Everyone uses Samsung SSDs when they need an SSD for a build (the 990 Pro), an upgrade (the 870 EVO) or portable use (the T7), and they fail like every other SSD does: seldom, quietly, and totally. Here’s the picture that relates specifically to Samsung SSDs, including Magician’s honest limits and what you can recover at the chip level with Samsung SSDs.
860/870 EVO (SATA) / 970 to 990 EVO & Pro (NVMe) / T5/T7/T9 portables; different interfaces, same failure signature: perfect yesterday; gone today; data intact behind the fault.
The classic “vanishing” (EVO or Pro) where one morning the drive was gone (no longer visible in the BIOS, tools couldn’t see it) but still showed perfect stats at the time it failed. This isn’t just a Samsung thing ... this is how most SSD failures happen: controller / firmware failure. What is unique about Samsung is the history behind these issues. They’ve had well-documented firmware issues (the old 840 EVO read slow-down & the more recent 980/990 Pro degradation) which have been fixed via firmware updates. These remind us that firmware currency is maintenance for healthy drives, never first aid for failing ones. Portables bring their own twist: T7s get tossed around on desks, under chairs etc., so a dead T-series may sometimes be due to a dead USB interface rather than the actual drive itself. Regardless of platform, one thing remains consistent... when an SSD dies suddenly, the data usually remains intact in the NAND, and can be accessed if you know how.
In the case of Samsung Magician, which we would say is the best of the vendor-specific dashboards, you have wear percentage, how many terabytes written (TBW) to the drive compared to its rating, temperature readings, and firmware updates all clearly explained. If you are going to use Magician at all, do so two times each year on those drives you count on, as detailed by our SSD health guide. And be aware of Magician’s blind spot: while the gauge will measure wear on a Samsung drive, very few Samsungs fail due to wear. Most of them fail from an unexpected electrical failure that no gauge could possibly predict. This is because the chip reporting the data is also likely the chip that has failed. As a result, there is a practical religious approach that takes shape: using Magician for maintenance, and backups for the failure Magician can’t see coming. On the day a drive becomes problematic, there should be no firmware flash attempts, no Secure Erase “reset” attempts, and no long diagnostic tests performed. Simply document symptoms, remove the drive from service, box the drive, and schedule the drive for diagnostic testing.
When the controller still responds to commands in a maintenance mode environment, firmware-based repair/rebuild procedures can recreate the file-to-flash mapping tables used by the drive. When it does not respond to commands in a maintenance mode environment, the path forward is straight-forward: read out the contents of the NAND packages and reconstruct Samsung’s wear-levelling interleave algorithms in software, which, given their highly integrated design architecture, can indeed be quite complex, and is done here. One of the complications associated with this process is that certain hardware encryption configurations found on some Samsung models may limit access to the chip-off path. For this reason, every Samsung job begins with a free 48 hour diagnostic test and a written firm-fixed price quote prior to commencing work, rather than a promise. Additional information regarding accessing NAND packages via a chip-off procedure on a Samsung model can be accessed through the SSD & NVMe page. The distinction between deleted files versus a completely dead drive (TRIM’s territory) resides within the failed-SSD guide.
Magician is good for monitoring items such as flash wear, temperature readings, and firmware versions. A ‘Good’ judgement call made by Magician indicates that there are no issues related to flash wear based on the wear ledger entries. There are other types of failures that occur suddenly without warning, however. These include controller failure or firmware failure (the type of failures experienced by most dead Samsung drives) since the chip that fails is usually the chip responsible for generating the report. Therefore, use Magician for maintenance purposes and make backups of your data for protection/insurance.
Perhaps not; certain families of Samsung NVMe have had documented firmware issues in the past (980/990 Pro health-drop issue, old 840 EVO slowdown) fixed by later updates. Two honest takeaways here: keep your firmware up to date on a healthy, backed-up drive, and never flash firmware as a rescue attempt on a drive already misbehaving with data you need. Fixes prevent; they don’t resurrect.
Generally. The controllers and packaging of Samsung make chip-level work more complex than average, and it’s intricate work we do: firmware-mode access where the controller still answers; direct NAND reads and reassembly when it doesn’t. One caveat asked about early was hardware encryption. Some configurations complicate the chip-off route which is exactly why the free diagnostic determines the plan before any money moves.
Free 48-hour diagnostic: EVO, Pro or T-series, firmware-mode through chip-level. Written quote prior to any work.