A steady click, once or twice a second
The classic pattern and the one people mean by the click of death. Heads failing to find their servo data, retracting and retrying indefinitely. Mechanical, and it will not resolve itself.
Heads
The click is the head assembly retracting to its parking ramp, recalibrating, and swinging back out to try again — over and over. It is a loop, not a one-off failure, and every cycle costs you something. Power it down before reading the rest of this.
The drive cannot lock onto its servo information, so it retracts the heads and attempts it again. Where the heads are damaged, each of those passes takes them back across the surface holding your data.
Everything below is worth knowing. None of it is worth the minutes the drive spends running while you read it.
A clicking drive is deteriorating measurably, and the amount ultimately recoverable tracks closely with how long it ran after the noise began. That is the one variable entirely under your control, and no equipment afterwards compensates for spending it.
So: power it down, unplug it, and leave it. Do not try once more to grab a file. Do not plug it into a different computer to see whether that helps. Both are the same action repeated, and both cost you.
Mechanically precise, and it explains why the sound is so regular.
Inside a drive, the read/write heads float a few nanometres above spinning platters on a cushion of air. To find anything they rely on servo data — positioning information written across the surfaces at manufacture, which tells the heads continuously where they are.
When a head cannot read that servo information, the drive does not simply fail. It retracts the head assembly to its parking ramp at the edge of the disk, recalibrates from a known position, and swings back out to try again. That retraction and return is the click. It repeats at a steady rhythm because it is a loop: seek, fail, reset, seek.
Which is why the sound is so distinctive and so regular. It is not the drive breaking repeatedly. It is the drive trying repeatedly, in the only way it knows.
You do not need the vocabulary. These are the descriptions we actually hear on the phone.
The classic pattern and the one people mean by the click of death. Heads failing to find their servo data, retracting and retrying indefinitely. Mechanical, and it will not resolve itself.
The drive tried, gave up and parked itself deliberately. Often firmware or service-area rather than physical head damage — comparatively good news, and sometimes recoverable without opening the drive.
Head-to-platter contact. The magnetic coating is being removed while it runs, and whatever is in the affected band is gone permanently. The one where minutes genuinely count.
Not the heads at all. Either a seized spindle bearing or heads bonded to the platter after a long spell unpowered. Different fault, same instruction.
Each is recommended somewhere online, and each makes a professional recovery harder.
There is a kernel of history here — on drives from decades ago, cooling occasionally shrank a seized component enough for one more spin. On anything modern, condensation forms on the platters as the drive returns to room temperature, and moisture where the heads fly is destructive.
The idea is to free a stuck head. What it actually does on a drive with damaged heads is drive them into the platter surface, converting a head fault into permanent scoring.
Software cannot address a mechanical fault, and a deep scan is hours of continuous reading on hardware with limited reads remaining. Drives regularly die partway through a scan intended to save them.
The commonest and the most costly. Whatever that file is worth, it is not worth widening the damaged region across everything else on the disk.
So you know what you are paying for, and why it is not a desk job.
The drive is identified before it is powered — family, firmware revision, head configuration — because a mismatched donor makes things worse rather than better. A matched donor is then sourced: same family, compatible firmware, correct head count. That is frequently the slowest part, and it is why an unusual or elderly drive takes longer without being any less recoverable.
The head-stack is replaced under filtered air, with the internal filter changed to capture debris the failure has already generated. The drive is then imaged read-only in short controlled passes with per-head timeouts, prioritising the directories you actually asked for — because a repaired drive may give one good read, and it is spent deliberately rather than tested.
Everything after that happens against the image. The original is never written to, and it is not expected to work again. In Manchester that is from £300 +VAT after a free 48-hour diagnostic, with a 50% deposit because donor parts and clean-air time are committed before anyone knows what the platters look like.
Because the heads cannot read the servo data that tells them where they are, so the drive retracts them to the parking ramp, recalibrates and tries again. That retraction is the click, and it repeats because the drive is looping through the same failed seek.
Not in a way that leaves it usable. Recovery means opening it under filtered air, fitting a matched donor head assembly and reading it once. The drive is not expected to work afterwards — the objective is your data, not a working disk.
You should not use it at all. Damage accumulates every rotation, and how long the drive ran after the first noise is usually the single biggest factor in how much comes back. It is the one part entirely under your control.
No. Condensation forms on the platters as the drive warms back up, and moisture where the heads fly is destructive. It was marginal on drives made decades ago and it is actively harmful now.
Usually better news. That pattern often means the drive detected a problem and parked itself, which points at firmware or the service area rather than physical head damage — and those are frequently recoverable without opening the drive.
From £300 plus VAT for a single drive, quoted in writing after a free 48-hour diagnostic. Clean-air and donor-part work carries a 50% deposit with the balance due only on success.