I wondered if anyone has seen the following error reported in PC3000 before:

"DISK SERVO WRT FLT PREDICTIVE OFFTRACK ERROR"

It comes about when trying to write an entire firmware module to the SA. Writing data starts normally, but then for larger modules produces this error. When making a check on the heads, when writing small amounts of data (a few bytes) to an existing module the (donor) heads read/write fine. It appears that the write failure occurs when there is a large sequential write of 100KB+.

I'm working on an FBLite. Originally opened up by a PC Shop so changes to the inside of the HDD may have been made. Heads had failed. I've tried multiple donor heads - the results are replicated. When using donor heads it is possible to calibrate the HDD to RDY by blocking access to the SA in ROM either by shifting regions or blocking mod 30. Then once RDY it is possible to access SA in the normal manner and read all critical modules. However, because the HDD cannot write fully the HDD will not calibrate when SA access is allowed.

Any suggestions?

Regards,
John

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

Hotswap it?

_________________
Adatmentés - Data recovery

Some curious results off the back of a hot-swap:

1. Donor HDD, good match including ROM/FW. HDD was manufactured 2 days after the patient HDD.
2. Preparation of writing patient 01 to donor, then the remainder of modules, followed by ROM as per the Ace TS blog guide was completed successfully.
3. Power off-on the donor HDD with patient ROM to prepare for hotswap and find that the donor HDD now has the same fault as the patient. HDD stays in BSY unless, and SA access is only possible by blocking in ROM. Once I have access to SA via ROM block and I try to write modules again I get the same error in the donor HDD: "DISK SERVO WRT FLT PREDICTIVE OFFTRACK ERROR"
4. During the hot swap preparation procedure I noticed that when the patient ROM was still being used then the Drive ID of the patient would be shown and HDD would enter ‘Normal’ mode not ‘Kernel’. Given that the date of manufacturer between donor and patient is 2 days, I think the ROM contents is extremely close. Consequently, I try to write the Donor ROM back to PCB, and power on the Donor HDD which has Patient SA. The HDD boots to RDY, full ID, Normal mode and it is possible to read & write normally to the SA. Obviously all LBA reads result in UNC.

This suggests to me that there is some fault relating to the patient ROM.


John

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

I'm just rebuilding the ROM from the SA data and comparing it to the original saved ROM - looks like there are differences. I will update shortly.

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

The original patient ROM and rebuilt ROM from SA match OK. The differences I saw were in a version that I had spared regions. As such, I am running short on options for this HDD.

- Donor head incompatibility is very unlikely given the circumstances.
- The sequential writing fault in the patient HDD resulting in the error "DISK SERVO WRT FLT PREDICTIVE OFFTRACK ERROR" is replicated in the donor HDD when patient SA and patient ROM is written to the donor.
- Disabling write cache, read-look ahead and other features does not make a difference.

Below is a screenshot showing the current usage when the HDD is powered on. Annotated with the number 1) is when there is no block to the SA in the ROM. The current drawn is higher, and the HDD makes a slight buzzing sound, I presume as the HDD is trying to make some write attempts. Annotated with the number 2) is when module 30 is blocked in the ROM. The HDD enters RDY and Kernel mode normally. This fault is replicated (including the buzzing sound) in the donor HDD when patient SA and patient ROM are written to the donor HDD. It is not replicated in the donor HDD when only using the Patient SA and donor ROM.

If anyone has thoughts on the matter then I would love to hear them!

Best regards,
John

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

Just out of curiosity, have you tried using a totally different PCB? I've seen some strange anomalies that magically disappeared after swapping the PCB for another one, sometimes using a different REV.

_________________
Data Medics - Hard Drive, SSD, and RAID Data Recovery Service Company

dunno what guidelines you followed, but you know HS is not that straightforward on FBlite as earlier families...

pepe

_________________
Adatmentés - Data recovery

I've tried three PCBs; Two 771959-000 REV A and one 771960-000 REV A. Results are replicated. For good measure I just put the original USB based PCB on, and again I can hear the slight buzzing noise as it tries to write.

Regarding hot-swap on FBLite - yes I am aware there are some peculiarities due to the adaptives stored in the ROM. Perhaps I have not fully taken these fully into account, or that it is still not possible to perform hot-swap on FBlite? I know Ace TS always advise simply that "it is not possible."

J

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

Would it make sense to patch the patient ROM with the 4F module from the donor that supplied the heads?

Alternatively, one could amend the hotswap procedure by restoring the donor's 4F to the hotswapped donor.

_________________
A backup a day keeps DR away.

It may be a red herring, but I found this WD patent dated July 7, 2009:

DISK DRIVE SERVO CONTROL TECHNIQUES FOR PREDICTING AN OFF-TRACK CONDITION UTILIZING AN ESTIMATION FILTER:
https://patentimages.storage.googleapis.com/82/a9/6f/34aa2838129a3b/US7558016.pdf

Basically it uses a filter algorithm to predict whether the next write will be offtrack, in which case it pre-emptively shuts off the write gate. It uses a matrix whose coefficients are determined at time of manufacture. I expect that these "adaptives" would be stored either in the ROM or the SA.

_________________
A backup a day keeps DR away.

Firstly, thank you for all of your replies, and for the interesting read form fzabkar on Servo Control. The role of ROM module 4F appears to be relevant. I have tested numerous combinations of patient/donor/HotSwap-donor ROM and respective 4F modules. It is interesting to see how the original fault seen in the patient can effectively be replicated in donor disks, to produce the slight buzzing noise and the same output in the 'power graph' of increased current requirement, as (presumably) the HDD is making multiple attempts to write to the SA of the disk but failing to write normally.

Unfortunately, it has got to the stage where whilst a solution feels 'close' that it is probably not going to be reached. There is also the nagging thought that knowing that the patient HDD had been previously opened by a PC Shop that something has been altered. Frankly, I don't see how a PC Shop could cause this sort of fault, so the PC Shop's exposing of the platters and moving of the head assembly is probably irrelevant. However, it would have been useful to know what the original reason for failure in the HDD was before the PC Shop moved the head assembly across the platter surface which inevitably caused the heads to fail.

If there are any final suggestions, then they are welcome. Otherwise, I might have to put this disk 'to bed'.

Best regards,
John

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

Guys, it's an FBlite, 4F is unique. Not many people I know can hotswap them successfully without some pure magic..

_________________
http://www.northwind.gr
SandForce SSD Recovery
Ransomware Reverse Engineering - NoMoreRansom! partners

Long shot - but if you have a donor handy have you tried a platter swap.

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

Very interesting Fzabkar. Matrix algebra was taught in the first year of engineering...what a wide application is what I see here apart from solving some silly linear equations.

some 4F fun I had long ago https://forum.hddguru.com/viewtopic.php?f=24&t=37795

Only if someone could JTAG the HDD and see how the code is executing while the write command is given and the error is encountered, it will be an eye-opener. Perhaps if that error could be bypassed with some hacks here and there.. who knows

Any way. Hope all of you are doing well!

--

_________________
THERE IS ALWAYS HOPE TILL YOU LOSE IT.

Just to rule it out, platters were swapped to a replacement HDA today. No difference. I didn't expect there to be given that the ""DISK SERVO WRT FLT PREDICTIVE OFFTRACK ERROR"" fault can be replicated in a donor HDD when the patient SA & ROM is written to it.

@sin - I read through your post. Your description, . Can you remember what type of noise it was? I realise that it's probably hard to recall! The noise achieved with the patient HDD here is a feint consistent buzzing when the disk is requiring about 20% more current, presumably as the write attempts are being made.

Unfortunately, I am going to have call it a day on this disk, and conclude that there is some interaction between the patient 4F and the SA which is causing the failure. It certainly does not seem to be a heads or other mechanical issue.

Thanks for all the input, it really does help, and is also very interesting.

Best regards,
John

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

IIUC, module 4Fh has servo adaptives. If these adaptives need to be matched to the heads, then is it possible that no donor heads were good enough matches? That could explain why your ROM+SA transfer produced the same error in the donor.

_________________
A backup a day keeps DR away.

Yes, there is the potential that this patient HDD is being fussy about the donor heads and that this is all down to donor part incompatibility. However, I've never known an FBLite to be so difficult for a match. I have a stack of FBLite disks. I've tried six donor head assemblies, including one which was manufactured a day after the patient.

J

_________________
CDR - Manchester Data Recovery Services
0161 408 4857
http://www.cheadledatarecovery.co.uk/

Hi John, thanks for giving it a read.
It was HF oriented sound....imagine white noise only with a hi-pass filter at about 8KHz.(you can use reaper DAW to simulate the same noise by inserting a signal generator and hi passing the master bus of the DAW)

The noise only came when SA was tried to be accessed if I remember.
No noise heard with the original 4F

re-reading your original post, I have also seen in my experience incompatible 47s will only allow you to read but not allow you to write nicely or write at all
so that is another point to look after.

Can you resize the module you are trying to write in RAM (which is big in size) and try to write just a limited amount of data?

it should succeed mostly if its a size constrain purely that is in your observation.

Similarly, you can try various combinations of 4Fs and 47s within the known donor stock (leaving the patient aside) try to do more observations...

Fzabkar has pointed out a few good things about just targeting the 4F parameters that pertain to H0 and H1 while experimenting etc.

It's been quite long people have not really messed around with these drives locally and most of the Fblites that I did after writing this post had original ROM

Sadly no donor stock to experiment on and not really many WD hard drives donated by customers to experiment with and carry on the fun stuff.

--

_________________
THERE IS ALWAYS HOPE TILL YOU LOSE IT.

Hello,
Block SA Using Shifting Regions , Then Load Loader "Mod 11 Etc " To Ram ,Load Dir Module To Ram Then issue a Soft Reset And Watch The DRD DSC Light if it comes ,Then Quit Factory WD Tool thats in kernel mode And Run it Again HDD Should Remain Powered Up You Will Have Access To Data Without Doing Anything Else .Otherwise i have a plan .

_________________
Regards
Amarbir S Dhillon , Chandigarh Data Recovery Labs [India]
Logical,Semi Physical And Physical Data Recovery
Website-> http://www.chandigarhdatarecovery.com

Hello ,
Wanting To Ask Did you try this ....

_________________
Regards
Amarbir S Dhillon , Chandigarh Data Recovery Labs [India]
Logical,Semi Physical And Physical Data Recovery
Website-> http://www.chandigarhdatarecovery.com