Camera cards · bench notes · PLY-2025-0911
The Crack Never Reached the Memory.
A year of watching forecasts paid off inside a single week — two clear nights, no moon, stars down to the horizon
— and not a frame of it had been copied anywhere. Then the reader took hold of the card and would not let go. Somebody in Penzance attacked the slot with a screwdriver, and it was the card that suffered: a chunk of the casing snapped off next to the gold contacts
. The broken piece went under a desk and was never seen again. Since then, no reader will admit there is anything in the slot at all
.
Sounds like yours? Talk it through.
0800 6890668
Why it happens.
An SD card is mostly not card at all. Almost the whole of it is moulded plastic around a single small block — controller and memory sealed together in resin — and that block reaches the gold pads through a handful of fine traces running along precisely the corner he had been levering. Snap the corner and it is the wiring that goes, never the storage. Think of a shop whose shutter has stuck: the stock is all still on the shelves, but the front will not open. Losing the fragment cost us nothing at all. Magnification shows how far a fracture ran, and this one had given up well short of the die.
The kit this job called for.
How the job runs →| The kit | What it did here | Why we keep it |
|---|---|---|
| Rusolut Visual NAND Reconstructor | Ran error correction, then returned the dumped pages to their proper sequence | Turns a raw NAND dump back into data: ECC, XOR, page order, then reassembly |
| IR preheater & BGA rework station | Gentle, controlled heat, held just long enough to clear the coating off the pads | Hot enough to lift the NAND and reball it, and no hotter than the data will stand |
| PC-3000 Flash | Identified the monolith from our ID library, then mapped its pinout, all with the power off | Bypasses the controller and reads the raw NAND, against a maker-ID library we keep up to date |
How it ran.
The fracture explained itself
The fragment being gone for good, the card had to speak for itself. A minute beneath the stereo microscope did it: part of the contact block missing, traces ending where that part had been, and a crack that faded away a long distance from the die. Once you have seen that, a card that looks finished becomes routine chip-level work.
The way in that the factory left behind
Turn a monolith over and a row of service pads faces you. The factory tests through them. Nobody publishes the layout and no two makers agree on one, so probing carried on until the map was clear. Wire finer than a hair then goes onto the handful of pads that count. Answers came back from the memory along a route the break had never come near.
Raw pages to begin with, pictures much later
A chip surrenders its pages in whatever order its controller stored them, and in no other. Error correction has to run first. After that the XOR mask comes off, the pages go back into their true order, and a filesystem gets built over the lot before any folder name appears at all. Bridging that distance is the job, and no reader on sale will do it for you.
How it ended up.
Both nights came back entire — every frame opened and checked, the calibration files with them. The card was thrown away and nobody missed it.
Pages people open after this.
More cases filed under cards.
Seeing the same thing?
Leave it switched off and post it in. Nothing happens until the diagnosis, which lists what can still be read and what cannot.