Dissecting a Hard Drive :: 17/02/10 ~00:15 :: Comments: 0
category: random/tech
Following on somewhat from my last post about my dissertation, I decided to destroy an old drive I had floating around. I got a proper screwdriver set today so I can take apart drives, since hdds tend to use torx screws and I had sod all to use on those. The drive I'm using here is an old Maxtor IDE drive, this particular one is a 6gb 3.5" one (model number is 86480D6 if you particularly care). And here it is!
Hadn't really thought to take photos before I started this, so I'd already taken the screws out there. There were 2 screws which had labels covering them in the middle. I would appear to have voided the warranty by tearing them. Oh noes.
So, there you go, just a standard (if very old) hard drive. The fact that it's IDE doesn't really affect anything, since the mechanics of data storage on a hard drive are irrelevant regardless of the connector type; it's still got platters and heads. This drive is pretty old, so the technology is a bit dated.
The board is a standard printed circuit board, there's not much of interest on there. As part of my work, I'm planning on trying to switch individual components about a bit from drive to drive, or at the very least the connectors. Bit worried about doing it though since this uses a lot of very tiny components; soldering on the base of one of these is going to ruin it because there's just no way I'm not gonna end up bridging over connectors, which will either make it just completely fail or just plain not even start up, I would imagine. The connectors at least are harder to mess up. That's a worry for later either way, since I don't have a soldering iron right now anyway, or the money to spend on one. I'll probably do a bit of research on my own into the different drive connector types, IDE/SATA/SAS/FC etc. I mean, the hardware has to be very different on the boards I'd assume since they're going to buffer and read/write and cache in different ways. But that's an assumption on my part. Not terribly relevant to my dissertation, but something I'd like to know more about anyway.
Not the greatest of photos that, you can't get any sense of depth. But that's the metal casing removed from it, exposing the platters and the heads. There were I think 3 platters in the drive (going from memory here since I've put it back together now). For the record, there are 3 screws on the disk on top of the platters there that hold the platters in. Those screws were pretty tight, and bloody hard to get any purchase on because every time I tried to turn the screwdriver, it just spun the platters. In the end I had to physically press on that metal disc to get it to keep still, which put a bit too much pressure on it. There are metal discs like that one at the top between each platter, since the heads need to fit between the platters. The noise of the platters spinning while the metal disc grinded into the top two platters was absolutely horrific. Anyway. There's a small circuit board-esque thing (technical, I know) at the top left of the drive there. That has pins on the back through the metal to the main circuit board, which looks to be the only way the drive actually communicates. I'd imagine that has to be a fairly common point of failure for drives really, particularly with the amount of movement that the drive heads are gonna see, and they're connected to that component by one of those awful thin plastic-y cables. Although to be fair, those cables are generally bloody resilient in my experience, so who knows.
That's the back of the main circuit board. As you can see from that, it's intricate and soldering is going to be pretty delicate work. The row of connections down the left side is where the connectors from that component connected to the drive heads connects. My god, I used too many variants of 'connect' in that sentence. Someone get me a thesaurus. But anyway, the board is just screwed down to the drive chassis. The only other point of connection between the circuit and the actual drive is those four parallel lines at the top left. No idea what they do, I would have to assume it's something to do with power, but I have no clue at all. It's actually quite surprising how uncomplex a hard drive is, at a hardware level anyway. I mean, the electronics behind it are really quite bloody impressive when you consider what this lump of metal actually is, but there's not an awful lot going on.
So this is the internals again. I took the top platter off - those two metal rings above it are the blocks that go between the platters to keep them apart. I apologise for the disgustingness of that top platter, I ended up having to put my fingers on it to unscrew it, and as with all these things, the second you touch them they get horrendously dirty, even if your hands are reasonably clean. The head itself has a single screw on it, on the right side of it (you can't see it in the photo). I unscrewed it and it did sod all; I have no idea how to disassemble that part of the chassis without trying to force it. Clearly there is a way, but a lot of it uses those awful metal studs which are entirely not designed to ever be removed, so I don't know if a random person can actually do much to that. As it stands though, the head obviously only moves as far as the platters go, which means it's impossible to get the 2nd platter out the drive without snapping the heads off. Not that I would really imagine there's anything interesting below there. I'd quite like to see what's at the base of it, if there's any electronics underneath the platters. The one thing that I'm curious about that I've not identified is just what controls the motor for the drive, but that's going to be one of the components on the main circuit board, and I don't know anything about that kind of thing nowadays - last time I played about with components, it was just resistors, capacitors, gates and LEDs.
So, that's a hard drive! Quite interesting really, or at least in my opinion. You may disagree, but you're wrong to do so. One of the main reasons I did this was because I wanted to identify, or I should say try to identify, hardware points on the drive that look like they'd be most at risk if the drive were to fail. From this, the only two obvious ones would be the heads (because they're bloody delicate), and that component by the heads. Of course, if the drive isn't obviously physically damaged, these are two components that are next to impossible to replace, so it'd be hard to really know what to do there anyway. One thing that I am very curious about now though is physical damage by stamping on a drive. The metal plate at the top is fairly thin, so if you bent that in you'd probably destroy the top platter and the heads, but the lower platters would be absolutely fine, and you'd have recoverable data to a degree. Additionally, if you stamp on a drive/throw it about/whatever, cause physical damage that only affects the main circuit board - that's clearly incredibly easy to replace, so I'd not be surprised at all to find that in quite a few cases, a drive chucked out a window or whacked a couple of times with a hammer will be fully recoverable. Partly because I'd assume a lot of people who want to destroy it will not understand the workings of a hard drive, they won't think to open it and wreck the heads and the platters; they'll see a circuit board on the back of it, and try to wreck that instead. Although it obviously depends on the severity of the damage to whether anything is recoverable anyway, but I'm quite looking forward to experimenting with this.
I spoke to the Systems Support guys at my uni, and they've told me they'll try to pull a stack of drives together for me, since they refurbish at least one lab every summer and they're not allowed to just throw out the drives anyway, so apparently they have boxes of old ones just sitting there anyway. They've told me that they'll have to run DBAN or whatever on them themselves since they might have sensitive information on them, which saves me from doing it myself anyway (I had planned on fully wiping every drive I use anyway, then a clean install of an OS and some files and applications, just so I know precisely what's on the disk and what I can expect to recover). So I'm hoping to have at least 15-20 drives to play with, probably 8 or 10 of which will get physically destroyed.
Good times.
Lostprophets - Last Train Home