The history here is the diagnosis; it just needed reading in order. A nearly new WD Elements 2TB began "making noises and the light is still on" — cables were swapped, nothing improved — and over time the owner watched "the function gradually decline and then unexpectedly stop". The suspicion landed on a connection problem, which is where suspicion usually lands, but connections have an alibi built into the story's shape: they fail like switches — working, then not, cable-position-dependent, instantly better when reseated. What fails like a fade — noisier, slower, less reliable, then done — is machinery. Gradual decline is the mechanical signature; a connection can't decay week by week, but a head assembly can, and its noises were the running commentary.
| Media | A nearly new WD Elements 2TB external — grew noisy with the light still on, function declining gradually before stopping outright; cables already eliminated. |
| Reported situation | WD Elements 2TB · noises with normal light · leads swapped without effect · gradual functional decline · then sudden stop · contents wanted. |
| Fault class | Progressive mechanical failure — a degrading head assembly or motor growing audibly worse until it could no longer complete start-up. Not a connection: the decline curve and the failed cable-swaps both acquit it. Platters ordinarily intact; reached by clean-air repair. |
| Equipment used | Not run further; out of the enclosure and profiled once on a current-limited bench supply, the noise localised toward heads or motor; laminar-flow open — heads, motor and platters examined under the stereo microscope, the degraded assembly replaced with matched donors or the mechanism relieved; alignment set; DeepSpar behind PC-3000 UDMA imaging; contents located in R-Studio and verified onto fresh media. |
Failure modes have shapes, and the shapes rarely lie. Electrical connections are binary creatures: they conduct or they don't, and their faults flicker with touch and cable position. This owner tested that theory properly — leads swapped, no change — and the theory failed. The story's actual shape, noise arriving first and function eroding over weeks before a final stop, is the shape of machinery wearing out in public: a head assembly losing precision read by read, retrying more and more (the noise and the slowdown are the same fact, heard and felt), until one start-up asked more than the mechanics had left.
'Nearly new' stings, and deserves an honest word. Mechanical infant mortality is real: manufacturing tolerances at the nanometre scale mean a small fraction of drives carry defects that surface in months rather than years. Warranty will cheerfully replace the hardware — manufacturers are efficient about that — but replacement is all warranty does; the data on the failing original was always the owner's problem alone. A nearly new drive also fails with one advantage worth taking: young platters, barely worn, that image beautifully once the mechanics above them are made whole.
The weeks of decline shape the bench work in one respect: a drive that ran noisy for a while gets its surfaces inspected with extra care, because prolonged mechanical distress can mark platters even without a single dramatic event. Surfaces confirmed sound — the usual finding — the degraded assembly is exchanged for matched donors in filtered air, and the drive is imaged before its luck is asked another question.
One current-limited profile to place the noise — heads or motor — then no further running. Under laminar flow, the platters take a careful survey under the stereo microscope, given the weeks of audible decline; the worn assembly is replaced with matched donors (or the mechanism relieved where that's the whole story), alignment set. Imaged on DeepSpar behind PC-3000 UDMA, walked in R-Studio, and the contents verified onto fresh media.
The decline halted at the bench rather than at zero — mechanics renewed, young platters imaged, data back. Assessment is free and reported inside two working days of arrival, then one fixed figure in writing: clean-air work, so half up front, balance only on proven recovery. Warranty can have the hardware afterwards with our compliments. It was never going to save the part that mattered.
A decline is a countdown, and the merciful thing about countdowns is that you can act during them. New noise from a drive is the day to copy everything off — not the month, the day — because 'gradually worse' has 'unexpectedly stopped' waiting at the end, as this owner found. Already stopped? Then the copying window has closed and the preservation window is open: no more power-ups, no cable-shuffling beyond what you've done, box it and send it. And log the lesson nearly-new hardware teaches harshly: age was never the thing to trust. Copies were.
Every case file on this site is written up from a genuine enquiry handled for customers across York and North Yorkshire, anonymised to keep client confidence. Each one sets out the diagnostic reasoning and the recovery method our engineers apply to that fault, on the equipment named.
Two machines that can't see it have ruled themselves out — the drive is read at its own level instead
Years on a shelf stiffen the mechanism, not the recording — freed in clean air, not spun until it gives in
Five machines can't all be wrong — power without presentation points at firmware or board, read past directly
Free diagnostic inside 2 working days of arrival, one fixed quote, no fix no fee on logical faults — start online or ring the freephone.