An accountant in Croydon noticed his backup drive making a noise in the second week of January, which is the worst possible fortnight for an accountant to stop and deal with anything. He carried on opening client working papers off it for another two weeks. It got slower, then it got noisier, and then it stopped opening at all. I wish I had stopped when I first heard it, he wrote, which is the most common sentence in this trade.
| Media | An external hard drive used for client working papers that clicked intermittently for about a fortnight while still in use, and then became completely inaccessible. |
| What was reported | External hard drive, clicking noise noticed and used for a further two weeks, gradually slower, eventually cannot be accessed at all, used on Windows, client working papers needed, owner regrets not stopping sooner. |
| Fault class | Mechanical. The heads had begun failing to read, which was the clicking; two weeks of continued use widened the damage until the drive could not ready at all. Two surfaces showed light marking, which set a limit on what could be recovered. |
| On the bench | Not run on arrival, profiled once on a current-limited bench supply, opened in clean air under laminar flow, heads and all surfaces inspected under a stereo microscope with the marked areas mapped, matched donor heads fitted and alignment set, imaged on DeepSpar behind PC-3000 UDMA with the marked regions deferred and approached last on a restricted head map, the volume parsed in R-Studio and documents validated by opening them. |
Clicking is not a noise a drive makes while working; it is a noise it makes while failing to. The heads cannot resolve their position, so they retract and try again. Each cycle is an attempt, and attempts have a cost when the head that is attempting is damaged.
The fortnight is the part that is actually expensive. A drive that clicks and is switched off usually has clean surfaces. A drive that clicks for two weeks under daily use has heads dragging over the magnetic layer hundreds of times a day, and eventually some of that layer is polished away. The data in a polished region is not recoverable by anyone.
Getting slower was the middle of the story, not a separate problem. Retries take time. A drive that has begun to struggle presents as a drive that has become slow, and the slowness is the sound of it re-reading things it could read first time a month earlier.
Marked surfaces make a job partial rather than impossible. Damage on a platter is regional. Everything outside the marked bands still reads normally once a working head stack is fitted, and whether that includes the files you need is a question of luck and layout. It is established by imaging, not by guessing, and the honest answer is given before the invoice.
It was not run. A single profiled start established that the drive was no longer reaching ready at all, and it was opened under laminar flow rather than asked again. Under magnification the head stack was worn through and two of the four surfaces carried light circumferential marking, which was mapped before anything was replaced.
Matched donor heads were fitted and alignment set. Imaging on DeepSpar behind PC-3000 UDMA took the undamaged surfaces first and the marked bands last, on a restricted head map at reduced speed, so that the majority was secured before the difficult part was attempted. R-Studio parsed the volume from the image and the recovered documents were opened rather than counted.
The great majority came back, including every current client file. What did not come back sat inside one of the marked bands and turned out to be an archive of returns from three and four years earlier, most of which existed elsewhere. He was told what was missing by name rather than by percentage.
One drive, £300 + VAT. This was mechanical work with the drive opened in clean air, so it fell outside no fix, no fee: 50% upfront, balance on a confirmed result, and the shortfall stated plainly before the invoice went out.
Switch it off then, not when it becomes inconvenient. The difference between a drive stopped at the first click and the same drive stopped a fortnight later is often the difference between everything and most of it. Do not copy files off it in the meantime, however urgent they are, because each read is another pass of a damaged head across the surface. Do not run a scan or a check, which is worse: those read the whole disk. And do not put it in the freezer. The advice circulates endlessly, it comes from a class of fault that barely exists any more, and condensation inside a drive is a genuine and additional problem. Unplug it, pack it, send it, and take the diagnostic before deciding anything.
These files are written up from genuine enquiries taken from households and businesses in Bracknell, across Berkshire and Surrey and into London, with names, firms and anything else identifying stripped out. Each one sets down how the fault was reasoned through, the method it actually needed and the equipment it was worked on. One of them is a job that was turned away, which belongs here as much as the rest of them.
A drive that announces itself to the port but never appears has a mechanism stuck short of ready
A fall damages heads mechanically whether the drive was running or not, and the clicking is the proof
A lit indicator proves power and nothing else, and a shop that has already bypassed the caddy has narrowed the job usefully
Examining it is free and takes 2 working days from arrival, the figure goes to you in writing and does not move afterwards, and a logical fault carries no fix, no fee. Start online or ring the freephone.