Geekhack Keyboard Proposal

[Shawn Stanford]

Not easy at all.  Buckling springs you're talking new membranes and the plastic key wells and a new controller for each key layout.  For mechanical switches you would need new PCB layout, controller.

Then you get into the case design.....

I'm actually not too worried about the case design. The case could be based off the existing Model M templates stretched or cut to fit the new key layout. Really, it could be done with a spare case, a Dremel and some Gorilla Glue. It's the controller and cuircuit board I was concerned about. What is complicated about mounting the keys? In my pea brain it just seemed like a matter of laying them into the mounting points and hitting them with some solder. Of course - as I mentioned - I'm an idiot when it comes to this stuff.

Didn't you do some of this when created your Frankenboard?

[msiegel]

I would help if you were Korean. And a dentist. And a little crazy.

XD awesome clip

now if only we had members with CAD qi, a CNC milling machine, and injection molding equipment...

[Rajagra]

Oh come on now, these are innocent play toys!

I blame it on subliminal corruption from the Nickelodeon channel.
NSFW

[Shawn Stanford]

now if only we had members with CAD qi, a CNC milling machine, and injection molding equipment...

One of the reasons I'm not overly worried about the case is because I have a friend who's dean of an engineering school. He has access to the design software and modeling equipment and his brother-in-law has done small-runs of injection molded parts. :becky:

[Shawn Stanford]

Awesome...

Okay, I see two circuit sheets and a rubber isolation sheet behind the plastic keywell frame, all of which bolts to a metal plate. So, the plastics would be the case and the keywell frame. Both of these could be modeled from existing stock modified for the new layout.

Does the rubber sheet somehow aid the action of the board, or could it be replaced with any rubber sheet of appropriate size, shape and thickness?

If the appropriate changes were made to the top of the casing, do you think the back of the casing could be replaced with another metal plate? That should increase weight and might be cheaper to manufacture.

Still, the biggest issue I'm seeing is the circuitry. This stuff looks kinda hard...

[JBert]

Still, the biggest issue I'm seeing is the circuitry. This stuff looks kinda hard...

Hard to make yourself, that's a fact.

It would take some retooling to make those for the proposal board...

[msiegel]

if we had conductive hammers, a flat model f -style pcb might be easier to work with...

[TheSoulhunter]

If others can solder their own CPU, doing a keyboard circuit shouldn't be that hard...

[quadibloc]

If others can solder their own CPU, doing a keyboard circuit shouldn't be that hard...

Who solders a CPU any more? Yes, embedded chips are routinely soldered on automated assembly lines - but the CPU in a home computer on the desktop goes into a zero-insertion force socket, just before you clamp the heatsink down on it...

But your illustration looks like a wirewrap board, which is complicated, but not soldered.

[quadibloc]

Not easy at all.

Well, it might indeed be necessary to give up on buckling springs. And it is true that a new controller would be needed with the kinds of layouts I'm suggesting, where there are novel special shifts.

I see that the wireless SK-7500 keyboard they sell is made by Maxi Switch (as the photos in another thread of Webwit's IBM industrial keyboards remind us, they made IBM's keyboards for a while too); I was going to note that they could use the same controller chip as a membrane keyboard uses on any layout with the same keys. For laptop-style keyboards that shift some keys into a numpad, there are standard controllers available.

But the controller is likely the easy part. So maybe the question should be: what's the second-best switch after buckling spring? Cherry Browns? Cherry Blues? Something by ALPS?

Unicomp does note, though, that the Space Saver keyboard they make - basically a more compact 104-key layout - is programmable at the time of manufacture if you get the PS/2 version. That may not offer much in the ways of capabilities of interest, though.

[timw4mail]

When it comes to Cherry switches, its a lot easier to be flexible. Since Cherry switches can be solder in replacements, a main switch decision isn't of as much importance.

ALPS seem like the worst idea, as the original design is dead, and Fukka seems to be the only good switch, on top of recent controller issues with ALPS switches.

I think most of us will be okay with any Cherry or similar design that isn't linear.

(Maybe even finding out how to get those "fake" and clicky Cherry clears?)

[quadibloc]

While this would require some custom coding at the computer end, I think, because it would be impractical to handle it all by shuffling scan codes around, unless one knew it was to be used with the U.S. layout only...

Code: [Select]


 -------------------------------------------------------
|   | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 0 | { | } | | |
|Tab| Q | W | E | R | T | Y | U | I | O | P | [ | ] | \ |
 -------------------------------------------------------
 |Bk | + | - | * | $ | _ | ! | ( | ) | = | : | " |      |
 |Spc| A | S | D | F | G | H | J | K | L | ; | ' |Enter |
 -------------------------------------------------------
|      | ~ | @ | # | ` | % | ^ | & | < | > | ? |       |
|Shift | Z | X | C | V | B | N | M | , | . | / |Shift  |
 ------------------------------------------------------
  |   |   |   |       |               |A  |1  |1  |    |
  |Ctl|Fn |Alt|       |               |a  |a  |A  |Fn  |
   ----------------------------------------------------


one can get even smaller than the HHKB.

A three-bank keyboard, like on very old typewriters.

But instead of a figures shift, there's only one shift key in the normal position.

There are three keys, marked Aa, 1a, and 1A, that set what the shift key does.

With Aa, you type small letters, and the shift key gives you capitals.
With 1a, you type small letters, and the shift key gives you punctuation marks.
With 1A, you type capital letters, and the shift key gives you punctuation marks.

However, when in 1a mode, the left half of a split spacebar is also useful. That puts you in Aa mode just for the next shifted character you type.

Note that the punctuation keys to the right of the keyboard just shift normally.

You get one control key, one Alt key, but two Fn keys. Note that backspace, on the lopped-off fourth row, replaces Caps Lock.

[1839cc]

(Maybe even finding out how to get those "fake" and clicky Cherry clears?)

Real Cherry clears FTW.

[Xuan]

While this would require some custom coding at the computer end, I think, because it would be impractical to handle it all by shuffling scan codes around, unless one knew it was to be used with the U.S. layout only...

Code: [Select]


 -------------------------------------------------------
|   | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 0 | { | } | | |
|Tab| Q | W | E | R | T | Y | U | I | O | P | [ | ] | \ |
 -------------------------------------------------------
 |Bk | + | - | * | $ | _ | ! | ( | ) | = | : | " |      |
 |Spc| A | S | D | F | G | H | J | K | L | ; | ' |Enter |
 -------------------------------------------------------
|      | ~ | @ | # | ` | % | ^ | & | < | > | ? |       |
|Shift | Z | X | C | V | B | N | M | , | . | / |Shift  |
 ------------------------------------------------------
  |   |   |   |       |               |A  |1  |1  |    |
  |Ctl|Fn |Alt|       |               |a  |a  |A  |Fn  |
   ----------------------------------------------------


one can get even smaller than the HHKB.

A three-bank keyboard, like on very old typewriters.

But instead of a figures shift, there's only one shift key in the normal position.

There are three keys, marked Aa, 1a, and 1A, that set what the shift key does.

With Aa, you type small letters, and the shift key gives you capitals.
With 1a, you type small letters, and the shift key gives you punctuation marks.
With 1A, you type capital letters, and the shift key gives you punctuation marks.

However, when in 1a mode, the left half of a split spacebar is also useful. That puts you in Aa mode just for the next shifted character you type.

Note that the punctuation keys to the right of the keyboard just shift normally.

You get one control key, one Alt key, but two Fn keys. Note that backspace, on the lopped-off fourth row, replaces Caps Lock.

Nice idea. But where's Esc?

I don't like the idea of A1, etc, looks worse than CapsLock.

[rdh]

Nice idea. But where's Esc?

 ctrl + [   perhaps?

[Shawn Stanford]

Well, it might indeed be necessary to give up on buckling springs.

There are already very compact layouts out there that use alternate switches. The appeal of this project to me was the BS keys.

Oh well. Damn you reality!

[TheSoulhunter]

Who solders a CPU any more? Yes, embedded chips are routinely soldered on automated assembly lines - but the CPU in a home computer on the desktop goes into a zero-insertion force socket, just before you clamp the heatsink down on it...

But your illustration looks like a wirewrap board, which is complicated, but not soldered.

http://www.stevechamberlin.com/cpu/2009/05/28/bmow-project-summary/

Soldered, wrapped, clipped, glued... The exact method doesn't matter.
I just wanted to point out that a real geek shouldn't be afraid of anything.
As complicated some stuff seems to be, there are geeks doing things 100x more complicated, its doable!

[lowpoly]

That would be this, right?

Show Image

In the bottom row, the three keys left and right to space should have the same width. Otherwise it is not possible to freely exchange the keycaps. If you use a Meta key to activate a 2nd layer it is really nice if this key can be thumb activated. Meaning it would have to move where the Alt key is right now.

[quadibloc]

I don't like the idea of A1, etc, looks worse than CapsLock.

The idea is that usually, three-bank keyboards had two shifts, Caps and Figs. But that meant that one of those shifts was awkward to reach. Instead, have only one shift, and control which two of the three rails are accessed in each shift state.

Also, I have a split space bar, like the one marked "3" and "2" on the IBM Executive typewriter. ("3" was a space the same size as a digit, so that columns of numbers would line up, "2" was a narrower space, the normal one between words.)

The left side of the space bar is used for the space preceding a capitalized word, so as to trigger shift leading to uppercase for one character. The idea is to make a three-bank keyboard as efficient and convenient as a normal four-bank keyboard.

There are already very compact layouts out there that use alternate switches.

True. There's the HHKB with Topre. But is there a reasonably-priced alternate switch that is almost as good as a buckling spring?

Looking in the Geekhack Wiki, though, I see that some companies made keyboards based on their own form of buckling spring mechanism - but in the one case where one was reviewed here, it was terrible.


Oh, and here's a really advanced keyboard design which indeed would not be suited to buckling springs:



When you push the Num Lock lever to the right, the rows of keys line up vertically.

[quadibloc]

I don't like the idea of A1, etc, looks worse than CapsLock.

The idea is that usually, three-bank keyboards had two shifts, Caps and Figs. But that meant that one of those shifts was awkward to reach. Instead, have only one shift, and control which two of the three rails are accessed in each shift state.

Also, I have a split space bar, like the one marked "3" and "2" on the IBM Executive typewriter. ("3" was a space the same size as a digit, so that columns of numbers would line up, "2" was a narrower space, the normal one between words.)

The left side of the space bar is used for the space preceding a capitalized word, so as to trigger shift leading to uppercase for one character. The idea is to make a three-bank keyboard as efficient and convenient as a normal four-bank keyboard.

[lowpoly]

Here's how Maltron hand-wires their Executive boards:



From this page.

No switch pcb necessary.

[Rajagra]

Oh, and here's a really advanced keyboard design which indeed would not be suited to buckling springs:



When you push the Num Lock lever to the right, the rows of keys line up vertically.

I was thinking of something like that for adjusting the pitch/direction of the stagger on keyboards, for comfort not numpad use. But if you have separate adustment for left and right it would achieve both results.

That wiring makes Tim Tyler's elegant in comparison.

"under construction" - looks like he only has half the wiring in place, it is easy to keep it tidy at that stage. :-)

The actual wires in the Maltron look good. Like gossamer threads of a spider's web.

[Rajagra]

That wiring makes Tim Tyler's elegant in comparison.

"under construction" - looks like he only has half the wiring in place, it is easy to keep it tidy at that stage. :-)

The actual wires in the Maltron look good. Like gossamer threads of a spider's web.

[Xuan]

Oh, and here's a really advanced keyboard design which indeed would not be suited to buckling springs:

That looks fun. The day I found my own keyboard factory, I'll contract you.

[msiegel]

i'd like to give these guys a special commendation for novel re-use of the right shift key

http://geekhack.org/showpost.php?p=95330&postcount=1

[quadibloc]

I'm wondering if "Happy with the array controller" might mean this: someone added a bunch of Cherry MX switches to make a top row of keys for an HHKB. But the picture of the circuit board shows that this is not the case, the keyboard is built as a unit.

Apparently, a keyboard matrix chip was used by CAP Lyon to build his own keyboard or do an extensive modification on an existing one. Note that the keys on the left side have very low contrast standard legends, even if the diagram shows them used as a cursor pad.

[lowpoly]

IIRC, that's a modded Filco using the LIMKB/AIKON controller.

[JBert]

YFYI: I saw that williamjoseph posted a picture of a keyboard which integrates the cursor T into the numpad.

A quick google search turns out that it is a rebranded Steelseries Merc. Here's a better picture of yet another variant:

The only problem with it according to aforementioned member is that the numpad isn't supported in hardware but in software, hence you get problems when the driver fails you. Also, they messed up the order of keys with regard to the usual 6-key cluster.

[quadibloc]

the numpad isn't supported in hardware but in software,

That's terrible, since there doesn't seem to be any excuse for that; those are all standard keys. The keys on the left side indicate the keyboard is a gaming keyboard, which, of course, is a special reason for not leaving anything to software!