deskthority

When I became interested in alternate keyboard layouts about 3 years ago and researched possible more ergonomic alternatives, I also stumbled upon geekhack and was immediately drawn to the idea of starting my learning process with a new keyboard, which of course had to have mechanical switches. A used cherry keyboard was easy to source here in Germany, and while I practiced learning Neo (an optimized layout, http://www.neo-layout.org) I was lurking the forums and admired what other people put together in their free time.

Finally, when I had realized how much the custom layout - by now I had switched to AdNW (very well optimized, http://www.adnw.de) - was making my day job as software developer more comfortable I decided to also get a better input device, and in lack of any interesting and affordable alternatives was soon tempted to build one myself (did I mention I'm a tinkerer? ) .

Based on ergonomic research and descriptions of mainly japanese keyboards on the internet and especially forums like these, I decided my custom board had to be based on non-staggered matrix-style columns, separated in two halves and equipped with a trackpoint. Also, several thumb-buttons for modifiers and space/backspace should be included - why waste two main fingers for just one key

The software part was the least of my worries, as notable members of GH had already utilized a Teensy for similar projects, and I ended up basing the first versions on some of their developments before rolling my own software. Hardware-wise, old cherry boards as switch-donors were readily available so my quest began...
]
So, without further ado, here are my progressions.

I started with two Numpads and Cherry ML switches. While I found them much better than their reputation around here had suggested, I longed for real MX keys. And the rather thick frame made using the thumbs a little awkward, but having thumbbuttons at all and especially a split keyboard was already a great improvement. Also, an analog joystick as mouse replacement did not work out for me, but by now I had bought some IBM trackpoints from old laptops I intended to use.
A second approach with sawn-off num-blocks from old some cherry keyboards with MX-blacks satisfied my tactile expectations much better, and it also turned out to be most comfortable if I tented the boards rather steeply around 45 degrees.
At this point, the announcement of the Truly Ergonomic had raised my interest - could having just an increased separation between both hands, paired with an angle and their column arrangement result in a equally comfortable typing? Also, carrying around the two parts connected by cable wasn't always as convenient, so I decided to design my own TE. Once again with MLs, it turned out to be quite nice - but I had taken measurements from a laptop keyboard and further shrinked the distance between the keys so I could build the casing in one part. Too small to type on for a longer time with the flat keycaps I had sintered, but a nice integrated trackpoint made it my daily driver for a few months.
If you wondered about the case: It is laser-sintered after a custom design I created in blender, and thanks to working for one of the major manufacturers of laser-sintering machines I was able to test several versions without producing too much garbage, and also could get in-house consulting on design specialities. But with readily available services like Shapeways this technology is a viable alternative for anyone to produce individual parts for custom projects within a reasonable budget and incredible timeframe.

To further leverage some advantages of this technology, curved or bowl-shaped frames were designed: The straight approach wasn't as comfortable as I thought, and when my smoothed design with datahand-like thumb crevice was done I wasn't able to get any small teensy controllers (damn PS3 hackers) - by now I have one again and I intend to finish that promising arrangement sometime, but as my latest design fits me so well I don't know when that will be...
So I went back to a more minimalistic approach once again, and finally also created the missing pieces like a foldable stand, integrated trackpoint and teensy housing. When a recent thread here promised dylon dyes to be a possible colouring option I finally put up all the finishing touches (no, the colours are still WIP, will probably settle for blue with orange accents - and don't get confused about the letters, they don't mean anything, just fit by their shape in the row used...)
Thanks to all the knowledgeable people around GH and deskthority that got me interested in mechanical keyboards in the first place and all the hackers and tinkerers that inspired me to incorporate so many features this is (and hopefully will remain for a while, too) my current setup that I have come to love. The layout is still AdNW, slightly modified for programming and also scrapping the number row, and I never regretted making that switch - it made maybe even more of a difference to getting an ergonomic setup than any hardware mod can probably get you. And the 3rd and 4th layers of Neo are any programmers dream come true, give it a try (their AHK drivers features querty as well)!

Well, enough bragging, just wanted to share these experiences with you, flame me with criticism for this long post or better yet with more ideas and inputs for future projects, thanks for reading this far - Let me know if you have any questions,

suka

Super cool! How do the different forms compare with more regular keyboards?

Are the switches mounted or fastened in any other way than with the small clips?

Brag away! It makes me glad to know there are people doing stuff like this and not just sitting at desks getting bored and miserable. Nicely done.

Minskleip wrote:Super cool! How do the different forms compare with more regular keyboards?

Thanks - I am not sure I understand what you want to know exactly, but in my opinion having the flexibilty of a split keyboard is probably the most important difference as separation and tenting angle can be adjusted to individual preferences. Having thumb buttons is another great enhancement, and I also like the straight columns much better as they lead to a symmetric arrangement. From what I read in the newsgroup of AdNW others also were very content with improvements of the TECK in this regard.

Are the switches mounted or fastened in any other way than with the small clips?

No, the MLs are press-fitted very slightly into a square frame and held by their pins, too, while all my MX designs hold their switches by the sintered frame that functions as a kind of plate-mount. As the plastic structures between key-cutouts are a little flexible in my setup, this holds the keys firmly enough to be able to pull the caps without ripping out a key, but pressing a little stronger from below ejects the switch.

Where have you been all my life?

Do you do requests?

I just had to look this one up:

Today I've refound a lot of very old very slick tenkey-less mods, that made me drool. But this is something I haven't come across at all, very nice you have those tools at your disposition, and the result: Absolutely wicked.

All these cases are made from the same shapeways-like plastic?

off wrote:All these cases are made from the same shapeways-like plastic?

Yes, it is their most inexpensive material "White, Strong & Flexible" or PA2200 sintered on an EOS Formiga P100. All colored parts were first tumble finished with stones and then dyed with fabric dye, with Dylon brands producing great results.

hoggy wrote:Where have you been all my life?

Uhm - I am already engaged

Do you do requests?

Depends - after all these projects I'm not sure my girlfriend is going to be happy if I commit myself to any further keyboard projects in the near future
But if you have anything specific in mind I might be able to assist you - it really isn't that hard at all, once you get going...

Great projects, very interesting!

you actually had it made by them? and that is the stuff you describe as being able to pop switches out of easily, yet also pull caps from without pulling the switch? If so, care to share around how much such board molds cost you?

And now I look again at the datahand/maltron/oven; it seems to me like it might be too tight to fit another switch pointing downwards from the top of that thumbwell, looks like it would interfere with the switch on the top side (one for the fingers).. but very creative&promising nonetheless.

off wrote:you actually had it made by them? and that is the stuff you describe as being able to pop switches out of easily, yet also pull caps from without pulling the switch? If so, care to share around how much such board molds cost you?

No, I work for the manufacturer of the laser-sintering machines they also use, and thus it is the same material and process - I do not know how much it would cost exactly to reproduce, but as you generally are charged by the volume built the flatter models will come out cheaper. If you're interested, I could upload my models and find out their quote.
EDIT: I just uploaded my latest design (the BlueCube one) and it turns out to be around 50€ for all necessary parts, but without any optimization on parts arrangement and volume.

off wrote:And now I look again at the datahand/maltron/oven; it seems to me like it might be too tight to fit another switch pointing downwards from the top of that thumbwell, looks like it would interfere with the switch on the top side (one for the fingers).. but very creative&promising nonetheless.

Very well spotted, one extra point for this keen observation - and you're right, indeed: The fourth switch points backward (and is thus the least desirable to use) and the design mandates custom keycaps and switches pre-wired prior to insertion.

Looking at that last pic, that material really looks nice, mesmerizing patterns in the 'fabric like structure';
big difference between the inside and outside of that shell, is that due to the tumbling or purely the lasering?
I'm guessing the second due to the difference in physical shape of the walls (more geometrical vs natural), but that's a blind guess.
Speaking of which, I suddenly realise that the walls you've utilised are very thin in comparison to for example the 'Humble Hacker' keyboard, this seems like 2 milimeter.. This is getting me very curious to the physical properties, does it flex much or is that, although there, mostly countered by the reinforcing nature of the shape?
Whichever it is, seeing such thin walls migh mean that a Shapeways sollution coud turn out decently 'cheap'; I would very much appreciate it if you'd get a quote on either of your designs, do make sure you don't 'sign over' the design rights or somesuch to them in the process

Just came to me that it really should even be doable 'on the really cheap' seeing how they've stated on their site a price of $1.40/cm3, which, lousy guess, can't really add up to much at all with these designs you're showcasing. Pricewise I mean; they've 'got the looks'.

disclaimer, I don't have the intention of blatantly copying any of your models. Yet.

Suka, I'm really impressed. Your ideas just gears my ones... and it seems like Off's too

@Off: Did you notice that one of the Shapeways offices is located in NL ?

off wrote:... mesmerizing patterns in the 'fabric like structure', big difference between the inside and outside of that shell, is that due to the tumbling or purely the lasering?
I'm guessing the second due to the difference in physical shape of the walls (more geometrical vs natural), but that's a blind guess.

And spot on! As the shapes are sintered in layers of 0.1mm thickness, what you correctly identified are the "stairs" approximating any not completely vertical face, and at the regular square shapes inside this leads to symmetric step sizes, while the natural curves on the outside with their variety of angles are only approximated with irregular widths.

off wrote:.. I suddenly realise that the walls you've utilised are very thin ... this seems like 2 milimeter.. This is getting me very curious to the physical properties, does it flex much or is that, although there, mostly countered by the reinforcing nature of the shape?

A 2mm strip of PA2200 can still be flexed a little by force, but as soon as any other support structures or edges are present it becomes about as stiff as any other plastic case. As for the "plate-mount" grid holding the switches: A base plate of 1.5mm thickness is mechanically enhanced on up- and downside with 2x2mm² bars between all switches. Only this flat part with all cutouts can be flexed with great force and both hands at thge edges - just pressing on an inserted switch in the middle as hard as you can is not enough to make it flex more than a millimeter downwards. Now add alll the switches, the frame and 2 support pillars and you can imagine it does not budge a bit.

off wrote:Whichever it is, seeing such thin walls migh mean that a Shapeways sollution coud turn out decently 'cheap'; I would very much appreciate it if you'd get a quote on either of your designs, do make sure you don't 'sign over' the design rights or somesuch to them in the process

As I mentioned, a whole set of all parts for the blue spit KB comes out at under €50, non-optimized for price at all. But the most promising alternative is to just have the keyswitch plate-mount sintered as desired, with a more regular outside wall and then creating a casing from aluminum profiles or wood or anything else, bringing that price down to probably a third or less as the building height in the machine will go down a lot ( flat inner parts!)

If you like, I'll gladly share the blender sources I used ( or can even send you some throwaway test parts to evaluate them) - any skilled CAD-designer would throw out something like that in a few hours anyways, and I like the thought of giving back to the community...

I find it surprising that you went from bowl shaped (like Maltron and Kinesis) to a hm...do we have a word for this yet?
Let's just call it a ball shape.
So what gave you the idea to do that and how does it feel in comparison.

And of course, *praise praise* awesome work!

You have to get these out to the community bud very very interested.

Just had my two teensy's arrive and it seems like you've been down the path I wanted to go, although I don't think I've got the skill to get anywhere near what you've acheived!

Wow this is great, brag away!!!

Have you posted this over at GH? I'm sure the guys working on the split ergo-dox project would be interested in this...

Thanks for posting this!

The main reason I ended up finalizing the flat halves was that it offers more flexibilty with tenting, compared with the bowl shapes - I still assume that the latter could be even better from an ergonomic standpoint as it minimizes finger movements further, but since the geometry is fixed any design error or wrong assumption on tenting angle or key placement will not be as easily correctable as with my current stand. In addition, as the frames are only about 14mm high production is cheaper, and the final keyboard assembly also less clumsy to carry around on a daily basis.
And after several weeks of usage the flat sides feel as natural as I ever wanted, so I guess the bowl shape might not be that much better after all.
I'll try to clean up my 3D-design files tonight and upload them somewhere for interested parties to have a look, and maybe also mention it on geekhack, too. But since the inception of this great forum my visits there have been rather infrequent, so that might take a while....

Ok cool, it's very impressive!

I've posted the link to this thread here over at GH. It's a similar kind of project, I think that they are currently trying to work out the best way of getting the two halves connected at the moment.

http://geekhack.org/showthread.php?2278 ... o-keyboard

I hope you don't mind me posting the link over there; I'm sure the ergo home brew guys will be really interested in this.

No, don't mind at all - I guess I had been following that thread for a while, too.
And yes, the connection was a tricky decision: I settled for a fixed 10 wire cable as I found out the variation I wanted would be little enough so I would not need to change it later on. For maximum flexibility I will try to get the current 6x4 Matrix hooked up via a shift register so I can use simple USB cables and connectors in an improved version. Or go with USB-3 or even HDMI plugs...

Someone needs to turn this into a group buy

Any chance this could turn into a workshop guide? I'd love to know the steps involved and his to do this myself.

With the upcoming Wiki competition, I might write something up if there is general interest. Then again, the steps themselves are really simple:
- Design your own mounting plate layout with any CAD-software or even Blender: For MX switches, 14x14mm² square holes in 1.5mmn thick plate fits fine. If you are getting funky in all 3 dimensions, you should simulate the keycaps also as not to get lockups on them later on.
- Integrating a case is tricky without having tested your initial design, if confident in your skills do it, otherwise delay it until later.
- Export your design in a format understood by Shapeways or any other 3D-printing service and wait until the order arrives
- In the meantime, order any of the USB-controllers tested by other fellow hackers and test if an existing library already serves your needs - humblehackers was the one I initially used, then rolled my own and by now would probably check out tmk_keyboard, which looks very mature and flexible.
- Hardware-wise, if you know how to solder and have the equipment only diodes and switches from a donor board are needed
- Put together a boring matrix, program, enjoy - see, it is that easy !

Of course if you are no electrical engineering person with a software development background some or even all of the above might become rather hard, but I bet there are plenty of people around here that will gladly assist with advice or even services. And as you can see from my initial descriptions, I didn't go through all these steps in one go either: Using the numpads allowed testing the software first, then adding a trackpoint and other features required some hardware modifications that finally cumulated in the design I use right now.

A group design would make this iterative and time-consuming process obsolete, and certainly cheaper and easier for a broader audience to achieve. Given the last discussions from either the ergodox project or some other custom ideas I stumbled upon in the last months, I think a lot of them have the potential of making a lot of things much better than I have solved them, but it takes time to finalize all the cool ideas floating around the larger the group gets. Maybe I could show some ideas that help others too - I guess the easiest way is putting some of the designs online so you can take a look and see what fits your needs - or should I put together a "Keyboard Prototype Try Bag" that gets send around like some did with different keys? Because before investing time and money in a rather unknown technology I'd want to see the possible reults for myself first, too...

suka wrote:or should I put together a "Keyboard Prototype Try Bag" that gets send around like some did with different keys?

..uh, yes.

and yes to this as well:

suka wrote:With the upcoming Wiki competition, I might write something up if there is general interest.

Also:

How are you going to use a USB cable for the connection? I thought about that but even with shift registers you will need 4 connections for the rows, 2 for power supply and 2 for feeding the shift register. Am I missing something?

Thanks a lot for sharing these.
You're awesome!
I want one (BlueCube)

namespace clash...

Icarium wrote:How are you going to use a USB cable for the connection? I thought about that but even with shift registers you will need 4 connections for the rows, 2 for power supply and 2 for feeding the shift register. Am I missing something?

You're right - I was thinking about an IO-Port expander using the SPI protokol. I ordered a MCP 23S17 lately and will see how that goes - at least that'll be a project that should keep my workspace clean and tidy and not turn into that battlefield of tools it has become recently

Very cool projects suka! I love the look of the "datahand thumb meets maltron bowl" and the other ones too!
You may want to look into I²C for the communication between the 2 halves. ic07 on geekhack have developed the firmware that will be used for the ergoDox with a MCP23018 and from the testing I did it looks quite good!

Getting something done that satisfy a big number of people is hard. Especially in the ergo scene.
Keep up your good work!

Warning: evil tease ahead

Here are some prototypes I have built in the last two years that I am willing to share with interested parties. While some parts will not be usable for anything but informational purposes some goodies I'd like to give away include a completely finished wing case plus cover and a copy of my final split design - these should preferably go to either people that I was able to base my work on or seriously determined folks that actually want to use them.

For the remainder I can setup a Tryout Package that gets re-mailed around this group, or sent individual pieces, preferably to Germany or Europe... Hope this thread is not getting flooded now, please hold back otherwise until details settle

And here is a simple early prototype blender design of my daily driver

When I first saw this I thought it was gonna be another person who will be here one day to show some pics of his awesome stuff and never be seen again but this is some really great info, keep up the good work.

If you are willing to share (perhaps under a CC license) I would love to see the blender files for the data-thumb variant. I can see that being a great leaping point for my split ergo I'm working on.