obstacle, not a reason to send somebody away with their photographs still inside
Diagnosis costs nothing and takes up to 2 working days from arrival, the quote is fixed in writing, and logical work runs no fix, no fee — start online or ring the freephone.