Data Recovery Case File · Memory Cards · The Second Attempt

Stopped at 80%: returning to a card another lab couldn't finish

Second-attempt cases arrive carrying their history, and this one's was unusually informative: a 400GB microSD that had died without warning in a phone — utterly dead, no acknowledgement from any device — taken to a recovery firm a year earlier, who reported a "NAND error" and an imaging attempt that climbed to 80% and stopped. Since then, a drawer. The enquiry asked the two right questions: is recovery still possible, and how does the process actually work? Both deserve public answers.

Answering the year, and the 80%

First, the drawer: a year unpowered cost essentially nothing — flash sitting cold is stable on these timescales, and second attempts aren't racing a clock the first attempt started. Second, the prior lab's result, read fairly: reaching 80% on a monolith this size means they did the hard part properly — accessed the chip beneath a dead controller and read the majority of it — and their stall says the NAND itself was answering unreliably in places, pages returning errors faster than a standard read strategy could absorb. That's not a verdict; it's a boundary of technique. Weak NAND can very often be read past a standard attempt's stopping point — the same pages asked again under varied read conditions, their answers compared and error-corrected across attempts, the flash's own redundancy voting out the noise — slower, more stubborn work than a first pass, and exactly what a second attempt exists to bring.

The recovery — and the process, as asked

The card was re-accessed at its test points against the VNR's references, and the read ran the stubborn way from the start: the cooperative 80%-class regions banked and cross-checked, then the unstable remainder worked in the multi-read rhythm — repeated passes, adjusted conditions, error-correction voting page by page — until the map of the chip closed as far as physics allowed. Reconstruction did the usual translation from raw chip to card — scrambling reversed, geometry established, the translation layer rebuilt — and the phone's photo library rose from it, organised by the files' own dates. Coverage landed where honest second attempts on degrading NAND land: the strong majority of the library recovered and verified, a minority lost to pages past all persuasion, itemised rather than smoothed over. And the process, answered publicly for her and every second-opinion enquirer: assessment free (a prior lab's report welcome — it saves diagnostic time, as theirs did); one fixed written quote before any chargeable work; 50% of parts and labour upfront with the balance only on success; recovered data returned on new media, encrypted.

Outcome

Most of a library its owner had written off twice — once at the failure, once at the 80% — delivered from a card that spent a year as a drawer ornament. The generalisation is the page's point: a stalled or partial first attempt is a data point, not a death certificate. Techniques differ between benches; degrading NAND in particular rewards patience-heavy strategies not every first pass deploys; and a card sitting unpowered keeps its options open indefinitely. If somewhere in a drawer there's media carrying an old "couldn't finish" verdict — the verdict aged; the data didn't.

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.