I've been having a problem with my WD SATA 250 GB hard-drive for some time.
I will describe symptoms, and I hope you can help me identifying the problem, and suggest me on data recovery (there's enough free software available, for Linux, DOS a Windows, so unless it's a hardware failure that cannot be fixed, I'd prefer using available tools).
I am using Windows XP SP3.
Several months ago, I noticed that after waking up the computer from Sleep Mode, one of the 3 SATA drives is missing from Explorer and Device Manager. BIOS detects the disk and it is visible in Linux (although recently it hasn't been mountable). It has 1 NTFS partition (fully on 250GB).
Sometimes, doing "Refresh" in the Device Manager, brings the disk back.
Mostly, I can see the full directory structure and list the files.
I can also copy some files off the drive, but when I am trying to write files on the drive, or read some files, the operation stalls, after a while I get an error message (from the task-bar error balloon) saying "Delayed write failed", and the device disappears. Again, it can be brought back by rebooting or refreshing device manager.
Using any data recovery software for Windows won't help, because as soon as it reached the problematic region of the disk, the error occurs, disk vanishes, and the recovery software is left with nothing to do.
Another thing that happens to this drive is heating. Compared to other drives, it is much warmer, especially on the electric circuit area, but on the back side as well.
Currently, the HD is standing in front of a big fan, outside of the computer, so it's quite cool.
I switched to Linux, and I am currently in the process of trying to recover everything with dd_rescue to another 250 GB empty drive (formatted for NTFS, but no data on it).
After 2.9 GB of data that was copied without any problem, the bad blocks started showing, and in the last 24 hours, it reached only 3.8 GB.
I am using default settings for block sizes and everything.
I am logging the bad block numbers, and it seems that there is a pattern.
In most occurences (and I haven't been through the entire 5.5k bad blocks log), It seems that there are 7 bad blocks in a row, followed by a gap (sometimes 50 blocks, sometimes 2000 blocks, etc.)
It's not always the case, but almost always it is.