Last week a client told me that a shop machine went down hard and that apparently the operators had been saving scripts locally to the machine instead of to the backed-up network drives. We replaced the hard drive and got the machine back up and running from a backup, but we still had scripts that they needed to recover, so I took their external drive to work on. Windows 7 wouldn't recognize the drive in an external chassis at all. I tried the standbys: freezing the drive, rapping it gently on the desk while it was spinning. No dice. The drive would spin, click, spin, click, spin, click and then shut down, never recognized by the OS.
This client doesn't have the budget for a Level 3 clean-room repair, but I told them I'd source another drive and see if we could do a hardware swap to get it running again. Initially I wanted to try a PCB swap, but the version I got was slightly different and so that didn't help. For my entire computer-interested life I've been told that hard drives are magical things that will, if ever opened outside a clean room, immediately become scrap. Of course I've taken apart my share of hard drives in search of strong magnets, but only on drives that had failed and only after I was sure that it was okay to destroy it. I finally proved to myself that this fear wasn't entirely accurate.
We talked about swapping platters, but after tinkering a bit I figured I'd try a head swap instead. First I took the cover off the replacement drive and lay it sideways over the open drive to try to protect the platters. With the head parked, I cut some rings of CAT5 insulation and inserted the loops between the actuator arms to gently hold the heads apart. I took off the top magnet that covers the voice coil and removed the stop pin to move the actuator where it would clear the platters. A few more screws were removed to allow the actuator to be removed. I replaced the cover and repeated the procedure on the dead drive, swapping the old actuator and head assembly with the new one.
Once I plugged the drive into the external chassis I started to make some progress, but Windows 7 couldn't load the primary data partition - probably due to some corruption. I fired up a Linux recovery environment and TestDisk was able to navigate the partition and copy the necessary data onto a good drive.
Lessons learned:
1. Ensure the replacement drive PCB is the exact same version as the one I'm replacing. Same drive model number is not sufficient. The new PCB wouldn't work on the old drive at all, and if the problem had been the PCB or a component on it then I'm not sure I could have completed the repair.
2. If I'm going to do this sort of thing regularly, build a clean glovebox with a clear plastic bin, a fan, air filters and some rubber gloves.
3. I need a wider selection of torx bits. I think I could have done a platter swap if I hadn't stripped out one of the platter retaining ring screws because a bit was slightly too small.
4. I need to keep more plastic tools around. Incidentally, a titanium spork did help when I was working around the magnets.
5. Modern drives are more robust than I had thought.