Forum latest

Guide: Open hard drive surgery
Written by Gizmo   
Sunday, 13 May 2007 21:00
Article Index
Guide: Open hard drive surgery
Identify the bodies
Expose the guts
Extract the vitals
It Lives!!!!
All Pages



What do you do when you have a dead drive with data that you really, REALLY need to recover?  Well, you can either send the drive to a data recovery outfit and spend a lot of bucks, or you can take the approach of the hardware hacker, and find another drive in your junk pile that is similar enough to swap platters.  This method can work with most hard drives.

Data Recovery

I recently had the misfortune of having to recover some data that had not been backed up from a company laptop (no, it wasn't mine).  The data on the drive were worth a considerable amount of money.  The data could be re-created, but it would take a significant amount of time to do so.  Unfortunately, the poor chap was caught between a rock and a hard place: he needed the data, but because he hadn't been following company policy (regular backups), he couldn't just come out to his boss and say that he needed to send the drive off for data recovery.  Neither could he afford to send it to a data recovery house himself.  Basically, he was looking at spending a lot of weekends and sleepless nights to re-create the lost information, and had to hope he wouldn't be called upon to produce the information until he had recovered it.

The drive in question was a Toshiba MK4025GAS.  This is a 4200 RPM drive with a 40 GB capacity and an 8 MB data cache in a 2 1/2" form factor, with an ATA 133 interface and a single platter.  The reason the drive failed was because he had pulled the drive out of his laptop to plug into his home computer system (he had a drive bay that he could plug the drive into) and had dropped the drive, whereupon it ceased to function.

Upon inspection, I determined that the spindle wasn't spinning, and shaking the drive caused a tell-tale rattle that wasn't particularly encouraging.  So, now what?

Well, the first thought that crossed my mind was to jump on E-bay and see if I could score another drive.  Unfortunately, there weren't any other drives available right then.  A quick search of the 'net revealed that there were a number of places selling replacement drives that were COMPATIBLE, but were not that exact drive.  There were some places that claimed to have the drive, but when you called, they were either out of stock, or selling a drop-in replacement.  Hmm...

Wait a minute!  As sheer luck would have it, I remembered that I had a Toshiba drive in my junk box that was a 2 1/2" drive.  It was an MK3021GAS, basically the baby brother of the 4025.  It was a 30 GB drive and a 2 MB cache, but other than that, it appeared to be the same drive.  The reason it was in my junk box was because the control board had died.  Being your basic tech pack-rat, I hate to throw anything away that might POSSIBLY, SOMEDAY, MAYBE have a use, so into my junk box it went.

The Victims




Right.  So first things first.  Do we have good electronics to work with?  If we don't, then all bets are off.  Best thing to do here is swap control boards between the two drives.  We can do this because they are from the same drive family, and we are counting on being able to use the electronics from the drive we are trying to rescue anyway.

Board View



The board swap is pretty straight-forward on this drive.  Everything mates together using either touch-contact pads or pressure-fit connectors, so removing some screws and a little careful prying is sufficient to to remove the board.  Note the gasket.  This is actually a heat conductive foam that acts as a thermal interface material for the components on the board, as well as an electrical insulator to prevent shorts.  Placement is not terribly critical during reassembly, just make sure it gets back on more-or-less in the same place.

Board Removed



Once I got the board swapped between the drives, I powered up the 30 GB drive with the 40 GB board.  The BIOS found the drive, indicated that there wasn't anything on it (naturally).  On a lark, I decided to see if I could format it.  Lo and behold, it took the format, although it began generating media errors as soon as I tried to store anything.  Oh well.  For this test, it was sufficient.  It proved that the contol board wasn't blown to smitherenes.

Ok, time to whip out the tools and open these babies up.  First thing is to get the covers off.  This proved easier said than done, as they were held on with #4 torx screws.  After calling every industrial and electronics supply store in the area, I finally got desperate and called Radio Shack.  They didn't have it, but the guy thought that Sears might.  Well, having nothing to lose, I figured 'What the heck?' and called Sears.  To my utter astonishment, they not only had the #4, they had a #5, #6, and #7, for the paltry price of $3 each.  A quick trip to the mall and I returned home, ready for surgery, torx drivers at hand.  Oh, did I mention that these things need to be demagnetized?  You do know that magnetic fields are a Bad Thing when dealling with magnetic media, right?  Fortunately I have one, but if you don't have one they can be found pretty cheaply.

Top Cover Removed



So, off come the covers.  First inspection of the guts looks promising; the model numbers of the circuit boards that interface with the drive heads are the same, which is a fair indication that the heads are the same.  This is important, because different drive heads have different characteristics.  Sometimes you can't swap media between even same modeled drives, because they change to a different head during production (usually for cost reasons, but sometimes for reliability as well) and they behave differently on the media.  The data recovery guys know this, and it is my  understanding they have deals worked out with drive manufacturers to ensure that they have working samples of all the different production runs, in order to allow them to recover the data.  I would imagine the CIA and FBI have access to much the same stuff.

Things are looking promising so far.  Next step is to get the platters out.  Doing this is tricky, for several reasons:

  1. You can't get your fingers on the media.  The skin oils you leave behind will destroy the media over time.  In addition, the flying heads run so close to the platter that the debris from your fingers look like boulders and can potentially cause a head crash.
  2. You've got to maintain the media orientation, i.e. which side is up.
  3. You've got to dismount the heads.  Even at rest, when the heads are parked, they are still over the outer edge of the disk.  This is made more difficult owing to the fact that the heads are spring-loaded by the actuator arm, and removing them from the parking carrier will result in the heads either smacking the platter or smacking each other with considerable force, potentially damaging the heads.
  4. Oh yeah, the hub screw is on bloody tight, and the spindle wants to rotate while you are try to turn the screw.

However, before we deal with all of this, maybe we ought to plug the thing into some power and see if it decides it wants to work?   Well, that turned out to be a no-go.  Plugged the drive in, watched the head seek out and back, but the spindle never rotated.  Still don't know what the rattle is, though, because the heads appear to be intact.  So, on to the next step.

Removing Hub



Taking the hub loose is a bit tricky, owing to the fact that (oddly enough) the spindle wants to turn with the screw.  The only way to deal with this is by apply counter pressure to the spindle to keep it from turning while you remove the screw.  Apparently, the manufacturer has to deal with the problem as well (fancy that).  See those 4 little holes around the edge of the hub cover?  They work as dandy fixing points; just stick something in there (in my case, a small jeweler's screw driver) to stop the rotation, and then undo the hub screw.

Hub and ring assembly

Once you've got the hub screw loose, you'll notice that the hub actually consists of two parts.  The top part is a clamping plate (which the screw fastens down).  Below that is a clamping ring.  The clamping plate is made of spring steel, and the whole arrangement is designed so that all you have to do is just tighten the screw down until it bottoms; the plate ensures that you don't have too much pressure on the platter and end up warping or damaging it. Now we've got the hub off, we need to do something real quick to ensure that we don't get the platter flipped by accident.  In my case, I just used a small marker and placed a dot on the disk inside the area where the hub ring rested.  Placing it here allowed me to ensure that I could identify 'this side up' on the platter, while also ensuring that I didn't do something to disrupt the data area.

OK, so now we've got the hub off, and the top side of the platter marked.  Now it's time to handle the platter.  I have a pair of cloth gloves for this; I would imaging rubber gloves would work just as well.

Key Parts

Uh, oh.  The platter won't come out because it can't clear the spindle.  Hmm... that means the heads have to come loose.  This is not a fun operation, but it turns out that we don't have to completely remove the heads.  If you look at the picture again, you'll note that the parking carrier is held in by a screw.  It turns out that with some care, the screw can be removed and the whole assembly moved downwards far enough to get the heads clear of the platter.  Depending on the drive model, you may actually have to remove the carrier as well.  If you do, the heads are going to come together due to the spring tension of the head actuator arms.  I've found that the best thing to do is get a piece of paper to slide between the heads before you remove them from the carrier; this provides a cushion to keep them from smacking each other.  Getting them back apart is a trick, though; I generally use a pair of wooden toothpicks for this.  The problem is that you have to be VERY gentle; too much pressure and you'll distort the actuator arms and the heads will no longer fly over the platter correctly.

Now that we've got the platter out, we just need to extract the platter from the other drive, and then insert this platter in its place.  As the second drive is virtually identical to the first, there's no need to detail that operation here.  In addition, assembly is pretty much the reverse of disassembly, with the addition that when you put the platter back in, you need to try to get it as free of debris as possible.  I used a can of compressed air to blow all of the debris off the bottom surface of the platter before inserting it into the drive, and then again on the top surface after inserting the platter.

So, control board swapped, platters swapped.  Before we actually put the cover back on, we'll go ahead and spin the drive up.  This will allow the drive to spin any fine debris off the platter without having to chance it getting trapped inside the drive.  Once the drive spins up, we put the cover back on, shut down, and finish bolting the cover down.  (I don't do this step with the power on, because I don't want to take any more chances with the drive than I have to.  It's safer to put the screws in when the drive isn't spinning.

Presto!  Bingo!  Power the drive up in the computer, it finds the drive and says there is something resembling real data on it.  Cool!  Stick it back in his laptop, and it boots up to his desktop, good as new.

So there you are; a buddy's career is saved, you are a hero, and it only took a couple of hours.

Of course, you do want make sure to lecture your buddy about doing regular backups of sensitive data.  :)

Let me know what you thought of this article, in the forums!

Don't Click Here Don't Click Here Either