Unfortunately it was not packed appropriately for a 21lb keyboard and got pretty hurt in shipping
Big connector
Caps removed
Under the caps was some material that has completely degraded. When it got tossed around in shipping it got outside of the assembly and all over the case. This stuff is worse than any model f foam I've dealt with. So insanely sticky and hard to get off. I filled a 2L jar with 99% isopronanol and soaked everything I could fit in it for days but even then it needed tons of elbow grease. And obviously most of the parts did not fit
Underside of the assembly. Middle part is one massive block of steel
Yes, that's a 120V transformer. For some reason they decided to run 120v ac through the same cable as everything else. The output powers the bulbs which illuminate the channels for the light sensing.
13 bulbs for 13 channels
The receiver circuit (oddly channel 13 (strobe) is not connected)
Lens for focusing the light onto the little photocells
My disgusting schematic (circles are the photocells)
Cranks removed. You can see the light channels carved into the metal block
All the cranks laid out in order. Each of the regular keys has a shutter in 4 out of 8 of the main light channels for 4-of-8 encoding. Each of the regular keys also has a shutter in channel 13 presumably for detecting that a key is pressed; however this channel was not connected on mine. The role of the shutters is to block light for that channel while the key is pressed. For example, if you pressed a key with shutters on first, second, fifth, and eighth channels then those channels would not receive light on the other end. You can think of it as the byte 11001001. This benefit of this 4-of-8 design is that you can detect clashes; any byte with more than 4 bits set would imply that more than one key is pressed.
There are also 6 modifier keys intended to be pressed in conjunction with other keys so they have 4 separate channels.
All cleaned up. Each crank hinges on a bar at the top like in a typewriter and the blue leaf springs provide the return force.
Midway through the travel, the crank comes into contact with a second set of leaf springs on the underside. These are held down (up?) by a big magnet running along the length of the keyboard. It takes a certain amount of force to get the leaf spring to separate from the magnet and once it does the attraction rapidly drops off, ensuring that that part of the travel distance is traveled fast. This part of the travel distance coincides with the actuation which is somewhat gradual. If not for this tactility then you would be able to hold the key in a position where the output is floating.
video:
https://i.imgur.com/4cobkcq.mp4
I would love to see a force curve for this keyboard, I imagine it would look something like this:
Regarding restoration I glued the keycaps where necessary with jb weld and am considering polishing them. Here is the worst one: