SSD & flash · case notes · EDR-2025-2918
Writes Failed First, Reads Followed Weeks Later.
The decline ran to a script. First the parish-hall laptop would not copy the new scans across
; weeks later, reading failed as well; now it produces only the bubble that says USB device not recognized
. And on the stick, nowhere else, lived a history society’s archive — photographed registers, magazine scans, interview transcripts — from a village out on the Somerset Levels.
Sound familiar? Talk to us.
0800 6890668
In plain English.
Saving is what wears flash out. A write taxes a memory cell far harder than a read ever does, so losing saving first and reading second is failure arriving in its natural order. The controller shuffles tired blocks for as long as its mapping tables still have somewhere sound to live; when that runs out, it stops presenting a device at all. The laptop’s ‘not recognized’ bubble reports that silence — and it carries no verdict whatever on whether the scans still sit safe in the chips behind it.
The tools this job took.
See how your case runs →| System | The part it played | Why we rate it |
|---|---|---|
| PC-3000 Flash | Identified the bare memory chip and read it directly, controller left out | Reads the memory chips themselves, matching each one against an up-to-date identification library |
| Rusolut Visual NAND Reconstructor | Unpicked the XOR masking and page ordering the dead controller used to apply | Rebuilds files from a raw NAND dump — ECC, XOR, page transforms and reassembly |
| R-Studio Technician | Rebuilt the FAT volume so every file kept the name volunteers gave it | Handles most filesystems; dependable RAID reassembly |
On the bench.
Decide which of the two parts went silent
With USB a dead end, the case opened literally — casing off, chip on the bench. The memory answered a direct probe and matched its database entry first time; the component no longer finishing its boot was the controller. Writes failing ahead of reads had predicted precisely that split.
Dump the chips raw, then undo the housekeeping
What comes off bare NAND is nothing like a filesystem: error-correction codes woven through the data, every page XOR-masked by the controller, the ordering set by a translation map that died with it. Strip the ECC, reverse the masking, rebuild the sequence — only then did anything file-shaped show.
Reassemble the volume; open every last file
Over the corrected dump the FAT volume went back together, and the archive surfaced under the filenames its volunteers had typed. Sign-off was a page-turning exercise — registers, magazine issues, transcripts, opened one after another — because an unopenable file counts as nothing recovered.
How it ended.
The archive went home complete on new media; the society’s next minutes record a new rule — three copies, one in another building. The worn-out stick stayed on as an exhibit.
More like this in the casebook.
Elsewhere in SSD & flash.
Sound like your drive?
Every case here points the same way: power the device down, take the free diagnosis, then make your decision.