Hard drive failures can be tough to bear, particularly for the average person who doesn’t back up. When it comes to data recovery, there’s always those bleating from the sidelines that it’s a job that must be left to the professionals. However, this is Hackaday, not HireSomeoneADay, so [Matt]’s video on do-it-yourself hard drive repair is just what we like to see.
The video begins with plugging the non-functional drive into an external caddy, and using a microphone to listen to the sounds it makes. Upon analysis, [Matt] concludes that the drive is not spinning up, and suspects the heads may be stuck, causing the problem. When tapping the drive fails to unstick the heads, the next step is disassembly.
Despite the best advice from armchair commentators, this can be achieved at home without a clean room. [Matt] opens the drive carefully, and notes that the head is indeed stuck to the platter, instead of sitting in its home position. Using a screwdriver, the platters are rotated in their usual direction while gentle pressure is applied to pull the head away, being sure to use a light touch to avoid ripping the heads off entirely. With this done, the drive is reassembled and powered up. Amazingly, the repair is successful, and data is able to be recovered!
It’s important to note that this is a highly risky procedure, and not guaranteed to succeed. Truly valuable data should be left to the professionals, but if you’re skint or simply unbothered, it doesn’t hurt to have a go yourself. Be sure to avoid dust entering the drive, and take care not to touch the platters themselves. Of course, if you brick the device, you can always scavenge it for parts. Video after the break.
I also did this once myself long ago. The data was backed up but there were a few things I would have liked to recover but wouldn’t mind if it failed.
I made an improvised clean-room in the bathroom by letting the shower run for half an hour. No idea if that really did anything though :P
Well the shower most certainly increased the room humidity, which is great for reducing possible ESD problems other then that I would not know how effective a waterfall is in reducing the number of dust particles in the air.
I’d think that is highly depend on the type of shower head.
It will remove quite a bit of dust when it’s one that saves water by mixing it with a lot of air, I assume.
The water taps around here all have aerators installed and I don’t like to get my drinking water from them because I often get the feeling of “junk”/dust getting stuck at/in the back of my throat when drinking such previously “aerated” water.
But since I’ve never verified that scientifically I’m not entirely sure it’s not imagination (thou my little almost-blind test did support my “findings”).
you need to run the hot water for steam to interact with the dust particles, it is even recommended during the analog/homemade processing of the photographic films during the drying stage
Hot water does nothing and it is not even steam. Steam is water heated to a gaseous state and in that state it is invisible but when it begins cooling you then see water vapor and how water hitting the air cools off in almost instantly. So the “interact with dust” statement would apply if you are mopping the floor with water but you do not know what steam is so how would you even know what dust does?
The point of the shower thing is to fill the room with water vapour and then wait for it to all fall out of the air. It’ll suck up much of the free-floating dust in the process. Not great, but not terrible.
Of course, it’s pointless if you leave the door open :)
I did this also several times back in the 5 1/4″ and 3.5″ times. Most times the drives contained commercially important but not critical data. Professional data recovery was, due to the costs, always out of question. I never had anything near clean room conditions, just removes all unnecessary stuff from my desk cleaned it, and locked the door to avoid someone coming in and moving dust around.
I powered up the patient and listened, if it did not spin up, I opened it. Once opened I looked for obvious things like arms in the wrong position or circular scratch marks (due to head crash). Than I carefully rotated the disks, most times I recognized a blocking at first, than free movement. Next I closed it, powered up again and copied the data. If there was a head crash I copied with dd until it failed and than dd’ed the rest by skipping some cylinders. So even in case of a head crash the data loss was minimal and unimportant. In one case I had to replace the pcb of the drive, I just swapped with one of the same type.
I did not count them but I think there where about 10 and I got all needed data from them. I can’t say anything about modern drives, but my point is it is absolute worth trying. With 3.5″ drives it was really simple, I’m just a normal guy with a not so clean admins office in that time. So be not afraid of killing something which is already quite dead (-:
check out ddrescue, it can skip all the faulty blocks for you, really helpful
Really really helpful, came here to ensure this was mentioned, ddrescue is a lifesaver those less fortunate clients…. Come-on guys backup your shit.
Come on guys, back up your DATA.
If you’ve got backed up shit, DIY treatments and over-the-counter meds are usually effective, but if that doesn’t shift it, you need to see a medic fairly fast.
This is definitely a trick to keep in mind.
It goes without saying once the lid is removed, all bets are off regarding the longevity of the drive, so once you put the lid back on, I’d have `dd-rescue` or similar ready and waiting to slurp as much data off that disk as possible before it fails permanently.
However, as a last-chance recovery method (assuming you don’t have the money for professional recovery), well worth trying.
….was hoping to read of a half hearted proper hack of swapping platters over to another drive, then slapping on a clear CD case cover with silicon sealant for looks. Heh, Perhaps next time.
I tried that hack once. Drive never worked again.
You have to support the head bearing and spindle otherwise the slight movements between them can error out the drive
If it’s a multi-platter drive, you must maintain their relative position…if one of them rotates even a little bit, the data will be unreadable garbage. If you succeed at transferring the platters without rotating them or destroying the “new” heads, you still have run calibration, the HDD will NOT do that automatically.
This is a very common thing to perform on older drives I have a few video of it myself
https://www.youtube.com/watch?v=vbIrgJGMJ-o
https://www.youtube.com/watch?v=Z-0O_1OP34M
https://www.youtube.com/watch?v=GSXjDUqnUhA
etc
If you choose to do this and you’re not a data recovery professional, know that you will either not get much help from a professional subsequent to you opening the drive casing OR it will cost you a massive amount more money if you do find someone who is willing to pickup after you have a play with it.
Also, if a dust particle gets between a head and platter you could cause far worse damage than you started with. Opening a hard drive outside of a clean room is playing Russian Roulette with it, sure you might get away with it, but the clean room takes that one bullet out of the six-shooter.
In my experience data recovery “professionals” typically do a half arsed job anyway, and very few actually have access to a proper clean room. if lucky they’ll have access to a clean box, or a decent dust filter system, but a clean room isn’t on the cards at most places, nor is it typically required. Stuck heads or dead spindle motors are actually a relatively rare failure state. Stuck heads are often recoverable without opening the platter enclosure portion of the drive as well.
I’ve done more than a few data recoveries for people who tried the professionals first and were told their data was unrecoverable, despite having paid >$1000 for the attempt. Unfortunately there are too many scam artists in the market who will take the “non-refundable” up front service fee, sit on the drive for a month and then return it saying it is unrecoverable.
This. Happened to one of my clients. Sent a server with two failed drives in a RAID array. Scammers wouldn’t even respond to him. They gladly took his $3000 though.
Never had to open a drive, but have had to clone crashed drives (particularly MacBook drives of a certain vintage) for repair from an image. I found ddrescue particularly good for doing a quick first copy, then slowly filling in the damaged areas, often taking days to do this, but effecting almost total recovery. For MacOS, DiskWarrior is still my go to tool. For NTFS disks (usually when someone has formatted over the boot block and/or deleted the partition table) I have used TestDisk and this https://github.com/Lazza/RecuperaBit project to reconstruct data structures successfully. Would be interested to hear what other people use!
I’ve used ddrescue for block level copying in a “professional” (written applications around it for work) and personal setting more than anything else. DD comes in second, but once I felt comfortable with ddrescue and mapfiles I never used anything else.
Thanks for sharing. I have a drive in a old Sony VAIO that is doing just what you described so I might as well give your method a whirl :-)
the problem i have is normally i get a call from someone asking about recovery after the drive has been clicking for days and they leave the pc on in hopes that it will just work it out. it really is amazing how much of a groove those things can dig in there……
nice video and link
This. we disassembled one dead drive at work, and found it looked like a vinyl record. A good inch of the platter’s radius was scored deep enough you could feel it with a fingernail.
However, most “totally dead” drives had PCB issues not hard drive, so a PCB swap with the same model (usually the same batch even!) worked to get data off.
I’d very much like to see the results of running ddrescue on that very drive. With this storage density I bet the results might not actually be that great.
Also, with sound like that, first thing to check is not the stuck heads, but the motor driver. A motor driver with faulty phase output will give exactly same symptoms as this drive.
I will only leave this here: https://www.youtube.com/watch?v=SzIS_q4yS88
Ah yes, yard drive recovery at home. Gotta love it. I’m surprised nobody has mentioned The Freezer Trick! I was skeptical, but it works. The fully story is on my blog: https://miscdotgeek.com/adventures-in-hard-drives/ But the short version is that the freezer trick saved me some effort from rewriting some scripts that were on my home server drive that failed JUST as I was replacing it. I also have some recordings of what dead drives sound like.
My favorite “dead” drive was a laptop drive that had all a customers real estate business data on it, and of course she had no backups. I was trying to mount it with Linux, and it just wouldn’t mount. While the mount command was still running, I put the disk on its side so a trainee could hear the clicking. And right then, it mounted! Got the data off of it and educated the customer about the importance of backups.
A year later, I went back to the same customer. No backups. ¯\(º_o)/¯
I have also successfully used the freezer trick twice. On one of the efforts, I had to freeze it twice.
I put the drive inside a plastic bag with the power and data leads coming out of the top. Then I taped around the top to prevent air exchange and condensation on the drive. I left it in the freezer overnight, then connected the drive and used file explorer to pull the files off the drive.
It’s a great first technique, since there’s little chance of making things work.
Typo, last line should have been “little chance of making things WORSE!!
yep. worked for me several times
I’ve successfully unstuck discs by holding the drive between two fingers placed over the axis of rotation, then tapping the long edge of the frame with the handle of a screwdriver. Inertia keeps the platters from spinning while the drive frame rotates.
I doubt that trick would have worked with [Matt]’s really stuck heads, but it’s worth a shot.
Also, the cleanliness is important in that a particle of dust can damage the platter surface, flying head, or both. But that’s more of a long term reliability issue. When you’re recovering data, you need it to last only for as long as it takes to dd the data off the drive; after which it’s time to open it and scavenge the magnets.
Something to try with those IBM Deathstars. ;-)
Sure, if you have any surviving the past 2 decades.
way back in the heyday of the Seagate ST-225, this problem was called “sticktion”, and resolved by dropping the drive a few feet above the floor
here is more info in the issue:
https://en.wikipedia.org/wiki/Stiction#Hard_disk_drives
Some Mac 68k era SCSI drives had a similar problem. The solution was to smack the opposite end from the connectors with a mallet while turning on power.
our computer shop didn’t have that tool around, but it does now
A calibrated shoe heel or 2×4 were acceptable substitutes
Lol @ ST-225
20 meg mfm or 30 meg rll
Yes that’s MB. This was indeed a very long time ago.
Sometimes you get lucky and this works. You really need to determine wether or not that a stuck head is the issue first, though. Fiddling with the arm on a drive that’s not spinning can do more damage than good. I’ve made tools to seperate the heads from the platter, which you can also buy but they’re usually ridiculously expensive. But with that, swapping R/W heads even on a multi platter drive is a snap. I’ve also swapped platters to an identical working drive to salvage the data.
Part swapping is a bit of a dark art, though. Yet it’s not as secretive an industry as it once was. You can find the information you need if you look hard enough. I once had a WD drive that smoked it’s spindle controller chip. I had to find a donor contol board that was not only the same model, but had to be the same revision, and had to have a serial number within a specific range as well. And I couldn’t just swap the board, as the adaptive calibration data that was stored in the boards memory when the drive was manufactured was unique to this drive. I had to desolder the spindle controller chip from the donor board and transplant it to the patient. It’s a very tiny, 8 pin SMD, and it was a nerve wracking process to do without damaging the chip or the traces. You have to really want that data. Needless to say, a backup a day keeps the panic away.
It’s important to note that after you do this you should back up the data and not use the hard drive afterwards.
Stuck heads seem to be pretty much a thing of the past. Heck, mechanical hdd’s seem to fading away pretty quickly, but back when I used to deal with a lot of mechanical drives, the solution to the problem depends on what is on the drive. If it is critical information, the disk went in for data recovery, failing that it got a really good flat on the floor slam. I never lost one doing that, but I always warned that there was a good chance the disk could be irreparably damaged. I prefer that to trying to pop a disk open. Even more so with the disks that have the top of the spindle bearing fastened to the lid. You can tell these by the hex nut in the center of the curved part of the lid.
How to repair with CRC error
Sometimes you really do get lucky with this hail Mary. I “repaired” an old laptop hdd like this and kept using it in my “for fun” server for four years without issue, 24/7. Was a broke student at the time, and treated the machine as disposable (VPN server mostly).
i had the opposite problems once platter and mechanics where fine but the board was fried. on most of the never boards there are some 0 ohm resistors and micro fuses but since I’m not really equipped for those tiny SMD chips i wasn’t sure how to repair the board and i mean that’s assuming one of the security actually blew and it was not a chip being dead. in the end i opted for the easy solution and bought the same drive again. since the capacity are so high nowadays no drive is the same and they all get optimized at factory so if you get the optimization factors wrong you could do more harm than good. this also mean a straight electronics swap wont do the trick. luckily the flash eeprom containing the data was intact so i just re-flashed the new board with the old data and voila the drive came back to live. was still quit a journey to get it working since the information about hard drives seems to be super sparse.
I have ripped a few 0 ohm resistors off of drive boards to get them working again, temporarily….
mmmm Love this article. I’ve used many of the tricks mentioned in comments. If the owner thinks the data is important enough to pay Ontrack the $1,000+ to recover it and there isn’t something I know I can fix I send it off. Otherwise with dead stuff I have a rule: if its dead I can’t break it worse. :-) Although I’ve never tried to open a drive and use it afterward. But you do want to try the least risky stuff first since something you do may make sure nothing else can possibly bring it back.
One thing I haven’t seen mentioned here is that drives used to have the platter motor’s spindle extend through the bottom of the housing. In those cases I’ve used a pair of heavy needle nose pliers to spin it and break stiction without shock force or breaking the seal. I don’t know if any drives have the spindle coming through the case anymore.
I also highly recommend a RAID level with redundancy!
Poor boys have poor boy ways, so you have to try something. A good reminder to keep multiple sources of backup.
Let me throw a wrench into the works (or at least my friends did). In the 80’s a railroad’s drive wouldn’t start up. It was a very odd computer, an early SBC 386 with a ST-506 drive. Since it’s the railroad (with no funding from congress) they didn’t have a repair budget. So instead of sending it out for recovery they opened it up, spun the drive and started the computer. It booted to MSDOS and ran for 3 more years in a busy track side hut.
Oh they did keep a backup of the data they needed. And I’ve actually seen this mess running in the hut. A hut hadn’t been cleaned since it was built in the coal era.
I’ve since worked with dying drives with stuck heads and stuck bearing. VCF has had work shops to replace the bearing on these old drives. So there are other ways to save the drive. The work shops were not held in the track side hut. ;-)
I’ve done a few home recoveries before, I forego the whole clean room thing cause once the drive crashes I dont trust it anymore, I recover the data then trash the drive. Unlikely the dust in that time is going to make or break the recovery
The risk can be taken but depends on importance of data. If your data is crucial then no one will suggest you to operate the drive by your own, in case of physical issue to the hard drive. Just few small particles of dust can cause permanent loss to your critical data. However, if there is any logical issue to your hard drive like corruption, lost partition or accidental deletion then one can use DIY Data Recovery Software. If needed you can try Stellar Data Recovery software to recover your lost or deleted data from any storage device. It’s 100% safe and you can recover your data without opening your drive physically.
Hey, i tried this with a recent drive that didn’t have very important data on it, the disc looks fine (no visible damage). The thing is that the head/arm won’t get onto the drive, it just goes back and forth from the platter to its parked position. Anyone know what that means? Is it an semi-easy solve or do i have to change the arm or head? Thanks
Did you ever figure this out? I have a WD Red 4TB drive that I suspect is doing the exact same thing. I haven’t opened it yet, but from the sounds of it, the motor spins up and it sounds as if the head/arm are just going back and forth from the parked position to the platter and back again.
I may just say screw it and open it up since it’s effectively useless to me at this point anyway. Seems like a handful of people above have had positive experiences attempting this. What’s to lose at this point I suppose.