Your Hard Drive Is Beeping.
Stop. Unplug It. Now.

Beeping is a symptom, not a diagnosis. Seagate puts insufficient power first on its own list: a single intermittent beep usually means the drive is not getting enough of it, and most beeping happens on USB-powered models with no separate adapter. Rule the power path out before anything else.

When the beep survives a known-good cable, port, and supply, the fault is in the drive, and several of them present the same way: read/write heads resting on the platter surface, a seized fluid dynamic bearing, or a power-rail fault on the drive's own board. No vendor publishes a mapping from a beep pattern to one of those. Which one it is comes from the bench.

Seagate and LaCie Rosewood-family drives (ST1000LM035, ST2000LM007), found inside Backup Plus and LaCie Mobile Drive enclosures, are the most common beeping drives we receive. A clicking hard drive is a different observable state: it reaches speed and then fails to acquire servo, so the actuator resets and retries. Neither sound settles what has to be repaired, and the two do not map to one fix each.

The beep is not a glitch that clears on the next try. The reasons each of the following actions is destructive are covered in detail elsewhere on this page; the short version is that hearing the sound again tells you nothing the power test would not, and every attempt costs data surface where the heads are already down. Do these things in order, then send the drive to a lab that opens it on a clean bench.

How to package a beeping drive for mail-in

A drive that will not spin is fragile in transit, because its sliders may already be resting on the platters. The goal of packaging is to stop any further shock and to keep the drive flat so nothing shifts inside. Follow our mail-in instructions and pack it like this:

1. Anti-static bag first. Slide the bare drive or the whole external enclosure into an anti-static bag. A clean zip bag works if you do not have an anti-static one.
2. Immobilize it. Wrap the bagged drive in two inches of bubble wrap or foam on every side so it cannot move or rattle.
3. Use a rigid box, not an envelope. A padded mailer flexes and lets the drive take impact. A small corrugated box inside a larger one is better.
4. Keep it flat and level. Mark the box so it ships heads-flat, the same orientation the drive sits in a computer. Do not let it tumble loose.
5. Label it a non-working drive for recovery. Note inside the box that the drive beeps and is being sent for data recovery, with your name and contact details, so it routes straight to the bench.

Restraint is the variable that sets your price. A single-event stiction case where the drive was unplugged on the first beep and the original heads survive the clean-bench unstick stays in the firmware tier at $600–$900. Once the sliders have been dragged enough to fail the read test, the job becomes a donor head swap at $1,200–$1,500, and if repeated power cycles scored the platters it escalates to the surface-damage tier at $2,000.

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. We diagnose the drive for free and tell you which tier applies before any work starts; the full service is described on our hard drive data recovery page.

The good news is that if you stopped immediately, these drives are usually recoverable. The bad news is that the heads often need to be replaced after the unstick because the slider surfaces get damaged during the crash.

Watch: Why Seagate Rosewood drives fail so often.

A drop is evidence in a way a sound is not. The impact can force the read/write heads into contact with the platter surface, where they bond to it, or it can damage the spindle bearing, and either one stops the platters from reaching speed on the next power-up. Which of the two the drive has is settled when it is opened, and the beep on its own does not establish either one, or even that the drop is what caused it.

The operational state at the moment of impact decides which mechanism takes hold. A drop while the drive is running applies severe shock that can overcome the aerodynamic lift of the air bearing, so the sliders impact the spinning platters. If the drop also disconnects power, the same shock can interrupt the emergency back-EMF retract cycle and stall the actuator partway. A drop while the drive is powered off is gentler, because the heads should be parked on the ramp already.

In both cases the result on power-up is the same. The smooth AlTiC sliders rest against the smooth magnetic coating, intermolecular adhesion locks them in place, and the motor stalls trying to spin. If the drop instead disturbed the fluid dynamic bearing, the spindle itself is what will not turn. The repair path depends on whether the original heads survive a clean-bench unstick, which keeps the job at the firmware tier ($600–$900), or whether dragged sliders force a donor head swap ($1,200–$1,500 plus donor cost). The decisive variable is how many times the drive was powered on after the drop, so unplug an external Seagate Backup Plus or LaCie portable drive on the first beep and leave it alone.

How do WD and Toshiba drives differ when they beep?

A bus-powered portable pulls everything through the host port, so a port sitting near its current allowance can sag the moment the spindle asks for starting torque, drop the drive off the bus, and hand you a repeating start-and-collapse cycle that sounds mechanical from the outside. The noise still comes from the spindle motor either way; a starved rail stalls that motor the same way a bonded head stack does, which is why the sound alone settles nothing.

Stiction and a seized fluid dynamic bearing don't care where the electricity came from. Both stall identically on every power source that starts a healthy drive of the same family, so a symptom that shifts when you change the power path is pointing at the power path, and a symptom that repeats unchanged on every supply is pointing at the drive: either the head disk assembly or the drive's own board, where a shorted motor driver or power-rail fault stalls the spindle on any supply you give it. That split is why we check the power path before an external drive goes to the bench. On a drive that is already beeping, that check is worth one attempt on a known-good supply, not a series of them.

What changing the power source actually changes

1. A self-powered hub. It carries its own AC adapter, so the drive's current comes from the wall instead of from the host port. The data path stays where it was; only the power path moves. An unpowered hub does the opposite: it has no supply of its own, and the USB specification caps what it may hand each downstream port well below what a spindle needs to start. Do not test through one.
2. A rear motherboard port. Rear ports sit directly on the board. A front-panel connector reaches that same board through an internal header and a cable run, which is a longer and more shared path. Testing on a rear port removes those variables without buying anything.

If the drive still beeps on known-good power, stop testing. The fault is on the drive at that point, and every further power cycle buys you damage rather than information. Each retry pulses the motor against whatever is holding the mechanism, and where the sliders are down on the platter surface it drags them over the magnetic coating; that is the mechanism that turns a single-event case sitting in the firmware tier ($600–$900) into a donor head swap ($1,200–$1,500 plus donor cost), and then into the surface-damage tier ($2,000) once the coating is scored. From there the job belongs on a clean bench with donor head stacks on hand, not on your desk.

This test rules a cause in or out. It is not a repair. A drive that spins up once it has adequate power is not fixed: the port, cable, or enclosure board that starved it is still in the chain, and the next start can land on the marginal supply again. Copy the data to another drive first, then decide what to do about the enclosure itself.

Here is what recovering a beeping drive actually looks like. This is a Seagate with stuck heads being repaired on our clean bench.

Stiction repair steps shown in this video

• Drive opened inside laminar flow bench with ULPA filtration
• Heads carefully unstuck from platter surface
• Spindle rotated manually to verify motor is free
• Drive powered on to test if heads still function
• If heads are damaged, donor swap performed
• Drive imaged immediately before further degradation

The equipment is real. The process is real. We document our work so you can see exactly what you are paying for.

The first step is figuring out which fault is holding the drive: sliders resting on the platter surface, a seized spindle motor, or a fault on the board. The fix is different for each, and getting it wrong wastes a donor. Before any of this happens, we read SMART attribute 0x0A (Spin_Retry_Count) where the drive will still respond on the SATA bus. A raw count climbing each power cycle confirms the firmware itself has been logging failed motor starts and narrows the fault to the spindle subsystem rather than the host interface.

We connect the drive to PC-3000 and attempt to issue a motor start command through the drive's diagnostic terminal. The terminal-level workflow, register access, and adaptive-ROM handling are covered in detail in our reference on what PC-3000 actually does. On Seagate F3 drives, this is the U command (spin up). On WD drives, the equivalent command goes through the vendor-specific ATA interface. If the motor does not respond at all, we measure current draw at the motor pins using a bench multimeter.

This tells us which procedure to follow before we open the drive. Opening without a plan means unnecessary exposure time in the clean bench.

The visual check at the breather hole is the one most consumers reach for first, and it is also the most commonly misread. Standard air-filled drives carry a small breather hole labeled DO NOT COVER, backed by a dense desiccant micro-filter that equalizes internal pressure with the room. The filter is opaque; you will not actually see the platters through it. What you can sometimes feel is gyroscopic resistance when the chassis is gently rotated by hand with the drive powered. A healthy drive at nominal RPM resists yaw the way a small flywheel does. Stiction and seized FDB drives feel inert because nothing is spinning. Helium-filled drives are welded shut and have no breather hole at all, so this check does not apply; helium recovery work, including the helium refill and platter transplant where the FDB has seized, happens in-house at the same Austin bench.

Why does leaving a beeping drive plugged in make it worse?

A cheap unmanaged USB hub keeps +5V on the bus regardless of the host state, and Windows or macOS will aggressively re-enumerate a device that keeps disconnecting. Leaving a beeping portable drive plugged into a laptop over lunch is not passive; it is dozens of automated stall cycles.

A recovery that was a $600–$900 firmware-tier unstick becomes a $1,200–$1,500 head swap, and if the sliders dragged across the platters between attempts it escalates to the $2,000 surface damage tier. 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.

Preamp voltage rails

The preamp amplifies microvolt-level signals from the TMR read sensors up to millivolts for the read channel, and multiplexes which head is active. The logic rail is typically +3.3V or +5V with a separately regulated bias rail for the MR head current. A 3.5 inch drive pulls the spindle and VCM from +12V and logic from +5V; a 2.5 inch drive runs everything off +5V.

If a buck converter or LDO on the PCB fails open-loop and passes unregulated +5V or +12V onto the +3.3V preamp rail, the preamp die burns. Swapping the PCB without addressing the short repeats the damage on the donor.

FLIR thermal inspection before reconnecting the HDA

When bench diagnostics show a dead short on a power rail, reapplying full voltage will burn PCB traces and can propagate damage into the HDA. We clamp a bench supply to roughly 1.0V at a low current limit, apply it to the shorted rail, and scan the PCB with a FLIR thermal camera. The shorted component lights up against the cold board. The TVS diodes on the +5V and +12V rails are designed to fail short to protect the rest of the board.

Shorted TVS diodes are a protection feature that sacrificed itself; the original overvoltage event is what matters. We replace the shorted component, re-inspect thermally under power-limited voltage, and only then consider reconnecting the HDA.

VCM magnetic latch failures that mimic stiction

When the drive is powered down, back-EMF from the decelerating spindle sweeps the actuator arm onto a plastic parking ramp, and a latch holds the base of the arm so it cannot drift back onto the platters during shipping. Debris, corrosion, or a drop that deforms the crash stop can jam this latch closed. On the next power-on, the spindle may spin up briefly, but the VCM driver cannot overcome the mechanical jam.

From the host side this looks identical to a stiction-locked drive. Disambiguation requires opening the HDA on the 0.02µm ULPA clean bench and visually confirming where the heads are. Sliders bonded to the platter surface means stiction. Heads still on the parking ramp but arm jammed against the latch means VCM latch failure. The fix for the latch is mechanical release and inspection of the magnet assembly and crash stop; no head comb, no donor HSA.

Current-limited bench supply workflow

Before any HDA connection, suspect drives are powered from a regulated DC bench supply set to the rail voltage with current hard-limited to a healthy drive's peak rating. By form factor:

• 2.5 inch mobile drives: +5V only. We clamp the limit above the healthy peak and watch for an instant pull to the limit with rail collapse, which indicates a PCB short and triggers immediate disconnect.
• 3.5 inch desktop drives: +5V for logic and preamp bias, +12V for spindle and VCM. We current-limit both rails independently.

Once the drive passes the short-circuit screen, it connects to PC-3000 Portable III or Express through the vendor-specific UART terminal. We disable Service Area background initialization so the heads do not thrash the platters during diagnosis, then issue spin commands manually (on Seagate F3, the Z spin-down and U spin-up commands) while watching exact current draw. If the motor current spikes to the stall limit with no rotational feedback, we abort the spin attempt in the same second it starts. That single detail is the difference between preserving a firmware tier recovery and forcing the customer into a head swap.

After the heads are free, we verify the motor spins, test the heads with PC-3000, and image immediately. If the heads fail the read test after the unstick (slider surfaces are often damaged from the initial crash), we proceed to a full head transplant using a matched donor.

When stiction has damaged the read/write heads, the recovery becomes a head transplant. A random drive of the same nominal model is almost never a usable donor. Modern HDDs ship across many internal revisions even when the label is identical, and a mismatched donor will fail to read. Sourcing the right donor is the most expensive single decision in hard drive data recovery, which is why the donor is matched on those parameters before we open the patient drive.

Once the donor passes that match, the donor PCB is also evaluated. If the patient PCB shows TVS-diode damage or burned preamp rails under FLIR but the HDA is intact, we transplant the donor PCB and re-flash the patient's adaptive ROM region onto the donor board using PC-3000.

A beeping drive can be recovered without a donor head transplant in three narrow scenarios. All three depend on what the customer did between the first failure and shipping the drive to our Austin, TX lab.

Single-event stiction with intact sliders

After manual head-stack release on the laminar flow bench, the original heads pass the read test on PC-3000 Portable III and the surfaces inspect clean. No scoring rings, no metallic debris on the platters. Imaging proceeds with the original HSA. This is the firmware-tier path at $600–$900 and is the best outcome for any beeping drive. It requires the customer to have unplugged the drive on the first beep and shipped it without further power-on attempts.

Jammed VCM latch with heads still parked on the ramp

Visual inspection on the bench shows the head sliders are still on the parking ramp; the actuator arm is wedged against the magnetic latch from corrosion, debris, or a deformed crash stop. Mechanical release of the latch and inspection of the magnet assembly returns the drive to a normal park-unpark cycle. No head comb, no donor HSA, no surface contact. Falls in the firmware tier when nothing else has degraded.

PCB-side power-rail short caught before HSA damage

Bench diagnostics with a current-limited supply identify a shorted TVS diode or motor controller IC before the preamp on the HSA flex has burned. The PCB is repaired at the microsoldering bench (or a donor PCB is swapped with adaptive ROM transferred via PC-3000), the rails verify clean under FLIR thermal scan, and the HDA is reconnected with the original heads.

The recoveries that do require a head swap are the ones where the original sliders were dragged across the platters during repeated retry attempts at the user's desk. The slider air bearing surface picks up magnetic debris and develops contact wear that prevents reliable read signal. At that point the donor head transplant described in our head-swap procedure walkthrough is the only path forward. The single most important variable in deciding which tier a beeping drive lands in is how many times the customer powered it on after the first beep.

Platter transplants are the highest-risk recovery we perform. They fall into the Surface Damage tier at $2,000 because of the time, donor cost, and precision required.

After opening the drive, we inspect the platters under magnification. Scoring patterns tell us two things: which heads were down on the surface, and how much data surface is destroyed.

Debris is the second problem. Scraped magnetic coating produces fine particles that settle on other areas of the platter. When new heads fly over these particles, they crash. Before imaging, we clean the platter surfaces using lint-free wipes with isopropyl alcohol to remove loose debris.

The drive controller reads sectors in order by default. We read the undamaged areas first, then work inward toward the scoring rings with slower read speeds and higher retry counts. This maximizes the amount of data recovered from the intact platter surface before risking the new heads on the damaged zones.