Replacing Space Cadet Keycaps?

[Suzuran]

Please excuse my unfamiliarity with keyboard stuff; I never really paid attention to them. I generally used whatever came with the machine unless it was exceptionally irritating, in which case it was replaced with whatever was at hand. I was sent here by recommendation.

I am working on the restoration of a set of Lisp Machines. I am short two keycaps; One is missing, one is broken. Does anyone make one-off keycaps like this that don't cost $Texas? Is 3D printing up to the task of making acceptable keycaps?

I was also warned to tell people ahead of time that I only have two keyboards for three machines, so I have none available for sale. I am actually short one.

[orihalcon]

You might want to post pics of the boards and where the caps are missing from. You should be able to find a Honeywell hall effect keycap with the same profile, then hand paint it. I don't think 3D printing will beat this option in terms of quality and it would certainly cost less. Just my 2¢

[Slom]

So you have one lisp machine for sale, right?

Seriously though, post a picture of what keycaps are missing. I have a bunch of honeywell hall effect keycaps arround, though probably not with the right legends.

How is the broken one damaged?

[matt3o]

I am working on the restoration of a set of Lisp Machines. I am short two keycaps; One is missing, one is broken. Does anyone make one-off keycaps like this that don't cost $Texas? Is 3D printing up to the task of making acceptable keycaps?

if it's okay for you to just have the keycap (without legend) resin 3d printing could be a totally viable solution. I could work on that having a lot of reference material (or better the actual keycaps).

edit: another solution would be to make a mold of a keycap and make a cast (still without legend though)

[abrahamstechnology]

I suppose you can cut out the stems on modern keycaps, then 3D print inserts and epoxy them into the keycaps, the Hall Effect keycap mount seems pretty simple and I have a 3D printer you can use if needed.

[Suzuran]

So you have one lisp machine for sale, right?

I wish; They were not stored well - open to sky for more than 20 years. Lots of rust and mold. If even one winds up being restorable it will be a major achievement. That said, someone once restored a PDP-11/45 that had a literal tree growing through it (it was stored in a dirt floor barn - A seed got underneath it and life, uh, found a way) so there is legitimate hope.

I will take and post pictures when I get home from work.

[XMIT]

3D printing is viable. Some of the newer resin based 3D printers have such fine feature sizes that you can't feel them, ideal for key caps. Of course you'll still need to build a 3D model for your key cap, and, it won't have legends (unless you use resins with dyes or some other approach).

[Darkshado]

How about 3D printing a legend "shot", putting it into a mold and casting the rest of the cap?

[Suzuran]

OK, so here's pictures, hopefully not too big/too many.

This is where the return key goes. The stem is shaped like a plus sign:




Note the lack of the "supporting" dead keyswitch on the right. Neither keyboard has one; I don't know why they didn't put one here. This is why the keycap failed.

This is where the macro key goes. Identical stem to return on the right keyswitch. The left keyswitch is "dead" - It doesn't generate a one or zero in the scan matrix. It exists only to support the key if you press it off center. It doesn't have a normal stem, it has only a round (felt?) pad where the plus-shaped stem would be.




This is the bottom of the broken return key. The round socket the stem goes into broke under sideways load. I don't have the missing piece so I can't just glue it back together. (I checked to see if it got inside the keyboard; it didn't)



Looks like there was intended to be some kind of support on the right there, but neither keyboard has it.

Here's the good set of keycaps from the other keyboard, top and bottom:



One keyboard is tested *mostly* working; All of the switches (that are supposed to) generate scan bits. The real problem with these is that the EPROM-based microcontrollers died of bit rot, and the Lambda keyboard program (which hasn't been dumped) differs from the CADR keyboard program (which has been dumped). I have to write a replacement program from scratch and reprogram them. I temporarily put an Arduino in where the interface sub-board would be to drive the microcontroller pins, and I've been working out the Lambda protocol. I have everything worked out except the special lead-in bytes for three of the four boot chords, which aren't documented anywhere and will have to be solved for by reverse-engineering the console firmware.

[Findecanor]

This is the bottom of the broken return key. The round socket the stem goes into broke under sideways load. I don't have the missing piece so I can't just glue it back together. (I checked to see if it got inside the keyboard; it didn't)

I think that you could repair that broken stem: Pour a small amount of moulding silicone into the good centre mount (or in the left mount on a good key from the same row) to get a small mould. Then insert that into the broken stem and build up the broken wall with epoxy putty.
It does not look as if the keycap's stem would need to enter inside the switch itself so it should not matter if you would fill the gaps around the stem with putty.

"Epoxy putty" is really a two-part modelling clay based on epoxy. I often use "Milliput" brand epoxy putty bought in a plastic model store. There are three grades, and the middle "fine" grade would be best, I think. There are several other brands that I have not tried but been told should be just as good: Magic Sculpt, Apoxie Sculpt and Tamiya's epoxy putty. Varieties from the hardware store, for plumbing or with metal contents that I have tried have cured too fast.

It could be difficult to mix the right proportions for a very small amount of moulding silicone, so I would make a larger batch and cast something else (such as making a mould for the missing key) at the same time.
Easiest to use would be a RTV silicone that mixes proportions 1:1 by volume and weight of components A and B. Do still use a very sensitive scale (1/100 of a g) to get the measurements right!
It would not hurt to make more than one small stem-moulding, just in case one (or some) got too many bubbles to be usable.

[webwit]

Don't know if this is of any use to you, I got mine without case and EPROM, but to counter NOS, and the guy I got it from (former employee) sent me the original controller source code, in LISP of course. I last used LISP in CS college way too long ago, I think to code something to generate fractals, so I forgot most about it and never analyzed the code, but maybe its logic is of use if you plan to program a replacement controller.

viewtopic.php?p=23781#p23781
(Post at 27 Oct 2011, 23:19)

[matt3o]

it takes a spacecadet to wake the duck!

[Darkshado]

Webwit, if possible you should put that source on a public git repo somewhere

[Slom]

https://www.ebay.com/itm/HONEYWELL-MICR ... 2589806048

[Suzuran]

Don't know if this is of any use to you...

That's "ukbd" - it's the CADR keyboard program, and is already available in several places.

The Lambda protocol would be easy to identify because it does not have all-keys-up codes.

[JP!]

I have everything worked out except the special lead-in bytes for three of the four boot chords, which aren't documented anywhere and will have to be solved for by reverse-engineering the console firmware.

What about a brute force approach if not able to reverse engineer?

[Suzuran]

What about a brute force approach if not able to reverse engineer?

It should be doable since there's only one byte to work out, but considering I already emulated the entire rest of the machine picking apart the console again is just a matter of getting the time to do it. (The emulator is how I've verified everything else so far)