Okay, as suggested by Ripster, I moved all these pics over here to their own article, and even took a few more for good measure. Here is the inside of my PC/AT Model F. I won't bother going through disassembly instructions, as Chloe's guide is more detailed than anything I'd be able to give. The only thing I'd point out is that you don't really need clamps, you can detach and reattach the two plates by pushing one end of the assembly into your floor at an angle, while pushing hard on the other end. Also, Chloe's article mentions something about needing to remove a screw before seperating the plates, but I think this is specific to the XT model.
Here's the keyboard in it's fully assembled state, front and back. In contrast to the verbose Model M labels, all you get is the serial number and the bar-code. (although Greenock-made ones have more detailed labels which sometimes include a manufacturing date) You can also see the feet which are spring loaded and adjusted by the two knobs on either side of the board. A million miles away from the rubber dome Dells in the computer lab in my college whose feet snap off at random, leaving the keyboard lopsided.
]
[/URL]
The inside of the front and back of the casing. Between the two of them, they weigh more than your average rubber dome board. The interior has a rough texture to it, as if it were made of really really fine sandpaper.
A common complaint when opening up the back is that you get a creaky noise and feeling, and many get the impression that it's about to break or snap. This is normal, a hard yank will separate the two sides.
Oh, and don't ask me what that white stain is - it was there when I got it, and I'd rather not think about it :P
The up-side of the assembly. You can see that the metal plate is actually painted a dark olive colour, not black. However, it catches the light and is quite shiny. It's apparent from the where the flash of the camera shines on the board, but you can see the shine between the keycaps when using the keyboard in a well lit room. A particularly good example is
this picture.The underside of the assembly with it's 'Model F' badge. I thought from the 'shop Date' code that my keyboard was made in 1987, but I have seen shop dates from similar boards that look completely meaningless, so I guess it's just some random numbers that only IBM understands. Also note the lack of plastic rivets - the two sides slot together.
The underside of the front plate, which is in a bit of mess with all the spring/hammers either upside-down or turned to their side. There's a layer of (extra crumbly) foam covered with what looks and feels like a thin layer of fake leather attached to the underside. You can also see the infamous stabilizer bar which can only be hooked into the spacebar when the whole thing is disassembled, and the latches which hook the bottom plate to the top. As you can see that from the two missing ones beside the space bar, the barrels are easily removed, in fact, nothing is holding them in place except for the back plate (when it's attached) which means that the whole thing falls apart when you open it up. You also have to be careful when reassembling it to ensure the springs don't fall out and get squashed.
Here's the PCB mounted on the lower plate. It appears to have two layers to it. The PCB is covered in a protective layer which ensures that the hammers never make direct contact with the switches beneath them. Each switch consists of two plates, and when you press a key, it causes a hammer to lie across the two and change their capacitance, which is picked up by the keyboard and registered as a key press. What's interesting is that some of the terminals are unusued, and therefore it could be possible to 'hack' the keyboard into a more standard ANSI or even an ISO layout. (EDIT: 'JohnElliot' has confirmed that this is possible. The unused contacts send out unmapped scancodes that could be mapped to the missing keys...)
A zoom up of the controller board. The large Intel chip is the controller, the black one beside it is a Hex Invertor, and the silver unit to the other side is more than likely the oscillator, which is needed for a capactive keyboard to work (see the Perfect Keyboard Controller thread). Most of the pins on the Berg connector are unused. Only the five topmost and five bottommost are used - for the LED panel and AT cable respectively.
Finally, a picture of what you see when you open the back of the keyboard when it's properly connected together. You can see that the black grounding cable from the controller placed on top of one of the screw holes in the casing, with a similar wire from the AT cable placed on top, and fastened together with a screw. Also of note is that the whole assembly is only secured with two screws (although the bottom of the assembly slots into little tabs that you can see at the bottom of the picture). I'm not sure if the XT's assembly was secured any better, but if I were designing it, I'd have it secured with at least four screws, especially considering that the key assembly alone weighs nearly 2.5KG
Thought I'd throw this in here - When I reassembled it after opening it up for the first time, I decided to check it out before I put the keyboard assembly back into the casing. I hooked up the cable into my laptop and opened a text editor. I press some keys and nothing happened. A second later, the keyboard started sending loads of random keypresses to the computer, causing the text editor to be filled with letters and numbers, and menus on the software to be opened. Figuring that I hadn't assembled it properly, I opened it up, and re did all my earlier work (a rather grueling task considering that when you take the the assembly apart, all the springs and barrels inevitably fall out, requiring you to remove all the keycaps, put all the barrels in, put all the springs in and then later put all the keycaps back in) and put it back together again. Same result. So, I tried again, and then again... The same result each time.
I had pretty much given up on getting it to work again, so I put it up on the desk and went to unplug it, when I noticed that it stopped sending the random keypresses. I typed on it, and it worked perfectly. I put it back on my lap and it started going off again. Suddenly it hit me - it was because I was wearing shorts. When I had it on my lap, the underside of the controller board was resting against my exposed knee, where the skin was causing it to short out and act as if all the keys were being pressed at the same time...
The irony was, that if I had never bothered "testing" to make sure it was ok, I never would have experienced the problem.