NVMe modules are the fastest thing sold to consumers, but they also stop just as fast. There is no warning noise. There is no gradual degradation in performance. The machine was working fine before it produced a "boot error" where the drive used to be. All of the M.2 sticks and PCIe cards from every manufacturer end up here, pulled out of home computers, gaming rigs, and the design and engineering offices around the city.
Every NVMe job is diagnosed free of charge. The price for any given job will be provided to you in written form prior to any tools coming out.
In most cases you only pay if we recover your data. The exceptions are electronic/mechanical failures, chip level work, DVR and forensic jobs. All pricing information can be found on the data recovery cost page.
Every single NVMe job begins by diagnosing the cause of the symptom. Twenty-odd years in, the twenty-five faults below cover nearly every one we have seen reach this bench.
A silent controller or firmware fault will clear the bus overnight. Behind that silence lies complete data, waiting for something to manage it again. This is the commonest way these modules fail.
Dead controller, bad power stage, corrupted boot code. Dead memory is the least probable of all four possibilities. The odds favour recovery. This module has no method of letting you know.
When the translator crashes, the module reports an unknown size. Corrected in service mode via a clearly marked path.
A bare module throttles, drops off under a load, and then will start scrambling its own tables over time due to hot weather. Imaged when it is cool, many (a lot) resolve their issues.
Windows update or vendor tool, same results: a module with an inability to identify itself. Reflashed here in areas where reflashing can occur. Full work-around in all other cases.
Flash has no record (diary) of its activity nor does it send out a notification. Faultless service until the moment it fails is the typical bio, and that's why backups count more in this case than anywhere else.
Thin boards get damaged the worst by upgrade levers and stiff envelopes. Track repairs occur when there are small breakage areas around the edges. Any damage outside of these parameters will be chip removal.
Surge-damaged internal components get repaired at the individual-component level prior to attempting to recover anything. The sequence of the entire process is what defines success or failure. Change the order of the steps and you've essentially destroyed the module.
Blocks being erased after they are "freed" operate as background processes immediately after their deletion. The only time the erasure stops running is when power is removed. Pull the plug and that clock will freeze right where it was. This is probably the best thing you can do for it.
WD's top-of-the-line model shipped with a version of firmware that would stall under certain loads then eventually disappear, until a patch arrived. Unpatched products still keep our booking sheet busy most months of each year.
It was determined that a corrective update was needed due to excessive consumption of NAND lifespan during operation of the 980 PRO and 990 PRO generation models. Units that never received this update still arrive with most of their NAND endurance already gone.
The exact controller and flash configuration a Kingston NV2 shipped with depends on whatever was going that quarter, and that will determine how two identically labelled modules get treated (i.e. they may require totally different methods). The method used is determined by the actual contents of each module.
Handheld devices like the Steam Deck bake small modules inside crowded shells, and there isn't much room to dissipate the heat generated by those modules. It is a tiny format, with a very conventional recovery process on the bench.
Modules which do not contain DRAM use memory from the host computer to store their own map information. If a single bad interaction occurs after sleep mode, or a crash, all of the memory used to store map information becomes corrupted. Service mode will correct this ledger.
Bare fast modules age quickly due to excessive heat generation. The initial symptoms of failure occur via intermittent dropout behaviour followed by complete loss of function. The window of time available for making an image of the module occurs during the period of dropout activity.
Budget drives aggressively hammer their pseudo-SLC regions until these areas become exhausted; at which point they will either stall completely or brick. The area immediately behind the pseudo-SLC region (the slow flash) typically responds perfectly well.
Screws holding heatsinks in place over-tightened, thermal pads damaged, boards flexed and joints below the controller cracked. These get reworked with the aid of a microscope, followed by imaging.
After a BIOS update, or if a marginal riser is involved, the module and socket can no longer agree upon communication terms resulting in the appearance that the drive has failed. These issues have been proven out using lab sockets known to be functional as opposed to attempting to troubleshoot based on cable swaps.
If a single failing die is encountered by the controller it can either cause the overall size of the disk to be smaller than intended (due to having fewer working physical blocks), scatter errors randomly across all of its respective stripes, or lock the entire disk in a read-only state. The imaging process will find alternative paths around the failed die, using the neighbouring dies to move your data off the disk.
When power is lost while performing garbage collection and the table is partially rewritten, the disk module may mount, loop back and forth, and act erratically. A service mode recovery process can recover your tables.
One command takes an NVMe disk and changes it into a different block format. In doing this, the mapping information is discarded. Once done quickly and quietly, the disk appears empty. Recovery now has to start with the raw memory, and then build a new block table based upon that.
Both commands were designed to be irreversible; both accomplish their goals. If a sanitize completed successfully, you will have no files remaining to recover. If it was terminated prematurely, a great deal sometimes remains. Since guessing is futile, the free diagnostic identifies which of the two occurred.
The documented loss of signal (dropout) occurs when the system enters aggressive low power modes on specific platforms/modules (especially Linux), and causes some drives to go to sleep permanently. It looks like a dead drive. However, the drive and/or module are frequently fine, and your tables simply need rebuilding after the failure that caused it to go down.
The problem lies when you have all blocks of space used within a drive. It does not allow the drive to do "housekeeping" as there will be nowhere to move the data. This can lead to performance failures, error accumulation, and some controllers will put their drive into read only mode rather than continue operation. Not ideal but helpful for creating images of the drives.
Depending on the machine, an M.2 socket will carry different signals, therefore a module may fit mechanically without having any wiring it can use as a communication interface. Nothing shows up, nothing is wrong, and hours go by trying to find a problem that does not exist. Rule this out before you write off your drive.
There are no mechanical components between an M.2 module and the processor; they communicate via PCIe, which results in both speed and sudden failure of the modules (there is rarely any warning or "limp home" behaviour). Stacked NAND is managed by dense firmware; once the controller or that firmware stops functioning properly, the module will be silent and your data will remain intact behind this silence. There are two routes in. Service mode, if the controller is still able to communicate with the rest of the system. Otherwise, the individual chips themselves, with the translation layer rebuilt from scratch. However, neither option will be improved upon with further dozens of reboots at your kitchen table.
2280, 2242 and 2230 sizes. PCIe plug-in cards. Gen3 up to Gen5. Samsung, WD, Crucial and Kingston. Kioxia, SK hynix, Sabrent, Solidigm. As long as your module will unclip and fit into an envelope it is acceptable. The one exception is soldered NVMe modules, and we'll refuse them; if you are not sure which kind you have, please call us before sending anything. Heat also deserves special mention. Using a bare stick on a high load without any heat sink, the device will throttle for a while, then drop off the bus. Repeated episodes like this may cause corruption to the device's mapping tables. We see a steady supply of these from small form factor builds and thin laptops. Regardless of why your device failed, our free assessment will identify it and give you a figure before you spend anything.
This is a PCIe bench that allows controller work, firmware work and chip repairs (chip surgery) to occur at PCIe speed. There are no bottlenecks to the process, and no job should ever wait for equipment:
When a module goes silent, it is brought up in the manufacturer's service mode. This enables firmware corrections, rebuilds the translation table, and creates an image of the memory prior to anyone attempting to use the memory.
If a PCIe module stalls, or wanders away from its operating parameters, it can be caught by timeouts enforced in hardware, using retry behaviours that no downloadable tool can access.
Everything from short (2230) drives up to full-length (2280) drives, add-in card drives and enterprise U.2 drives is supported via native connections, with forced cooling options available for any that fail when warmed up.
Packages that are beyond the reach of service mode are removed from the board and accessed one package at a time. This is a longer route than others to arrive at the same end point.
Interleave, XOR and ECC are worked out from the dump to get a mapping to exist again, with your volume sitting on top of that mapping.
Before we even attempt a read, each burnt power stage will be completely rebuilt one component at a time, every cracked board will have been reworked on our workbench using the microscope, and all the ripped-out traces will have been fixed.
All of these reputations were earned the hard way. WD's initial firmware of the SN850X wobbled under constant use, but eventually WD released a fix. Samsung modules from the 980 PRO and 990 PRO generations that never received their update still turn up with most of their endurance spent. Each NV2 is also a luck of the draw, and what the NV2 contains determines our method. Gen4 SSDs fitted somewhere a heat sink was judged optional will slowly cook, then disappear. We support all sizes from 2230 to 2280 as well as add-in cards, U.2, Gen3, Gen4 and Gen5. But there is one catch. The module has to unplug so it can be sent by post. The two exceptions here are soldered-down NVMe drives (if you are not sure which kind you have, please ring us first), and phones and tablets.
Get a quote first by telephone or use our free online diagnostic. Most of what you see on this bench will then be delivered via tracked, insured post. This is one of the safest ways to move a failing drive and packages handed in at a North-East post office are normally delivered here the next working day.
Still attached to a laptop, desktop, MacBook, iMac, server or CCTV/DVR unit? Remove the hard drive or SSD first. The bare drive travels on its own; when removing disks from a server remove all the disks and label each disk by its bay. This lab does not take machines apart. Any storage that has been soldered onto a motherboard (as in T2 and Apple Silicon Macs and a couple of thin laptops) will not be worked on: if it will not unbolt, we cannot work on it. If you're not sure what type of machine you have, ring us first. Phones and tablets are also not taken on.
↓ Print the shipping & booking-in form (PDF)
Your parcel should be addressed to Edinburgh Data Recovery, approximately two and a half hours (and 130 miles) from Aberdeen via the A90; alternatively it can reach us the next working day by tracked post if this journey time / distance is prohibitive to travel. As soon as your drive has been scheduled in we will contact you.
Still unsure about what should be placed inside of the packaging? Please call us first on 0800 689 0668, or use our free online diagnostic tool to assist you in deciding which items belong inside of the packaging before sealing.
Free diagnosis, one fixed price quoted in writing, and in most cases you only pay if we recover your data. Start online or phone it in.