Flash & SSD · job logged · CDF-2025-0939
Plugged In, and Not a Flicker Anywhere.
The stick worked at one pond and was scrap by the next. An ecology practice outside Bridgend went round the whole office with it — three laptops, two desktops, and a hub with a power supply of its own — and nothing even notices it going in
. No chime, no entry in Disk Management, and it might as well be a paperclip
. On it sat the photographs from a whole season of newt surveys.
Does that match yours? Call us.
0800 6890668
Why that happens.
Five computers and a hub between them, and not a reaction from any: that acquits the machines and locates the fault inside the casing. Two components in there are capable of it. Under the metal connector are solder pads, flexed a little on every insertion; alongside them sits the controller, which speaks for the memory in all dealings with a host. Neither part keeps a photograph. Both merely stand in the way of them. Flash keeps its contents whether or not anything survives that can ask for them, and that is where a lab begins.
What we used on this one.
Every step, in order →| The tools | What it did here | Why it helps |
|---|---|---|
| PC-3000 Flash | Went straight at the bare package, with no controller anywhere in circuit | Bypasses the controller to read the NAND itself, against a manufacturer-ID library kept up to date |
| Rusolut Visual NAND Reconstructor | Turned the raw dump back into a logical image: ECC, XOR, page and block order | Turns a raw NAND dump into something readable: ECC, XOR, page order, reassembly |
| IR preheater & BGA rework station | Took the BGA off without damage and dressed it for the socket adapter | Enough heat to lift the NAND and reball it, but never enough to lose the data |
How it went.
Start with the cheap fault
First job was the stereo microscope, because a hairline crack in the joints behind the connector is an hour at the bench with a fine iron and no worse. Nothing was cracked. Every track ran clean from end to end. With power on it, the controller pulled current while its data lines said nothing at all, which confirmed what the office had already found.
Go to the memory, leave the controller where it lies
So the package came off, had its pads cleaned up, and was dumped raw through a socket reader. The rest of the job stands on one fact: nothing leaves with a controller when it fails. Whatever was written is still sitting in the silicon, arranged however the chip chose to arrange it at the time.
Unpick the controller's layout and the folders return
Rusolut pulled the dump apart and built it back up: error correction applied, the XOR pattern worked out, pages and blocks put back into the order they belonged in, until the layout the dead controller once served up existed again. Up came a filesystem, with the survey folders in it.
Signing it off.
The set went back on fresh media, three frames short, and in time for the licence application. Those three all came from the final pond visit, all three came out damaged, and the report says so.
Pages people often read next.
Next to SSD & flash.
Is that what yours is doing?
Power it down, get it to us, and hold off on any decision until the diagnosis says what can still be read.