Drive failure and SMART relationship

[kanenas]

Hi all.

A WD 1600JD SATA disk developed a problem a few days ago. The drive had checked clean just the day before (I was checking all the drives while fixing another SATA disk - WD SE16 5000KS SATA II).
It might be because of a failing PSU which I just replaced just in case, or a failing mobo (Asus P4C800-E Deluxe) which I still have.

Things as of now are: 2 IDE disks on an addin Promise Ultra133, 2 CD drives sharing a channel on the mobo, one WD SE16 SATA II (brand-new one which replaced the failing one - the drive is jumpered to behave as a normal SATA) and the year-old WD 1600JD SATA.
The two SATAs are connected on the ICH5R connector. No Raid configuration and the mobo's Promise Raid controller is disabled.

When WD Diagnostics identified a SMART error on the 1600JD and couldn't fix it, I run SpinRite and HDD Regenerator on it. Both programs took ages trying to fix something near the 92% point of the first partition so I interrupted them.
After that, BIOS would complain about SMART Status BAD, Backup and Replace.

If I leave the warning enabled, WD Diagnostics refuses to run.
If I disable it, WD Diagnostics hangs up soon as I tell it to run the test.

All the tools I've tried (I'm booting from CDs), see the two equal NTFS partitions on this disk, the position of the MFTs and on the second partition a corrupted backup Boot sector.
I tell them to replace the bad copy with the good one, they do it without complaining but nothing happens. The bad copy remains.

Recovery tools get error after error while scanning for the filesystem (NTFS) and they don't seem to be able to identify anything.

I find it hard to believe that the disk would get so badly corrupted so quickly.

I suspect it's either the SMART counters that shot past their thresholds after SpinRite, or worse, maybe a bad SMART chip (is it an actual chip or just firmware?) on the drive.

Is there any utility for WD drives that can reset the SMART counters to defaults? Presumably there's one for Maxtor. They use it and if the counters increase again out of range, they RMA the drive.

Also, what exactly is SMART doing?
Does it just watch what happens or do requests go through it and depend on it before any data gets back?
If the last one is true, a corrupted SMART subsystem would raise errors on the drive even though the actual drive and data are good. Can I get around it in this case and access the drive directly?
Well, not me really. Just some utility that hopefully exists.

Thanks in advance for any pointers.

[kanenas]

I'll add a small note in case somebody knows of a utility to handle it.

I just had a look at the two failing partitions on the bad SATA disk using Partition Table Doctor.

In partition 1, the MFT and its mirror were identical except I was getting an error when I was reading in the mirror.
In partition 2, the same situation except that now the mirror was fine and the main MFT was getting errors.

In both partitions, I backed up their good MFT and wrote it over their bad one.

The writes were successful but the sector was still giving read errors when accessed.

Why is that? Is it a different mechanism for reads versus writes? Or is it just the SMART subsystem like I suspected?

Couldn't the bad MFT be remapped somewhere else with some utility? It looks to me like it's a matter of copying the sector somewhere and adjusting a couple of pointers.