Hard Drive Resurrection

Follow along as reader [anonymous Gort] swaps the guts of two hard drives to bring one back from the land of the dead.

Someone at work had a laptop computer they never backed up. They traveled 1000 miles to give a presentation, using the laptop

235 thoughts on “Hard Drive Resurrection

  1. i’ve also had alot of success with freezing the hard drive. take care not to let ice crystals form on it (…or inside it). it usually has a reduced read speed and it eventually stops working again, but freezing the drive gives you plenty of time to get the data off, or even boot from it if thats what gets you happy. i’ve tried this three times with success each time.

  2. First, this is a good thing. Data is more important than hardware. Remember that. Always.

    Second, I’ve been thinking what one could still do with old (damaged) drives. Indeed, practice, reconaissance (learn from how details stick together before taking action) and swapping parts.

    If you take it the way it is stated, it looks pretty simple, but messing with valuable data is indeed tricky. The last (7th) statement should be repeated endlessly: it might be your last option/chance/attempt…

    – Unomi –

  3. Often (well, in my experience at least) the internal platters, read/write heads, motor, and solenoids etc. of a “fried” hard drive are all perfectly fine, but the logic board on the bottom will have had a component fail. Luckily – these can be removed and replaced with identical logic boards from a good drive without exposing the platters. In the past I’ve saved hard drives from the trash this way.

    I assume however, that in this case you knew some internal mechanism (though not the platters themselves) had failed?

  4. As anyone who has taken a hard drive apart knows, the platters are held together by physical force of a top mount clamp held by torex, there is no key to keep them aligned, so if one platter is out of allignment with the other, then the bottom of the last platter’s sector loction information should not be in allignment with the other platters, and all but the one platter with the sector alignment data should be unreadable. There is no way to take all the platters out as a single unit. OF COURSE there could be a newer standard of sector location data prefixing each sector on the disk, in which case, this _might_ work. Either way, the screws that hold the platter bobbin and spindle motor to the case are found under the platters, and necessitate their removal (with consequential misallignment). Given the lack of real specifics on this treatment, I think that this operation is fake.
    In addition, what went bad on the drive… the spindle motor? the actuator mechanism? the platter surface?
    Freezing the drive seemed to help me once, a better idea is to put the drive on a linoleum (sp?) floor covering concrete, and put a heavy block on top to stabilize the drive against vibration (if that is the mode of failure).

    1. Now an cylindrical shaped instrument is available which clamps around the platters, maintaining their relative position. Then you unscrew the top screws and lift the multiple platters out as a unit. It’s called a “platter removal tool”.

  5. In response to #5, the drive in question was a laptop HD which I take only consisted of one platter in which case the operation was not fake. I suspect your alignment concerns are valid for multi-platter drives.

  6. When removing the platers from the disk, how do you stop the head from touching / scratching the disk. I have found that the “clean room” does not have to be so clean if you work quickly. I ran a 6GB disk with the cover off (running a contious scan disk) for a whole weekend before bad sections of the drive started showing up.

    What does the cold do to the drive that helps recover the data?

  7. This is such a load of horseshit. Um, there’s no way to realign the heads on the platter so they start at the correct sectors. I’d say you could have recovered the info if you had bought an identical drive and swapped out the pcb on top of the drive. That works a lot of the time, but, there’s no way in hell you took apart a drive and recovered the info by swapping the platters to another drive.The platters themselves spin at thousand of revolutions a second. What do you think would happen to your alignment if there was the tiniest piece of dust that impacted the head, that microns above the surface of the platter. Why the hell do you think they’re assembled in a green room to start with?

    Sorry, but this kind of misinformation pisses me off.
    And no. Freezing the drive will do doodly squat in reviving it. Most times that clunk you hear in your HD is the sound of the drives head tearing itself free. Unless the pcb itself is gone, there is virtually no way to ever recover the information outside of a professional setup.

    1. I had two hard drives (at different times) on one of them when it spun up you could hear the slightest “clicking” or … banging noise – the other no noise at all … totally silent … I could feel both spin up tho – took each drive put in freezer for 4 hours – one SATA one IDE – first attempt failed… still could not see either drive – second attempt 4 hours later this time in freezer for 5 hours……… sata drive came up as did the IDE drive I pulled data from both …
      had everything I “thought” I lost forever… thankfully I did NOT read Darrick Terry’s (above post first) if I had … I might not have even tried …………..

  8. If the allignment truly is a problem, might there then be a way to raw-copy the entire drive with respect to the platter indexes? Thereby creating separate raw image files for each surface, copied sequentially from the outer to inner track (or vise-versa, I forget the proper order). And when it finishes, let some custom software examine the individual platter structures looking for inter-sector gaps and other physical properties of the drive that would have been copied as data to effectively re-align the platters and then produce a functional disk image.

  9. Whenever you have to make any sort of presentation, always always ALWAYS print out lecture notes and overhead transparencies, and keep those in your carry-on luggage (but not in your laptop case). It’s cool that they were able to hack out a solution to this, but never be without a technology-less backup.

  10. raw-reading the un-aligned data is unlikely to work. modern drives store the actual data in a very different format that what it looks like from the outside.
    for instance: from the outside each track has the same number of sectors (at it used to be on real drives, and still is on floppy disks), but the drive will probably put more sectors on the outer tracks, because they are longer and therefore can store more data per revolution.

    aligning the disks may not be a problem at all (at least when reading a single platter), because they probably use the data on the platter to get the alignment info anyway. The motor is not accurate enough to predict the exact location of the disk, and a separate sensor that generates a sinc pulse every revolution is an (expensive) extra part, and probably also does lack the needed accuracy.
    I think you can at least read some of the data this way.

  11. a couple of (small) dust particles are not a big problem. most drives circulate part of the inside airflow over a filter that picks up any dust particles, and even if dust is slowly sanding the heads and/or platters, you can probably get the data off before it stops working.
    He mentions a lot of noise after the swap. this is very likely because he did not do any balancing of the platter assembly. also if the platters are mounted off-center the heads will need to vibrate very fast in order to keep following the track. but i think a drive is capable of following an off-center track, just like a cd-player. have you ever bend a cd and watches the cd players lens move up and down to keep the focus correct ?

  12. joe and Tercero, Have you actually tried what you are describing and failed or are you both talk out the side of your necks? The platters do not stop so that the the heads can read them, instead the platter spins, the head just grabs what it wants when it comes by, the drive should easily be able to “seek” for the particular data, even if it has to wait multiple revolutions for the right sector to come by.

  13. #9 freezing works. I’ve done it. I have friends who have done it. It is a well documented solution. What it does (I’ve heard…not sure of this) is shrink the parts a tiny bit, which can, in some cases, allow you to use it while it stays cold. Though I know someone who had that fix it perminantly. I mean, considering that I’ve managed to fix things by opening them, deciding I couldn’t fix them, and closing them up again, I wouldn’t discount any theories.

  14. Having just had a WD 160g die on me, i’m rather interested in this article. Reading through the comments, esp about alignment concerns me though. The drive powers up and it sounds like the head arm is whacking against the body for about 90sec and then it just stops and the bios doesn’t recoginize a drive there.

    Would this be worth cracking open? Regardless, does anyone know legitimate, reputable, trustworthy, etc companies/sites that can perform proper repair? Many sites offer no charge unless data is recovered, but since the drive was part of a raid/lvm array, the costs seem to double.


  15. #16 if it was part of a raid array then fu*k it who cares about the data. its on the other drives as well unless you used raid 0. if you did use raid 0 then youre screwed because the data on the other drive is mostly trash as well.

    but anyways it sounds like the bearings in the head have gotten sloppy so freezeing the drive should allow you to back up the data onto another drive. next time you should use raid 1 or raid5

  16. ‘The hard drive got fried and all there photos, movies…’

    ‘there’ should be ‘their’ ;)

    Interesting way to recover data, though – I’ll have to keep this in mind if/when one of my clients’ drives give up the ghost.

  17. Will this work for the infamous IBM Deskstar drives? i had one fail about a year ago, it’s still sitting in my room someplace because it had 120 gigs of data on it that i really miss. if i can get another, working deskstar of the same size, can i drop the new head/arm on the old drive or pull the platters out as the article poster did and drop em in the new drive?

  18. This is a great fix but should only be used as a last resort. I recommend
    1) Tap all drive sides sharply against a table. – This will often free the drive if something is binding.
    2) Freeze the drive in a zip lock sack (to reduce chances of condensation.) – This will cause shrinking and increase molecular density very slightly.
    3) Replace the PCB with an exact match. This will rule out electronic components and can be done without opening the housing.
    4) Open drive and transfer platters to a second drive. On many drives the platters can be removed as a unit.

    I used to work in a computer repair shop and had a number of customers willing to try anything to get their data back so I’ve had several HD’s to try these steps with. I’ve had successes and failures with all of them. However, I haven’t tried a platter removal where the platters couldn’t be removed as a single unit so I can’t speak for or against the possibility of misalignment in that situation.

    Remember – Some Data is always better than No Data


  19. I call BS, too. With modern drives, they assemble the platters and lock them to the spindle, then at the factory (and never again) the servo tracks are written to that particular alignment of the platters and spindle. The position of the tracks is not obvious to the heads; they seek the servo track info to find the tracks. So there’s no chance of moving platters to a new spindle and preserving the alignment of the platters to the spindle. Imagine how hard it would be to re-align concentric circles within a few microns. They might be able to do this at the $1000+ data recovery places, but I don’t think it can be done in a “clean room” on a desktop.

    Swapping logic boards – yes, that’s easy and often successful.

  20. I hope all of you have learned to keep backups of the data you care about. It’s always a gamble whether these recovery methods will work, but if you have a backup, it’s a SURE THING. Every time.

    Back your data up. Keep a backup off-site of your most critical data. Never trust any single hard drive to keep your data safe. Ever.

    If it sounds like I’m preachin’, that’s because backing up is a religion that everyone should follow.


  21. I am trying to resurrect a dead hard drive but i need help… WHERE DO YOU FIND LOGIC BOARDS? so far i have only found one but they dont respond so i can order one. if you know where please email me gorac369 at gmail dot com

  22. After almost spending 2500 bucks on HD recovery, but having the initial fee waived because it was unrecoverable by a place one step down in price from drivesavers, I stomaeched the PTSD once again and kissed my 145th gig of lost data goodbye.

    I have lost 2 HDs and quite a few gigs just to formatting and enthusiastic youthful error, but nothing is worse than not having the recoverable state to work with. This operation is executable only if the drives platters are good and there are few errors on them.

    This is an excellent tutorial, but more detail and specific scenario recovery tactics need to be developed and disclosed. At a minimum more detail is reccomended, as this is more of a seeded concept with pictures than a convention at this point.

    I strongly reccomend that anyone who has a hard drive fail turn off their computer and leave the L1 and L2 recoveries to dedicated machines, as attempting to recovery is more challenging than it is presented here.

  23. To chinakow…

    Yah. I’ve done the freezing route. Doesn’t work. I did have success with replacing the pcb, and got the info out that way.

    There’s also a number of software tools you can use to access the drive. Do a google for a software called PC3000. It’s made by a Russian company, it’s very expensive, but it’s made for recovering info from supposedly dead drives. Also look for firmware that will re-intialize your pcb if it’s gone funky.

    Diskexplorer, Easyrecovery, GetDataBack…there’s dozens of other types that will work on a drive that’s been wiped…but not a dead drive.

    Lot’s of people who are throwing ideas here have never tried to recover the info, or they wouldn’t repeat what doesn’t work.

    //14 years at IBM, Toronto/Ontario/R&D…

  24. I heard a guy once who froze his drive to 0 Kelvin, then took it apart (with gloves), transferred the platters to another drive which was kept at 30 Kelvin. He then flipped it upside down and tapped at the exact center 5 times (very lightly) with a hammer. He then put it in a paper sack, buried it in his backyard for a week. Some animal dug it up though so we don’t know if it would have worked or not.

    Just kidding. These are all good ideas, and why not try them on cheap old drives. Discounting the post based on what you “may know” is silly. Unless you try it don’t knock it. It’s all heresay in this post.

  25. As in # 15, the drive does wright and read on the fly. This is the cause of fragmentation. Also when formated, each sector are identifyed. So I guest that it could be possible to run the HD for a few hours. But it might mecanily suffer a lot and die because of the extra work neede to read sector in order.

  26. The freezing thing is for drives that fail with the click of death once they warm up. I have never put one in the freezer but I have used ice packs many times on drives that are failing in this way. I have also use the method of swapping logic boards but apparently you need to have a drive of the same make/model created in the same batch to even hope that it works. I’ve had more failures on this recovery type than success.

  27. I recently had an important 200gb drive crash on me due to a corrupt Master File Table. Does anyone know if it’s at all possible to rebuild said file table? I know the data is still accessable because I have run sector reading software with a 95% success rate.

  28. the operation works i tried it on a seagate baricuda that lost a head luckily the platters sufferd no damage and the drive worked long enough to get my valuable porn off of it clone it is a good utility to use also for doing the data transfer

  29. I have had success with a logic board swap. I had a RAID 5 array of 4 120GB WD drives that I bought from Dell for a steal. First, one drive failed, and then another. Data lost. Not too happy about that but life goes on. I think it was a bad batch of drives. My suspicions were pretty much confirmed when a friend complained that their computer wasn’t working. I took a look at it and their hard drive had died. I took the HD home to see if I could get any data off of it. No luck. Then I looked at the label on the drive. Hmm… 120GB WD drive! I matched the part number/batch number exactly to the drives I had in RAID 5. I was able to do a board swap and get the data off. I suspect that would have worked with the two drives of mine that died, but I didn’t have enough logic boards. Logic board swaps work!

  30. From my experience while in the industry of fixing computers, there are 5 solutions to fix drive problems. Listed in order of price.

    1) Put drive in another machine, run data recovery app on drive.

    2) Open drive. Spin the platters with fingers. Fixes bearing issues where they have seized. Freezing the drive _sometimes_ also works.

    3) Image the drive (ghost,etc) and place the image onto another (good, working) drive. Use data recovery app to retrieve data.

    4) Replace electronics (PCB) with identical model.

    5) Take platters out. Smash them into the customers face, because he is a moron for not backing up his data.

    The only thing I have not done, and know nothing about is swapping platters. From my limited knowledge, it’s possible, as in the 80’s, you could take the platters off, spit on them, wipe’m clean, put’em back… and it would work for another year or two. Rinse and repeat. Old bud of mine STILL does this. None of this platter allignment mumbo-jumbo. My BS flags are going up on that one.

    The clean room thing is bullshit if you are trying to recover the data off the drive.

    Granted… if you plan on using the drive for any length of time (i.e. longer than it takes to get the data off), you kind of need a dust free area… and don’t do it in your back yard. The kitchen table is fine.

    These are the methods I know work, as I’ve used them personally a lot more than once.

  31. In response to all the freezing comments, I’ve had good luck in restoring bad drives by throwing it in a firewire enclosure, then throwing that in the freezer. The condensation issue never happens, and ice crystals never form because there’s a dessicant in the drive anyway (silica gel, anybody?). I’ve restored 2-3 drives this way, and each time, they ran for a decent amount of time.
    (One 200GB drive ran for about a week, and a small 10GB server OS hard drive managed to survive for about 3-4 days.) Luckily I’ve been able to get away with not managing to rust my enclosure’s circuitry (thank god for good assembly/solder paste).

    If the drive starts dying in the freezer, you can usually powercycle it, and it’ll come back for x amount of time before it dies again.

    Whatever ailment the freezing route fixes, I’m not sure. But it’s worked in my cases.

  32. “if you plan on using the drive for any length of time (i.e. longer than it takes to get the data off), you kind of need a dust free area… and don’t do it in your back yard. The kitchen table is fine.”

    Acually the kitchen will still probably be dusty if it has open entrances. For a “true” clean room simulation you want to use the bathroom. Turn on the shower until the room completely fills with steam, turn shower off, let the steam settle. The steam brings all the dust to the floor/toilet/sink surfaces and gives you as close to a “dust free clean room” as you’ll get in your house.

    I’ve used this method when windowing hard drives and it’s worked great-i’ve opened drives, put them back together and have them still running to this day. Of course all this should be done very carefully and only if you really care about your data.

  33. “if you plan on using the drive for any length of time (i.e. longer than it takes to get the data off), you kind of need a dust free area… and don’t do it in your back yard. The kitchen table is fine.”

    Acually the kitchen will still probably be dusty if it has open entrances. For a “true” clean room simulation you want to use the bathroom. Turn on the shower until the room completely fills with steam, turn shower off, let the steam settle. The steam brings all the dust to the floor/toilet/sink surfaces and gives you as close to a “dust free clean room” as you’ll get in your house.

    I’ve used this method when windowing hard drives and it’s worked great-i’ve opened drives, put them back together and have them still running to this day. Of course all this should be done very carefully and only if you really care about your data.

  34. To the two jokers who claim this is impossible. OnTrack and a dozen other companies offer this exact service professionally for thousand sof customers every year. This is EXACTLY how you recover data in a last resort situation. Since the platters SPIN, there is no need to “align” it any more than you align a record.

  35. For the people who question about what happens when you freeze a drive, here’s my guess:

    The effect freezing has on matter is that it slows down particles/electrons. I think that by slowing down the operandi of the drive, the chance that it finds the correct data is getting higher. It is not going to the max at high speed, but just at a lower speed making it more accurate.

    I have no clue what I’m talking about, since I’m not a specialist at all. It is just what I think what could be logical. Anybody else an opion on this?

    – Unomi –

  36. To all the idiots who refute any of this is possible, you clearly don’t realise that each seperate problem has it’s own unique fix. Simply because something didn’t work for you it doesn’t mean that it doesn’t work. Either you tried the wrong fix or you were stupid or unlucky. To the fool who started to confuse people by introducing the term “align” instead of “synch”, take solice in the knowledge that you are 100 times more respected than the fuckwit who decided to spell lame someone.

    All the methods used have worked in the past. Platters are locked together as a pair and can be removed as such. If anyone claims that they cannot it simply means that the poster was unable to do so.

    I worked in IT for bluechip comapanies for over 2 decades. I have worked for disaster recovery and data recovery companies so I do know what is done and what the outcome can be.

    Can someone explain why half the respondents here want to shoot people down in flames when it is patently obvious that they haven’t got the technical accuity to install a new drive into a PC, let alone take one to bits?

  37. The first time I opened the chamber of a hard drive was in 1983, was a 15MB Tandon MFM drive, and it had been clobbered by UPS. That one turned out to be stuck heads (meaning jammed so the positioner couldn’t free them). After I freed them and put a clear lid on with double-sticky, the drive worked for over a month before starting to show bad sectors…

    Since then, I have literally “fixed” thousands of drives working inside the chamber; but, never by swapping the platters because I found there was microscopic slop on the spindle and I worried about never being able to get them centered properly.

    My technique has always been to move the head assemblies from an identical drive (after the PCBA, of course) and the reason it was fabulously successful was because the final, analog, read/write IC is on the head assembly, prone to being blown by power surges.

    It is often necessary to tweak the position of the heads on their shaft (sometimes it takes hours to get it right); but, when the drive performs it’s POST and restores to track 0, a quick cloning gets the data onto a good drive.

    No, the drive won’t last very long; but, even when the only tool for cloning was Norton, which took forever, it worked 70+ percent of the time.

    One drive, in the late 80’s was 8 platters & 16 heads, all recessed down inside the chamber. Now that was tricky!

    Having said all this, I should add a big note; recent models of both Maxtor and WD drives fail by corrupting the Grown Defect List which is on the media and, because there is no blown hardware, just a confused drive, none of these techniques will work. I have tried……

  38. I’ve had several drives die in the past month and just started toying with this process earlier this week. Anyone have any advice on moving the heads off of the platter? The last drive I played with had a single platter, but heads on both sides. In order to get the platter out, I had to drag the heads across the platter and off where the heads proceed to slap together. To get them back on the platter you have to somehow get them spread and slipped back on. That drive ended up dead…

    Is it bad to drag the heads across the platter since they normally float? Or is it acceptable since the platter isn’t spinning? I guess I need some sort of little block to keep the heads separated so that I don’t twist them trying to get them back on the platter.

  39. I worked in a PC repair shop for 6 years and now have a degree in computer engineering. I saw all kinds of messed up hard drives come in. I have repaired many hard drives in my day using various methods. This is a last resort in the case the drive IS spinning but not being read by the BIOS.

    Older (<60G) drives
    4/4, 100% success in retrieving data

    Newer drives (20G laptop, 160G)
    0/2, 0% success

    I am inclined to agree with the folks saying alignment is critical with multi-platter drives. This swap technique may only work with single-platter. I recommend swapping the board/chips first if the drive still spins up.

    If it doesn't spin up, I agree with #22's steps. Freezing is a stupid idea in all, but I have seen it work before. Of course the drive imploded 5 min later.

    A note on cleanrooms:
    I stole this from a FAQ on mushroom growing. To create a ghetto cleanroom try these steps in a bathroom:

    1. Seal/Tape off all vents (and windows if necessary). You do not want ambient breezes stirring dust.
    2. Clean work surface and tools with lint free cloth. Set up workspace, leave HDs in static bags.
    3. Turn on shower with hot water for 10+ min. Leave door closed for 20-30 min. The steam will cling to dust and settle it out of the air. The humidity will also reduce the potential for static buildup.
    4. Move very slowly back into the room and close the door gently. Try not to stir or breath heavily.

  40. i am supprised it worked because from what i heard the platters have other servo tracking dta on them so that they can sync up the data and if one platter is off by a fraction then the hardware cant track the data when data is spanning multiple platters.

    single platter drives you may have better luck with.

  41. I am rather sceptical about tranferring the platters to another drive. I do agree with freezing and finding the identitcal “top board” those 2 will work and actually I have had a rather high success rate using those two methods. When I say identical I mean identical
    all the way to the second decimal of the Rev on the board.

  42. Generally – platter swapping will NEVER work on current 3.5″ drives unless it’s a single platter (and REAL low capacity). The reason this worked so well on the laptop drive is because most of them have a “park” position that is completely off of the platter surface – this makes removal of the platters very easy (although still not the right way to do it).. Ever since 120 to 160 gb and bigger 3.5’s starting coming out, the track density has been greater, and alignment and flying height MUCH more critical. I work for a Data Recovery shop and we do these day in and day out – moving heads across the platters while not spinning will destroy the heads and most likely the platters as well. Freezing is by far the most correct first step for the layperson. However 90% of what we see lately is FIRMWARE related. Clicking – firmware, not posting – firmware, bad sectors (CRC errors) – firmware. The comment above about PC3000 is also correct – great package (if you’ve got an extra $6k to $8k and read fluent broken russian) but this utilizes a hardware component not just software and only works on 1 out of 10 drives because of newer firmware incompatibility.

  43. why not use dd to make a disk image, then recover files from that?

    I’m sure you coudl slave a bigger drive than 20GB, and then you could recover more files, possibly. Also, after the drive died, youw oudl have a perfect virtual copy of it.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.