Hard Drive Platter Swap: Lab Walkthrough

The Problem: Seized Bearing in an Old Toshiba IDE Drive

The drive arrived grinding and unable to spin. Older drives like this use fluid dynamic bearings where the oil oxidizes over 10-20 years, thickens, and eventually seizes the spindle. The motor tries to spin, draws too much current, and can blow the protective fuse on the PCB before anyone gets a chance to read data.

The technician applied controlled heat to the spindle motor area to warm the bearing fluid and reduce its viscosity. The platters started spinning, and PC-3000 detected the drive as ready. That confirmed the diagnosis: bearing failure, not dead motor or PCB damage. But the bearing was too degraded to support a stable read, so a platter swap was the only viable path.

The Swap: Heads, Platters, and Scotch Tape

Inside the laminar flow cabinet (localized ISO 14644-1 Class 5 equivalent, ULPA-filtered positive pressure), the technician removed the read/write heads using a head comb to keep the sliders from sticking together magnetically. Without a head comb, the sliders attract each other and tear when separated; that destroys them.

Multi-platter drives have no alignment markings on the platters. If the spacing or rotation shifts during transfer, the data becomes permanently inaccessible. The industry-standard solution is scotch tape applied to the outer edges of the platters, holding the assembly together as a single unit during extraction and transfer to the donor drive. It sounds crude for data stored at micrometer-scale precision, but it works because the tape just prevents relative movement between platters.

The matching Toshiba donor drive was sourced from eBay. The platters went onto its functional spindle, the retaining nut was tightened, the tape came off, and the platters spun freely. The donor heads were installed, the drive was reassembled, and on first power-up it came up silent; no grinding, no clicking. PC-3000 detected it as ready immediately.

The Recovery

Head two was sluggish, so the technician disabled it in the head map and imaged from the three remaining heads. Read speeds hit 30 MB/s, fast for an IDE drive of this era. The full 80GB clone completed successfully with all data intact.

The recovered image contained an HFS partition; this was an old Mac drive. The file system structure was fully intact. All user data was returned to the client.