Mechanical noises (clicking, beeping, grinding) mean stop powering the drive if the data matters — that is recovery-service territory. A drive invisible to the BIOS is usually a cable, port, power, or 3.3V-pin problem before it is a dead drive. Visible-but-unmounted is usually fixable yourself. Verdicts per symptom below, live replacement prices further down.
A steady click-click-click on power-up is the read/write head assembly failing to find its servo data and resetting — over and over. The heads sweep, miss, park, and try again, and each cycle is a chance for physical contact with the platter surface. This is the most serious common symptom a drive produces, and the folk name is earned: drives that click rhythmically at power-on rarely return to service.
The one honest instruction for a clicking drive with data you need: stop powering it on. Every additional attempt risks turning a recoverable head-failure into unrecoverable platter damage, and professional labs price jobs partly by how much post-failure runtime the drive has accumulated. If the data matters, this is the recovery-service scenario, full stop; if it is backed up, the drive is a warranty claim or e-waste.
A beep or buzz from a drive (distinct from a system speaker) is usually the spindle motor trying and failing to spin the platters — commonly seized bearings or, in drives that suffered a fall, heads stuck to the platter surface. The electronics power up, the motor strains, and the sound is the strain. The drive typically never appears to the system because the platters never reach speed.
Like clicking, this is a mechanical fault with no field repair. The folklore fixes for stiction (tapping, twisting the drive) trade a small chance of one last spin-up against a real chance of finishing off the platters — a gamble that only makes sense when the data is worthless and the alternative is the bin anyway. For data that matters, spindle and stiction work is routine for professional labs.
Invisible to the BIOS/UEFI means the drive is not answering the bus at all — and the honest first suspects are not the drive. Swap the SATA cable, try a different port, try a different power connector (or a different USB enclosure/cable for externals). A surprising share of 'dead' drives are dead cables, marginal ports, or an overloaded power splitter, and the swap test costs five minutes.
One false alarm deserves its own paragraph: the 3.3V pin. Many enterprise drives — including nearly every drive shucked from an external enclosure — implement the Power Disable feature, which holds the drive in reset when pin 3 of the SATA power connector carries 3.3V. Desktop PSUs with older cables supply exactly that, so a perfectly healthy shucked drive plays completely dead. The fix is famous and simple (Molex adapter, or masking the pin — details in our shucking guide), and it has resurrected countless 'failed' drives. Rule it out before mourning.
If the drive stays invisible on known-good cable, port, and power — and it is not the 3.3V case — the fault is on the drive's PCB or internals. That is replace territory, or recovery-service territory if the contents matter: board-level faults are among the more tractable jobs for labs.
A drive the firmware sees but the OS does not is usually a logical problem — and logical problems are the ones you can safely fix yourself. Check Disk Management (Windows) or lsblk/Disk Utility (Linux/macOS): a drive present there but unmounted may need a drive letter, carry a filesystem the OS cannot read (an ext4 NAS drive in Windows, say), or show as unallocated because the partition table is damaged.
The one discipline: if the data matters and the partition table looks damaged, do recovery reads before repair writes. Tools that scan for lost partitions or carve files work read-only; 'initialize disk' and quick-format do not, and they are the standard way people convert a recoverable situation into a paid one. Copy first, repair second.
A drive that intermittently drops from an array and returns after reboot is telling you it can no longer answer within the controller's timeout — from deep error-recovery cycles on a degrading surface, an incompatible desktop drive doing minutes-long internal retries where a NAS drive would report the error promptly, or genuine electrical flakiness. NAS platforms respond by kicking the member, and each kick means a rebuild.
Check two things in order: the SMART table (any 05/C5/C6 activity means the dropouts are the surface failing — see the SMART decoder), and the drive model itself (desktop drives without error-recovery control in arrays cause exactly this pattern even when healthy). Either way, a member that has dropped twice has lost its seat: arrays are the wrong place for a drive you no longer trust, and rebuilds are the most stressful workload you can hand the remaining members.
The quiet symptom: transfers that stall, a system that hangs on file access, an OS that intermittently freezes with the drive light solid. Underneath, the drive is spending seconds of error-correction on reads that should take milliseconds — the signature of a surface accumulating weak sectors. The SMART table usually confirms it: reallocated (05) climbing, pending sectors (C5) appearing.
This is the failure mode that grants the most warning, and the honest move is to spend that warning on an orderly exit: full backup at whatever speed the drive manages, then replacement, then — only if curiosity demands — diagnostics on the empty patient. The trap is the opposite order: hours of surface scans and repair passes on a drive full of unbacked-up data, adding load to a surface already failing.
Live replacement options · prices updated every 4-5 hours · last checked 15 min ago · CMR-family drives only (RAID-safe)
| Drive | Cap | Cond | Warranty | $/TB | Price | |
|---|---|---|---|---|---|---|
| WD Ultrastar DC HC520 12TB SATA | 12TB | REFURB | 5 yr | $19.17 | $229.99 | Buy → |
| MDD 12TB SATA 6G 7200RPM NAS | 12TB | NEW | 5 yr | $19.17 | $229.99 | Buy → |
| HGST 12TB SAS 12G 7.2K | 12TB | REFURB | see listing | $21.85 | $262.20 | Buy → |
| HP 12TB SAS 7.2K LFF Enterprise | 12TB | REFURB | 5 yr | $22.62 | $271.40 | Buy → |
Every one of these survives in search results because it occasionally appeared to work a long time ago. On modern high-density drives they range from useless to actively destructive, and the destructive ones share a trait: they convert recoverable failures into unrecoverable ones.
Almost always mechanically failing, yes — the rhythmic click is the head assembly repeatedly failing to calibrate. Whether the DATA is lost is a different question: professional recovery labs regularly recover clicking drives via head-stack replacement. The drive itself, however, is done as a storage device even if recovery succeeds.
Often not. External enclosures fail more often than the drives inside them (bridge boards and power supplies are the weak points). Shuck the enclosure and test the bare drive on SATA directly — remembering the 3.3V pin issue that makes healthy drives play dead on desktop power cables. If the bare drive works, you need an enclosure or a NAS bay, not recovery.
Professional mechanical recoveries commonly run from several hundred to a few thousand dollars depending on damage and lab. That is the honest range — and the honest advice is to spend it only on data that justifies it, decide early, and stop powering the drive while you decide.
Rarely, on modern drives. The PCB carries adaptive calibration data unique to that drive; a donor board without a ROM transfer usually won't run the patient. Board-level work with ROM transfer is standard lab fare — as a DIY fix it mostly worked in an earlier drive generation and survives as outdated advice.
A mechanical wear symptom — marginal bearings or head geometry that gravity is tipping in and out of tolerance. Treat it as a failing drive that has offered you a convenient orientation for the final backup: copy everything in whichever position works, then replace.
If it's in warranty: both questions are covered in our warranty section — the RMA gets you a free replacement in weeks, buying gets you one today. For array members the math usually favors buying now and turning the RMA return into a cold spare; the RMA-vs-replace guide runs the numbers live.
Match or exceed the failed capacity, stay CMR (everything in our picks is), and let $/TB choose among brands — reliability differences between the major enterprise lines are small next to the price spread. If the drive is going into an array, read the failed-RAID guide first for the extra rules.