Data Recovery Case File · Solid State & Flash · The Chip That Is the Whole Drive

System-on-Chip storage decoded: when controller and NAND share one package, platform support becomes everything — the home firms' workup honoured, the SoC anatomy explained, and the support-before-shipping sequence an overseas enquiry deserves

The enquiry travelled further than most, and arrived better-documented than almost any. Writing from overseas about his laptop's SK Hynix SSD — "specifically of the System-on-Chip (SoC) type" — he relayed the diagnosis of the recovery firms he'd consulted at home, and it's a competent workup by any standard: the SSD "does not identify on the M.2 NVMe interface"; "all power sources supply the necessary voltage, and the chip heats up normally"; "a reflow was performed to ensure the integrity of the solder joints, but this did not resolve the issue"; and — the crux — "given that it is a SoC, they were unable to access the NAND memory chip separately… the problem is likely firmware-related. Unfortunately, this specific SK Hynix model is not supported by the equipment available to them." This page does three things his diligence has earned: honours that workup and builds on it rather than repeating it; decodes what SoC architecture actually changes about recovery — why "one package" collapses every fallback into a single question of platform support; and sets out the only honest sequence for an international case: the support check comes first, and shipping comes second, or not at all.

The decode: what SoC changes, why support is everything, and the sequence that protects him

The architecture, and what it forecloses: a conventional SSD is a small team — a controller chip managing separate NAND memory packages — and that separation is recovery's great fallback: when the controller or its firmware fails beyond repair, the NAND chips can be read directly, off the controller's corpse, and reconstructed. A System-on-Chip SSD fuses the team into one sealed package: controller logic and NAND storage on shared silicon, no separate memory chips to lift, no "read the NAND directly" road at all — which is exactly what his home firms meant by being "unable to access the NAND memory to verify its integrity." On SoC storage, every road to the data runs through the controller's own technological modes: the manufacturer service interfaces that recovery platforms implement, family by family, through research. Foreclose the chip-off fallback, and one question remains standing alone: does a platform exist that speaks this exact controller's service dialect? Their diagnosis — power good, joints reflowed, chip warming, no identification — points precisely where they said: firmware. And firmware on SoC means platform support is not a factor in the odds. It is the odds.

Why "unsupported by their equipment" is a statement about equipment, not the drive: the crucial reframe for his decision: recovery platforms cover controller families unevenly and expand continuously — support is a moving frontier, added variant by variant as research lands. A model outside one lab's platform can sit inside another's current coverage, or arrive in next quarter's update. So the home firms' honest "not supported by the equipment available to us" closes their door without closing the case — provided the next step is taken in the right order.

The sequence an overseas case deserves: and here is the practical doctrine his situation writes: never ship internationally on hope. The exact model string and, where known, controller variant travel first — by email, at zero cost — and are checked against current platform coverage; the answer comes back in writing: supported (with the route, the figure, and no-recovery-no-fee terms stated before dispatch), or not supported (with, where possible, an honest pointer toward who might, and the same keep-the-drive counsel this archive gives every frontier case — SoC support lists grow, and an unsupported variant today is a watch-list entry, not a verdict). Customs, carriage and weeks of transit are only ever spent on a confirmed yes. His enquiry, arriving as a question before a parcel, had the sequence right by instinct.

On the bench

The protocol ran exactly as the page describes — paperwork before parcel. The precise model and controller variant were checked against the PC-3000 portfolio's current SoC coverage, and the answer went back overseas in writing before any shipping decision existed: the support status as of today, and with it either the confirmed route — firmware-mode access to the controller's service interfaces, stabilised imaging on the Atola TaskForce 2's native NVMe ports, one fixed figure and no-recovery-no-fee terms stated across borders in advance — or the honest not-yet, with the variant logged for the support watch-list and the drive counselled onto a shelf rather than into a courier's van. Either answer costing him nothing but the email; only the yes ever costing him carriage.

The outcome

His enquiry answered in the order that protects an international client: architecture decoded, the home firms' workup credited and built upon, the exact variant's support status established in writing first — and shipping reserved strictly for a confirmed route under fixed written terms. Free assessment of supportability, one fixed written figure where the route exists, 50% of parts and labour upfront with the balance only on success. The doctrine, posted for every SoC owner anywhere: when controller and NAND share one package, chip-off dies as a fallback and platform support becomes the entire question; support is a moving frontier that differs between labs and grows by the quarter; and the only sequence worth following across a border is his — confirm first, ship second, or keep the drive and watch the lists.

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.