Data Recovery Case File · Electrical Damage · The Smoking Tower

The drive that vetoed two computers: a smoke event decoded, a customer's cable experiment ratified — and the database file brought out from behind the short

His enquiry contained the rarest thing in this archive's electrical genre: a completed, correctly-designed experiment. The event: "I switched my tower computer on and smoke started coming out of it." His response: drive out, installed as a second drive in another PC — which then wouldn't start. And then the test that isolated everything: "The second PC will start if the data cable is attached but the power supply cable is not attached." Data lead: fine. Power lead: total veto. He'd considered trying a legacy power supply and — wisely — hadn't yet. The cargo, with his own punctuation preserved because it says everything: an SQL database file, "as I didn't have a recent backup (!)". The drive: a Hitachi Deskstar 250GB, vintage 2007. His experiment gets ratified below, his legacy-supply idea gets gently retired, and his database comes home.

DeviceHitachi Deskstar 250GB SATA (2007 vintage) — working database storage; an SQL database file the priority, no recent backup
Reported eventsSmoke from the host tower on power-up → drive moved to a second PC → second PC refuses to power on with the drive's power cable attached; starts normally with data cable only — owner's own elimination
Fault classShorted drive electronics — the power-side short convicted by the two-cable test; platters untouched
Equipment usedComponent-level board diagnostics and repair · ROM preservation · ACE Lab PC-3000 Express + Data Extractor · database-file verification

The decode: what his experiment proved, what the smoke was — and why the legacy supply stays in the drawer

His two-cable test, ratified formally: a modern power supply monitors what its rails feed, and a device demanding the impossible — the signature of a short circuit — trips the supply's protection instantly: the PSU refuses to start rather than pour current into a fault. His experiment split the drive's two connections perfectly: the data cable carries conversation and drew no objection; the power cable carries current into the drive's board, and the second PC's supply vetoed it on contact. That is a textbook conviction of the drive's power circuitry, specifically — the fault localised to one region of one board by a man with two cables and a methodical streak, and it's the same elimination this archive performs at the bench, done on a kitchen-table budget. The smoke, placed in the story: smoke on power-up means a component burning — and the drive's board shorting is entirely consistent with being either the smoke's author or its fellow casualty in the tower's electrical event; either way, the visible drama lives in the electronics, and the platters — unpowered magnetic surfaces — sleep through smoke the way they sleep through everything electrical, protected further by every protection trip that kept current out. His database's odds were excellent from the moment the second PC refused. And the legacy-supply idea, retired with respect: the instinct (try a different power source) is the natural next step, and it's the one step this genre begs owners to skip — every supply a shorted board meets is another delivery of the original fault, and older supplies bring looser protection to the party. Two PSUs had already voted. The case needed a bench, not a third ballot.

The recovery — the board repaired, the database verified as a database

The bench confirmed his conviction with instruments: the board's power circuitry shorted at component level — repaired, with the board's ROM, the drive's unique identity, preserved throughout as the one part no donor can supply. On current-limited first power the 2007 mechanism spun up cleanly — thirteen years old and mechanically sound, the electrical event its only injury — and the PC-3000 imaged the full drive without a distressed region. His priority got priority: the SQL database file located, extracted, and verified as a working database — opened, its structure intact and consistent, not merely present as bytes — before the rest of the drive's estate followed on new media in duplicate. The report closed with his own exclamation mark answered: the missing backup, now existing in triplicate, with the database-specific addendum his cargo earns — live databases deserve scheduled dumps to a second location, because a database is the one file type that's being rewritten even while you're not looking, and "no recent backup (!)" is a sentence this page exists to retire.

Outcome

Full recovery, the database verified and working — and his method filed as the genre's model. Run his elimination if a drive is suspected of killing machines: one drive at a time, and if you can, his refinement — data cable versus power cable — which localises the fault to the board's power side in one move. Then stop where he stopped: no further supplies, legacy or otherwise; two protection trips are a verdict, not a losing streak. Read smoke as electronics news, not data news: the burning component and the magnetic surfaces live in different worlds, and the veto that frightened him was also the mechanism protecting his platters. He did the experiment, drew the right conclusion, and asked before the third attempt. The database — unbacked-up, exclamation mark and all — never stood a chance of being lost to anything except one more power source it never met.

Smoke, then a drive that kills computers

Unplug everything and stop: a PC that won't power on with a drive connected is a shorted drive board being vetoed by the supply's protection — the diagnosis is complete, and further power sources only re-deliver the fault. If you must test, one careful pass of data-cable-only versus power-cable-attached settles it. Never open the drive, never try older or "stronger" supplies, and note what smoked and when. The board repairs at component level; the identity chip on it is the recovery. And databases: schedule the dump tonight.

Drive vetoing every machine it meets?
The short is in the board, not your data — call Belfast Data Recovery on 028 9002 0144 before it meets another supply.
Request a quote online →

Our case files are drawn from genuine enquiries received by our laboratory over the past ten years, anonymised to protect client confidentiality. Each one describes the diagnostic and recovery procedure our engineers apply to that fault, using the equipment listed.

Call us — 028 9002 0144
Mon–Fri · 9am–5:30pm · No fix, no fee
Start a free diagnostic →
028 9002 0144