Attempt at hdd data recovery

give it ago

And the original poster wanted to recover some photos.

The "learning value" only came in later, or from one of the repliers.

But what does one learn? That it can be done? Something for the future?

Most people will never have a need for doing this a second time, even if successful the first time.

And given the poster was going to follow some text, I doubt he can learn from the experience, because if he's successful all he's learned is how to follow the instructions, if he fails I doubt he has any safety net to help him figure out why.

I would urge everyone to open up a hard drive they are about to toss out. They can grab the magnets, and that first hand viewing of the insides is a good learning experience. It costs nothing but a few minutes of time. But after that, the cost is going to go up.

Michael

You could change the heads on some older 8" and larger hard drives, in the 5 and 10 MB per platter days. There were places that sold rebuilt heads and new platters so you could rebuild a drive for a minicomputer. I knew somone who used to rebuild drives for Data general minicomputers in a home made plexiglass box that was presurized with air that was passed through a pair of HEPA filters so any dust was blown out the front of the box. It was crude, but worked on the crude drives of the day.

I've fixed a few drives by changing boards, but it was never for the typical clunk clunk clunk symptom. These were cases of boards that were damaged by someone shorting it against the case, a bad motor driver, or an IC that was obviously bad. Andy Cuffe

snipped-for-privacy@gmail.com

It depends on the drive. I've opened older drives that would run for quite a while. The few semi-working modern drives I've opened didn't last more than a few seconds before totally failing. The higher the data density is, the less tolerant of dust they are.

Does anyone have any inside knowledge of what professional data recovery companies actually do? I've opened plenty of bad drives that have no obvious damage to the heads, or disks.

Andy Cuffe

snipped-for-privacy@gmail.com

In message , Michael A. Terrell writes

Actually, I have recovered data this way on several (5 IIRC) 3.5" IDE drives. I've replaced head pre-amp chips but with limited success. It doesn't fix everything but it's worth trying for 5 minutes of work.

It's more likely to work on a modern IDE drive IMHO, the boards are more generic across a range and are not 'matched' to one HDA, lots of the old MFM/RLL drives had 'SOT' components which were designed to match the board to the HDA. ESDI and SCSI drives were more tolerant of a board swap but drives of that era had boards with obtainable components so a repair was also possible.

It's worth a try, if the donor spins up and your drive doesn't, I'd take a shot at it.

Most people don't have a clue how difficult it is to carry out this kind of work with the kind of precision required.

You could change heads and platters on 5.25" drives as well, I've done it at a repair house I worked for lots of years ago in the 50Mb per platter days as well.

And microcomputers, we used to import heads and platters from the US for Seagate, Rodime, Micropolis and Maxtor drives.

As an experiment to see if newer drives can run without the cover on them I took apart a 40 gb diamondMax plus 8 drive. It is currently running in the computer that I'm using and has been working for the last 45 mins.

It is similar to the laptop drive in that it doesn't use the top of the first platter. So any dust that lands there won't be directly in the path of a head.

I think I'm going to try this after I pratice on some old hdd's. Who knows might even work, but like I said I'm not really expecting it to.

You can't, don't even bother. The precision required is orders of magnitude greater than anything you can accomplish at home. There's 60 billion bits on those tiny platters, the alignment has to be mind bogglingly accurate to read them.

The heads may well be just fine, often the result of a head crash is the heads plow into the platters and destroy the surface, too much damage and the drive can't read the control data and doesn't know where to put the heads. All you'll succeed in is destroying a perfectly good donor drive.

That does work occasionally, or did. Years ago I swapped the controller on a 1.2gb Connor drive, at the time it was reasonably big, used the drive for the next 3 years on the transplanted controller with no issues. I've tried it one other time on a newer 30GB drive and it didn't work.

The drive was dropped while running.

On Sun, 8 Oct 2006 22:00:16 -0400, "Michael Kennedy" put finger to keyboard and composed:

I would think that the platters and head stack would have been low level formatted as an assembly. The heads in a replacement stack could never match the alignment of the original, so I would think that the voice coil servo would twitch every time there was a head switch. If the drive uses an embedded servo, then the user might notice some degradation in seek times, but if the drive has a separate servo head (are there any that still do?), then a head stack replacement will never succeed.

As for replacing PCBs, I would think that this may also fail if the flash EEPROM is used to store SMART data and defect maps. In fact there is at least one data recovery website that states that certain WD Caviar drives cannot be recovered with PCB swaps.

Another point with which I take issue is the claim that a drive can be "low level formatted":

This hasn't been possible since the very early days, some 15 years ago, and especially not since the advent of embedded servos. Today the term means little more than wiping all data from the drive.

See

"You cannot format the disk drive at low level. The servo, sector layout and defect management information contained in the low level format is designed to last the life of the drive. As such, this information cannot be overwritten outside of the factory."

- Franc Zabkar

One thing that does puzzle me, is even if the control pcb is replaced, the replacement controller won't have the correct flaw map ! Ppl may not know about this stuff if they've never had to low level format a drive. These days it's done at the factory you see.

Graham

They certainly have clean rooms and keep a large stock of drives for the spare parts they'll need.

Graham

In message , Eeyore writes

Surely this is stored on the HDA and not the controller PCB?

Not many people know the joy of prodding around using debug to find the LLF code entry point on a controller you have no documents for... Ahh, happy days.

Michael Kennedy ha escrito:

as

The article cleverly avoids to explain something very important: You remove the platter stack to replace the heads, ok, but when try to reasembly it, how do you realign each platter with respect the others? I think that if any platter is shifted even a 0.000001 deg. with respect to the other(s), the data will be unreadable. You can=B4t mark the platters, and even if you can, the mark will be too coarse to be efective as an alignment guide. Even worse: how to handle each platter to avoid any damage to it=B4s magnetic surface? If even handling the magnetic disk of a disquette damages it, and in those disks the data density is relatively low (only 80 tracks per side), how to avoid damage to a glass disk with a density thousand of times greater than the density of a diskette?

No, I don=B4t think the person who wrote the article has ever attempted to do what he claims, or if he has done it, he did it in a very different environment and using different tools than those described in the article.

But the pre-IDE drives gave you a lot of access to the drive, since the controller was separate. MOve the controller to the drive, as is the case with IDE drives, and you don't have that low level access.

Pre-IDE, the bad sectors on the drives would be written on the drives, and you'd tell something (I guess the formatting software but I never really had any experience with pre-IDE drives) to avoid those sectors. With IDE drives, the drive itself keeps track of such things.

ANd this is what came of the thread some months back. It was pointed ou that the eeprom on the board was likely keeping track of those bad sectors, and when the board was swapped, it no longer matched the state of the new drive.

Michael

An IDE drive is the only one I've done it on and succeeded, as advanced as modern drives are though I would expect it to be a lot less likely to work, this was a rather old IDE from around '96.

Thats becasue you don't remove the platters in the defective drive. You remove them in the donar drive and carefuly slide the head stack over the platters in the defective drive.

- Mike

Michael Kennedy ha escrito:

The article cleverly avoids to explain something very important: You remove the platter stack to replace the heads, ok, but when try to reasembly it, how do you realign each platter with respect the others? I think that if any platter is shifted even a 0.000001 deg. with respect to the other(s), the data will be unreadable. You can´t mark the platters, and even if you can, the mark will be too coarse to be efective as an alignment guide. Even worse: how to handle each platter to avoid any damage to it´s magnetic surface? If even handling the magnetic disk of a disquette damages it, and in those disks the data density is relatively low (only 80 tracks per side), how to avoid damage to a glass disk with a density thousand of times greater than the density of a diskette?

No, I don´t think the person who wrote the article has ever attempted to do what he claims, or if he has done it, he did it in a very different environment and using different tools than those described in the article.

That's very true. In fact, using a bulk tape eraser on a working drive is a very effective way of rendering it useless. As soon as the low level format has been erased it will start clunking like a drive with a crashed head. Andy Cuffe

snipped-for-privacy@gmail.com