Case file · Logical & corruption · MHD-2025-6311
One Cyclic Redundancy Check After Another.
For years the Toshiba had sat in a USB dock in the yard office. A power cut changed its manner: any sizeable copy would halt on the same three words — cyclic redundancy check
— odd files first, then whole folders. He tried a different cable, then the office laptop, and set the Windows check going twice, and it ran all night both times
. Behind all that sat the working memory of a boatbuilder on the Thames above Maidenhead: every drawing and every cut list from thirty years of boats
.
Much the same fault? Call us.
0800 6890668
In everyday terms.
The wording sounds like something a program did, and that is precisely how it misleads people. A cyclic redundancy check is really a receipt — the disk gave something back, and it no longer matched the figure filed next to it. A filesystem in disarray on a healthy disk produces that message. So does a surface beginning to lose its hold on what was written there. The two faults need opposite handling, and picking wrongly carries a cost. Windows will interrogate a weak patch until either it yields or the drive does, and two overnight checks were two nights of exactly that.
What the bench used here.
How a job is handled →| Tool | Why we used it | What it brings |
|---|---|---|
| DeepSpar Disk Imager 4 | Imaged the Toshiba on a fixed retry budget, weak zones noted and left for later | Head map first, then head-by-head reads, holding resets, timeouts and the power line steady |
| R-Studio Technician | Every repair was carried out on the image, not the disk | Its file-system coverage is wide and its array rebuilds hold up |
| ACE Lab PC-3000 Express | Interrogated the firmware first: attribute history, the grown defect table, and the long power-on sessions logged | Reaches the vendor-specific commands that ordinary ATA has no way of issuing |
The lab work.
Ask the disk's own counters what is wrong
No sector was read until the disk had been interrogated at firmware level, and what came back was unambiguous. The reallocation count was high and still climbing, with a further queue waiting to be moved. Two vast power-on sessions sat in the log, lining up with the nights he had left the Windows check running. That is a surface failing, not a filesystem tied in knots.
Treat the retries as money and spend them well
The rules for the pass were settled before it began: timeouts kept short, a firm ceiling on attempts, anything obstinate written down and walked past rather than argued with. The awkward zones were revisited briefly at the very end. It is an arithmetic question above all. Forcing one difficult sector into submission can lose you the thousand sound ones waiting behind it.
Leave the disk powered down and use the copy
With the image taken, the Toshiba never received power again. On the copy the picture improved: the directory tree had come through intact, and what was lost sat in the band where the weak zone ran across a set of scanned sheets. Every casualty was recorded by filename — an itemised list of drawings rather than a general apology — so he could see exactly which sheets the surface had held back.
How the job closed.
On the phone he had named the folder that mattered most — steelwork drawings for hulls still in the water — and that folder was checked and opened before money was discussed. The archive went home on a new disk, whole apart from the scans on the list. Two of those he can draw again from paper in the yard; the rest belong to boats built before his time, whose paper went years ago.
More of the jobs we've closed.
Also on the Logical & corruption shelf.
Does any of that match your case?
Nothing needs deciding until the diagnosis is back: switch the device off, send it in, and let the findings settle it.