Camera cards · lab notes · HUL-2025-0911
The Crack Stopped Short of the Memory.
Twelve months of checking the forecast came good inside one week — two nights, no moon, stars right down to the horizon
— with not a frame of it backed up yet. Then the card stuck fast in the reader. Someone in York went at the slot with a screwdriver, and the card came off worse: a lump of the casing broke away beside the gold contacts
. That piece rolled under a desk and stayed lost. Ever since, no reader will accept that there is anything in the slot
.
Sounds like yours? Ring us.
0800 6890668
What causes it.
There is not much card in an SD card. Most of the thing is moulded plastic wrapped round a single small block — the controller and the memory sealed into one piece of resin — and the only connection from that block to the gold pads is a few fine traces that run along the exact corner he was levering against. Break that corner and the wiring is what goes, not the storage. Picture a shop whose shutter has jammed: everything is still on the shelves, but the front is shut. Losing the fragment cost us nothing. Under magnification a fracture shows how far it travelled, and this one had stopped a long way from the memory.
The tools this one needed.
The steps we follow →| The gear | What this one did | Why we have it |
|---|---|---|
| Rusolut Visual NAND Reconstructor | Applied error correction, then put the dumped pages back into sequence | Makes sense of a raw NAND dump: ECC, XOR, page order, then reassembly |
| IR preheater & BGA rework station | Low, controlled heat, held only long enough to lift the coating off the pads | Warm enough to take the NAND off and reball it, never warm enough to risk the data |
| PC-3000 Flash | Named the monolith from the ID library and mapped the pinout, all before it saw any power | Goes round the controller and reads the NAND itself, matched to a maker-ID library kept current |
The stages.
The fracture told the story
With the fragment gone for good, the card had to account for itself. One minute under the stereo microscope was enough: part of the contact block missing, traces stopping where it had been, and a crack that faded to nothing well before the die. See that, and a card that looks ruined turns into ordinary chip-level work.
A back door left by the factory
Flip a monolith over and there is a row of service pads on the back. The factory uses them for testing. Nobody publishes the layout and no two manufacturers agree on it, so this one was probed until the map came clear. Then wire thinner than a hair goes onto the few pads that matter. The memory answered along a path the break had never touched.
Raw pages first, pictures later
A chip gives up its pages in whatever order the controller stored them, and nothing else. Error correction comes first. Then the XOR mask has to be stripped, the pages sorted back into their real order, and a filesystem built over the whole lot before a single folder name appears. Bridging that gap is the work, and no reader you can buy will do it.
How it finished.
Both nights came back whole — every frame opened and checked, and the calibration files with them. The card went in the bin, and was not missed.
What people read after this.
More cases under Memory cards.
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.