Hi, I have had this 4TB drive for a few years, I noticed recently that some of the files I tried to save to it were getting corrupted. I did a surface can and ~50 bad sectors were found. I imaged the drive immediately and managed to recover 99% of the data from it so no data recovery is needed. Removing this drive from my system also greatly improved my boot speeds so it was likely causing system hangs too.

I wanted to see if I could fix the bad sectors, so using a combination of hirens boot cd and ultimate boot cd I ran a bunch of programs to see what the result would be. I moved the drive from my main PC (since the scans take hours and I didn't want it occupying my main PC) to an older laptop and connected it up using a combination of esata + power from a usb3 sata adapter.

full reformat and chkdsk /b dropped the bad sectors down to ~25
wd bootable lifeguard wipe dropped this down to ~20 (both the quick and extended tests in this software passed despite the bad sectors?)
hddregen dropped the bad sectors down to ~11 and this also seemed to fix the slow boots. I changed the bios from SATA to ATA for hddregen to run without errors.

I probably should have stopped here since the drive seemed to be useable again and all the bad sectors were located in the first 2GB so I could have partitioned the drive to ignore this area.

I then tried using hdat2 to do the same thing as hddregen but it seemed to get stuck straight away and would give a "drive not ready" error. I made a huge mistake here by seeing what creating uncorrectable errors would do. After a few seconds I stopped it since it seemed to be going really fast and swapped back to using hddregen. This took a really long time to scan each sector this time, after a few hours it had only managed to cover about 100. I then decided to start over and reformat again. After 2 days it was at 3% done so something had really gone wrong now.

I decided to cancel the reformat and look at the other tools again to see if anything could be done, but after a reboot the drive is no longer accessible. It shows up in the bios with the correct name. Windows can see the drive in device manager, but disk management does not let the drive initialise and just shows it as having 0 capacity. None of the linux tools from either of the live cds detects the drive anymore, they sometimes come up with errors saying drive reporting 0 sectors, using 63 instead.

When the bios is set to PATA the drive spins up on boot, after a minute or so it makes a noise like it is spinning down and then up again straight away, another minute later it spins down and does not wake back up again.
When set to ATA the drive will spin up whenever it is accessed, but will freeze the pc until it spins down again.

Is there any advice on how to proceed? If possible I would like to get back to the stage I had hit before attempting to use hdat2. As stated at the start the drive has no data I need to access, I mostly want to do this to learn what tools are useful in this situation.

Thanks for any advise!

To summarize : You had an unstable failing drive spitting out bad sectors, you ran a bunch of tools which stress the drive even more in an attempt to hide them, it reassigned some of these bad sectors but started producing more. Eventually the died from the strain and now isn't recognised at bios level.

It's dead Jim. It was already dying and you thrashed it to death, let it rest in peace now.

_________________
Data Recovery Services in the UK.
https://www.usbrecovery.co.uk/

Is there a way to diagnose what went wrong with it? I still want to learn from it if possible.

They are pretty complicated devices, but isn't it still just a chain of different tech and one of the links has broken?

I don't get the impression the motor failed since the motor still seems to work.

I understand some of these drives store important information on the platters and specialised tools are needed to access this area, but can tools like hdat2 modify this area to break the drive?

I realise there is no point opening the drive up since I don't have any of the tools or experience for dealing with the platters themselves, but I do have stuff like a usb->serial cable + arduino if they can access the drive in some other way (the 8 extra pins next to the sata data connector?)

Most modern HDD's automatically move data from one place to another if the sector is getting weak (eg: bad sector).
By running those tools you would have caused the HDD's own internal data moving process to effectively fight what you were running in software. This is extremely stressful and will wear a drive out pretty quick.

Those tools were designed for much older drives.
Bad sectors are not repaired; the HDD simply remaps the data they contain to a spare area on the drive and the brains of the HDD will not allow the computer to read that physical sector again. If your tools says it reduced the amount of bad sectors, it really didn't. The same physical sector is bad, the HDD is simply going elsewhere on the surface of the disk and reading / writing another physical location.

I would guess a head has failed or is very weak; most likely caused by the excessive amount of R/W done via software multiplied by it's own sector reallocation processes.

To further this case, you'd need tools to upload firmware components and then edit the head map in RAM. I'm fairly sure hdat2 cannot do that.

sorry for the necro, but I am still working on the same hard drive...

I just put this drive aside until recently, I thought I would try using some AI tools to see if they can help make progress. I have had success but some issues are still present that the AI is unable to provide useful help with. The AI did give some stupid suggestions such as doing destructive operations on the donor drive to test if they do anything...

I paid for a month of WDMarvel, this was able to access the drive with the sata port set to ide (old pc) and with driver verification turned off in windows 10.
no modules visible, when searching only 8 returned.
these 8 modules closely matched the modules present in ram -> presumed the sa on the platters was inaccessible in current state.
tried a bunch of different features, nothing could access the platters.
I did have a similar drive with different firmware, I got the packages from this donor drive.
Ran a full SA format followed by running the full ARCO process (took 4 days) using packages from the donor drive.

The ARCO process found 5689 defects, the final drive capacity is larger than before by almost 250MB. A low level format returns an additional 88 bad sectors, I ran 5 more low level formats after this with no additional bad sectors found. I get the feeling some of the defects/bad sectors are part of the extra space that was originally in the p-list (which has been wiped).

The drive is giving smart errors due to the amount of bad sectors, but they currently are not growing, is there a way to move the g-list to the p-list and clear the smart data so only new bad sectors from this point forward show up?

My current process is to run defect operation move g-list to p-list, then run the translator regenerator (this breaks module 01), then run rebuild dir.

This breaks module 02 and once the drive is power cycled it becomes unusable again. I can easily restore it by doing an SA format and using a restore point from after ARCO finished.

What is the correct procedure for moving bad sectors into the p-list?

Exactly the same: transfer defects from the g-list to the p-list, run the internal format with a regeneration of the translator. There are some bugs in the program itself or in the hard drive firmware that could cause the p-list to be completely cleared after this, or corrupt the modules on the drive.

If the drive refuses to cooperate, you can manually extract the defects from the g-list, clean it, or restore it from a restore point, export the p-list to a file, insert the defects from the g-list, and then load all this work back onto the drive. There shouldn't be any problems, you can even power cycle the hard drive. And then, as usual, the internal format with regeneration of the translator.

Simply put, you need to replace the p-list on the drive with a manually edited one. The INI file format is very simple. With 90 defects, you can either enter them manually, ask the AI ​​to do it for you, or write a script. There were already ready-made programs for this somewhere, but now it's much easier to ask the AI ​​to create one than to search online.
Whenever we talk about a donor drive, we always assume it could be damaged in favor of the one being restored. But with a backup restore point and a copy of the service area tracks, you can always restore the WD hard drive to its original state, even after wild experiments.

As for a first attempt, passing the ARCO test is a significant success; before the days of AI, by this point you would have ruined about 10 disks and spent a couple of months reading hundred-page forum threads, pestering everyone with requests for help, and obviously not getting it, hearing only "RTFM, read the forum."

It's not the g->p I have been having issues with, its the regeneration step that does not seem to work.

Using the Buttons layout, Manage SA and flash->Defect operations->Move G-List to P-List(cmd) seems to work fine for moving (they are present in the P-List editor afterwards), but when I use SA surface->Format SA->SA Translator regeneration (ID30) the directory structure for the modules is lost. I am not sure if I am using the right option.

One other thing, if I set the capacity size to the one on the drive label (shrinking the capacity by 270MB), do I need to do anything so the unused non-defective space goes back into the spare capacity for bad sector replacement?

Why are you doing this to the service area translator? You need to regenerate the USER area translator. Yes, these are two different entities - the service area also needs to somehow convert physical clusters and tracks into a list of sectors, and it can also have defects. But that's not what you need now. The service area is fine, we need to hide the defects from the user area.

It's not surprising that problems with modules occur in this case.

If you are using a button layout, this option is hidden here.
To save time, you can select the "Skip write of all LBAs (PDList regeneration)" option. The program will pretend to do nothing for a few minutes, but the result will be exactly what you need. If any errors occur, simply try the internal format.

No, sectors for replacement are taken from spare tracks inside minizones or spare sectors inside a track - it doesn’t really matter. This doesn't affect anything. It is advisable to configure the capacity as on the label just for cases when you have to clone a disk or use it in a raid array - this was a problem in the days of 40 gigabyte hard drives, when due to extra or missing megabytes it was necessary to look for identical drives.

thanks for that, the internal format works if I do not touch the capacity that ARCO set, but if I set the capacity to that on the label (and the same as the donor) then the internal format returns an error "FMT_ERR_NEW_BAD_TRKS" at the end.

In this case, it's probably easier to perform an internal format first and then adjust the hard drive capacity. Consider the following statement misleading, but as far as I know, to ensure the capacity is set correctly after the ARCO test, you need to edit it in the script before the test. It's best to ask on the WDMarvel forum; people there run self-scans, ARCO tests, and TREX on their hard drives every day; they'll definitely tell you the reason.

By the way, the drive capacity can be adjusted in either direction. In fact, there's much more extra space than 200 MB; you can check the zone list to see how many spare blocks there are.

Another point why you shouldn't rush to adjust the capacity is that all these tests are designed to be run on new, factory-assembled hard drives. They shouldn't be run on drives that have already been used, and especially not on drives that have sustained any damage. When a hard drive is damaged to the point of "trying to revive it or it's time to throw it in the trash," then yes, these tests are used because manually hiding tens of thousands of defects or correcting sector allocation tables is simply impossible. The correct drive repair method in this case would be different: try to preserve the original service area data to avoid losing factory calibrations, defect lists, and so on. Then, manually identify problem areas and enter them into the g-list either as individual sectors, with a few good sectors in reserve, or as entire tracks.

Most often, such drives appear to have no further defects, but if you write data to the drive and then leave it unused for several weeks, new defects may reappear when reading.

I did wonder about re-running the ARCO with the correct capacity, but I could not find a way to add a specific capacity size in the interface provided.

I don't mind re-running the tests, if they push the drive to actually fail then it will at least give closure on sending the drive for electrical recycling. I would say it's been hammered in the past couple of weeks with these ARCO/self-tests/reformats and as far as I can tell only the same bad sectors are being rediscovered.

I thought the number of defects was quite high too, but I checked the p-list on a few of my healthy drives and a couple of them had a lot more, one had 11276 with nothing added to the g-list after 12 years in use (I surface scan them every 6 months).

Thanks for responding so far, if I end up paying for another month of WDMarvel then ill see about asking on their forum as well. A month really does not last long with how much time gets eaten by some of the drive operations.

throw it in the bin and move on with your life.

I suggest getting a new hard disk and copy the old disk as it is obviously seen better days. Posting a screenshot from CrystalDiskInfo should make it clear.

I suggest that redundant copies of important data should be made.

_________________
USB 3.5" enclosures are $19 USD postage paid from China, I own several

I think I have managed to do it. The internal format might have been failing due to the firmware mismatch. If I moved the g-list to the p-list, then used Skip surface formar (recalculate PDList) and then did a low level format using a different program (so the drive itself isn't conducting the format) then it succeeds.

Since getting to the "fixed" state the drive has been through 3 formats, 3 full surface scans and a zero fill+verify (8-10 hours each...) without any additional errors appearing. Prior to this the drive has done the full arco+self test along with about 20 various full scans, I doubt the heads are wearing out otherwise they probably would have failed quite fast during all this. The P-list defects have a pretty close spread across the heads so it doesn't look like there is a particular weak one (unless I am reading it wrong).

Obviously I need to remain suspicious of the drive and not use it for anything important and I will be running full surface scans once a month to keep track of things. In its current state the size is correct, there are no reallocated or pending sectors in the smart data. The drive being connected does not cause any slowdowns to the system.

Thanks ora for the input, I have learned a lot about hard drives during this. It's a shame there isn't a more widespread database of module dumps for all the different drive models. I got lucky with already owning a nearly identical drive.

3 formats and surface wipes suggests the disk may be stable but I still strongly recommend backups galore.

I suggest the old disk might be good as a backup for maybe a SDD to modernize

_________________
USB 3.5" enclosures are $19 USD postage paid from China, I own several

In my experience, the lifespan of drives with damaged surfaces is limited.
Even if the head on the damaged surface is disabled, the dust gradually damages the others.
There are many HDD resources online, but not all are openly accessible or free.

I understand about damage having the potential to get worse. I have a lot of mechanical drives and have had my fair share of failed drives, all my Seagates are dead, I still have some working IBM "deathstars" too though (worthless 120GB IDE drives I scan occasionally, just interested to see how long they last with glass platters).

From my understanding the p-list is basically surface damage from the factory due to it being impossible to manufacture an absolutely perfect drive surface.

I am mostly running this drive to experiment with it, its planned usage is a cache for online content to get around slow download speeds. If it fails the content is already backed up in multiple locations, but is less convenient to access.

I did run into another issue that has since been solved, The drive was stuck at SATA2 max speed and features such as NCQ were not working.

I tried using HDAT2 to toggle NCQ back on without success.

I compared module 02 to the donor drive and found a lot of different data despite the donor providing the packages to rebuild it. Replacing module 02 with the one from the donor and then adjusting settings such as the serial number has fixed the drive so its detected as supporting SATA3 and NCQ works again (much faster random seek speeds). A further 2 surface scans have been done since this without errors or new bad sectors appearing.