Data Recovery Case File · Cameras, Drones & Cards · The 64GB Card Showing 32MB

Sixty-four gigabytes presenting as thirty-two megabytes — and warm to the touch: the wrong-capacity signature decoded, the heat read as the warning it is, and the footage recovered past the failed controller

His report carried two observations, both diagnostic gold. A SanDisk 64GB microSD, corrupted after filming video: "I can get it to show up on a laptop using Mac Disk Utility, but only as a 32MB drive that needs formatting — not the full 64GB." And: "the card seems to heat up when connected this way." An offer to drop it in locally, welcomed. The two symptoms are independent witnesses telling the same story. A card announcing a tiny, wrong capacity is a controller that has lost its configuration and fallen back to a bare default — it isn't showing a damaged filesystem; it's showing that the manager no longer knows what it manages. And a card that heats up is reporting an electrical fault in progress — the one symptom in the flash world that upgrades "stop using it" to "unplug it now and don't reconnect it." This page decodes both, and the chip-level road that recovered the footage without ever asking the failing controller for anything again.

The decode: two symptoms, one dead controller — and the heat's special urgency

The wrong-capacity signature: when a card presents a fraction of its true size — sixty-four gigabytes appearing as thirty-two megabytes — it isn't miscounting its contents; it's revealing that its controller has lost the plot entirely. A flash controller learns the card's real geometry from configuration data it maintains internally; when that fails — through firmware corruption or the controller's own decline — the chip falls back to announcing a bare, built-in default: a tiny generic capacity, unformatted, with none of the card's actual structure attached. The "32MB drive that needs formatting" is that fallback: the controller's blank ID card, shown when it can no longer read its own records. Two consequences follow. The data isn't in that 32MB — his footage lives in the real NAND the controller can no longer address, intact and simply unrepresented. And the format must be declined, as he rightly did by instinct: formatting the fallback device recovers nothing and risks writes to a card in electrical distress. There is no version of this card that returns to health; the 32MB apparition is a symptom to be read, not a drive to be used.

The heat — the symptom that changes the instructions: a storage card has no business getting warm, and one that does is reporting an electrical fault in progress — a failing component drawing excessive current, cooking itself a little further with every session. That upgrades the standard advice in a specific way: where most failed cards can be tried in a reader without much harm beyond futility, a heating card should not be connected again at all. Each powered session pushes current through the failing part, and heat is precisely the mechanism by which a recoverable chip-level case degrades toward an unrecoverable one — silicon damaged by its own dying component. His observation of the warmth was the most valuable line in the enquiry: it moved the card's status from "stop trying readers" to "stop supplying it power, full stop." From that moment, the card's next contact with electricity should be on a bench, on its own terms.

On the bench

The card was never powered conventionally again. It went straight to chip level: the monolith's technological pads exposed and probed on the PC-3000 Flash Spider Board — which addresses the NAND directly, bypassing the failing controller and its overheating circuitry entirely — with the pinout confirmed against the Rusolut VNR monolith database. The raw dump ran the full reconstruction: XOR descrambling, ECC correction, and rebuild of the translation layer until the card's true sixty-four gigabytes stood up as a filesystem — the footage he'd filmed sitting exactly where the fallback capacity had never been able to show it. The video was recovered structure-aware and verified by playback, clip by clip, and delivered — the heating controller left permanently out of the conversation.

The outcome

The footage recovered from the NAND past the failed controller and delivered verified by playback. 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 two signatures, decoded for every card owner: a tiny wrong capacity is the controller's blank fallback — the data lives untouched in the memory it can no longer represent, and the offered format recovers nothing — while heat is an electrical fault burning in real time, the one symptom that means unplug it and never reconnect it. He read both correctly, declined the format, and brought the card in with its options intact; the Spider Board did the rest.

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.