Your Hard Drive Is Not Detected.
It May Not Be Dead.

The sound your drive makes, or does not make, is the most important diagnostic clue. A silent drive, a spinning drive with no data, and a clicking drive each point to different failure categories. Before you download any software, listen closely and match the symptom below.

A drive goes undetected for one of three reasons: an electrical short on the PCB, corrupted firmware in the Service Area, or a mechanical fault such as a seized motor or failed heads. Each failure requires a different repair path and falls into a different pricing tier.

PCB / Electrical Failure

A shorted TVS diode or failed voltage regulator on the circuit board prevents the drive from receiving stable power. The platters and heads are typically undamaged. Repair involves replacing the failed component or transferring the ROM chip to a donor board. This falls into the firmware pricing tier ($600–$900) for hard drive recovery.

Firmware / Service Area Corruption

The drive spins but cannot complete initialization because its internal microcode is corrupted. It may report 0 GB capacity, display a wrong model name, or hang during the BIOS handshake. PC-3000 vendor-specific terminal access (Seagate F3, WD COM) is required to read and rebuild the corrupted firmware modules.

Mechanical Failure

Failed read/write heads or a seized spindle motor prevent the platters from spinning or the drive from reading its own Service Area. The drive may be silent, clicking, or beeping. Recovery requires opening the drive in a 0.02 micron ULPA-filtered clean bench and transplanting donor parts. This is the most expensive tier ($1,200–$1,500 plus donor cost).

Logical / File System Corruption

The drive hardware is healthy but the partition table, MBR, or file system metadata is damaged. The OS sees the drive as RAW or Unallocated. This is recoverable without opening the drive and falls into the lowest pricing tier ($100 to From $250 for standard recovery).

"Not detected" is not a single failure. It describes three distinct breakdowns: at the BIOS/UEFI hardware layer, the operating system driver layer, or the file system metadata layer. BIOS-invisible drives need PCB or firmware repair with PC-3000. OS-invisible drives may need a firmware patch. File-system-invisible drives often need only logical reconstruction.

BIOS/UEFI Not Detected

The motherboard firmware cannot see the drive on the SATA or NVMe bus. No entry appears in BIOS storage settings. This means the drive fails the initial identification handshake entirely. Common causes: dead PCB with a shorted TVS diode, seized spindle motor, or firmware corruption in the Service Area that prevents the drive from reporting its model string. No operating system or recovery software can address a device the BIOS cannot find. Hardware-level diagnosis with PC-3000 is required.

OS Not Detected

The BIOS sees the drive and reports its model and capacity, but Windows Disk Management or macOS Disk Utility does not list it. The drive responds to the SATA/NVMe handshake but stalls during initialization. For NVMe drives on 11th Gen+ Intel systems, Intel VMD can hide drives unless the RST driver is loaded. If VMD is not the cause, the firmware translator module or defect list is corrupted and requires firmware-level repair.

File System Not Detected

Windows Disk Management shows the drive as Unknown, Unallocated, or RAW. The hardware works and firmware is intact; only the partition table or file system metadata (NTFS, APFS, exFAT) is damaged. This is the most favorable scenario for data recovery. Professional imaging followed by file system reconstruction can recover data without opening the drive. Do not format the drive or run chkdsk.

These videos show the real process of diagnosing a not-detected Western Digital drive and recovering a locked Seagate Rosewood firmware. No marketing footage, voice-overs, or stock footage; actual on-bench work with PC-3000 terminal access, ROM extraction, and Service Area patching.

Why a dead drive might just be a PCB issue.

Fixing a locked Seagate firmware.

If the drive becomes visible after swapping cables but Windows prompts you to format it or shows it as RAW, the file system has sustained logical damage. Stop using the drive and review the steps for corrupted hard drive recovery before taking any repair actions.

Before assuming an external drive is dead, rule out the USB bridge chip in the enclosure. A failed bridge looks like a dead drive: no enumeration and a flickering LED, so the bridge is worth ruling out first, even though mechanical damage to the drive mechanism itself is the more common cause of external drive failure. WD My Passport drives use a Native USB PCB with no separate bridge, so shucking does not apply.

The canonical procedure is to extract the bare drive from the enclosure and connect it to a known-good SATA port on a desktop motherboard. If the drive enumerates on direct SATA when it would not enumerate over USB, the bridge was the failure and the drive itself is fine.

Bridge replacement or transplanting the bare drive into a compatible enclosure resolves the case. If the drive remains undetected on direct SATA, the failure is on the drive itself: PCB, firmware, or mechanical.

A drive that enumerates normally on direct SATA but not through its case is an external hard drive data recovery case.

From there the diagnostic tree below applies. Spin behavior matters at this stage too. A silent drive that stays silent on direct SATA points at PCB or seized-motor territory, the same path covered on our hard drive motor failure page.

Differential Procedure: Bare Drive on a Known-Good SATA Port

1. Open the external enclosure carefully. WD Passport and Seagate Backup Plus housings use plastic clips, not screws; a thin pry tool along the seam separates them without breaking the latches.
2. Identify the bridge PCB versus the drive PCB. The bridge PCB carries the USB connector and the JMS-series, ASM-series, or OXFW971-family bridge chip. The drive PCB is the controller board screwed to the bottom of the bare 2.5 inch or 3.5 inch HDA.
3. Disconnect the bridge from the bare drive. Most enclosures use a direct board-to-board SATA edge connector; some 2.5 inch units use a short SATA data and power ribbon.
4. Connect the bare drive to a desktop motherboard using a known-good SATA data cable and SATA power from the PSU. Use a desktop, not a USB-to-SATA dongle, so you are not just adding a second bridge chip into the chain.
5. Power on and check BIOS/UEFI storage settings. If the drive enumerates with the correct model and capacity, the bridge was the failure point. If the drive still does not enumerate, the failure is on the drive itself and the diagnostic tree below applies.

Device Manager & Disk Management answer two different questions. A drive on a direct SATA port listed under Device Manager's Disk drives node answered the identify handshake, so the bus built a device object for it. A disk row in Disk Management means the storage stack got further still: it read the disk's reported capacity & went looking for a partition table on it.

BIOS/UEFI setup is the third reading, & it is the earliest one. It lists what answered at power-on, before a single Windows driver loaded. Which of those three screens the drive reaches, & which one it stops at, is the first branch of this whole diagnosis.

Read all three. The pair of answers you end up with already names the layer that failed. A drive still inside a USB enclosure reads differently, because the bridge chip is what answers the host, so an entry can appear under Disk drives while the drive behind the bridge is dead.

Observation Order for Device Manager and Disk Management

1. Press Win+X & choose Device Manager from the menu that opens. Reading that list shows what the storage stack already enumerated; it sends nothing new to the drive.
2. Expand the Disk drives node & read the model strings under it. A model matching the drive in question means that drive answered identify. Note any entry carrying a warning marker.
3. Expand the Storage controllers node & read what is listed there. A controller that is missing or flagged moves the problem toward the host, & the drive's own entry is not the only thing that can be absent.
4. Press Win+X again & choose Disk Management. Read three things off the disk list: the row for the disk, the size it reports, & whether a partition entry sits inside it.
5. Compare the two screens against the table above, then close both. That comparison is what identifies the branch.

None of those steps updates a driver, uninstalls the device, rolls a driver back, scans for hardware changes, or opens the Volumes tab in the device's Properties & presses Populate. Each of those sends commands or I/O to the drive. Populate queries the disk for its partition & volume layout, a rescan or a device reinstall re-runs the whole enumeration & read sequence, and on a drive with degraded heads every one of those retries is more time the heads spend working weak media.

A silent drive that never reaches BIOS/UEFI setup is a drive that won't spin up. One that reports a capacity that doesn't match its label is service area firmware corruption.

A drive shucked out of a USB enclosure continues as external drive recovery, & a 2.5 inch drive pulled from a notebook as laptop hard drive recovery. Every drive-side branch gets imaged on a DeepSpar Disk Imager or a PC-3000 Portable III before a file system is touched, & we don't charge to look at it first.

Once the BIOS reports the drive's model & its labeled capacity, the drive has finished its own firmware initialization and answered the identify command. That is not proof the user area reads cleanly. Windows then shows one of three states: a volume marked RAW, a partition sitting there with no drive letter, or a disk reported as unallocated or not initialized.

Disk Management is worth opening here as a read-only instrument, nothing more. You are reading three fields off the screen: the capacity the disk reports, whether a partition entry exists inside it, and whether that partition carries a recognized file system. Those three fields separate a mount-layer problem from file system damage on the platters, and they separate both of those from a drive that is misreporting its own geometry.

Volume shows as RAW

The partition entry was found. The boot sector or file system header inside that partition is not readable, so Windows can't name the file system & falls back to the RAW label. The partition table is what got Windows this far, so that structure reads. What does not read is the file system metadata inside the partition, either because it is corrupt or because the sectors holding it are physically unreadable.

Partition present, no drive letter

The disk row, the partition entry, & the file system label are all listed, but no letter is assigned, so nothing shows up in File Explorer. On a disk reading that way, the fault sits at the mount and assignment layer: the operating system parsed the partition and named the file system, it just hasn't handed the volume a mount point. Of the three states, that is the one that leaves the most structure standing.

Disk shows as unallocated or not initialized

Windows found the disk but no partition entry inside it. Different faults land here. The partition table can be gone, which is a host-side event, or the drive can be reporting the wrong capacity because its firmware address-translation layer is damaged, which is not something the operating system did. Compare the size Windows reports against the capacity printed on the drive label; a mismatch points at the drive rather than at the host.

What Happens If You Accept the Format or Initialize Prompt

Windows offers to format a RAW volume, & it offers to initialize a disk that reads as not initialized. Accepting either prompt is how a recoverable drive turns into a much harder recovery. Both operations write fresh file system or partition structures onto the disk, and what they overwrite is the damaged original that a reconstruction would otherwise have been parsed back out of.

CHKDSK carries a different risk & deserves its own answer. It runs at the file system layer over standard ATA commands, so it can't reach or rewrite the drive's service area firmware or its translator. The damage it does on a mechanically failing drive comes from sustained read retries that work degraded heads across weak media for hours, which is the argument laid out on our page about running CHKDSK on a dying drive.

When to Stop OS-Level Checks and Image the Drive

Reading Disk Management is safe. Everything past reading it stops the moment the drive shows a physical symptom, because reallocated sectors & pending sectors are media & head problems. Stop & get the drive imaged if any of the following is true:

• The drive clicks, grinds, ticks, or spins up and back down.
• Enumeration is slow, intermittent, or the machine hangs while the disk list populates.
• The reported capacity does not match the capacity on the drive label.
• SMART reports reallocated or pending sectors, or read errors climb during use.

Assigning a drive letter is not the free action it looks like. Handing a volume a mount point mounts it, and mounting an NTFS volume lets the file system driver replay its journal, which is a write to the volume. On any disk that trips an item on that list, the next step is a sector-by-sector image on a DeepSpar Disk Imager or PC-3000, not another attempt at mounting it. That imaging-first order is the same one we apply across every hard drive case we take, and it's why we don't charge a diagnostic fee to look first.

When a hard drive won't enumerate on the SATA bus, consumer diagnostics are useless because they depend on the operating system seeing the drive first. PC-3000 connects at the ATA register level and communicates with the drive's controller directly, bypassing BIOS and OS entirely.

Four-Branch Diagnostic Tree

Every non-detecting drive falls into one of four failure categories. PC-3000's first job is determining which branch applies, because each one requires different tools, donor parts, and pricing.

Hot-Swap Detection for Drives That Won't Enumerate

When a drive's firmware corruption is severe enough that even PC-3000 cannot get a response through the normal ATA interface, technicians use the hot-swap procedure to force the drive onto the bus. This works by borrowing a known-good donor drive's initialization, then physically switching to the patient drive while maintaining bus power.

1. Connect a compatible donor drive (matching model family, firmware revision, and head count) to the PC-3000 and power it on.
2. Wait for the donor to reach the DRDY (Drive Ready) state and complete a full Service Area backup.
3. Issue the ATA Standby Immediate command (E0h) through PC-3000. This stops the donor's spindle motor while the SATA bus remains powered and the ATA link stays active.
4. Without disconnecting the SATA or power cables, unscrew the PCB from the donor's HDA (Hard Drive Assembly) and mount it onto the patient's HDA.
5. Spin the patient drive up under the donor's PCB. From there firmware repair and imaging can begin without the patient board ever having to initialize on its own.

Healthy vs Corrupted PC-3000 Boot Trace

The serial debug trace that comes out of the drive's diagnostic port during spin-up is the technician's first read on which branch of the diagnostic tree applies. PC-3000 captures the trace through a TTL adapter on the F3 (Seagate) or UART (WD) header. A healthy drive emits a deterministic sequence of state transitions in a few seconds and lands at the operator prompt. A corrupted drive hangs at a specific transition that names the failed module. Knowing what to read the trace as turns a guess into a diagnosis. For broader background on what the tool does at the protocol level, see what PC-3000 does and the architectural overview of hard drive firmware.

Seagate F3 Terminal Subcodes and SysFile 28 Translator Rebuild

When a Seagate F3 drive cannot complete the ATA handshake, a serial TTL connection from PC-3000 opens a terminal on the drive's debug port. The terminal output identifies which part of the boot sequence failed. BSY means the drive halted while parsing the Service Area microcode. 0 LBA means the drive enumerated but the translator returned zero user capacity, which points directly at SysFile 28, the primary translator. A recurring LED:000000CC FAddr:... string in the terminal loop is an Init SMART Fail or bad-translator condition, the 7200.11-class busy bug, and it blocks the terminal from accepting commands.

On ES.2 and comparable F3 families, the technician interrupts the LED loop with CTRL+Z during the brief window where F3 T>appears. From there the sequence is:/2 to drop to Level 2,Z to stop the spindle, reconnect the PCB to the head stack assembly, U to spin up, then drop to Level 0 and regenerate the translator. This procedure is specific to older F3 drives; running it on a Rosewood or other SMR family wipes the Media Cache Management Table and is destructive, which is why a family-aware diagnosis precedes any terminal write.

SysFile 28 is rebuilt by parsing the P-List (factory defect list) and G-List (grown defect list) and recomputing the LBA-to-PBA offset across each Zone Bit Recording band. ZBR stores physically more sectors per track on the outer zones than the inner zones, so a single corrupted zone boundary shifts the mathematical offset for every sector past that point. Deleting a P-List entry misaligns the entire table at once and makes every file unreadable, which is why the P-List and G-List are both backed up before any write operation touches the Service Area.

Western Digital uses a different file naming convention. On WD SMR drives the dynamic translator is Module 190 (the T2 Translator); Module 32 holds the relocation list. The repair sequence mirrors the Seagate workflow in principle: extract the corrupt module from the Service Area, sort the internal nodes by LBA, identify missing or overlapping nodes, rebuild the valid T2 data, and load it into RAM so the user area becomes addressable.

Seagate F3 Diagnostic Port Lock and Tech Mode Unlocking

Modern Seagate F3 drives lock their diagnostic port at boot, rejecting terminal connections even after a Ctrl+Z interrupt. The drive stays in a BSY state because the firmware security subsystem blocks UART access before firmware modules can be read. PC-3000 dumps the original ROM and applies a Tech Mode unlock patch that restores diagnostic-terminal access for Service Area repair. It does not decrypt user data and it does not clear an SED or ATA password lock.

The Diagnostic Port Lock is a firmware-level gate that activates during the initialization sequence on Rosewood and newer Barracuda families. When the drive powers on, the controller asserts the lock before the F3 terminal prompt becomes available. A standard TTL adapter connected to the UART header receives no response to Ctrl+Z; the drive remains in BSY and standard ATA commands time out because the controller never reaches DRDY.

PC-3000 handles the unlock through a five-step workflow.

1. Open a UART serial connection at 38400 baud. PC-3000 connects to the drive's COM port through a TTL adapter. F3 drives broadcast their diagnostic codes at that rate.
2. Extract the original ROM. The ROM is dumped over that same serial link. It contains the RAP, CAP, & SAP adaptive parameters unique to the patient drive.
3. Generate the Tech Mode unlocking preparation patch. Inside PC-3000 Seagate F3 Utility extensions, the technician builds the patch from the ROM image. It restores diagnostic-terminal access for Service Area repair; it does not decrypt user data & it does not remove an SED or ATA password lock.
4. Issue the unlock command. The technician uses the "Unlock Tech, drive prepared by utility" menu item, which restores terminal and Service Area access. Restarting the drive means running that function again.
5. Access Service Area and User Area. With the terminal unlocked, SysFiles can be read, patched, & rewritten. The translator can be rebuilt & the drive imaged through DeepSpar Disk Imager or Data Extractor.

SMR Rosewood Media Cache Recovery

LED:000000BD on a Seagate Rosewood signals an MCMT exception. Power loss during CMR-to-SMR cache migration desynchronizes System File 348, leaving staged data orphaned in the cache zone. Recovery means locking the drive's background processes out through PC-3000 first, then reconstructing the cache table without erasing the physical platters.

SMR drives write incoming data to a Conventional Magnetic Recording cache zone first, then migrate it to overlapping shingled bands during idle time. System File 348, the Media Cache Management Table, tracks which sectors live in the cache zone versus the shingled bands. When power is lost during a background flush, the MCMT points to bands that were never written, & the firmware enters a BSY panic on the next boot.

CMR Cache Zone

A Conventional Magnetic Recording area where incoming writes land first. Sectors are written side-by-side with normal track gaps. The host sees full write speed because adjacent tracks don't need rewriting.

SMR Bands

Overlapping tracks written like roof shingles, where each new track partially covers the previous one. Data migrated from the CMR cache to SMR bands during idle time gains density but loses random-write capability. If the MCMT loses track of cached data, the drive can't serve reads.

The recovery workflow targets the RAM-resident copy of the MCMT, not the physical platters. Data staged in the CMR cache zone is still physically present; only the table that points to it is corrupt. The goal is to freeze background activity, rebuild the pointer table in RAM, & image the drive before any internal housekeeping can run.

1. Backup SysFiles 1B, 28, 35, & 348. PC-3000 reads each System File from the Service Area before any modification. If a later step fails, the original modules can be restored without additional damage.
2. Lock out the background processes. A firmware-level write lock through PC-3000 halts the drive's own auto-repair and CMR-to-SMR cache migration, so nothing overwrites the orphaned cache data during imaging.
3. Reconstruct SysFile 348 in RAM. PC-3000 rebuilds the Media Cache Management Table in controller RAM without erasing the cached data on the physical platters. The new table maps the existing cache-zone sectors to their correct logical addresses.
4. Image sector-by-sector through the cache layer. Once the MCMT is valid in RAM, DeepSpar Disk Imager or Data Extractor reads the user area through the rebuilt cache mapping. The technician images the drive before power-cycling, because the RAM-resident rebuild disappears on shutdown & the MCMT would have to be rebuilt again.

Locked-Drive Recovery Starts With Family Identification

Western Digital SMR drives with Self-Encrypting Drive locks and modern Seagate F3 drives with diagnostic port locks both block terminal access, and the two are not worked the same way.

The difference is not cosmetic. It determines which PC-3000 utility is loaded, which adapter is connected, and which donor parts are needed. Family identification is the first step in every locked-drive recovery, which is why the drive is identified before any terminal command is issued.

SED-Locked WD SMR Drives and Translator Repair

ACE Lab added an SED unlocking option to PC-3000 for the newer WD SMR families. For additional background on how WD SMR translator corruption develops, see our page on WD SMR translator failure. The firmware repair process is covered in depth on our hard drive firmware reference.

The translator repair itself is the same work as on any other WD SMR drive: extract Module 190 (T2 Translator) and Module 32 (relocation list) from the Service Area, rebuild the corrupted translator nodes, load the repaired module into RAM, and verify that the drive reports the correct LBA count before imaging.

Firmware-tier pricing for this work is $600–$900 ($600 for CMR, $900 for SMR). Rush is available: +$100 rush fee to move to the front of the queue. No diagnostic fee; no data, no fee.

Toshiba MK-series drives fail into a different pattern. Bad sectors trigger a loop of G-List growth and SMART table updates that the drive never finishes, so it never reaches Ready.

Firmware-tier pricing for this work is $600–$900 ($600 for CMR, $900 for SMR). Rush is available: +$100 rush fee to move to the front of the queue. No diagnostic fee; no data, no fee.

Handoff to DeepSpar for PRML Read Channel Imaging

Once the translator rebuild completes and the drive reports the correct LBA count, the user area still has to come off without further damaging the heads. OS-level tools and desktop cloning software wait on the drive's internal retry timers, which on a marginal head means seconds of re-reading the same damaged track. That continuous contact generates heat at the head-platter interface and is how a recoverable drive becomes a platter-damage case.

DeepSpar Disk Imager bypasses host timeouts and talks to the drive's ATA registers directly. The workflow is multi-pass. The first pass reads only fast-responding sectors and skips anything that does not return within a short configured window, building a per-head bitmap of good reads. The second pass targets the skipped sectors, often reading them in reverse, because asymmetric head wear sometimes lets a sector read cleanly when approached from the opposite direction. Per-head isolation means data on the three healthy heads of a four-head drive is imaged first while a failing head is left for last, minimizing exposure.

Modern drives encode data through a Partial Response Maximum Likelihood (PRML) or Extended PRML read channel. The analog waveform from the head passes through a Continuous Time Analog Filter and an FIR digital equalizer, then a Viterbi detector resolves the most probable bit sequence. After a head swap from a matching donor, the signal the new stack delivers no longer matches the channel the firmware was calibrated against. PC-3000 retunes the read-channel adaptives, the FIR equalizer tap coefficients, gain, and target response, so marginal reads off the donor heads resolve.

On Western Digital drives the microjog and servo adaptives live in ROM Module 47, shadowed in the Service Area. Donor head work fails when the patient heads are driven with the donor's Module 47 values: the read channel cannot lock onto servo tracks, and the drive clicks or reads at kilobytes per second. The adaptives are transferred, or the ROM is regenerated, before imaging starts.

For the physical connection, PC-3000 Portable III handles SATA, PATA, USB, and NVMe connections through a laptop-based interface. PC-3000 Express is the lab-based PCIe card with multiple SATA and PATA ports. For a BIOS-invisible HDD case the Portable III runs the terminal and the initial diagnostic pass, then the stabilized drive is handed to DeepSpar for the bulk imaging phase.

Head-swap tier work is $1,200–$1,500 plus donor cost. Donor drives are matching drives used for parts. Typical donor cost: $50–$150 for common drives, $200–$400 for rare or high-capacity models. We source the cheapest compatible donor available. Rush is available: +$100 rush fee to move to the front of the queue. Our HDD work routes back to the hard drive data recovery page for full tier detail.

PCB ROM Extraction and Donor PCB Matching

On post-2008 hard drives the SOIC-8 SPI flash chip on the PCB stores adaptive parameters that are unique to the head stack inside that specific HDA. Voice Coil Motor coefficients, preamplifier gain matching, thermal calibration tables, and zone mapping all live in this ROM. Swapping a donor board WITHOUT transferring this ROM produces a clicking drive even when the silkscreen board number matches, because the donor PCB is calibrated to a different head stack and cannot drive the patient's preamplifier inside its tuned range. For the full breakdown of what each component on the board does, see hard drive PCB components.

ROM Transplant Workflow

1. Identify the SPI flash on the patient PCB and confirm its package before any heat goes near the board.
2. Apply a thin layer of liquid no-clean flux around the chip leads. Flux is non-negotiable on lead-free assemblies; without it the joints will not reflow cleanly and the chip body will lift before the leads release.
3. Reflow with hot-air rework at controlled temperature. We use a Hakko FM-2032 micro-precision iron on an FM-203 or FX-951 base for any pad rework, and an Atten 862 hot-air rework station for the chip lift itself. Nozzle sized to the SOIC-8 footprint, with surrounding components shielded by Kapton tape.
4. Lift the chip with vacuum tweezers once the joints liquefy. Place it on a programmer adapter, dump the contents to a binary file, and verify a clean read with no bit errors before moving on.
5. Reflow the same chip onto the donor PCB with the same flux and temperature profile. Verify orientation by the dot marker; SOIC-8 reversed will short the supply rails and cook the chip on power-up.
6. Mount the rebuilt donor PCB on the patient HDA, power on, and confirm the drive enumerates with the correct model and capacity in BIOS. From there the normal imaging path applies.

PCB Silkscreen and Revision Matching

ROM transplant alone is not sufficient if the donor board revision differs from the patient. Preamplifier impedance, motor driver coefficients, and the PWM tuning for the spindle are all revision-tied. Three silkscreen identifiers on the donor must match the patient exactly before the ROM transplant has any chance of working:

• Board part number. The primary identifier printed in large type, e.g. 2060-771852-001 on a WD board or 100717520 on a Seagate board.
• PCB revision. WD prints REV A, REV P1, or REV P2 in small type near the part number. Two boards with the same part number but different revisions are not interchangeable.
• Motor controller part number. STMicroelectronics SMOOTH, LSI, or TI motor controller part numbers are revision-tied; a mismatch means the drive may spin briefly and then drop out as the motor controller fails commutation.

When a wrong-revision board is used even after a clean ROM transplant, the drive can fail to enumerate or start clicking, and clicking on a previously recoverable drive is how a firmware-tier case becomes a head-swap case. That is why we verify all three silkscreen identifiers before the ROM transplant touches the donor.

Hot-air rework on HDD pages is the documented exception case in our equipment list. Lifting and replacing an SPI flash on a controller PCB is genuine board-level rework, not bench soldering for diagnostic noise, and the work routes through calibrated rework stations rather than improvised setups.

Why PCB Swaps Fail Without ROM Transfer

The ROM chip on a hard drive's PCB stores factory-calibrated data that is unique to the specific head stack and platters inside that individual drive. This is not generic firmware that can be downloaded.

Seagate F3 ROM: RAP, CAP, and SAP Adaptive Parameters

Seagate F3 drives store three sets of adaptive parameters in the SPI ROM on the PCB: RAP (Read Adaptive Parameters), CAP (Controller Adaptive Parameters), and SAP (Servo Adaptive Parameters). They are head-specific. If any of these are lost or mismatched during a PCB swap, the heads cannot locate servo tracks and the drive clicks or reads at kilobytes per second.

WD Marvell ROM: DIR and Loader Modules

Western Digital ROM chips contain the DIR (Directory) and Loader modules that bootstrap the drive's controller. When the Service Area on the platters is too corrupted to read, PC-3000 uploads a loader into controller RAM so the controller can address the Service Area tracks and the modules can be repaired.

This is why the common eBay advice to buy a matching PCB and swap it does not work. The ROM chip must be desoldered from the original board and transplanted onto the replacement, or read via SPI programmer and written to the new board's ROM. Without the original ROM data, the drive will click, spin down, or report incorrect capacity. We perform this ROM transfer as part of every PCB failure recovery.

Before the head disk assembly is ever opened, a drive that will not enumerate gets a bench electrical read. Powering the drive on a current-limited DC bench supply and watching the current draw on each rail tells us whether the fault is on the board or in the mechanics. That single measurement decides whether the drive routes to board rework or to clean-bench work, and it prevents a shorted board from being destroyed by an uncontrolled supply.

Why a current-limited bench supply, not an ATX PSU or USB adapter

A consumer ATX power supply and a USB-to-SATA adapter have no fine current limit. If the board has a dead short, they dump every amp they can into it until something on the PCB burns open. A bench supply lets us set a hard ceiling on current and a target voltage, so a shorted board clamps at the limit instead of cooking. The drive either draws a sane amount of current and starts its spin-up sequence, or it pins the limiter and tells us the board is shorted before any further damage is done.

A 3.5-inch drive uses two rails. The +12V rail feeds the spindle motor and the voice coil actuator. The +5V rail feeds the microcontroller, the ROM, and the read-channel preamplifier that sits on the head stack inside the sealed enclosure. A 2.5-inch drive runs on a single +5V rail and steps it down internally, so on a laptop drive both the logic and the motor read on that one rail.

Reading the current-draw signature

The number on the supply and the sound from the drive together point at the bench the drive belongs on.

The seized-motor and stiction cases are mechanical work covered in the mechanical non-detection section of this page; the electrical read is only how we sort the drive into that bucket without opening the head disk assembly first.

The TVS diode short test with a multimeter

When the supply reports a dead short, the next read is done cold with a digital multimeter in diode or continuity mode. Most desktop boards carry two transient voltage suppression diodes: one protecting the 5V rail and one protecting the 12V rail. In series with each rail is a zero-ohm resistor that acts as a fuse.

Healthy TVS diode

Reads open-loop in reverse bias. A working suppressor is invisible to the rail under normal voltage and only clamps when a surge exceeds its threshold.

Shorted TVS diode

Reads near zero ohms in both directions. A TVS diode is built to fail as a dead short when a surge overwhelms it, acting as a crowbar that shunts the rail to ground, which is the dead short the bench supply caught.

Blown fuse resistor

The zero-ohm resistor fuses are probed as well. An open reading across one means it sacrificed itself to break the short, which can leave the rail dead even after the TVS is dealt with.

When the multimeter cannot localize which component is pulling the rail down, a FLIR thermal camera does it visually. Injecting a low, controlled current into the shorted rail makes the faulty part dissipate that power as heat, and it lights up on the thermal image while the rest of the board stays cool.

Stabilize to image, not repair to reuse

In a recovery lab the board only has to live long enough to clone the data. A shorted TVS diode is frequently just removed, which breaks the dead short and lets the drive enumerate, with the understanding that the board now has no surge protection and may only ever see a regulated bench supply afterward. A permanent fix swaps in an equivalent surface-mount TVS, but that is repair work for a drive going back into service, not recovery work. The goal here is a stable rail for one imaging pass, not a board that lasts another five years.

Motor controller IC versus a blown preamplifier

The two chips fail in different ways under instruments, and telling them apart decides whether the work is on the board or inside the head disk assembly.

Motor and VCM controller IC

The combo chip that drives the three-phase spindle and the voice coil actuator sits on the PCB. It is the part labeled SMOOTH on many Western Digital boards and a Texas Instruments part on many Seagate boards. When it fails the drive usually goes silent or hums faintly and does not click; a shorted output stage shows a massive spin-up current draw. That is a board-rework call.

Read-channel preamplifier

The preamp lives on the actuator inside the sealed enclosure on the 5V rail. A blown preamp lets the drive spin up normally, but the heads cannot read the servo wedges, so the actuator recalibrates against the crash stop and the drive clicks in a rhythmic pattern. A drive that spins and clicks with a PC-3000 log showing it cannot read the Service Area points at the preamp, which means a matched head-stack swap with donor head matching on the clean bench, not board work.

This electrical read is what sets the recovery tier before any quote is given. A shorted TVS or a dead motor controller that an imaging-grade board fix can solve routes the case into the firmware and PCB tier of our hard drive data recovery workflow at $600–$900, while a confirmed preamp failure that forces a head-stack swap moves it into the $1,200–$1,500 head-swap tier.

Hand-off to imaging once power is stable

Once the rail is stable, whether the original board was patched or the original ROM was moved onto a donor board, the drive goes to a hardware imager such as the DeepSpar Disk Imager or PC-3000, never through the operating system. The drive's background routines are disabled and short read timeouts are enforced so a marginal head cannot hang the channel. A fast first pass grabs the readable areas, slower retry passes recover the rest, and a per-head map lets the healthy heads image first. File-system reconstruction happens only on the finished clone; the patient drive is never written to and never mounted.

Certain hard drive families fail to detect far more often than others due to their firmware architecture. Seagate Rosewood, WD Palmer, and Toshiba MK each require a different PC-3000 workflow. Applying the wrong procedure to the wrong family will destroy data permanently.

When an SSD disappears from your BIOS or Disk Management, there are no sounds to diagnose; SSDs fail silently. However, if your SSD is scalding hot to the touch immediately upon booting, disconnect it. Do not attempt to recover data via software. You have a shorted PMIC. Otherwise, an invisible SSD is commonly a controller lockup, firmware corruption, or a motherboard configuration issue masking the drive. Try these steps:

1. 1
   Reseat the M.2 drive at a 30-degree angle and check the standoff.A loose connector accounts for many "dead" SSD reports. Never screw the drive directly flat to the motherboard without the proper standoff; bending the drive will crack the solder joints under the controller and permanently destroy it.
2. 2
   Check if the drive appears in BIOS but not in Windows.If you have an 11th Gen or newer Intel CPU, your drive may simply be hidden by Intel VMD. You must inject the Intel RST driver during Windows setup or disable VMD. Otherwise, visible in BIOS but absent in Windows points to a partition or firmware issue.
3. 3
   Try a different M.2 slot or a USB adapter.M.2 is a shape, not a protocol. A SATA M.2 drive in an NVMe-only slot will not be detected. Also, motherboards often share PCIe lanes, disabling M.2 slots if certain SATA ports are in use. A second slot rules out these conflicts.
4. 4
   If you ruled out configuration issues and it is still invisible, the controller is dead.If VMD is disabled, the slot is correct, and the drive is not shorted, a completely invisible drive has a dead controller. Software cannot help. Professional firmware-level tools like PC-3000 are required. See our SSD data recovery service.

SSD Controller Firmware Failures That Cause "Not Detected"

When an SSD disappears from BIOS, the root cause is almost always a controller firmware panic, not a cable or driver problem. Each controller family fails in a distinct way, and each requires a different PC-3000 recovery approach for SSD data recovery.

Silicon Motion SM2258 FTL Corruption (Crucial MX500, ADATA SU800)

When the Flash Translation Layer corrupts, the SM2258 drops to 0 bytes and enters ROM mode or a BSY safe state. The drive reports 0 bytes in Disk Management but responds to SATA identification. Recovery requires PC-3000 SSD to inject a loader into the controller SRAM and rebuild the block mapping tables. Software cannot scan a 0-byte drive.

Samsung Phoenix Controller (970 EVO, 970 EVO Plus)

The drive may cycle between detected and undetected states on each reboot, or disappear entirely after a power cycle. There is no PC-3000 FTL-reconstruction path for the Phoenix controller; it is board-repair-only work, which means keeping the original controller alive because it holds the encryption key. This routes into NVMe recovery.

Phison E16 PCIe Gen4 Initialization Failure (Corsair MP600, Sabrent Rocket 4.0)

The Phison E16 Gen4 controller can fail its PCIe link training sequence, causing the drive to be invisible to BIOS despite intact NAND. Unlike the SATA Phison controllers that fall back to SATAFIRM S11, Gen4 NVMe controllers provide no fallback identifier. The E16 is one of the controllers PC-3000 SSD can rebuild at the firmware level, so a drive that will not link-train still has a path that does not start with lifting NAND off the board.

See also: SSD Shows 0GB or Wrong Capacity

We charge based on what the problem is, not a flat worst-case rate. Firmware repair, head swap, and logical recovery each fall into a different pricing tier. The table below shows our published pricing and exactly what each tier covers.

We provide a firm quote after free evaluation. If it turns out to be firmware instead of heads, you pay the firmware price, not a flat "worst-case" tier.

Each failure type carries a different amount of bench work, which is what hard drive data recovery cost by failure type separates out.