deskthority

I thought I'd create a new thread for my design project instead of cluttering up the big ergo thread with more posts.
I'd like to give it the codename "Rivertron", as a nod to "Maltron" (and to Firefly ).

Goals:

• Create a keyboard design based on several ergonomic variables, and not a single one at the expense of others.
• It should suit touch typists and non-touch typists alike. You should not have to learn a new layout or typing style to use it.
• Build at least one keyboard (for myself), but offer kits to others to keep costs down.
• Create a solid design for a keyboard that could be mass-produced for profit in the future, if I got the means to.

Design points (fixed):

• Split into two separate keyboards: one for each hand.
• All modifiers on thumb-keys.
• Big control keys (Enter, Backspace, Tab, Delete) are on the innermost operated by the index finger.
• Each half-keyboard is adjustable in wrist angle (side to side) and slope (forward/backwards).
• Built-in wrist rest

Construction details (not fixed):
Switch: Cherry MX frame-mounted, with diode.
Switch frame: 3D design cast in polyurethane resin, epoxy/fiberglass or some other alternative.
Circuit: Wired and soldered by hand, if I can not find a better method.
Controller: One Teensy 2.0 in each half. Firmware not decided.

Status
2011-08-20: Second mechanical key well layout prototype

Cool, love to see that prototype.

Are you making the plate yourself? (ref: prototype)

Design ideas
I have spent some time looking at my hands when the fingers are opening and closing a kind of grasping.
When you close the fingers, but not your entire hand then the fingers tips arrange themselves in almost a straight line close to one-another. When the hand is half-open, there is more difference in height, and the finger tips are often more apart.

A problem with most keyboard layouts that I wanted to alleviate somewhat, is that keys closer to the edge are harder to reach. You may be required to lift and move your wrist to reach them. I don't think that making farther keys larger is the way to go. Instead, those keys should be moved closer to the centre by having them lifted and/or angled. Keys in a group intended to be used by the same finger could also be made smaller than others.

I noticed that on a flat QWERTY keyboard, the middle, ring and little fingers are not pointing in the switch vectors, but slightly slanted towards the middle. On the Maltron and Kinesis, these switch vectors and the directions of the fingers are more aligned. However, key wells are quite deep: these keyboards are constructed using more expensive methods than normal flat keyboards and you are practically required to touch-type: the height differences hinder you from using a finger outside its designated column.

I have also looked at keyboards with a layout like the Burboa Type Writing Machine: for instance, the Japanese M-system keyboards, flat Maltron, etc. The layout is flat, but the keys are laid out in a kind of curve, mimicking the position of the finger tips when you lay the hand on a desk with fingers bent 90°.

I therefore decided for a blend of the two approaches: columns offset both in Y direction and height, but not as much of either. The lack of difference in height should be compensated for by more Y-offset, and vice versa.

First iteration:
This design has been scrapped, but I have learned some things.

I had made this mockup by pushing ALPS key caps into modelling clay. I have posted this picture in the ergo thread.


I had intended to use a flat PCB -- identical for both the left and the right hand, to save costs.
The offsets between columns on the PCB was based on a 1/sqrt(2) of the offsets in the Burboa layout and then evened out
I intended to sculpt and cast key caps of different heights myself. (placed on top of low flat Tipro key caps).

I then built a prototype for a right-hand key well from MY switch assemblies (without any actuation mechanisms) and using unmodified Cherry-compatible key caps of different heights.
It did not feel right: there was not enough offset between the columns. It was especially the little finger column that needed to be more offset. Reaching with the index finger towards the middle did also not feel right.
I realized that the columns needed to be more offset from one-another in height and/or in shift.

Another member had also remarked that the keys with the highest caps would probably wobble a lot. When I got a Kinesis keyboard, I noticed that it has four very high key caps that I could test with. I found that a Cherry MX key with a high key cap does not wobble as much as it has friction when the key is pressed off-centre/in the wrong angle.

Dang. I had to scrap the idea of using a flat PCB.
Back to the drawing board ...

Second iteration:
My muscle memory is much too used to the QWERTY keyboard to use the Burboa layout completely. I think it feels weird, especially if you press on the other index column.
I have noticed that if you compare the QWERTY and Burboa layouts, you can see that the columns for the middle finger and little finger have practically the same offset from one-another in both layouts The biggest difference is that the (square) key caps are rotated differently. I have therefore opted to use a middle-ground between QWERTY and Burboa for the X/Y layout. I am also varying the height.

I would have preferred the keys to be staggered for familiarity with older keyboards, but (of course) staggered in a uniform and symmetrically between the hands.
I made this model (again) from pressing key caps into clay, and then I edited the image:
However, sculpting and making my own key caps is not easy. Not only would they be slanted inwards, they would also have to be smaller, or wider apart than normal so as to not rub against one-another.
I think that I could use stock Cherry-compatible key caps if I just rotated the keys a bit so that they are all in the same angle ...
Next: Prototype made from MY switches

Again, I made this key well prototype using Cherry MY switches. The heights and angles, I got by cutting up and measuring the clay model. The MY switches are just glued onto pieces of plastic -- there is no electronics. I wanted to test the feel. It would have been even better to test with clicky or highly tactile switches, but it is just one step on the way.

I am not entirely satisfied with it... A few minor tweaks are needed, but I think that I am headed in the right direction.
This is not a matrix layout. It is staggered!

I have a silly idea that I am going to check next... Would it be feasible it to mount the switches on multiple layers of PCB instead?

It's not staggered, there's no zig zag, just a healthy curve/wave to help the shorter pinky finger.
I like what you did a lot! It's sort of a flatter Kinesis keywell with a wave added, seems like an improvement to me, I'd love to try something like that in practice. Looks very promising!

I'm curious how customized these layouts are for the particular shapes and dimensions of your hands. Do other people report exactly the same overall ergonomic as you do when trying to "type" on these layouts? If they do vary, does it correlate with how similar their hands are to yours, and are certain measures (space between this or that finger, whatever) more significant while others have minimal impact on changing the custom design? No doubt even an imperfect fit with a better ergonomic will still be better than traditional rectilinear keyboard layouts, it might be difficult for you to accurately compare what amounts to being qualitative and highly subjective data.

In short, are you making a keyboard layout which is ergonomically perfect for you alone or for everyone?

I am absolutely making a keyboard for myself. I'm not trying to make a "perfect" layout -- because I don't think that a perfect layout is even possible. I am trying to improve on QWERTY, but without straying too much away from what is familiar.
I have tested only with my own hands, and I do think that my hands are quite average male hands. Once I have got a clicky prototype, I would like to get as many people as possible to test it.

The layout of the keys in the plane were plotted as mathematically right in-between flat "Sholes" (row-staggering) and "Burboa" (column staggering) layouts.
The keys on the home row to the right of the middle finger are inclined by 5 degrees, because Maltron mentions in their documentation that they adjust for pronation by 15 degrees and I found 7.5 degrees to be a little bit too much with the curved rows. The bottom row has only a very gentle height-curve and heights of the keys on the rows above the home row are adjusted accordingly in the other direction.

ohh sign me up for prototype testing =D

I'd like to be put next on the list for prototype testing

(Edited silly typo)

[Account and posts deleted on request]

I'd be really happy to help with the testing!

Very nice!
Where have you been all of my life?

I meant to post earlier, but I was so surprised and impressed, I've been digesting some of the nice ideas you put into it.