Thread
Transplanting Hard Disk Platters?
How possible is it to transplant hard disk platters, at least for the purposes of data recovery when a drive mechanism is otherwise physically dead?
In particular I have a 40MB 2.5" SCSI Conner HD and an old style IWM Hard Disk 20. I would imagine if it i possible, I would need working drive mechanisms to sacrifice. Once cracking the drives open, I would assume that I have basically destroyed their reliable useful life?
In particular I have a 40MB 2.5" SCSI Conner HD and an old style IWM Hard Disk 20. I would imagine if it i possible, I would need working drive mechanisms to sacrifice. Once cracking the drives open, I would assume that I have basically destroyed their reliable useful life?
It is quite possible, and in fact it is one of the methods used by data recovery services, so your intuition is spot-on (as usual).
If you are able to open the drives in a clean environment, lifetime need not be dramatically reduced. You could probably rig up a "clean box" to create a mini-environment that is dust-free.
I once cracked open one of my many stiction-afflicted Quantum 40MB 3.5" drives, and for fun, I hooked it up to a Mac, just to see how long it would run with the cover off. It ran, and ran, and ... I lost patience, went out to dinner, and came back to a very crashed, shrieking drive. So it ran for several hours with no cover.
I then made a Plexiglas cover for another one of these drives (inspired by Ze Cube transparent Classic), and substituted it for the original. I did all of this on the kitchen table, and took no particularly extreme measures to keep things perfectly clean. That was many years ago, and the drive still works. It's fun to watch the mechanism in action (once it decides to start up).
Now, a couple of rudimentary experiments hardly constitute a reliable data set, but it would seem that instant destruction is not an inevitable consequence of opening up a drive (of course, it helps that these old, low-capacity drives have generous track dimensions and head-platter separations; I doubt that modern drives would be anywhere near as forgiving). That, in turn, gives us some hope that a "clean box" that a weekend experimenter could economically construct just might suffice.
If you are able to open the drives in a clean environment, lifetime need not be dramatically reduced. You could probably rig up a "clean box" to create a mini-environment that is dust-free.
I once cracked open one of my many stiction-afflicted Quantum 40MB 3.5" drives, and for fun, I hooked it up to a Mac, just to see how long it would run with the cover off. It ran, and ran, and ... I lost patience, went out to dinner, and came back to a very crashed, shrieking drive. So it ran for several hours with no cover.
I then made a Plexiglas cover for another one of these drives (inspired by Ze Cube transparent Classic), and substituted it for the original. I did all of this on the kitchen table, and took no particularly extreme measures to keep things perfectly clean. That was many years ago, and the drive still works. It's fun to watch the mechanism in action (once it decides to start up).
Now, a couple of rudimentary experiments hardly constitute a reliable data set, but it would seem that instant destruction is not an inevitable consequence of opening up a drive (of course, it helps that these old, low-capacity drives have generous track dimensions and head-platter separations; I doubt that modern drives would be anywhere near as forgiving). That, in turn, gives us some hope that a "clean box" that a weekend experimenter could economically construct just might suffice.
You would need an exact model (right down to firmware and revision) donor hard disk/PCB electronics for this to work. Anything slightly different, and it won't work.
JB
JB
And once you have those matching disks, one maneuvre to try first is to transplant the electronics board from the working to the non-working drive. If that doesn't work, transplant it back and then try the platter transplant.
Also have you tried the other low tech methods of relieving sticktion? ie Freezing and thawing in a moisture proof bag, shaking, knocking on the table etc.
Also have you tried the other low tech methods of relieving sticktion? ie Freezing and thawing in a moisture proof bag, shaking, knocking on the table etc.
Take some other defective drive mechanisms to practise the procedure of dismantling and mounting. There are strong magnets inside for the actuation of the reading heads. You can rip off or bend the heads with near to no force. Attempt to do this task of platter transplant with enough time to work patiently.
Another thing to keep in mind is that you'll most likely want to remember what order the platters are in - installing them differently may result in it not working.
I'm pretty sure that "may" is actually "will"
As am I, actually. 
What tomlee59 said above. I opened a 40MB one, planted it in a Quadra 605, and installed 7.1 just to watch it working away. A full OS install and some app installs later, I put the lid back on, and left the drive - it still worked OK a few years later.
I have no idea how careful you'd need to be with the heads, or removing platters - that's something I've not done - but as far as simply exposing platters to the elements, for a drive that age & size it worked for me.
Dana
You'd need to be very careful with the heads - the metal that they are on is very weak and easily bent, and bending it just a millimetre is enough to render that head useless.
I heard someone made a clean room out of his bathroom, something about the steam weighing down all the dust. I doubt the moisture would be good for the drive, but might want to read more into it. Especially if you need the drive. Those old drives are getting hard to come by.
My idea for that (never been tested) is to put a box fan next to your work area(given that it is clean) and just keep the fan running so the air is always moving thus keeping dust away. Just a theory.
it would also be pulling dust though... which is problematic. Could try something with a closed container, a wet filter, and a fan. Have ports with gloves fitted in them. Like the thinks they keep premature babies in.
Now why didn't I think of that?
Macintoshme has correctly identified the logical flaw in your theory. Moving air doesn't just blow dust away, it also blows dust to. Much better to have still air.
Let's go back to step one.
You mention a 40 MB 2.5" SCSI drive and a 20 MB, 3.5" Rodime (proprietary) drive. Do you mean you want to swap platters between those two? Or just that you want to perform this on those two completely separately?
Assuming you mean to do it completely separately, it would be possible. But, for the love of $deity, don't sacrifice a working HD20!!! (I have two dead HD20s that I need to rip open and try to fix, but I would never even consider ripping open a known-good one in an attempt to get either of my broken ones working.)
If the information is really that valuable, pay a dedicated data recovery company to recover it.
Mac128 knows what he's doing; he is not proposing to swap a 2.5" platter with one from a 3.5" drive, or vice versa.
Just one question: would swapping the drive platters even work? When all is considered, a hard drive is an analog device with some electronics converting that analog signal into a digital format that a computer can handle. Any misalignment would likely distort the signal beyond use.
(Yes, I know that certain SyQuest drives were essentially removable hard drive platters. On the other hand, SyQuest drives were designed with the removable bit into consideration and they probably made some compromises to get it working.)
That said, nothing ventured nothing gained.
If you know it's something like stiction though, I would give the drive a few good twists around the rotation axis then try recovering the data immediately. If that doesn't work, I'd go onto opening the case and rotating the platters at the hub then trying it. I would not consider removing the platters since it is to intertwined with the drive heads to remove without the risk of damage.
If you don't know what the problem is, try swapping the electronics board. I don't know how dependent the firmware level is. I would imagine that the wrong firmware (or mixing up the plate) would leave you with a readable drive, but the OS would have trouble sorting out the blocks in order to reconstruct the file system.
I would not use the fan, for reasons already mentioned.
I would not weigh dust down in a steamy bathroom, since you are just asking for condensation.
Good luck. And if your data isn't important, try to have fun.
(Yes, I know that certain SyQuest drives were essentially removable hard drive platters. On the other hand, SyQuest drives were designed with the removable bit into consideration and they probably made some compromises to get it working.)
That said, nothing ventured nothing gained.
If you know it's something like stiction though, I would give the drive a few good twists around the rotation axis then try recovering the data immediately. If that doesn't work, I'd go onto opening the case and rotating the platters at the hub then trying it. I would not consider removing the platters since it is to intertwined with the drive heads to remove without the risk of damage.
If you don't know what the problem is, try swapping the electronics board. I don't know how dependent the firmware level is. I would imagine that the wrong firmware (or mixing up the plate) would leave you with a readable drive, but the OS would have trouble sorting out the blocks in order to reconstruct the file system.
I would not use the fan, for reasons already mentioned.
I would not weigh dust down in a steamy bathroom, since you are just asking for condensation.
Good luck. And if your data isn't important, try to have fun.
The alignment question is a good one. In very old mechanisms, the location of the tracks is defined as in a floppy drive. Later drives use a feedback positioning method. Swapping produces no problem in the latter -- the servo will find the tracks and center the head over them.
In the case of the older open-loop drives, you may have to perform an alignment. It's easy, though, and the method is the same as in a floppy: Rotate the head-positioning stepper motor until goodness occurs.
That said, swapping platters is definitely an operation of last resort. As others have suggested here, other methods should be tried first (freezing, thwacking, swapping electronics, etc.)
My guess would be ditto on which way up they are [I'm pretty sure that "may" is actually "will"
] ]'>
I've pulled tops off of several brands of hard drive, to varying degrees of success.
Quantums are really easy, because they're (usually) the toilet-bowl design with the stamped metal top that easily screws down, with no odd protrusions into the lower bit of the case. Most of them work fine with or without their tops, but some single-platter drives won't operate properly exposed because of the peculiar implementation of the AirLock system.
Conners can be problematic, as it's easy to hit the platters when removing the bowled top casing. Other than that, most of the ones I've run exposed have exhibited no problems. In fact, several have run without their tops when they wouldn't run at all otherwise.
Seagates are a mixed bag: some work fine, some don't. Medalists are usually fine, but (sub-10GB) Hawks and Barracudas typically freak out and won't even spin up if the top is off. U Series drives usually work nudified, but they're difficult to disassemble with both the sealing tape and the shock-absorbing "condom".
WDs work fine open, but older ones are almost always sealed with tape, which is close to impossible to remove in one piece. When I reseal the taped drives (of any brand), I use thin strips of duct tape.
IBMs are also very cooperative, even the notebook models. A couple of odd UltraStars are the exception, because their casing is split down the middle.
Fujitsus, Maxtors, MiniScribes, and Toshibas typically run exposed well with no real issues.
As for positioning systems, there have been many different approaches over the years. Early drives often just used stepper motors, counting the steps the motor moved the heads to tell where on the disks they were positioned.
Some later drives, Quantum and Sony for sure, used an optical tracking system. These drives employed a small attachment to the armature, a small piece of glass or plastic with a bar code-like grid on it, read by an infrared LED and pickup. The electronics would count the movement of the gradients on the tracking grid, and thus keep track of where the heads were positioned.
As a side note, on Quantum ProDrive LPS drives (from 40 to 230MB, employed quite frequently in Macs), the little "tick-ticka-tick-ticka-tick" sound emitted every 15 to 30 seconds is the sound of the electronics recalibrating the heads, just in case it missed counting a gradient.
Then, modern drives use embedded servo information. Some drives would dedicate an entire platter to maintaining servo information, where others embed the servo data alongside the usable data.
I've never tried swapping platters or head assemblies, though, as most of my problems have been easily solvable (stiction, damaged PCBs) or unsolvable (extreme shock or water intrusion).
As for swapping PCBs, older drives (<4GB) are picky about firmware revisions and all, but modern drives work just fine so long as the boards are from the same model and capacity.
Quantums are really easy, because they're (usually) the toilet-bowl design with the stamped metal top that easily screws down, with no odd protrusions into the lower bit of the case. Most of them work fine with or without their tops, but some single-platter drives won't operate properly exposed because of the peculiar implementation of the AirLock system.
Conners can be problematic, as it's easy to hit the platters when removing the bowled top casing. Other than that, most of the ones I've run exposed have exhibited no problems. In fact, several have run without their tops when they wouldn't run at all otherwise.
Seagates are a mixed bag: some work fine, some don't. Medalists are usually fine, but (sub-10GB) Hawks and Barracudas typically freak out and won't even spin up if the top is off. U Series drives usually work nudified, but they're difficult to disassemble with both the sealing tape and the shock-absorbing "condom".
WDs work fine open, but older ones are almost always sealed with tape, which is close to impossible to remove in one piece. When I reseal the taped drives (of any brand), I use thin strips of duct tape.
IBMs are also very cooperative, even the notebook models. A couple of odd UltraStars are the exception, because their casing is split down the middle.
Fujitsus, Maxtors, MiniScribes, and Toshibas typically run exposed well with no real issues.
As for positioning systems, there have been many different approaches over the years. Early drives often just used stepper motors, counting the steps the motor moved the heads to tell where on the disks they were positioned.
Some later drives, Quantum and Sony for sure, used an optical tracking system. These drives employed a small attachment to the armature, a small piece of glass or plastic with a bar code-like grid on it, read by an infrared LED and pickup. The electronics would count the movement of the gradients on the tracking grid, and thus keep track of where the heads were positioned.
As a side note, on Quantum ProDrive LPS drives (from 40 to 230MB, employed quite frequently in Macs), the little "tick-ticka-tick-ticka-tick" sound emitted every 15 to 30 seconds is the sound of the electronics recalibrating the heads, just in case it missed counting a gradient.
Then, modern drives use embedded servo information. Some drives would dedicate an entire platter to maintaining servo information, where others embed the servo data alongside the usable data.
I've never tried swapping platters or head assemblies, though, as most of my problems have been easily solvable (stiction, damaged PCBs) or unsolvable (extreme shock or water intrusion).
As for swapping PCBs, older drives (<4GB) are picky about firmware revisions and all, but modern drives work just fine so long as the boards are from the same model and capacity.
Once I had to recover data from a harddisk drive that was submerged into water by accident. Water was gurgling inside the case. Never attempt to connect a drive in this state! To repair the drive I removed the metal top and used plenty of deionised water to rinse all the innards of the drive thoroughly. Afterwards I put the drive into a plastic bag, accompagnied with some bags of drying agent (silica gel) and left it for several days. Once the drive mechanism was dry I rotated the platters a bit to resolve stiction. After reassembling the case and connecting the drive to a computer it spun up and allowed to copy the complete contents to another drive. This harddisk was quite usable for a few week, still, until the water initiated corrosion caused fast growing defects.
Glad to hear the happy ending to your tale, register!
On the other hand, think for a moment of fume cupboards in chemical laboratories, or better, laminar-flow microbiological work hoods. Given that the work to be protected is at the bottom of the contained space, horizontal airflow above it is protective if the size/mass/velocity of contaminating particles is such as to prevent them from falling to the bottom of the space within the time taken for the air's transit of the space.
Still air demands total filtration/cleanliness, or the work can be contaminated by the effects of gravity on contaminating particles. Laminar flow allows for some contamination to be endured if the mathematics of risk are correctly evaluated. In conjunction with filtration of the inlet air, it can work a treat.
de
Yes -- a lot certainly depends on the location (and nature) of the contaminants. In a typical home lab (yes, I am making assumptions here, but ones based on visits to a good number of home labs; dust seems to be a common companion), there's likely to be a fair amount of deposited dust on many surfaces. Blowing air in those environments virtually assures a redistribution of that dust, to the detriment of the disk platter. In still air, the rain of dust will be much more moderate, given the assumptions.
I agree that the best arrangement is filtered air, with a gentle circulation to scrub out particulates that evolve within the protected space. Semiconductor fabs are arranged this way, with the primary source of contaminating particulates in that situation -- humans -- properly isolated within protective suits.
I wish I had observed more interest in this thread earlier...
The first removeable disk drive that I used was the DEC RL02 pack, followed by various Syquest designs. I never met an Iomega design before the Zip drive. But from all of those drives we can learn a little about how a home brewer might build an enclosure in which to disassemble and reassemble disk drives. This is a first pass at an idea, so corrections are welcome.
1. Build a box that is fully enclosed (ie capable of sustaining slightly above atmospheric pressure without leaking). The walls of the box need to be "closed" (ie the fabric of the box will not generate dust in itself). The box needs a filtered air output (easily cleaned) and a chemically clean pressure system (ie no lubricant leakage) which is again filtered. Flow for the pressure system must be laminar, particularly at the inlet and outlet to the box.
2. An enclosed box is useless to work in. So you are going to need external/internal gloves (as seen in all videography of nuclear power plants) that allow you to access the box from outside.
3. The box will need a decontamination zone by which tools can be introduced. The same rules apply for the decontamination zone as for the enclosed box.
4. Insert your drives and run the pressure system for hours to drive dust out of the system.
5. To clean the box out initially, run the pressure system high to generate turbulence. After that, don't crank up the positive pressure because you may recirculate undisturbed particles.
The first removeable disk drive that I used was the DEC RL02 pack, followed by various Syquest designs. I never met an Iomega design before the Zip drive. But from all of those drives we can learn a little about how a home brewer might build an enclosure in which to disassemble and reassemble disk drives. This is a first pass at an idea, so corrections are welcome.
1. Build a box that is fully enclosed (ie capable of sustaining slightly above atmospheric pressure without leaking). The walls of the box need to be "closed" (ie the fabric of the box will not generate dust in itself). The box needs a filtered air output (easily cleaned) and a chemically clean pressure system (ie no lubricant leakage) which is again filtered. Flow for the pressure system must be laminar, particularly at the inlet and outlet to the box.
2. An enclosed box is useless to work in. So you are going to need external/internal gloves (as seen in all videography of nuclear power plants) that allow you to access the box from outside.
3. The box will need a decontamination zone by which tools can be introduced. The same rules apply for the decontamination zone as for the enclosed box.
4. Insert your drives and run the pressure system for hours to drive dust out of the system.
5. To clean the box out initially, run the pressure system high to generate turbulence. After that, don't crank up the positive pressure because you may recirculate undisturbed particles.
Given that other people, in this thread and elsewhere, have rebuilt drives on their kitchen benches, and the objective is just a one-time data retrieval, isn't that a little overboard?
If it's all for self-entertainment or for learning, more power to the elbow of the experimenter. If it's for a panic-stricken recovery of unbacked-up data (naughty, naughty), it's understandable, even if it is, as Bunsen writes, perhaps overkill. But a one-shot recovery must work first time, and who can say in advance how elaborate that one shot needs to be?
de
de
Agreed, overboard for a quick data exchange. Like others, I've run naked drives for experimentation and it is quite hard to deliberately kill them. But original hard drives are becoming difficult to find, so a home brew clean room environment is worthy of consderation. Would you want to rebuild a Lisa widget or a Profile or an ST-506 without making the effort to maximise success?
one thing about changing platters on a voice coil drive.
if its a multi platter drive, FORGET IT. its easier to change the heads than the platters. because if you undo the platters, if the two platters slip any at all (which they will, without a piston compressor or platter tool). you will loose the cylinder and servo track alignment, and there goes your possibility to recover data... down the drain.
so you need to watch for that. its far far easier to change the head/actuator and swap between two identical drives than it is a multi platter.
the tolerances between platter to platter alignment tracking and head pressure bering is a micron or less.
if its a multi platter drive, FORGET IT. its easier to change the heads than the platters. because if you undo the platters, if the two platters slip any at all (which they will, without a piston compressor or platter tool). you will loose the cylinder and servo track alignment, and there goes your possibility to recover data... down the drain.
so you need to watch for that. its far far easier to change the head/actuator and swap between two identical drives than it is a multi platter.
the tolerances between platter to platter alignment tracking and head pressure bering is a micron or less.