Hard Drive Failure *Cries* - Engineers to the rescue?

[Megamox]

Thanks for the advice guys! I too have seen conflicted opinions about just how viable it is to open a hard drive up and get it working again. Naturally I understand companies need to be in business but I'd hope they'd also be trying to give out good advice too. Suppose you just never know. I was thinking of the idea of making the portable clean room like in the video below. Sadly the version this guy made didn't work, but I respect the effort.

https://www.youtube.com/watch?v=GZ1aAKaHaMc&list=UL

I've also thought that if I do exchange the head, I could flip the hard drive upside down and let it spin up for a few minutes. The debris would fall downward due to gravity and then I'd put the top back on quickly. How useful this might be, I'm not sure. It certainly would work in the macroscopic world.

Megamox

 

[()blivion]

Watched the video, that person totally fails.

Way to care free about how he/she handled the drive and it's contents. Way to much contact, way to hard on the arm and heads. It's not surprise it didn't work. If you have to keep out 0.3μm sized particles, what do you think happens when you touch the disk surface even a little bit? It wasn't going to work at all the way they did it. No amount of clean room in the world is going to substitute for terrible technique.

1/10

 

[Megamox]

That's true, although I must admit I had a great time watching it!

Okay faulty drive arrived today, here are some preliminary tests:

My Dead hard drive: Seagate Model st9320320as, Firmware version DE05, Date of manufacture (09304 = 23rd Jan 2009)
Ebay Test Drive (Faulty): Seagate Model (Same), Firmware (Same), Date of manufacture (09297 = 19th Jan 2009) - 4 days apart, what luck!
Seagate date code calculator here: https://www.bugaco.com/calculators/seagate_date_code.php

First test: Connected Ebay test drive to see what it's fault actually was, no description given in ebay listing. It turned on, revved up fine, detected by windows, popped up in my computer. But the drive (supposedly 320GB) only showed up as 70GB. I suppose this is the fault they mean. Dragged a few files back and forth, seemed to work fine. For a drive that doesnt show its full space, I'm not sure where the fault really lies, anyone?

Second test: Took the board from my bad drive and put it into the ebay drive. I wanted to see if my board was really faulty. The ebay drive spun up and was detected by windows, hoorah! The USB notification popped up saying device was ready to use, but sadly nothing appeared in device manager or windows drive management. I know obviously the adaptive rom memory is different from board to board, but I think this is a good sign that the board is possibly okay.

Third test: Put the ebay board into my dead hard drive instead. Same clicking, non recognition. Expected as much since the heads do sound faulty.

Put everything back in their own boards and drives and my faulty drive still clicked as usual and the ebay drive still showed up as 70GB as it did when i started. Checked the files I had transferred to it earlier and they read fine. So the tests above seem to have had no damaging effect on the components. Great.

Next tests, to test the effect of torque tightness on drive functionality. These seagate 2.5 laptop drives seem to have 5 torque screws around the outside and one under the sticker. Here's the test results:

Undid all 5 torque screws around the edge of the drive. They all felt the same in terms of tightness and not alot of tightness really. Just enough so you hear a little click when you undo them. Tested the drive with these 5 screws removed. Drive fine, data readable.

Undid the 6th torque screw under the sticker which I believe is of obviously more importance as it connects to something in the middle of the drive as opposed to the edge chassis. Turned on, drived WAILED and SCREAMED. Turned it off quickly and replaced the torque screw under the sticker. Drive now fine, read and writes fine (with just this one torque screw holding the top together). Conclusion: this screw under the sticker is very important, yet I paid no attention to special tightness, just generally kept it how I found it.

Last test: Removed all screws from the top. Lifted the cover off just enough to dislodge it to undo any 'memory' the lid had of being connected to the drive. To throw things off I suppose. Rescrewed the 'Important' screw under the sticker and turned on. Drive fine. Reads and writes fine. Conclusion: Removing the lid for this particular model and dislodging it doesnt seem to kill the drive as long as you keep the screw that was under the sticker on there. In terms of the advice given in the Gillware Tech video, perhaps he meant you need to be careful with torque tightness on some models. My experiment here shows it doesnt necessarily apply to this seagate model i'm working with in particular. That's all I can really say. Results may vary when I decide to take the top off completely, perhaps it'll work fine without the lid and this particular screw altogether.

Thoughts?
Megamox

 

[kubeek]

The middle screw holds the head assembly from the top, so no surprise it has a lot of influence. Looking forward to your next tests

 

[alec_t]

For a drive that doesnt show its full space, I'm not sure where the fault really lies, anyone?

Have you checked for a hidden partition?

 

[Megamox]

Alec_t, you just beat me to it! Yes there's a unallocated partition that is about 250GB. So it appears this drive, sold as faulty, is really working fine. Feel really bad about experimenting on a working drive, but it was only £10 and needs must. Essentially, I want to test if opening up the drive is worth the data loss vs data recovered.

Okay results of the new experiment in. This one I'm sure will be very interesting. This experiment is to show the effect on data integrity by opening up a hard drive and exposing it a non clean room environment for a period of 3 minutes (A time I feel is sufficient for a head swap, with practise).

Step 1: Loaded the drive with 70GB of data, movies, mp3s and text files to have a good spread of filetypes to test later.
Step 2: Opened the drive up completely in a clean-ish area, and exposed the platters for a period of exactly 3 minutes. Did not perform any work on the drive. Just took a picture of the internal.
Step 3: Closed the hard drive and hooked it up. Detected fine, windows reported all 70 GB used up as it was before I opened it.
Step 4: Transferred 70GB back to my home hard drive C:\ to check for errors.
Step 5: Windows reported every so often an error in copying/reading the data back. The errors seem to be of error checking CRC nature.
Step 6: 2 hours later using (USB 2.0) have copied as much data back to the hard drive that could be read.
Step 7: Cyclic Redundancy errors noticed by the copying program on transferring data back to the home drive. These files were skipped to measure the difference and error rate.

Conclusion: For this model of hard drive exposed to 3 minutes in a non clean room environment with no attempt to shield or protect the platters results in a data loss of approximately 34.5%. (65.5% data recovered successfully). This was for 70GB of data.

This is my worst case bench mark for a repair. I wonder how much a data recovery company would charge for almost 70% recovery when it seems this level can possibly be done by the individual. I'm sure they operate in the high 90 percentiles though. Or at least I hope they do for a drive with no physical internal damage. I reckon I could do the head swap in under a minute and use the hard drive cover to keep the platters shielded during this time. Perhaps this might make a difference. I would say the result here supports the conclusion that exposing the internal of your drive to a non clean room environment will degrade the data no question but at least this quantifies it a little. Bear in mind I have not attempted to run the hard drive through any type of data recovery program to restore the unreadable data either, if thats even possible. So essentially, 3 minutes of exposure = 34.4% data loss. Also at no time did i notice even the minutest speck of dust or debris land on the platter. So this is unavoidable. Also during the data transfer, the drive did no make any different sounds than normal, sounded just fine.

Summary: The world will not end if you open up this model of hard drive. If you unscrew it, take the top off, put it back on and read the data again. The drive will not explode. However data loss is inevitable this way without a clean room almost at a rate of 10% per minute exposed. However it can be done and with refinements may be done well, depending on how much risk you want to take. As this is a 'faulty' drive and my first ever oafish attempt, it's just practise here. Almost 3 hours after opening and closing, the drive sounds fine and apart from the obvious sectors of unreadable data, looks and spins up like any ordinary laptop hard drive and is still going strong. Looks like it'd be surviving long enough to back up data from it if this was a real scenario. I'll of course post an update if it dies horribly later on down the road.

What do we feel my chances are for a head swap in significantly less time than 3 minutes, with at least some sort of aid to protect dust settling on the platters might be? Worth the risk? Thoughts? Advice?

Megamox

**broken link removed**

 

[()blivion]

Well... if you have the head comb and lock, and were extraordinarily careful, I don't see a reason you can't try it. But without a clean room, things almost certainly WILL get damaged. The question is, will you be able to get it working enough to get your data off of it? You also need to be super careful with taking off the arm magnet. If you hit the back of the arm with a tool while you are taking it off or putting the new one on, there could be problems.

If the data isn't worth the recovery fee, but you don't really want to throw it away, you might as well try and do the repair yourself. Good learning experience. Just don't have any delusions that it is going to work out well. You should surrender to the idea that your data is totally gone before you even try.

 

[()blivion]

Honestly, if this happened to me, the biggest problem would not be recovering the actual data, but just having a list of what I had. If there is a script I wrote or project file that I would need to recover, just knowing the name or general contents would most likely be good enough for me to recreate it in it's entirety. I would probably end up improving on it too as some of my stuff is from a while ago, I have learned many things since then. Though I admit, I would rather not have to re invent my own wheels if I don't have to. Finally there is such thing as data priority. The majority of the HDD space I have taken up is freeware or installs for software I have backups for in the form of the original disk. Such things that I could just download/reinstall again if need be. My critical (unique) files would only take a few DVDs at most.

Most things that take up HDD space exist elsewhere. So sometimes just knowing what you lost is as good as any backup. Custom data are usually small in comparison.

 

[MrAl]

Hi,


It would be interesting to build a home clean room.

When we say "clean room" it doesnt have to be a huge room like they use for silicon wafers where several people have to be in the room at the same time. All we need is a 'clean room' big enough to hold a hard drive and two hands and forearms. Maybe not impossible nor expensive.

Starting with a box made of plexiglass, glued on all edges so there are no air leaks. Medium size box should be ok but that's up to the individual. Cut out a rectangular hole in the top, drill several holes and tap them for small bolts. Cut a piece of plexiglass that fits over the hole with enough overlap, drill holes for the bolts, use weather stripping along the edges so when the top is bolted on it seals the box up.
Cut two smaller holes one on each end, one smaller than the other.
Cut two large holes in the side for the hands to fit through. Tape two large clear bags (maybe regular plastic food storage bags) inside the box over the holes on the side, being careful to seal them to the box completely.

Place the hard drive into the box from the top as well as any replacement parts, screws, and any tools that are needed, bolt the top on.
Pump charcoal filtered air into one side, maintain a small pressure inside the box, letting a small amount of air escape through the other (smaller) hole in the other side.
As the box is starting to 'fill up' with clean air, place the hands into the two bags. Wait a few minutes and then see if you can manage to take the top off, replace any parts, and reassemble.

You can also look for better filters for reducing particle size.

 

[()blivion]

I said pretty much the same thing in Post #60.

 

[MrAl]

I said pretty much the same thing in Post #60.

Hi,

Oh ok ha ha. Well sorry about that. I didnt read every single post in this thread. You also provided a few more details which was nice.

I agree with most except for the gloves. I think it may be hard to find gloves that would give free enough movement and have long arms for full movement. I suggested plastic bags because they are quite flexible, but there's always room for improvement. Robot arms anyone ?

 

[()blivion]

Understandable, my post was TL;DR anyway.

But yeah, the gloves are one of the harder parts. Professional such devices usually have folded over parts and the ends of the cuff are belled out quite a bit. This makes it somewhat telescopic and lets you attach the cuff to a much larger hole so you can move in all directions more freely. We used to have a sandblasting box that was like this. Another idea is just to make a rubber sphincter like orifice and just cover your arms up good. non powdered nitrile long cuff laboratory/exam/clean room gloves or similar may work for that and would provided unmatched dexterity. It definitely would be better from a contaminant perspective if it was hermetically sealed by design. You could possibly use the same gloves and seal them if they were the beaded cuff type.

IMO, the biggest problem would be air filtration. I think it would be hard to miniaturize an effective true HEPA filter and still get the air turnover rate needed to purge the whole box. Though admittedly, it would be a lot easier to turnover a smaller volume of air. It would just have to be a jerry rigged design which lowers the chances of success significantly.

 

[MrAl]

Hi again,


Miniaturize? Why Miniaturize? Why not just pipe some of air in and let some escape, maybe even use the escapement pipe as pressure regulator so there's not too much pressure in the box.

My biggest doubt was finding a filter that gets better than 1u particle removal. I think charcoal is known for 1u but im not sure about better. Maybe a better filter material.

 

[jpanhalt]

On Glove Boxes and Such

"Disposable" glove boxes are available: https://www.glascol.com/product/subproduct/id/58 . We just call them I^2R bags based on their original manufacturer.

The first time I used one was about 1965. The prices have certainly gone up since then. Despite the clumsy appearance, you can handle milligram amounts of materials for setting up air/moisture sensitive reactions and purifications. I have disassembled drives just to see how they are made. I think one could easily do it inside such a bag. Current pricing might be a barrier. Of course, list price is probably inflated at least 30%.

You would need a filtered air source. A small cylinder of nitrogen might suffice.

If you know anybody who works in a pathology or molecular diagnostics laboratory, you may get access to use one of their HEPA-filtered hoods. There are hoods that flow toward the user and ones that flow to an exhaust. You want the type that flows toward the user.

John

 

[()blivion]

HEPA filter size.
I would want to minimize hard seals that needed to be made into the side of the box, so I would want the HEPA filter totally inside the box. Thus I imagine it would need to be made small. I could be wrong though. I suppose you could always put it directly outside the box. Then It could be almost any size.

< 1u particle removal
Activated carbon (charcoal) filtering is really really good at absorbing "chemicals". That is... odors, fumes, caustic gases, volatile organic solvents, these sort of things. What it is not so good at though is removing particles. It's a common misconception that sub micron filtering uses a sieve action to remove particles. It's actually more ballistic, or kinetic in nature. This means that the physical construct of the filter and it's fluid dynamics are very important to sub micron particle capture. Activated carbon also uses physical structure for filtering action, but it's method is different than what is most effective for 5-0.1 micron filtering.

Re: jpanhalt
Those glove/boxes are a nice find. Yeah, that would work nicely I bet. It's all about the filtered air source at that point. Maybe just any normal source of canister air ran through some kind of mini HEPA filter. With a post filter, such as a fused activated carbon membrane, or a carbon aerogel filter. An electrostatic precipitator is still an option too.

I like the idea of a recirculating filter though. Seems like you would get better particle capture if the air in the box was run through the filter several times before you actually cracked open the HDD. Then again, if the filter had any statistical chances of spitting out particles then the point would be rather moot.

 

[MrAl]

Hi,

I like the idea of the recirculating filter too, just in case the filter can miss some particles and then catch them during the next pass. I'd think there was a pretty good chance of that making at least some improvement.

Funny i was thinking of a garbage bag even without gloves, because if the plastic was thin enough you could just grab the item inside the bag through the plastic itself. I couldnt think of where i could find a clear bag though, that was also thin.

I have a filter myself and i could easily add something to it that would pipe the air into the box/bag, perhaps letting some escape as we discussed. So that's no problem really. Duct tape comes in handy as does hot glue and hand moldable low temperature plastic

I am also wondering how much we have to worry about excess moisture.

 

[jpanhalt]

For large, clear, heavy weight bags, I would suggest "barrel liners." Bags used by dry cleaners for cloths may also work, but are thinner.

One underlying, yet unproved assumption is that the lost data in the opened drive was due to particulate contamination from the air. We don't know anything quantitative about the conditions, such as how dusty the area was, humidity, sources of EM, fomites from the experimenter, or anything else that could affect the drive.

I would be a bit less concerned about suspended particles in the air than about other aspects of the process. For example, the outside of the drive is certainly dusty. And, there is potential for metal dust/scrapings from the screws, etc. A gentle flow of filtered air/nitrogen might be sufficiently laminar and provide adequate protection compared to the other sources of error.

John

 

[MrAl]

Hi,

Yeah lots to consider. What would help is if we could find someone who worked in a shop that did this for their business and ask them for some tips. There might be some tricks like a magnet to pick up stray metal (presumably iron based) particles (not too close to the platters of course). Maybe turning the thing upside down, etc.

 

[atferrari]

Hi,

Oh ok ha ha. Well sorry about that. I didnt read every single post in this thread. You also provided a few more details which was nice.

I agree with most except for the gloves. I think it may be hard to find gloves that would give free enough movement and have long arms for full movement. I suggested plastic bags because they are quite flexible, but there's always room for improvement. Robot arms anyone ?

MrAl & Oblivion

I wonder how is the actual cleaning of the boards used to build the box/room. Just filtering seems the last part of it. But a deep cleaning (whatever is that) prior start filtering seems a must.

From a friend who has a friend who does data recovery, it appears as a quite standarized process with a usual high percentage of success. But when I asked him to recover data of my last failed HDD, he replied that for that brand he needed to buy something that did not justfied the expense. (?)

I do not know what to think of that.

 

[Megamox]

Hey I'm really liking the information on the DIY filter boxes, I'll have to look into that! Doubt I have the expertise to build but perhaps one of you may attempt your own and show us? Love the little bag thing, maybe that'd be easier to fabricate. Not sure how much those cost.

Anyway, here's some more test results. We're going to test the effect of removing and re-attaching the magnetic plate which sits on top of the actuator. How delicate is it, especially without the right tool?
- Opened up hard drive and pulled off magnetic plate by hand. Not very difficult, slightly grasping the plate from both ends and gently lifting. The magnet holding it underneath is quite strong though. The professionals seem to use a magnet sucker tool thingie, which is better.
- Replaced metal plate. Lined it up and it snapped back, no damage seemed to have been inflicted on the drive heads due to this motion.
- Closed the hard drive. The whole process took about 10 seconds.
- Drive detected by windows, another 2 hours of reading = Data Integrity exactly the same = 65% recoverable. No head failure issues. Is one test meaningful though? Nope!

New Test, since the value of one removal does not statistically mean much:
Repeat the above test but with FIFTEEN consecutive removals and re-attachments. Again being careful not to be sloppy but in no way going for pinpoint accuracy on the removal and replace. Whole procedure of 15 remove and re-attaches took about 30 seconds. Nothing else done to the drive during this process. Drive lid closed and hooked up to windows for another test:
Results: No change in data integrity (another 2 hour process).
Conclusion: the plate which is magnetically held on top of the acutator can be removed by hand during a repair procedure without damage to the head as long as you're not sloppy. Hard drive cover was used to sheild platters during this experiment and minimise the effects of any other variable. Although there's probably a million microscopic variables I havent even accounted for as previously mentioned with the screw fragments etc.

Next test: Drive head removal and reattachment and its effect on data integrity. The big one!
Step 1: Removed all torque screws from the test ebay hard drive. There's also one underneath the drive from the PCB side which holds the actuator arm. You must not remove this actuator screw by turning the hard drive upside down because the actuator arm will not be connected to anything and crash into the upper lid.
Step 2: This is going to sound ridiculous but.. took a party balloon and rubbed it furiously against the wall and cellotaped it down on my work area. To attract dust!
Step 3: Flipped the hard drive lid but kept it so it partially covered the platters to aid sheilding. Started the stop watch and positioned the balloon overshadowing my work area.
Step 4: Undid the ribbon screws and decided to use the orange ramp as my portable comb for the heads. However the comb and heads are not connected in any way, the slightest movement of one or the other either way and the heads will flick loose and this apparantly is very bad. Game over if this happens, although it may be fun to test this next
Step 5: Grabbed two cocktail-sticks/matchsticks and snipped them to length. Used them to bolster a connection between the actuator and the orange ramp (ie my new head comb). Had to use tweezers for this, was very fiddly but no where near impossible. Undid the orange ramp torque screw and lifted the whole thing up out of the drive. Examined the heads to make sure they were still aligned (about 10 seconds) and dropped the whole lot back into the drive and secured it.
Step 6: Reattached everything making sure not to even touch the platter at all. Closed the lid, screwed everything back down. Stopped stopwatch, total time 1min 48 secs.
Step 7: Plugged device back into windows, recognised as 70GB as usual.
Step 8: 2 hours of copying the data back - Data recovered = 56.3% (A loss of 9%).

Conclusion: Head replacements are possible! However my ghetto methods are probably no where near efficient. But a ~10% loss of data for a first hard drive head swap isn't bad, had this been a full recovery I'd be looking at 63GB recovered out of 70GB (Assuming the failure was purely related to the head). Still, 7GB loss is not going to win any awards Also this process may have degraded data which was already degraded and so I would not have been able to account for this.

Better ways to improve the method:
- That head comb tool would be so much easier than bolstering the actuator to the orange ramp, it was shoddy at best and would have surely fell apart under any kind of tension or torsion. This solution is really ugly. Tweezers definitely important. However this method is possible. Maybe someone can knock up some sort of head comb on one of those programmable model making CNC machines? Would have made the method almost childsplay.

- The balloon I hope helped to attract any dust particles away from the platter. Also gave it a spray with furniture polish because it's supposed to attract dust too. Anyways I've ended up with a shiny balloon to play with. Seriously though, was this just crazy?
- If you're attempting this, please test it on a test drive first. You learn so much and you really don't see how difficult it is until your trying to manipulate tiny delicate components first hand. I'm no expert and I'm hardly a practical genius, but practise makes perfect. I recommend playing with a dead one to learn, have fun!
- Obviously if any of you are creative enough to get your own clean room going (I dont think I am), I think you'd have a shot at a pretty good recovery.

I did spot that some of the 'corrupted' files were only partially corrupted. For example a few were 200MB movie test files which only skipped a few times during playback. Obviously if this file were two smaller 100MB files instead, the recovery percentage would have been higher as one of them would have possibly have been recovered. If your hard drive is full of tiny files you'd probably get a better margin of recovery. I set my windows copy program (Bust Copy) to skip any file completely if it had a transfer problem. Maybe data recovery programs would be able to do a better job but havent looked into that. I loaded the test drive with a range of files to model my original dead one, so whether or not the recovery is truly accurate or not I'm not sure. The good thing that even though only 56% of the drive is recoverable at this point, I can pull off a entire file listing for the whole 70GB drive so at least I have a complete file reference to compare it to what's missing.

All experiments done in the kitchen, on a clean area, but not much else really done to prepare it. My laziness at least gives you a bench mark for what you can improve upon, or at least that's my excuse

Conclusion: Had this particular drive been suffering from a failed head and assuming head transfers between similar models is the correct method to fix this (Seems so from the youtube videos, I believe they swap heads rather than try to fix them) then I would have recovered 70GB -7GB = 63GB = 90% recovery (complete estimate). Is home repair of hard drives possible? One experiment on one hard drive at one time isn't really telling. Doesn't look impossible though and I'm not using any specialist equipment. The balloon doesn't count

Next experiment ideas? What if your attempt goes horribly wrong and the heads make contact, is it game over? Effect of head contact with a) something foreign, b) two heads together? The heads actually look like two smooth black rectangular surfaces, almost the size of 1mm. Maybe they can be cleaned with a squirt of compressed air? That head is a work of art though, very compact, very serious, very opposing to adjustment. Maybe eventually I'll get around to a platter removal and replacement, that looks really difficult though.

Does any of this mean anything, or is it just beginners luck? Im trying to give clear scientific results with no bad advice or bias

Megamox

**broken link removed**
**broken link removed**