Computer Repair · October 26, 2026 · ComputerKing Team

SSD Not Detected? The 72-Hour Assessment-or-Give-Up Guide

After your SSD dies, the next 72 hours decide whether your data can still be saved: stop, triage the three kinds of 'not detected', recover your encryption keys, then choose assessment or give-up.

<a id="summary"></a>Key Summary

Your SSD is not detected — do not rush into repair attempts. 72 hours is not a data expiry period, only a decision window. This article uses a three-way classification plus a 72-hour decision tree to show you how to stop, triage, recover your encryption keys, and then decide whether to send the drive for assessment or give up.

<a id="s1"></a>1. First, Be Clear: 72 Hours Is Not a Data Expiry Period

Many people hear "72 hours" and assume the data will automatically disappear after three days, so they panic and try anything. No public source says that. In this article, "72 hours" is only a decision window: time to calm down, stop using the drive, work out which kind of "not detected" you have, recover your encryption keys, and obtain a written quotation before deciding. It is not a technical deadline. It does not mean that after 72 hours recovery is impossible, and it does not mean that frantic attempts within 72 hours are safe.

There are two genuinely time-related risks, and they point in opposite directions, so they must be kept separate:

First, after deleting files or formatting, TRIM and garbage collection may erase data within a short time. So after an accidental deletion or a mistaken format, the correct response is to power off immediately and stay powered off until assessment, avoiding further writes and avoiding a normal shutdown that gives TRIM another chance to finish.

Second, on a drive whose controller or firmware has already failed, repeated power cycles and restarts can make things worse. So after a failure, the first principle is not "try a few more tricks quickly" but "stop first, classify, then choose a path."

One sentence sums it up: stopping is the fastest rescue; random attempts are the slowest form of damage.

<a id="s2"></a>2. Hours 0–1: Stop, Characterise, Do Not Start by Repairing

When a drive fails, many people's first instinct is to keep rebooting to see whether it comes back on its own, or to click "Format" the moment Windows suggests it, or to run chkdsk or First Aid immediately to "fix" it. All of these are high-risk actions, especially when the drive holds important data.

Why not keep power-cycling? Each power cycle forces the controller to rebuild its mapping table from NAND. If that table is already damaged, repeated power cycles may overwrite residual data that a professional would need to rebuild, making laboratory recovery more complex. Every failed handshake and every re-insertion can trigger further unnecessary work. So "try a few more times and see if it comes back" is not a check; it consumes the drive's remaining chances.

The first fork should not be "which trick to try" but "how important is the data?":

• Data is important, with no backup, and this is the only copy: stop immediately, go to Section 3 to classify the failure, then follow Sections 5–7 to understand the red lines, encryption, and the assessment process, and use the closing decision tree in Section 8 to decide between assessment and giving up. Do not run repair tools, do not format, do not initialise, and do not update firmware.
• Data is unimportant, or a complete backup already exists: general repair, a replacement drive, and restoring from backup can be considered. In this case the goal is "get the computer working again," not "rescue the data."

For example: suppose a user finds that an external SSD has suddenly disappeared from File Explorer, and it holds the only copy of years of photographs. In that situation the right first step is not to download recovery software, but to power off, note down what happened just before the failure (for example, whether the drive was dropped, exposed to water, updated, or disappeared during a large write), and then classify the failure using the three-way method below. This is only a general illustrative scenario, not a real customer case.

<a id="s3"></a>3. The Decision-Tree Backbone: Three Kinds of "Not Detected" Must Be Separated

"Not detected" is not one kind of death. It is three completely different situations, with completely different triage. Getting the classification wrong means every later step may go wrong. Warning signs before failure and triage after failure are two different matters; Further reading: SSD Failure Symptoms: 7 Warning Signs, SMART & Data Recovery.

3.1 Type 1: Completely Invisible in BIOS/UEFI

The definition is simple: boot into the BIOS/UEFI storage device list, and the drive is nowhere in that list. This most likely points to a physical fault or a connection problem, such as controller failure, firmware damage, power-circuit problems, or interface or cable problems.

Note that controller failure can present in many forms; it does not necessarily mean the controller chip itself has burned out. Firmware, power delivery, NAND access, and the translation layer can all produce similar symptoms. So at this stage, do not diagnose which component died from a single symptom. Classification is only for deciding the next step, not for confirming a diagnosis.

Safe preliminary isolation checks apply only while it is still unclear whether the drive, the cable, or the slot is at fault, and they must be done after a complete shutdown: for an external drive, try another cable, another USB port, and another computer, and see whether the problem follows the drive. For an internal drive that would require opening the machine, do not open the SSD yourself or improvise inside the machine; leave that to experienced personnel.

Triage conclusion: if the drive is invisible on multiple machines and in different enclosures, and the data matters, stop. Do not try further repair tools. This level generally requires laboratory handling, and home-level software options are effectively exhausted.

3.2 Type 2: The System Sees It, But Cannot Read or Open It

This covers many screens: BIOS sees the drive but Windows or macOS does not; Disk Management shows "Not Initialised," "Unallocated," or no drive letter; the drive shows as RAW or asks to be formatted; or boot reports no boot device.

This type is the easiest for users to damage themselves, for an ironic reason: precisely because the drive is "visible," people assume running a repair tool is safe. In fact, "detected" only means the controller can still handshake with the host. It does not mean the data in NAND is intact, and it certainly does not mean the drive is healthy.

This branch has two absolute red lines:

• On a drive holding important data, never initialise or format it. Initialisation destroys the file-system index and can cause irreversible data loss.
• Do not click through a format prompt just because Windows keeps showing it. Formatting is a way to make a drive usable again, not a way to rescue data.

On Windows, a drive visible in Disk Management but without a drive letter sometimes simply has no letter assigned. RAW, unknown status, or wrong capacity usually indicates file-system damage or an unsupported file system. "Unallocated" means there is no volume, so File Explorer cannot see it. None of these states by themselves mean the data is already gone, but the order of operations matters: if data must be rescued, rescue methods must copy it out first, and only then can formatting be considered.

On macOS, a drive visible in Disk Utility's sidebar but greyed out means the system sees the hardware but has not mounted it. That is actually relatively good news, not proof of a dead drive. The order of operations must first split by how important the data is: if the data matters, if this is the only copy, or if the drive already shows intermittent or failure signs, skip First Aid, power off, and consider imaging first (only when the Section 5 conditions are met) or sending the drive for assessment. First Aid is only for cases where the data is unimportant or a complete backup already exists, and where the drive is merely unmounted with no other failure signs — and then only once (select the top-level drive first, then each volume below it). If Disk Utility cannot see the drive at all, check System Information under USB or Thunderbolt to see whether the hardware layer sees it; if the hardware layer cannot see it either, the problem leans towards connection, power, or hardware.

3.3 Type 3: Intermittently Detected — Sometimes There, Sometimes Gone

Examples include: disappearing from File Explorer or Disk Management during use; reappearing after a reboot; or disappearing during a large write, with SMART or vendor tools unable to read data.

This is the most dangerous do-it-yourself branch, because "sometimes it works" tempts people to keep retrying, keep copying, and keep scanning. But one intermittent reappearance does not mean the drive has "recovered." How that reappearance should be used depends on a decision already made: if you have already decided to send the drive for assessment, do not keep powering it on to "see if you can copy something off"; if you have already decided not to send it and accept the risk, only then consider copying just the most important files to another drive during a window when it can still be read, accepting that this itself may worsen the damage.

A recent background item deserves careful citation: after the Windows 11 KB5063878 update and related preview updates in 2025, testers reported drives disappearing during sustained sequential writes, SMART becoming unreadable, some drives returning after reboot, and some requiring vendor tools or warranty handling. These are tester reports, not proof that all SSDs were affected. Early lists of affected models were inconsistent, because firmware versions, motherboard BIOS versions, and other variables all affect reproducibility. So do not read this as a settled claim that "a Windows update killed SSDs," nor as something applying to all brands. A safer reading is: there were such reports, the symptom was intermittent disappearance, and triage should first isolate whether the cable, slot, power supply, or host is involved.

Separately, searches for 2025–2026 found no single public recall list covering all brands and all consumer SSD models. The events below are all specific models or specific contexts and must not be generalised: for example, after the Windows 11 24H2 update, some WD NVMe SSDs showed persistent blue screens, with a mechanism related to Host Memory Buffer (the host lending memory to the SSD), and the vendor later provided firmware updates through its Dashboard; HPE SAS SSD advisories and firmware notices for particular server configurations belong to the enterprise/server world and are not general home-SSD recalls. Firmware version numbers in secondary sources are also inconsistent, so readers can check their own machine with the official Dashboard; this article lists no supposedly definitive version table. Dashboard checks and updates apply only to drives still working normally with no unbacked-up important data at risk; if the drive is already in a Section 3.1–3.3 failure state and the data matters, never update firmware — stop and send for assessment.

3.4 Note: Two Easily Misread Special Displays

First, a wrong name or a tiny capacity, such as SATAFIRM S11, 0 MB, or a few MB. This usually means the drive has fallen into a firmware safe mode because it cannot load its own configuration; the data is often still in flash memory. Initialising or formatting in this state writes to the drive and may destroy data needed for recovery. Community discussions mention re-flashing procedures for particular controllers that are themselves destructive: they erase data and rebuild the mapping table, and the drive may only revive briefly. If such procedures are mentioned at all, it must be stated alongside that they erase data and are not data-recovery methods.

Second, a sudden switch to read-only. When flash memory reaches its life limit, the controller may lock into a read-only state to preserve data: files can be read but the drive cannot be formatted or written to. This is an end-of-life state, mentioned here in one line only: on seeing read-only, do not force a format; first copy any readable important files to another drive, then decide the next step based on how important the data is.

<a id="s4"></a>4. The 72-Hour Decision Tree at a Glance

The table below is the backbone of this article. Readers should follow it in order after a failure, without skipping steps.

Logical Damage vs Hardware Damage: Quick Comparison

A reminder: this table only indicates direction, not a diagnosis. Even when the signs lean towards logical damage, if the data matters, the order is still rescue first and repair second, never the reverse.

Factors That Can Fake a Dead Drive

Some "not detected" cases are not dead drives at all, but host or configuration problems. These should be ruled out during hours 1–24:

• SATA: common causes are loose cables, bad cables, or a disabled SATA port. M.2: common causes are a drive not fully seated or a slot incompatible with that drive type. Some M.2 slots support only NVMe, only SATA, or both; inserting the wrong type is a common cause of a "new SSD not detected" report and does not mean the drive is dead.
• Some M.2 slots share bandwidth with SATA ports: after installing an M.2 drive, one or two SATA ports may be disabled. If a SATA SSD is connected to a disabled port, it will "suddenly disappear," which is actually the bandwidth allocation stated in the motherboard manual, not a dead drive.
• Older motherboards may not natively support NVMe booting and may need a BIOS update before recognising NVMe as a storage device (only for new-drive triage, not a failed-drive rescue step). When Windows installation cannot see NVMe, the cause may be a missing storage driver or VMD/RAID mode that needs to be turned off following the motherboard manual (switching the storage controller from RAID to standard mode). Changing controller mode may affect an existing boot drive or RAID array; if unfamiliar, do not change it yourself, and this passage applies only to a "new drive not detected" lookalike, not to an already-failed drive holding important data. BIOS changes of this kind, if unfamiliar, should be left to experienced personnel rather than changed at random.
• Heat: the normal NVMe operating temperature range is generally 0–70°C, and overheating usually causes throttling (slowing down) first, not instant death. Light everyday loads rarely trigger sustained throttling, so "it feels hot" should not be equated directly with "it burned out." A heatsink can itself be the problem, for example if the thermal pad misses the controller, the protective film was not removed, or excessive thickness bends the drive. For external SSDs, insufficient power, a bad USB hub, a bad cable, or a bad USB port can all produce abnormal displays; triage includes connecting directly to a computer USB port, changing cables, and trying another computer.

<a id="s5"></a>5. Red-Light List: Actions That May Leave No Way Back

Each item below is a red light on the "data matters" branch. That does not mean these actions can never be used in any situation; it means that when data must be rescued, doing them may destroy the chance of recovery:

Chkdsk deserves a specific explanation of its TRIM interaction: chkdsk is designed for file-system consistency on healthy drives, and on a failing SSD it writes to the drive while attempting file-system repair. When chkdsk marks clusters as free, Windows issues TRIM commands for those addresses; during garbage collection the controller erases the corresponding NAND pages, and once erased they are permanently gone. The macOS equivalent is the same in principle: repeatedly running First Aid on a damaged APFS volume can make its structure harder to recover.

Firmware updates are not a data-rescue method. The update process stresses the drive and may trigger total failure on an already-failing drive; a firmware update writes to the controller and may permanently kill a marginal controller. Vendor diagnostic tools such as Samsung Magician and Crucial Storage Executive, and the third-party SMART reader CrystalDiskInfo, are for confirming health or warranty status, not for rescuing data. If data matters and the drive is already unstable, do not use diagnostic tools for repeated scans; a normal SMART reading from a diagnostic tool does not prove the data is safe.

On free tools, one principle governs everything: image first, analyse second; never analyse the original drive. The general workflow is: if a drive may be failing, make a byte-level image onto a healthy drive first, then scan the image, not the original hardware; recovered files must be saved to a different drive, never back onto the same drive.

GNU ddrescue is a cloning tool for read errors: it copies the easy-to-read parts first, skips slow or unresponsive parts, and retries them later, and its mapfile (log) allows resuming. But its conditions are narrow and must all hold before it is considered: the drive must be visible in BIOS, and the situation must involve slow reads, bad sectors, or partition damage. If BIOS cannot see the drive at all, skip it and go to a laboratory. The destination drive or image file must be at least as large as the source, and writing to the destination will erase everything on it, including its partition table. If the drive itself is already disconnecting intermittently, ddrescue retries may add to the controller's burden. So ordinary users should not treat ddrescue as a universal tutorial; if the conditions are not met, stop.

The TRIM and accidental-deletion branch needs two situations kept apart. After deleting files or formatting, the operating system issues TRIM commands telling the SSD which blocks are no longer needed, and the controller then erases those blocks in the background. Unlike a hard drive, an SSD does not leave deleted data sitting there waiting to be overwritten. Some sources say TRIM triggers within seconds to minutes of deletion on modern systems; that claim comes from commercial recovery companies and should be written as "possibly very fast," not as a rule for every system and every setting. The correct action after an accidental deletion or format is to power off immediately, avoid a normal shutdown, and stay powered off until assessment. But this TRIM risk applies only when the controller is still working normally and files were deleted or formatted. If the controller itself is dead, a dead controller is not executing TRIM at all, and the data remains in NAND. So "the SSD died" does not equal "TRIM has already erased everything"; the two scenarios must be judged separately.

<a id="s6"></a>6. Encryption: No Key Means No Rescue, and Reset Means Erasure

This section is the easiest to get wrong, so the four encryption types must be kept separate. After a failure, the first task is not to repair the drive but to recover the key on another device and note down the recovery key ID. Before the key is found, do not format, reinstall, or reset anything, because those actions will not remove encryption; they will only destroy data.

BitLocker (Windows)

Without the 48-digit recovery key, even Microsoft Support cannot retrieve, provide, or reconstruct a lost BitLocker recovery key. The proper places to look include: the device's recovery keys under your Microsoft account, matching the first 8 digits of the recovery key ID shown on screen; IT or Entra ID for a company machine; and the printed copy, USB drive, or OneDrive location where the key was originally stored. If partition metadata is damaged and there is no 48-digit key, sources state that the encryption key is, by design, mathematically destroyed and the data is permanently inaccessible. That is a consequence of Microsoft's design, not a matter of insufficient skill.

FileVault (Mac)

FileVault recovery is only possible if decryption is possible first: a login password, a recovery key, or an Apple ID that holds the key. Without one of these, no laboratory can break the encryption. Apple's support documentation warns, in effect, that if FileVault is on and you forget the login password, cannot reset it, and forget the recovery key, you cannot log in and the files and settings are lost forever.

T2 and Apple Silicon add another layer: storage and encryption keys are bound to the logic board. On newer Macs with soldered SSDs, removing the drive and reading it elsewhere does not work; the logic board must be repaired to a state where it can read the data. Without the password or recovery key, even Apple itself cannot read it. So Mac users should ask three questions after a failure: Do I still remember the login password? Where is the recovery key? Can I still sign in to the Apple ID? If the answer to all three is no, then "give up the data and repair only the machine" is an honest and reasonable option; nobody should be believed if they claim they can break it.

TCG Opal, SED, and PSID Revert (the Most Dangerous Item)

A self-encrypting drive (SED) encrypts data with a media encryption key inside the controller; an ATA or BIOS password locks the key that unlocks that media key. Performing a PSID revert with the PSID printed on the drive label returns the drive to factory settings, deletes the previous password, and erases all data.

So this must be written in red: a PSID revert is not an "unlock"; it is an erasure. Anyone who tells you to use PSID to "rescue data" misunderstands it. PSID applies only when the data has already been given up and the only goal is to make the drive usable again. The PSID itself must be treated as sensitive information: anyone holding it can destroy the data on the drive.

VeraCrypt and LUKS (Principle Only)

Sources list BitLocker, FileVault, LUKS, and VeraCrypt together as a category where the recovery method depends on the encryption type and on whether key material or a recovery path exists. Without a key, no generic cracking method should be expected. This article states principles only and gives no detailed steps for VeraCrypt or LUKS. The principle is one line: with a key or a backed-up header, there is something to discuss; without one, stop.

<a id="s7"></a>7. Professional Data Recovery: Concepts Only (No Prices)

This article cites no specific price from any Hong Kong or overseas laboratory; all prices are written as "quotation required on a case-by-case basis." What follows is conceptual only, not a promise of recovery.

The first step of any professional process is always assessment or diagnosis: separate logical failure from physical failure, because the two require different work. The PC-3000 SSD concept is to place the drive into the controller's vendor service mode, correct firmware, rebuild the mapping table, and only then make a complete image — never ask the drive to do anything else before imaging. The chip-off concept applies when the controller is completely damaged or unrecoverable: NAND chips are removed from the circuit board, read directly with a dedicated reader, and reassembled in software (interleaving, XOR, ECC, and so on) to restore the file system. Other laboratory concepts include hardware imagers that apply hardware timeouts to stalling drives and retry according to each drive's condition, and PCB repair stations that first repair a burned power section, since an image interrupted by power loss is meaningless. On the working environment, sources mention working in an ESD-safe environment and never writing to the failed drive.

But a laboratory is not magic: files already erased by executed TRIM, a completed Secure Erase, and physically destroyed NAND are categories sources list as genuinely unrecoverable. For encrypted drives, a laboratory cannot bypass encryption without a key (see Section 6). Do not read it as "send it to a laboratory and everything will come back."

Before sending a drive, keep the following: the original drive, the original machine (especially for soldered Mac drives, and both parts of a Fusion Drive together), encryption keys or password hints, and a record of what happened just before the failure (power loss? update? water? drop?). If something was already done wrong (for example, chkdsk was run, or the drive was formatted or firmware-updated), stop immediately and tell the assessor honestly what was done; assessors need to know which structures may already have been rewritten. For how to read a quotation in parts, see SSD Data Recovery: Read the Quote First — 10 Questions on Diagnosis, Imaging, Data Handling and Warranty.

Pre-Assessment Checklist (Written Quotation Is the Standard)

1. Is diagnosis or assessment charged? Will the diagnostic report state the fault category in writing (controller, firmware, NAND, power, logical)?

2. Is the quotation staged? Will the diagnosis result and recovery plan be quoted first, with work starting only after the user's written agreement? Is there a written statement of what happens if recovery fails?

3. Will the original drive be written to, or will only an image be made and work done on the image?

4. Is the work done in their own laboratory, or passed to a third party? If passed on, where will the data travel?

5. Will the original drive be returned? On what medium will recovered data be handed over? After handover, will the other party keep a copy, for how long, and how will it be destroyed?

6. How is encryption handled: are BitLocker recovery keys, FileVault passwords, or PSIDs required? How will they be stored after provision? If there is no key, will the other party honestly say recovery is impossible rather than charging to try?

7. What is the order of warranty versus data recovery: warranty replacement erases data, so decide first whether you want the data or the replacement; do not send both together.

8. Does the quotation include parts, a destination drive, and handover media, or are those charged separately? (Ask only; this article states no amount.)

9. Is the turnaround time stated in writing? Is there a distinction between urgent and standard jobs? (Time itself requires an individual quotation; never write "it will definitely take X days.")

10. Is there a written receipt listing model, serial number, capacity, fault description, and condition on receipt? Is photographing the drive on handover a recommended practice?

How to verify a quotation: a verbal quotation does not count. Ask the other party to state in writing the diagnosis result, the plan, the charge items, what happens if recovery fails, and the original-drive return arrangement. Do not compare only the total price; compare whether diagnosis fees, start-up fees, post-success fees, destination drives, and handover methods are on the same basis, otherwise a cheaper price may simply cover less.

Warning signals (these are reminders, not source facts): guaranteeing recovery before seeing the drive, or guaranteeing a particular success rate; telling you to try a PSID revert, format, or firmware update first; asking you to send encryption keys through insecure channels without explaining storage; refusing a written quotation or refusing to explain the original-drive return arrangement. If the data is extremely important and the first shop's explanation is vague, a second written assessment can be considered for comparison; but each handover involves transport and power-on risk, so do not shuttle the drive endlessly.

A Hong Kong information gap must be stated honestly: this research found no public, verifiable price list or success-rate statistics for Hong Kong local SSD data-recovery shops, so this article names no Hong Kong shop, no Hong Kong price, and no Hong Kong success rate. The following items have no public source and are not stated: average Hong Kong charges, average Hong Kong turnaround times, success rates by fault type in Hong Kong, and specific complaint figures for data recovery.

<a id="s8"></a>8. The Closing Decision: Send for Assessment or Give Up

• Data backed up or unimportant: no need to send for assessment; replace the drive, check the backup, and improve backups afterwards.
• Data important, no encryption key: find the key first; if it cannot be found, honestly face the possibility that recovery may be impossible, and do not pay someone to try breaking encryption.
• Data important, key available (or no encryption), drive completely invisible: stop and send for assessment; do not try further.
• Data important, drive intermittently visible: do not keep retrying; decide once: send for assessment, or accept the risk and copy only the most important files when a read window appears (accepting possible further damage).
• Data important, drive visible but unreadable: do not initialise, format, or run repair tools; image first (only when the conditions in Section 5 are met), otherwise send for assessment.
• Something already done wrong: stop immediately and tell the assessor honestly what was done.
• Decision to give up: handle the old drive safely; do not throw it away or sell it directly. If encryption applies, PSID revert or Secure Erase applies only when the data has already been given up, and must follow vendor guidance.

Giving up without sending the drive can equally be a reasonable decision, and this article should not create anxiety about it: for example, when the data can be fully retrieved from the cloud, email, a client, or another device; when the drive has already been deleted or formatted with TRIM and there is no other copy; or when the user has accepted the data loss and the risk of further power-on testing outweighs the benefit. In those cases, move on to replacing the drive, rebuilding backups, and disposing of the old drive properly.

How to Verify Recovered Data

1. Copy first, use second: recovered data should first be copied in full to another healthy or new drive. Do not edit files directly on a temporary drive handed back by a recovery company, and do not write the recovery result back onto the original failed drive.

2. Spot-check different file types: open documents to check for missing text, broken formatting, or damaged images; view images for mosaic artefacts, truncation, or pixel anomalies; play videos for stuttering, missing frames, or missing audio; watch for internal errors when extracting archives; use built-in integrity tools for databases.

3. Check counts and lists: verification should compare against an expected list. Counting can reveal a whole missing group, but a correct count does not prove every value is correct; key files must also be checked for openability and key relationships must be complete.

4. Checksums answer only one question: if file hashes (for example MD5 or SHA-256) were recorded beforehand, a match proves consistency with the recorded file. Without a prior hash, checksums cannot prove completeness; only "it opens and spot checks show no anomalies" can be used.

5. Check file metadata: sudden anomalies in size or modification date are warning signs, but normal metadata does not prove the content is undamaged; it must be combined with opening and spot-checking files.

How to Handle the Original Drive: Iron Rules

Do not format the original drive immediately: until recovered data has been copied to a new location, passed spot checks, and a second backup exists, the original drive must stay untouched — no formatting, no writes, no daily use. Do not treat the original drive as "fixed" and return it to daily use: even if it suddenly becomes detectable again, that does not mean it is reliable again; its correct status is "evidence drive pending handling," not "spare drive." Before disposal or recycling, if the drive once held sensitive data, safe erasure or physical destruction must be considered; a quick format only deletes the file index and does not mean the data has been erased, but the specific safe-erasure method for an SSD must follow vendor tools or professional advice, and this article gives no generic erasure steps.

<a id="s9"></a>9. After Replacing the Drive: Brief Prevention, Not a Second Article

This section is brief by design. After replacing the drive, the 3-2-1 backup rule is worth adopting: 3 copies, 2 different media or devices, 1 copy off-site or in the cloud. Further reading: Hong Kong Backup Strategy Guide: The 3-2-1 Rule Done Properly — Time Machine, Windows, Cloud and Offline Drives. Time Machine is already one layer, but it cannot be the only layer. Time Machine keeps hourly backups for the past 24 hours, daily backups for the past month, and weekly backups for earlier months, deleting the oldest snapshots when the disk fills; Apple recommends a Time Machine drive at least twice the size of the Mac's internal storage. An advanced version adds one offline or immutable copy and aims for zero backup errors, meaning restores are tested regularly rather than set and forgotten.

Backup frequency should follow how much recent work you can afford to lose: active work may need hourly, continuous, or daily protection, while stable archives can be less frequent; automation beats relying on memory. Test restores monthly by restoring a few files and checking whether they open; "backup job completed" does not mean the data is usable. A few further points: do not leave a backup drive permanently connected; sync is not backup, because a sync service may propagate an accidental deletion or corruption to all devices, so choose a service with file versioning and recovery, not just a sync folder; and encryption passwords or recovery keys for backups must be stored separately and safely. After replacing a drive, note the model and firmware version, use vendor tools to check for updates, and stop and back up immediately on noticing abnormal slowness or disappearance. Further reading: SSD Upgrade Cheat Sheet: What to Buy, How to Install, What to Watch Out For.

<a id="s10"></a>10. Common Misconceptions, Cleared Up in One Place

First, the freezer trick: it is simply wrong for SSDs, and it carries risk. The core reason is that SSDs have no moving parts, so freezing cannot realign shifted components; the freezer method was a hard-drive-era folk remedy for a stuck head and does not apply to SSDs. As the drive returns from cold to room temperature, moisture in the air condenses, and extreme temperature change itself carries potential damage to any electronic component. Laboratories report receiving drives that went from "still bad" to "completely unrecoverable" after freezing; even for hard drives, freezing is no longer recommended. The correct comparison is: the freezer trick was a hard-drive-era folk remedy, now not recommended even for hard drives, and even less suitable for SSDs.

Second, repeated power-off and restart "wake-up" attempts: this is a misconception this decision tree targets directly, because users with intermittently detected drives fall for it most easily. Each power cycle forces the controller to rebuild its mapping table, each failed handshake may worsen firmware damage, and each plug-and-unplug triggers extra work. One intermittent return does not mean the drive is well again.

Third, swapping the circuit board: even for hard drives this is already complex, requiring full compatibility and professional knowledge, and a wrong swap can permanently lose data. For SSDs, controller, firmware, and mapping table are bound together, so it is even less a "swap the board and it works" concept. The direct source for board swapping concerns hard drives, so when cited it must be stated as an analogy, not as a tested conclusion about SSD board swaps.

Fourth, "recovery software can fix any drive": software can only handle logical problems and cannot communicate with an undetected controller; repeated reads on a failed controller add burden.

Fifth, hair-dryer heating, tapping, and similar tricks: this research found no reliable source supporting these methods. They can only be written as "no public source supports them; do not try them," without inventing principles or success stories.

Sixth, "data is still there after formatting": do not think that way on SSDs. On a hard drive, a quick format usually only deletes file-system references and the data itself is not immediately erased; on an SSD, a quick format usually issues TRIM across the whole drive, and data may begin to be erased within minutes (a commercial recovery-company claim, not the same for every system and every setting). The traditional hard-drive rule (quick equals index-only, full equals sector-by-sector overwrite) does not apply to SSDs; the key variable on an SSD is whether TRIM executed, not whether the format was quick or full. Exceptions exist — for example, a USB enclosure that does not pass TRIM, an operating system or file system without TRIM support, some RAID configurations, or immediate power-off after the event — where data may not have been erased. But it must be stated that an exception does not guarantee recovery, and readers must not be told to gamble on it.

<a id="s11"></a>Conclusion: Remember Three Sentences

Stop, triage the three kinds of "not detected," recover your encryption key and obtain a written quotation, and only then choose between assessment and giving up. 72 hours is a decision window, not a data expiry period; with important data at stake, do not format and do not repeatedly power-cycle, because one wrong step may be irreversible.

<a id="faq"></a>Frequently Asked Questions

Q: My SSD is suddenly not detected. What should I do first?

A: Stop. Judge how important the data is and whether a backup exists, note what happened just before the failure, then classify layer by layer: BIOS, system disk tools, File Explorer. If the data matters, do not run repair tools, do not format, and do not repeatedly power-cycle.

Q: BIOS cannot see the drive at all. Does that mean there is no hope?

A: Not necessarily no hope, but it leans towards a hardware or connection problem, and home software options are effectively exhausted. For an external drive, after a complete shutdown, try another cable, another port, and another computer. If it remains invisible in multiple environments and the data matters, stop and send it for assessment; do not try further repair tools.

Q: The system sees the drive but prompts me to format it. Can I click through?

A: If the drive holds important data, absolutely not. Formatting is a way to make a drive usable again, not a way to rescue data, and initialisation destroys the file-system index. The correct order is rescue data first, then deal with the drive — never the reverse.

Q: The drive appears and disappears. Does that mean it has fixed itself?

A: No. One intermittent return does not mean it is reliable again, and it certainly does not mean it can return to daily use. If you have already decided to send it for assessment, do not keep powering it on to try copying. Only if you have decided not to send it should you consider copying just the most important files during a read window, accepting possible further damage.

Q: After 72 hours, does the data disappear automatically?

A: No. The 72 hours is only a decision window, not a data expiry period, and no public source says SSD data automatically disappears after 72 hours. The genuinely urgent cases are TRIM risk after deletion or formatting, and worsening damage from repeated power cycles on a failed controller — two separate risks.

Q: Without a BitLocker recovery key, can a laboratory break the encryption?

A: No. Without the 48-digit recovery key, even Microsoft Support cannot retrieve, provide, or reconstruct it. FileVault is the same: without the login password, recovery key, or an Apple ID holding the key, no laboratory can break the encryption. After a failure, look for the key on another device first; before the key is found, do not format, reinstall, or reset.

Q: Is a PSID revert a way to unlock and rescue data?

A: No. A PSID revert is not an unlock; it is an erasure. Using the PSID printed on the drive label returns the drive to factory settings, deletes the previous password, and erases all data. PSID applies only when the data has already been given up and the only goal is to make the drive usable again.

Q: What should I ask a shop before sending a drive for assessment?

A: At minimum, ask in writing: whether diagnosis is charged, whether the quotation is staged, what happens if recovery fails, whether the original drive will be written to, whether the work is done in-house or passed on, whether the original drive will be returned, how copies are destroyed after handover, how encryption keys are stored, and the order of warranty replacement versus data recovery. All charges and times are subject to the shop's written reply; this article states no price, and charges require an individual quotation.

Q: Does the freezer trick work for SSDs?

A: No, and it carries risk. SSDs have no moving parts, so freezing cannot realign shifted components; condensation as the drive returns to room temperature may damage electronic components. This was a hard-drive-era folk remedy, now not recommended even for hard drives, and even less suitable for SSDs.

Q: After data is recovered, can I format the original drive and use it again immediately?

A: No. Until recovered data has been copied to a new location, passed spot checks, and a second backup exists, the original drive must stay untouched. Even if it suddenly becomes detectable again, that does not mean it is reliable; do not treat the original drive as a fixed daily drive.

<a id="sources"></a>Sources

All sources below come from the research reports; no URLs were invented for this article:

• DataRecovery.com: SSD not detected troubleshooting
• Gillware: signs of SSD failure
• MDRepairs: SSD not showing up guide
• Intel: memory and storage support article
• Stellar: fixing an SSD not detected by BIOS or OS
• MDRepairs: SanDisk SSD errors
• Rossmann Group: chkdsk on a failing drive myth
• Rossmann Group: DIY hard-drive data recovery
• Gentoo Wiki: Ddrescue
• Microsoft Support: finding your BitLocker recovery key
• Intel: SSD support article
• Windows Forum: Windows 11 KB5063878 SSD issues
• MDRepairs: SSD data recovery
• Norwich Data Recovery: SSD recovery
• Darwin's Data: why put a hard drive in a freezer
• DataRecovery.net: data recovery myths
• TeraDrive: deleted-data recovery from SSD and the TRIM function
• MDRepairs: does TRIM make SSD recovery impossible
• Dell Support: locating the BitLocker recovery key in your Microsoft account
• MDRepairs: FileVault data recovery
• SOS Data Recovery: Mac data recovery, SSD, T2, FileVault
• iBoysoft: SSD not showing up
• Sekin: fixing an SSD not showing up on Windows, full guide
• Denver Data Recovery: SSD failures guide
• HPE: SAS SSD advisory
• OWC: Time Machine is just the start, how to build a 3-2-1 Mac backup plan
• Lord AI: building a 3-2-1 computer backup system

Additional note: some sources above are articles from commercial data-recovery companies, which have an incentive to steer users towards sending drives for repair and are not neutral standards; Intel, Microsoft, and similar sources are closer to primary vendor or official sources. Commercial-source claims in this article are used only to describe what the industry commonly says, not as proven general facts. No public price list, success-rate statistics, or turnaround times were found for Hong Kong local shops, so all such items are left blank and charges require an individual quotation.