Data Recovery Case File · Cameras, Drones & Cards · The Crack You Only See Later

Cracked MicroSD Card: the Crack You Only See Later

Her enquiry connected two things she had assumed were separate, and the connection is the diagnosis. A PNY 16GB microSD card used in a DSLR: "for a while my card has been playing up — issues with saving and accessing photos — and on closer inspection there is a crack which travels from side to side. It does not go all the way through the card and is easily missed. I have just noticed it, so don't know how it happened or how long it has been there." Now nothing on it is accessible. The weeks of odd behaviour and the crack are the same event: a card that has been partially severed inside behaves exactly like that — intermittently, then not at all. This page explains why a crack that does not go all the way through is still enough, why the card kept half-working for so long, and how the memory is read when the card's own wiring has been cut.

The decode: why a partial crack is enough, and why it took so long to fail

What a microSD card actually is: not a container with a chip rattling inside, but a single moulded block — a monolith — in which the memory die, the controller and the wiring between them are set in resin, with fine internal traces running out to the gold contacts at the edge. There is no slack anywhere. Those traces are microscopically thin and they run close to the surface, which means a fracture does not need to break the card in half to break the card: it only has to cross the traces. A crack travelling side to side does precisely that.

Why it half-worked for weeks: this is the part that explains her whole experience. A cracked trace is not necessarily a clean break — it can be a fracture that still makes contact under some conditions and not others. Temperature, the pressure of the card in a slot, and even the orientation of the camera can be enough to close or open the gap. So the card works, then does not, then does again: photographs save correctly one day and fail the next, files appear and then cannot be read. Owners almost always attribute that to the card being "flaky" or old, because the crack is invisible unless you look for it — hers, as she says, was easily missed. Then the fracture propagates a little further, the intermittent contact becomes a permanent one, and the card stops entirely. Her timeline is the textbook version.

Why the photographs are probably fine: the encouraging part. The crack severs connections; the memory die that holds the images is a small, robust piece of silicon, and unless the fracture runs directly through it — which is rare, and visible in assessment — it is untouched. The card cannot deliver the photographs because its wiring is cut, not because it has lost them. The route, therefore, does not attempt to repair the card: it goes around the damage entirely, reaching the memory through the internal technological pads that sit beneath the surface, behind the broken traces, and reading the die directly.

The one thing that makes it worse: flexing. Every insertion into a slot applies pressure across exactly the span the crack runs along, and every attempt to read the card risks extending the fracture — potentially into the die itself. A cracked card should be handled flat, kept still, and not tried again "just to see". The intermittent phase is over; nothing further is to be learned by pushing it.

On the bench

The fracture was assessed first — its path traced across the card body and checked against the position of the die, to establish whether the damage was confined to the interconnect or had reached the memory itself. It had not. The card was then prepared as a monolith and its internal technological pads exposed and probed on the PC-3000 Flash Spider Board, with the pinout confirmed against the Rusolut VNR database — the broken traces and the edge contacts left out of the process entirely. The raw memory read out, and reconstruction ran the full pipeline: XOR descrambling for the controller family, ECC correction, and translation-layer rebuild until the card's filesystem stood up. Her photographs were extracted, validated by rendering rather than counted, and delivered on fresh media.

The outcome

The photographs recovered through the card's internal pads and delivered validated. Free assessment, one fixed written figure including VAT; where a chip has to be removed, 50% of parts and labour is payable upfront with the balance only on success — otherwise no recovery, no fee. The decode, for anyone whose card has been misbehaving for a while: a memory card is a solid moulded block with fine internal traces, so a crack does not have to go all the way through to cut the wiring; that explains the weeks of intermittent saving and reading that usually precede total failure, because a fractured trace makes contact under some conditions and not others; the memory die itself is small and robust and normally survives; and once you have seen the crack, stop inserting the card — flexing is what turns an interconnect fault into a broken die.

Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.