Solid state drives stop without making a sound, and that is the part owners struggle with. Fine on Thursday, invisible on Friday, nothing in between by way of notice. SSD hard drive recovery is a separate trade from disk work — no platters, no heads, no clean bench — and it runs here every day for households, for the labs and departments at the two universities, and for the offices between the waterfront and the Baltic Triangle.
Every ssd job is diagnosed free. One fixed figure follows in writing, agreed before a screwdriver comes out of the drawer.
No fix, no fee all jobs except electronic and mechanical failures, chip level work, DVR and Forensic jobs. Every band is listed on the data recovery cost page.
Matching the symptom to the fault underneath is the first job on any ssd, and after twenty-odd years these thirty account for very nearly everything that arrives.
The management chip stops responding and the drive disappears from the bus while every file stays exactly where it was written. What has gone missing is the part that speaks for the flash. Supplying that is what a firmware bench does.
No manufacturer ever sold a drive by that name. It is a Phison controller trapped in its own boot loader. The data behind the wrong label is untouched, and service-mode work walks it out.
Classic translator failure. The drive answers politely and quotes a size that belongs to no product on earth. Rebuild the translator in service mode, then take a complete image before asking it for anything else.
Boot code, power rail or controller. From the outside all three produce the same blank screen. Only a bench separates them, and that is the first question the free diagnostic answers.
End-of-life protection has stopped writes to protect what is already stored. That is a nuisance on a desk and close to helpful on an imaging rig, because the drive can no longer make its own situation worse.
Flash tends not to give notice. There is rarely a warning noise and rarely a slow decline. That is why so many calls to this number open with the same sentence: it was absolutely fine yesterday.
Interrupt a write and the tables that map logical addresses onto physical blocks can tear. A proper surge takes the power stages out as well. Both land here regularly, and neither is a software job.
Worn NAND reads back one way, then another, while SMART grumbles quietly in the background. Only the first full image is worth trusting, so it gets planned properly and taken once.
Freed blocks get wiped whenever the drive has power, regardless of what anybody has decided yet. Unplug it and that process halts. Leave it connected while you weigh up the options and the options cost you files.
Thin boards come off badly against screwdrivers, upgrade attempts and stiff envelopes. A clean break near the edge is a track repair. Anything worse goes to chip-off, which takes longer and reaches the same files.
Samsung's MZ-76E500 sits in a huge number of British machines, so the full range of its endings passes across this bench. The typical one is sudden. The drive leaves the SATA bus mid-session and returns as a sliver of its capacity, or not at all, and swapping cables changes nothing. The MJX controller has mislaid its translator while the V-NAND behind it is sound. Service-mode repair, then a full image, is the way home. Owners who cut the power at the first stumble get the best result.
First-run MZ-77E1T0 drives were reallocating sectors fast enough that Samsung published a firmware fix. Unpatched examples still arrive here. The flash reads perfectly well. It is the map above it that has to be rebuilt.
Crucial's MX500 is known for reporting a pending-sector figure that worries owners more than the drive deserves, right up to the day that figure comes with genuine read failures attached. Separating the harmless case from the real one happens on a bench, not by guesswork.
WD's SATA Blue has a documented habit of vanishing until the firmware gets some attention. A familiar patient with a familiar route, and the memory is rarely at fault.
When the translator on an A400 SA400S37 or one of its cousins corrupts, the drive shrinks to a few megabytes or disappears altogether. Service mode is where that gets fixed, and the flash underneath is generally blameless.
These halt partway through a write and then decline to say anything more. A proportion wake up in service mode. The rest get read at chip level, which reaches the same place by a slower road.
SSUBX and SSPOLARIS blades from machines built between 2013 and 2017 fail in the ordinary ways and image here on native adapters. People are often surprised that no donor Mac is needed at any point.
Without a DRAM cache the wear levelling is crude, so a BX500 asked to do heavy work wears unevenly and early. When service mode will not have it, the packages come off and the contents come out.
Marketplace fakes loop their writes around a small piece of genuine memory. Whatever landed inside the real capacity comes back. Anything written once the loop had begun was never stored at all, and nobody recovers what was never there.
The plastic tongue breaks during an otherwise routine swap and lifts a few pads with it. That is board-level repair, and afterwards the drive behaves as though nothing had happened. Board work takes 50% upfront rather than running under no fix, no fee.
Years of heavy writing eat the rated life and the drive starts locking up or reading unpredictably. What remains will support a single imaging attempt, so that attempt gets planned properly rather than begun in hope.
The manufacturer's tool stopped partway and left a drive that can no longer say what it is. Depending on what the controller still admits to, it gets reflashed on the bench or bypassed entirely.
QLC flash is quick until the fast cache fills, after which it slows dramatically. Owners read the slowdown as a crash and pull the lead out, and it is pulling the lead that causes the damage.
Every read nudges the charge in neighbouring cells a little. On a drive full of material that gets read often and rewritten never, those nudges add up until the ECC runs out of headroom. Slow repeated passes recover a great deal more than one confident sweep.
Unpowered cells leak. A drive last used in 2019 can come back reading as static. Repeated-read imaging, with ECC given the time it needs, returns a great deal that one quick pass would write off as unreadable.
Warming up and cooling down, season after season, cracks the BGA joints under the controller and under the memory packages. Reworking them restores contact for as long as one clean image takes, which is all that is asked of it.
Datacentre and enterprise SATA units carry capacitors that keep the drive alive long enough to flush its buffers when power disappears. Capacitors age. Once they have failed unnoticed, the following power cut carries the mapping tables off with it, and nobody was aware the net had gone.
Almost every modern SSD encrypts on the way in whether the owner asked for it or not, and the key lives in the controller. Damage that key store and the memory reads perfectly while meaning nothing. We say so directly instead of selling hope by the hour.
More than one manufacturer has shipped code that halts the drive at a fixed number of power-on hours. The hour comes round and the drive does not. Service-mode repair, then an image straight afterwards.
A glitch in the security state shuts an OPAL drive. The PSID code on the label unlocks it by erasing everything, which is precisely the opposite of the point. So it stays unused here, and it should stay unused everywhere else.
The address your computer asks for and the place your file actually sits are two different things on an SSD. Wear levelling moves data around the NAND packages according to whatever the controller decides is sensible at the time, and the one record of where everything ended up lives inside that controller's firmware. Lose the controller, or let the mapping table go corrupt, and the drive falls silent on the bus even though every byte is still sitting in the memory in perfect order, with nothing left able to point at it. Two legitimate routes exist from there. One goes in through the manufacturer's service mode and persuades the controller to rebuild what it lost. The other lifts the NAND chips off the board, reads them directly, and reconstructs the translation layer in software, which takes longer and does not care in the least whether the drive ever answers again. No downloaded utility performs either, for the simple reason that all of them require a drive Windows can already see.
Everything comes through: Samsung, Crucial, Kingston, SanDisk, WD, Intel, SK hynix, Kioxia and the removable Apple blades, in 2.5-inch SATA, mSATA and M.2 shapes. Which models turn up most is simply a record of what builders and repair shops around Merseyside were fitting in any given year, so the shelf usually holds a Samsung 860 EVO or the 870 EVO that replaced it, a Crucial MX500 or BX500, a Kingston A400 or KC600, a WD Blue SA510, a SanDisk Ultra 3D. The 860 EVO earns its own paragraph, since one particular failure on that model sends more people to this page than anything else on the list. An MZ-76E500 that looked completely healthy drops off the SATA bus in the middle of a session and comes back announcing a lost or corrupt signature, a capacity of a few megabytes, or nothing at all. What has happened is that the MJX controller has lost track of its translator. The V-NAND has not worn out, and the distinction is the whole point, because a drive with no map has nothing sensible to hand over while every file remains exactly where it was written. Samsung published firmware revisions across that generation aimed directly at the problem. Where a drive is already too far gone to take one, the translator gets rebuilt through service mode and a complete image is captured before that drive is asked to do anything else. Two categories are refused here, and saying so before you pay postage is fairer than saying so after: storage soldered to a mainboard, and phones or tablets, neither of which forms any part of this service. Aggressive TRIM and hardware encryption both make a job harder than it first appears, and you will be told that at the assessment rather than discover it on an invoice.
Mechanical disks give notice before they go. There is a tick, a slowdown, a sector that has to be asked for twice, and an owner who is paying attention can usually buy a few days out of those signs. A solid state hard drive offers nothing comparable, because there is no moving part in it capable of making a sound. Either the controller replies or it does not, and the switch between those two conditions happens between one perfectly ordinary Tuesday and the next. It is the reason so much solid state hard drive data recovery lands here with an owner insisting the thing was working fine the day before, and they are nearly always telling the truth. Worn-out memory is a genuine phenomenon but it is seldom what has actually happened. On a consumer drive, a power cut, a bug in the firmware, or a controller that has simply stopped responding will account for a lost translator far more often than exhausted write cycles do, that last one being the failure everybody expects and the one this bench sees least. Whichever it proves to be, the diagnostic costs nothing, takes two working days from the parcel arriving, and any single SSD is £300 + VAT, with M.2 and NVMe modules charged at the same figure.
From the first hour, solid state recovery is firmware work and chip surgery. The bench here is equipped for nothing else:
Wakes a silent drive in its own manufacturer's service mode, repairs the firmware, rebuilds the translator, and gets a complete image off before the drive is asked to do anything further.
When a drive stalls, timeouts are enforced in hardware and the retry pattern is chosen for the unit in front of us rather than lifted out of a manual.
Every removable solid state format gets a proper connector, 2230 stubs and Apple blades included, with active cooling for modules that turn awkward once warm. No hopeful USB caddies at any point in the chain.
When the controller cannot be woken at all, the memory packages leave the board and are read individually. Slower, and it arrives at the same address.
Interleave, XOR and ECC worked out from raw dumps until the mapping exists again in software and the volume can stand on top of it.
Burnt power stages and cracked boards get repaired before imaging begins. A drive that browns out halfway through an image has bought nobody anything.
One drive defines this page, and it is the Samsung 860 EVO. MZ-76E500 units went into machines by the thousand through the years when everybody was swapping a hard disk for something faster, so ten years later that whole wave of upgrades is expiring at once. The usual ending is a translator failure above healthy V-NAND, and translator failures recover. The 870 EVO generation behaves in much the same way. Around those two sit an A400 SA400S37 that has mislaid its map, an MX500 fretting over pending sectors, an SA510 that left without a word, a BX500 aged early, and an NV2 whose insides change from one quarter to the next. Anything that unplugs is workable, in 2.5-inch, mSATA or M.2 form, including drives that have shut themselves read-only. One SSD is £300 + VAT and the diagnostic before it is free. Two things we cannot take: flash soldered to a logic board, and phones or tablets.
Almost every job on this bench came in by tracked, insured post. It is the calmest way to move a drive that is already struggling, and a parcel handed in anywhere on Merseyside usually reaches the bench the following working day.
Still screwed inside a laptop, tower, MacBook, iMac, server or a CCTV recorder? Take the drive out and post that on its own. We do not strip machines here, and any computer shop will do it in a few minutes. The one job nobody can take on is flash soldered straight to a logic board, as on Apple Silicon Macs and a handful of very slim laptops: if the storage will not unbolt, there is nothing to send.
↓ Print the shipping & booking-in form (PDF)
Address it to Manchester Data Recovery. It is roughly 35 miles from Liverpool along the M62 if you fancy the run, and next working day by tracked post if you do not. We ring or email the moment it is booked in, and the free diagnostic closes 2 working days after that.
Not sure what belongs in the parcel? Call 0800 689 0668 before you tape it shut, or step through the free online diagnostic first.
Diagnosis free, one figure written down, and £300 + VAT covers a single drive or SSD. Start online or ring.