http://geekhack.org/index.php?topic=52241.0
Ok so I recently got an IBM displaywriter keyboard on eBay. Thanks to all the people that helped me get this once it was found. (The listing didn't say "keyboard" anywhere in the title or description, so it may have slipped past some people's radar.)
Pictures were taken with my galaxy s3 in a moderately lit room, so some are blurry, If you point them out, I'll rephotograph them. Full resolution is on my Flickr, link at the end.
2602678
1982 week 11. Most parts are dated from 1981
Lock key down. This key has no external spring, just the "beam spring". Pushing it down sets a mechanism. Pressing either shift raises it back up. The left shift is "tunable" with a screwdriver. (see later on)
Left shift lock key raise tuning system. The plastic acts as a spring you press against. You can set teh height of the plastic spring by adjusting the nut on the bottom and screw on top
Spacebar stabilizer. I really like this one! the wide plate moves forwards when the stabilizer is pressed. Grease from IBM is still on the mechanism. (Grease not shown.)
Controller. Closeups showing all the chip labels are on flickr.
Using Webwit's "unique" beamspring naming scheme, we should call this the "IBM Clare Division Keyboard".
(Image is of the label on the cable) 
OK on to the cool PCB shots!
Notice on this one there are 8 strobes. 4 on each side run the full width of the KB. The need for this becomes apparent when we see the back. The up down alternation is critical to the design. There are 12 sense lines.
Right of PCB. Notice the alternation on the far right function keys.
This is the back, so the far left corresponds to the right of the previous image. Notice they are in parallel! When I first saw this I was at first dumbfounded. That would mean this KB was a low KRO and makes no sense whatsoever. Why did IBM do it this way, when they had a good already established circuit for the rest? Then I flipped it back, noticed the alternation and it all became clear. The alternation is so they can have parallel sense lines for the keys. I do not know how this affects things, but I feel like keeping the sense lines short is a good idea.
Longest sense lines are for the spacebar and keys on that row. They are not paralleled like others. Makes sense to me, though I am only starting to understand IBM capacitive circuits.
Obligatory backlit PCB image. I'll get my roommate to scan the PCB in properly sometime.
Fly plates and associated hardware
Ok that was fun. Feel free to ask questions! Keycaps are cleaning now in a mug on top of the frige.
edit: corrected sense/strobe terminology.
