I’m assuming this is your board:
http://www.htpcw.com/uploadfile/20110118155423687.jpg
If freezing the board rather than the whole drive is what gets it going, then I suspect the board will have a problem in the vicinity of the Marvell MCU (U5) or the preamp connection points (J1).
Oxidisation of the preamp connections is a common problem in those models affected by materials changes necessitated by RoHS. The solution is to lightly shine up the contacts with a soft pencil eraser. Alternatively, a cotton bud and metal polish such as Brasso may do the trick. I’ve used Brasso on other electronic jobs, but haven’t tried it on HDD PCBs.
A can of spray freeze should help you narrow down the problem if it is related to a specific component.
Your board uses a Marvell 88i6740-LFH1 MCU. FWIW, I know that WD’s “Tornado” family were afflicted with MCU faults whose symptoms mimicked head or media faults. Affected model numbers included WD5000AAKS, WD5000AAVS, WD3200AAJS, WD5000AAJS, WD5000KS. These had a Marvell 88i6745-TFJ1 or 88i6545-TFJ1 MCU. I suspect that the fault may have been in the read channel logic.
If you would like to measure the various onboard supply voltages (-5V, +3.3V, Vcore), I can help you locate the test points.
BTW, location U12 on the pictured PCB is vacant, so the “adaptive” data are stored within the MCU. This makes a board swap very difficult.
That said, you may be able to read the flash contents using one of the following utilities:
http://nazyura.hardw.net/000006.htm
These make use of WD’s Vendor Specific ATA Commands.
My HDD IC database may also help you:
http://www.users.on.net/~fzabkar/HDD/HDD_ICs.txt
