Author Topic: converting a Model M to capsense  (Read 1476 times)

0 Members and 1 Guest are viewing this topic.

Offline SpaceGhost

  • Thread Starter
  • Posts: 90
  • Location: SoCal
converting a Model M to capsense
« on: Thu, 02 April 2015, 15:19:33 »
I did an interesting experiment awhile back and I thought it might be an interesting topic for discussion. I put a Model F-style spring & flipper assembly into a upside-down scrapped Model M barrel frame, alongside a Model M-style spring & pivot plate assembly. Then I laid in place a Model M steel plate with blanket and membrane layers in between and temporarily clamped the layers together with a keystem installed for each of the 2 actuator types so I could actuate each switch and compare them.

The Model F-style flipper was not from a Model F, but rather a 3D-printed version from Shapeways that was designed by someone else. It seems like a fairly accurate replica, and I attached a Model F spring to it for the experiment. The results were as I expected; the Model F-style flipper was significantly louder and crisper than the Model M-style pivot plate.

So now I wonder if one could replace the membrane and blanket layers inside a Model M with a thin Model F-style capsense PCB (with no blanket) that could be bent into the required curvature. A capsense controller would be required, and the barrel frame would be loaded with Model F-style flippers. The assembly would be secured using either a bolt mod or screw mod method. The capsense PCB would need to be designed with enough through-holes to accommodate an adequate number of screws, in the same locations as the original plastic rivets.

It may be difficult to design a capsense PCB with a through-hole for each and every original rivet location, but I think how many would be required for long-term reliability would depend on how stiff the capsense PCB is. If thin and flexible enough so as to exert minimal resistance to bending -- and therefore minimum added pullout stress to the screws -- it may be safe to use 50-75% of the original rivet-count, as long as they are evenly distributed across the area of the steel plate.

Of course, another option would be to pre-attach the PCB to the steel plate, bending it into the required curvature in the process. Metal rivets or screws around the perimeter of the PCB could be used, but this would require drilling through-holes around the edges of the steel plate. Not a problem for the determined, I'm sure; bit perhaps a bit difficult for the casual DIY'er. This method would greatly reduce the stress on the screws, so far fewer could be used, making the capsense PCB layout easier to do.

Some means to secure the controller board inside the Model M housing and to provide a cable connection would also need to be implemented.

I think much of the crisper click and feel of the Model F when compared to a Model M has to do with the increased size and mass of the F-style flipper vs. the M-style pivot plate, and also because it smacks down on the capsense PCB with no intervening rubber blanket when actuated. This has been suggested by others, and I agree after doing the simple experiment I described above.
« Last Edit: Thu, 02 April 2015, 15:56:16 by SpaceGhost »

Offline E TwentyNine

  • Posts: 884
    • Some of My Keyboards
Re: converting a Model M to capsense
« Reply #1 on: Thu, 02 April 2015, 15:56:37 »
I think you're basically describing the next phase of the XTant project.   Building a new layout into the XT case was first as it's fewer variables and could be slipped into the existing architecture, but allowing for any layout, including M or SSK which would necessitate more hardware than just a new pcb and controller, would be the logical followup.
Daily driver: SSK or Tenkeyless IBM AT
1984 Model M Industrial Prototype ⌨ 1992 Black Oval Industrial SSK ⌨ 1982 5251 Beam Spring ⌨ 89 Key "SSK" ⌨ M13 triplets

Offline 0100010

  • Posts: 1127
  • Location: DFW, TX, US
  • Not Sure
Re: converting a Model M to capsense
« Reply #2 on: Thu, 02 April 2015, 18:39:15 »
This thread should be relevant to your idea : http://deskthority.net/workshop-f7/ibm-model-f-62-and-f-107-mod-ideas-t9324.html

Hardest part about your idea with the shapeways flippies, is that they would not work with a capsense PCB.
  Quoting me causes a posting error that you need to ignore.

Offline SpaceGhost

  • Thread Starter
  • Posts: 90
  • Location: SoCal
Re: converting a Model M to capsense
« Reply #3 on: Thu, 02 April 2015, 18:59:13 »
This thread should be relevant to your idea : http://deskthority.net/workshop-f7/ibm-model-f-62-and-f-107-mod-ideas-t9324.html

Hardest part about your idea with the shapeways flippies, is that they would not work with a capsense PCB.

Yes, actual IBM flippers would be required, or possibly Shapeways flippers painted with carbon-loaded polymer ink on their undersides, the ink being similar to what is used for printed resistors in PTF substrates (Polymer Thick Film). Resistivity measurements would need to be made on IBM flippers to determine the proper ink, which comes in a range of cured-sheet-resistivity values.

If a manufacturer such as Unicomp were to try this hybrid construction, they could make Model F flippers exactly the way IBM made them, so they'd be sure to work. Just a thought...

Offline SpaceGhost

  • Thread Starter
  • Posts: 90
  • Location: SoCal
Re: converting a Model M to capsense
« Reply #4 on: Thu, 02 April 2015, 19:06:18 »
This thread should be relevant to your idea : http://deskthority.net/workshop-f7/ibm-model-f-62-and-f-107-mod-ideas-t9324.html

Hardest part about your idea with the shapeways flippies, is that they would not work with a capsense PCB.

Yes, it would be best to use actual IBM flippers; although Shapeways flippers might work if painted on their undersides with carbon-loaded polymer ink such as used for printed resistors on PTF (Polymer Thick Film) circuits. The correct sheet-resistivity ink would have to be chosen from the ranges supplied.

Of course, if a company such as Unicomp were to evaluate this idea for production, they would plan to make more of the IBM flippers using the same plastic composition and conductive filler (probably carbon).

Just a thought, although not entirely original I'm sure. It would involve minimal new tooling, as the traditional Model M-style barrel frames could be used along with the standard Model M curved-steel backplates. If the capsense PCB is pre-attached to the steel and thereby locked into the proper curvature then there would be no additional stress introduced, and the plastic-rivet assembly method could be retained, although I prefer a screwed-together construction.

Offline SpaceGhost

  • Thread Starter
  • Posts: 90
  • Location: SoCal
Re: converting a Model M to capsense
« Reply #5 on: Thu, 02 April 2015, 19:08:16 »
Sorry about the double-reply -- I got an error message when I tried to post the first one, so I dutifully re-did it and tried again. Apparently both went through despite the error message...