A dissertation, a payroll run, the one surviving copy of the accounts, all riding on a stick that caught the edge of a desk or stopped halfway through a copy. Work of this kind travels light, which suits the geography. An envelope handed in around Liverpool today is on the bench in Manchester by the following morning. Broken necks, dead controllers and folders deleted by accident all have routes back, and nearly every one of them begins with the owner giving up on trying yet another socket.
Every usb stick / flash drive 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 usb stick / flash drive, and after twenty-odd years these thirty account for very nearly everything that arrives.
A bag strap, somebody's knee, a chair shoved back under a desk without looking. Leverage wins every time. Where the board allows it the plug gets rebuilt under the microscope, and where it does not the memory is read directly.
The stick enumerates and then talks nonsense. That points at the controller, or the firmware running on it, rather than at anything happening to the files. Nothing has moved off the memory.
The controller has started up holding no map of the memory sitting next to it. Read those chips directly, rebuild the mapping in software, and the two are introduced to each other again.
A torn file table sitting on top of intact files. Refuse it, take the stick out of the port, and post it in exactly the condition it is in now.
Recoverable until the freed space gets used again, and not a moment beyond. From here, a drawer is a much better address for the stick than a USB socket.
Something has shorted and the stick is slowly cooking itself. Take it out of the port. Every minute it spends hot is taken out of whatever margin remains.
Leave a stick standing upright in a socket and it becomes a lever with a long arm and all the time in the world. Solder joints crack. The memory chips, as a rule, never notice.
There is nothing to lift off one of these. Probing the mapped test points takes longer and arrives at exactly the same files.
Twice over, in one memorable case. Given a proper clean and a proper dry on the bench, laundered sticks do far better than their owners expect. Plugging a damp stick in to check is the one move that ruins it.
One spell in a shared machine and every folder has been swapped for a lookalike link. Underneath the deception the genuine folders are usually still there and untouched.
The marketplace classic. As soon as the genuine chip fills, the writing loops back over its own beginning, so everything before that point carves out cleanly and everything after it was never written down anywhere.
The SDCZ50 has two familiar endings: the neck snapping where the plug meets the body, and a sudden drop to zero megabytes. There are standing routines here for each, and neither counts as unusual.
The Exodia DTX, and the DT100 G3 before it, lock into a controller state that responds to nothing whatever. There is no point arguing with it, so the memory gets addressed directly instead.
Whatever silicon was cheapest that quarter is inside, and the logo printed on the case tells you nothing about it. The controller that turns up decides the method, which is why the diagnostic comes first.
Bend a stick and let it spring back and there are usually hairline cracks in the tracks underneath. They get repaired under magnification, and then the stick is read end to end.
The SDCZ430 is built to live permanently in a port, and three years in a hot laptop socket is a hard address. When the controller gives up, monolith probing takes over.
The SDCZ73's metal sleeve works loose and takes a few tracks with it on the way out. The flash bench reworks the board first and images afterwards.
The Verbatim office staple ends up carrying live work instead of shifting files about, which is several years of writing it was never rated for. It declines gradually right up to the moment it stops.
The long-bodied DTX works against tight front sockets until the joints crack. Reflowing them under magnification usually puts it right, with nothing further needed.
Grit works into the retractable slider on a Courier. The Neon just gets old. In both cases the controller goes long before the memory does, which is the good news.
Damp pockets and years of thumbs leave the contacts oxidised, and the stick starts appearing and disappearing at random. They get dressed properly first, then imaged in one continuous pass instead of a dozen attempts.
A dying power supply, or weather on the mains, sends a spike out through the front sockets of a tower, and whatever stick was living in one takes it. The controller is cremated. The memory is generally untouched.
Flash leaks its charge away in storage and old archives come back faint. Careful imaging, with ECC given the time it needs, keeps a great deal that a quick confident read would have thrown away as unreadable.
Somebody accepted the offer to fix the drive, and Windows handed back a folder of numbered fragments carrying no filenames at all. Those fragments get ignored completely and the real layout is rebuilt from a raw image.
Too many wrong PINs and they erase themselves by design, and no laboratory argues one out of that. Where the PIN is known the job is an ordinary one. Worth knowing before anybody starts guessing.
Bend a stick far enough and the silicon can fracture even when the board looks straight and every joint is sound. It will never enumerate again, because the part responsible for enumerating is in pieces. Taking the memory off and rebuilding the mapping in software is the way through, and it works.
Booting a live system off a stick, or keeping an installer on one that gets rewritten every few months, is a workload nothing handed out at a trade counter was built for. The wear levelling goes first and whole blocks stop replying after it. Material written earlier is usually still within reach.
Hit an error mid-write and some controllers flip permanently into a protected state. Every utility then reports the disk as write-protected, and registry tips copied off a forum achieve nothing, because that flag is held inside the controller and not by Windows.
Somebody finds a low-level format utility for that controller online and runs it hoping to cure a fault. The tool builds fresh tables and throws away the map that recorded where everything sat. Getting the contents back then means reading the raw flash and working the arrangement out again, which is what this bench does all day.
An error appeared, the offer to format looked like the practical response, and somebody accepted it. Quick formats are survivable. Full ones far less so. In both cases the thing to do is stop, since every file saved to the stick from then on settles on top of the material we would otherwise be recovering.
Call them memory sticks, pen drives or thumb drives, the failure point is the same on all of them: that solder joint where the metal connector meets the small board behind it. One knee pressed against a laptop on a Merseyrail train is enough to do it. Under a microscope, torn pads can be rebuilt and broken tracks bridged back together. When the board has gone too far for that, the memory chip is taken off it and read on a programmer instead, which means nothing on the stick's own circuitry needs to work ever again. The regulars on the bench are whatever people actually buy: SanDisk Cruzer Blade and Ultra Fit, Samsung BAR Plus, the small metal Kingston DTSE9 with the DataTraveler DT100 G3 beside it, Verbatim Store 'n' Go, Integral and PNY. Behind those labels sits a short list of controllers, with Phison, Silicon Motion, Alcor and Chipsbank the usual suspects, and establishing which one you have determines how the raw dump gets decoded once it is off.
Two versions of this turn up constantly. Somebody deletes the wrong folder, or every document on the stick turns into a shortcut icon after it has been in a machine at a print shop. The second is far less alarming than it looks, because that class of infection hides files rather than destroying them, and a deletion simply flags the space as available without scrubbing anything off the chips. Which makes the advice short and rather dull: put it in a drawer, unplugged, and stop handling it. On the bench a complete copy is taken from one end to the other and your documents are carved out of that copy, never out of the item you sent. Sticks that just stopped one morning with no drama attached go the same way, straight to the chips. The newer monolithic designs, which seal controller and memory into one moulded part, are approached through test points instead of a soldering iron, and that takes longer. Whichever kind yours is, the assessment will have worked it out before anyone asks you to make a decision.
A stick is chip work from the moment it is booked in, so it stays on the flash bench for the whole of the job:
Reads the NAND packages directly and reconstructs in software the translation a dead controller can no longer perform. It is the starting point whenever a stick returns nothing at all.
Controlled hot-air and infrared profiles, so a memory package leaves a broken board without disturbing the charge held in its cells. Temperature discipline is the entire trick.
Small sticks are often one sealed slab with no chip to remove. Test points get located against our pinout records and hand-wired under magnification. It is slow, and nothing else reads them.
The XOR, ECC and wear-levelling puzzle solved step by step until a raw dump comes back as named folders. Dull arithmetic that ends well.
Snapped plugs, lifted pads and torn tracks rebuilt under magnification. A good proportion of sticks declared dead elsewhere need this and no more.
Sticks are read through a blocked interface throughout, so the original memory leaves exactly as it arrived.
A startling amount of Merseyside payroll, dissertations, tender documents and club accounts lives on a Cruzer Blade, an Ultra Flair, an Exodia or an Attaché, and most of it is copied nowhere else. Every model fails true to type. Slim metal SanDisk bodies run hot and shed their casings. An SDCZ50 either breaks at the neck or drops to nothing. Long budget Kingstons fracture their joints against tight front sockets. Store 'n' Go sticks are pressed into use as live working drives until the wear levelling gives up. Couriers and Neons sit in office drawers and keep going until their controllers stop. Trade-show freebies are a lottery, and the only useful advice about one is never to let it hold the sole copy of anything. Every stick here is read at chip level. The price is £250 + VAT, and the free diagnostic that chooses the method reports two working days after the stick reaches the lab.
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.