Logical & corruption · bench log · HUL-2025-0932
Every Name Was There; Every Size Was Zero.
She took the stick out of a laptop in Scunthorpe partway through a save — I never waited for it, I just pulled it out
— and gave it no further thought that day. Next morning the folder still held everything: the portfolio, both assignments, each name exactly where she had left it, each size showing 0 bytes
, the type column empty, every entry down as a bare File
. She tried opening a few of them. Her ward mentor then told her to leave it alone and get somebody to look
.
Sounds like yours? Ring us.
0800 6890668
What causes it.
A FAT volume stores its catalogue away from the goods. One short table carries each file’s name, its type, its length and the cluster where it starts. The data itself sits further along the chip and cares nothing about whether that table is right. Interrupt the table mid-update and the entry ends up half written: length zeroed, type missing. Meanwhile her coursework is still in its clusters, exactly as written, untouched. Windows reads out what the table tells it, without argument. A full folder is therefore reported as empty.
The tools this one needed.
The steps we follow →| The gear | What this one did | Why we have it |
|---|---|---|
| PC-3000 Flash | Read the chip sector by sector, taking nothing on trust from the filesystem | Goes round the controller and reads the NAND itself, matched to a maker-ID library kept current |
| R-Studio Technician | Restored broken entries by checking them against FAT’s second copy | Handles almost any file system, and puts RAID sets back together |
| UFS Explorer Professional Recovery | Signature carving for the files whose entries were past saving | Opens the volume formats most tools cannot: APFS, ReFS, XFS, ZFS, Btrfs |
The stages.
Image the chip before touching anything else
Power a flash stick up and its controller begins housekeeping of its own — wear levelling, garbage collection — with no interest in which blocks a recovery needs. So the opening move was a flat read, sector by sector, onto bench disks. After that copy existed, nothing we tried could cost her a file.
Work from whichever table copy survived better
The allocation table is kept twice on FAT, and the interrupted write had treated one copy worse than the other. Where the healthier one lined up with a directory entry still intact, the record went back whole: name, length, and every cluster in the chain. Files returned like that are the originals, not something built to look like them.
The remainder had to name themselves
A handful of entries were beyond repair, so those files had to identify themselves. We look for a Word or PDF header in the raw clusters, join the bytes that follow it back in sequence, and put a name on the result only if the header, or the first legible lines of text, justify one.
How it finished.
Most of the portfolio came back under the names it started with. A smaller pile was carved, so it arrived with names generated by software, and she spent an evening working through those. The lot travelled home on a new stick, with a single instruction: wait for the light before you take it out.
What people read after this.
More cases of corruption and deletion.
Is yours doing the same?
Turn it off. Send it to us. The diagnosis comes first. It says which files are still readable and which are not.