Controller and firmware
By far the most common serious failure. Missing from the BIOS; the model string is blank, or the capacity is absurd. Flash memory behind it is generally good.
Most likely
Many of these are encrypting "at rest" by default — often with BitLocker to add another layer on top — and will decide if recovery is possible before anything technical happens. It is established at the diagnostic level, rather than discovered after a bill.
Since Samsung produces the controller, the firmware and the flash memory, everything must go through Samsung's stack. Most of what decides these jobs is not the hardware but whether encryption sits between you and the result.
The thing that most commonly terminates a Samsung job, and one should check for themselves first.
Samsung SSDs generally support hardware encryption — Class 0, which encrypts at rest using a key stored within the controller and does not require entry of a password, and TCG Opal, which adds a layer of authentication. The portable T series drive supports encryption as standard.
You are unlikely to be limited just by Class 0, as the controller will handle decryption in the background. It's users setting their own passwords or enabling BitLocker that stops the job from completing. In addition, if you're using Windows, BitLocker could have used your storage device's built-in encryption hardware without you realising it; thus the recovery key has as much importance as it would have had with any other encrypted volume.
Find out whether the recovery key exists before trying anything else. It is probably stored in Azure AD or Intune on a company owned computer and accessible via IT within minutes. We'll attempt to recover the password where the recovery key really doesn't exist and should succeed when the password was weak, reused or partly remembered, though it fails against a long random one.
Faults in these drives are distributed as follows.
By far the most common serious failure. Missing from the BIOS; the model string is blank, or the capacity is absurd. Flash memory behind it is generally good.
Most consumer-grade units do not have the necessary power-loss protection capacitors. When power is interrupted while writing data, the mapping table becomes corrupted, and the unit will refuse to provide data that it does not trust or can't verify.
The 970 & 980 series runs hot when installed in thin laptops & under graphics cards. Often intermittent at first – works well cold, fails when warmed up – provides some advance notice as long as you notice it.
True cell degradation. A less frequent occurrence than most believe; and the case where the prospects of recovery are weakest.
Manufacturer tools are useful before a problem exists but may be hazardous when used during a problem.
A lot of things in Samsung Magician will be useful when you want to keep track of how well your drive is performing (wear, temperatures, etc., and firmware version). When a drive starts acting up there are several features that Samsung Magician provides that would be counterproductive.
Firmware updates rewrite the service area, which is exactly where a recovery needs to read to get to the mapping table. If a firmware update fails on a drive that was already in a fallback state then the error may be permanent. Secure Erase and PSID Revert are even worse; the second completely wipes the drive by design and is provided as a means to gain access back into a "locked" device.
Maintain your firmware at the most recent levels as proactive maintenance on drives that are still functioning. Don't go looking for them after one of yours starts malfunctioning.
That's alarming to see and better news than they look.
A drive is never manufactured with just one megabyte. So when a BIOS says 1MB, 2MB, or 8MB, the controller is using an intentional placeholder because it cannot load its own firmware or its translation tables, instead of trying to show you what it could not guarantee as accurate.
You can reach these from the manufacturer's diagnostic mode. Diagnostic mode lets you by-pass the standard interface and read the service area & translation tables directly. Those tables are all there really is — if you do not have those, then the NAND looks like unformatted data. This is due to wear levelling, which completely randomises the physical layout compared to the logical layout.
Removing flash packages is usually reserved for extreme cases rather than being the first thing done — removing flash packages is generally slower, riskier and if you're dealing with an encrypted device will give you back an exact replica of something you were unable to access. The cost of SSD/NVMe recovery in Manchester is from £300 +VAT after we've provided a free 48-hour diagnostic.
Frequently. Most of the time serious faults occur where the controller fails before the flash does; thus the NAND still contains your files but they are hidden behind a layer of non-responding hardware — and many times it is possible to get at it via the manufacturer's diagnostic mode.
Generally Class 0 does not prevent recovery, as the controller has the key and can encrypt/decrypt transparently. The most common reason for an encrypted job to end is a user-set password or a BitLocker layer above it. First, check for the recovery key, since on a work machine it is very likely escrowed with your IT department.
Magician’s monitoring is fine; however, its repair functionality is bad. When the firmware of a drive is updated, the service area which a recovery would need to read is rewritten. Additionally, a PSID Revert operation is intended to cryptographically wipe the drive and therefore should never be performed after a problem occurs with the drive.
The controller is in a deliberate fallback state because it cannot load its firmware or mapping tables. No drive is that size – it’s simply a placeholder and the flash memory behind it probably remains intact.
This could be due to heat; and it is also a good indicator when problems begin intermittently. Copy your files off the drive now – a warm drive that disappears and then comes back on cool indicates that the same thing will happen again.
Prices range from £300 + VAT after a free 48-hour diagnostic. There is a 50% deposit for controller-level work. If the diagnostic cannot access the data because it has been either encrypted without a key being provided, or had the TRIM command issued to it, you will be informed at that time.