Data Recovery Case File · Solid State & Flash · The Chip Is Already Off

The Memory Package Was Removed for the Wrong Reason and Is Now in the Right Place

His enquiry describes work already carried out. A stick that stopped being recognised during a transfer of many documents, where "I tried to repair it by transferring the memory to a donor board, but it is not recognised by the computer." The donor board was never going to work — and the removal, which was the risky part, has already put the memory package where chip-level reading would need it anyway.

The decode: why the swap fails, and what his removal actually achieved

Why a donor board cannot read this memory: the controller decides where everything goes. It distributes content across the memory by its own scheme, remaps regions as they wear, and maintains the map relating a file's apparent position to its physical one — built up over the life of that particular device.

Why a matching board does not inherit any of that: the map lives with the controller, and a replacement has its own. Fitted to this memory it finds an arrangement it did not create and cannot interpret, so the assembly presents as unformatted or not at all.

Why this differs from a spinning drive, where board swaps do work: there, the drive's own configuration is transferred with the board. Flash has no equivalent transferable configuration — the arrangement is the controller's working knowledge, and it does not exist separately.

Why the attempt was nonetheless not wasted: the memory is off the original board. Chip-level reading is exactly what a recovery does, and the removal of the memory package — the delicate, irreversible step — has already been performed.

What now has to be established: whether the removal and remounting damaged anything. Heat, mechanical stress and repeated soldering are the risks, and a NAND package subjected to two operations rather than one has had more exposure than necessary.

Why that is checked before anything else: a damaged connection on the component itself changes the approach. The package contacts are read chip-level rather than through any board, so their condition is what matters now.

What the actual route is: the memory package read on NAND equipment designed for it, producing raw content, and the translation layer reconstructed in software by working out how the original controller must have arranged things.

Why the interrupted transfer is worth noting as the likely cause: a large document transfer is sustained writing. The controller updates its maps continuously during that, and a failure or interruption mid-update leaves them inconsistent — which stops the device presenting at all.

Why documents recover comparatively well: they are small and self-contained, with recognisable structures. Where the arrangement can be partially reconstructed, individual documents come back whole rather than as fragments.

What must not happen: no further boards fitted, and no reheating the component. Each soldering cycle is thermal stress on the only thing holding the content.

On the bench

The donor board approach was assessed as inapplicable and abandoned — a flash controller determining placement across memory, remapping worn regions and maintaining the map between apparent and physical position, all of which is the controller's working knowledge and does not transfer, unlike a spinning drive whose configuration accompanies a replacement board. Memory was examined for damage arising from removal and remounting, then read chip-level on NAND equipment rather than through any board.

The outcome

The donor approach abandoned, the memory package examined for damage from remounting, and read chip-level with the translation layer reconstructed in software. 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: the donor board could not have worked, because a flash controller's arrangement is its own working knowledge and does not transfer. But the removal has put the memory package where chip-level reading needs it.

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.