Katy keyboard (or K80CS (Key80 Contoured Split))

[vvp]

I was asked to create a thread for this project. The description is here:
  http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html
You can comment there or here (if you do not have account at deskthority). I'll try to answer questions if any. Maybe incorporate your ideas if any I like.

[dorkvader]

that case looks super expensive to 3d print. Is that the case?

Is the internal PCB designed to be for the embedded display and various connenctions (modular connector to other hand, etc.) and microcontroller?

[vvp]

The material cost for the case (both sides) is about 7€. But I'm sure that a commercial 3dPrinting service will charge you arm and leg for it.

The internal PCB is designed for LCD and connection to the other side and ports to the keywell. The keywell itself is hand wired. Anyway, the first prototype v.0.6 will not have this PCB yet.

[Oobly]

That looks very nice, but it needs tenting added. Very nice concept and I like the positioning of the pinkie keys.

I would assign the thumb buttons in vertical pairs that can be pressed together, such as Ctrl and Alt or Shift and Fn for easier combinations. The outer thumb keys look like they'd be a bit of a stretch, so I'd use them for quite seldom used keys, but the inner 4 thumb keys all look nicely accessible.

Certainly better than an ErgoDox!

Bravo, good to see some more ergonomic designs being DIY'ed!

EDIT - you could try inverting 2 of the thumb key keycaps for a comfortable thumb cluster. See my avatar pic for how I've done mine.

[vvp]

Tenting of the whole keyboard is expected to be done by 3dPrinting an appropriate stand (into which the keyboard can be put). Alternatively, the top and bottom part of case is screwed together so the screw from the bottom may hold some legs which would tent it too. I did not make it integrated part of the case to save height if I would decide to print keywell cover and move it around. The highest point (including DCS keycap) is 52 mm above table currently. Kinesis Advantage is about 75 mm tall.

As for as the 4 thumb cluster keys, one could achieve similar shape as you have by rotating the bottom keys by 180° and switching the top keys with row2 DCS keycaps (again turned by 180°). Row2 because their height is lower compoared to row4. Of course the whole thumb cluster is not tilted like you have it. But tilting it that way would add height and I'm not sure if all 8 thumb cluster would be still easily reachable afterwards.

Compared to Kinesis the thumb cluster is about 2 mm lower, about 3 mm nearer along X axe, about 4 mm nearer along Y axe. The top 2 keys are 9 mm nearer along X axe (it is more since they are shifted). All the thumb cluster keys are easy to reach with my hand. But I can easily reach all the Kinesis thumb cluster keys too, well except the Alt key. Reaching that one on Kinesis is not that easy but acceptable.

It is also possible to rotate keycaps by 180° horizontally on the bottom keywell row.  Well, it makes sense for the first 3 of them near the thumb cluster. Then they can be pressed by a thumb(*). It is not perfect. The shape is optimized so that thy are pressed by fingers.

(*) But not really by a finger any more.

[Oobly]

That's great! I like the idea to keep it flat and have feet to create the tenting. I think your thumb clusters are fine and the bottom row of the normal areas is also fine to keep for fingers to press. The only improvement I am suggesting is to rotate the middle row of thumb keycaps by 180 degrees and use a different row profile, so it makes a nice contour for the thumb to feel and would differentiate them more by feel, so you have some more positional reference without having to look at the board. This is actually something that can remain as a preference, it doesn't change anything in the design really. A user can do this themselves.

Again, I love the design. I think it's better than all the available Ergo boards and I would love to try it out sometime

[vvp]

Again, I love the design. I think it's better than all the available Ergo boards and I would love to try it out sometime

Cool. Maybe it will get more users than me and a few people I know locally
It will be available in some form. Either I get a company to provide parts or it will be all opened or something in between. Still, I need to completely finish the prototype v0.6 first so that I'm providing something fully working.

[jacobolus]

In my own testing, I've found that I prefer having the finger keys mostly oriented the same direction (that is, not pointed inward toward the center), but with the position adjusted so that they sit along a shape curved to reflect finger motions.

Have you done any testing of that?

I find that the Kinesis in particular really does the switch orientation in a way that I don’t like. The Maltron is better, but still not perfect.

In particular in relation to your design, I’d test orienting the top 3 rows of finger keys the same way, but actually increasing the height (i.e. distance from the table) of the top two rows, so that there’s a ~2–4 mm height difference between the home row and the row above, and another 2–4mm up to the top row. I found this makes it extremely easy to reach keys on further away rows, and very comfortable to press them. [On the other hand, this may not be ideal for all use cases; I’m mainly concerned with typing, so it’s possible someone playing video games or similar would prefer less height stagger]

[tilmann]

This design looks great!

I went back to my ErgoDox when my DataHand started to fail but I'm not too happy with it (tenting helps only so much),  so I've been looking for a curved design hoping that this will increase typing comfort.

I've been thinking about buying a Kinesis Advantage LF but the fact that it's not a split keyboard makes me hesitate to go that route.

Your design looks like just what I need and I would be very happy to test it and contribute towards a working prototype (I have a 3D printer and I can definitely help out with the firmware), do you have any plans to make the case model available even before completing the prototype?

[vvp]

Like on mechanical typewriters? I did not like typing on them. Though I'm sure it was mostly because how stiff their keys are.

I did not try it with my prototypes so far. The different versions I tried modified mostly the innermost and outermost column, the bottom row, column stager and column height difference. The rest was guessed good enough at the first try (best on my testing and testing of few friends).

I like the curving. Pressing the top row is just finishing the unbending movement which starts when changing position from the home row to the top row. So no need to change finger direction (first unbending to move finger forward (and up) and then down press). The muscles pressing down are stronger but switches are so light that it does not bother me.

But you are the second person who thinks it is better so I may try it with some next prototype. Now I want to finish the current shape to a fully working keyboard (including firmware) first. And maybe make few copies for friends to get some feedback about it.

I tried to flip keycaps on the top row to decrease the angle change and I did not like it. Maybe because the height stagger needs to be bigger and the keys should be even more horizontal. But I'm sceptical based on this small test. I think it would be beneficial if the switches would be stiff. It sure looks funny:

[vvp]

I've been thinking about buying a Kinesis Advantage LF but the fact that it's not a split keyboard makes me hesitate to go that route.

I do not mind that much that Kinesis Advantage is not split. I use it for 12 years now. But its firmware is far from what it should be and it has insufficient keys, especially the modifier keys (e.g. layer shift, windows key) on the thumb clusters. The rubber keys do not count. I do not mind them (as some other people seem to) but they are unusable without looking. Only two of them can be pressed reliably without looking: the very first one (Esc), and the very last one (Prog). But the Prog key by itself is not useful and it cannot be remapped. So from the 18 rubber keys only one is useful   Otherwise it is a fine keyboard for me.

Your design looks like just what I need and I would be very happy to test it and contribute towards a working prototype (I have a 3D printer and I can definitely help out with the firmware), do you have any plans to make the case model available even before completing the prototype?

Not yet. I'll post a message here or PM you if it changes.

[vivalarevolución]

Great design.  Yet another great innovation in the ergonomic keyboard sector.  In my opinion. the Maltron is the gold standard in terms of ergonomic design.  It leaves something to be desired in terms of feel, but the position of the keys is near perfect. 

An important part of the Maltron design is making the thumb keys lower than the main key cluster.  Oobly also did a nice variation on the thumb cluster by making it into a small curved thumb bowl.  Making the thumb cluster the same height as the rest of the keys is not ideal.  At least I have developed that opinion from using the Maltron, Kinesis and Ergodox.

[CPTBadAss]

Thanks for sharing. I had the same thought that you did in regards to the thumb cluster being too far away on the Ergodox.

I was wondering how you decided on the locations of the MX switches in regards to the curved keywell. Is that just a copy of the Kinesis? There's another project that I've been following that has been trying to figure out how to make a curve keywell like that.

[vvp]

An important part of the Maltron design is making the thumb keys lower than the main key cluster.  Oobly also did a nice variation on the thumb cluster by making it into a small curved thumb bowl.  Making the thumb cluster the same height as the rest of the keys is not ideal.  At least I have developed that opinion from using the Maltron, Kinesis and Ergodox.

Height difference between K key and the middle of thumb cluster on Kinesis is about 28 mm. Thumb cluster is higher.

On Katy, it is about 18 mm 12 mm. You can make it smaller by almost 3 mm very easily by using row2 keycaps instead of row4 for the thumb cluster. Using some custom keycaps it can be at most 4.5 mm lower. If you would try to make it even lower there would be a problem: when the two bottom right thumb cluster keys would be pressed then their keycap top would be lower than the top surface of the palm rest. Which is not good since you would lose option to press the key with your hand completely resting on the palm plate. This was the design limitation then setting the thumb cluster height. As low as possible but the bottom two keycaps (when pressed) cannot be lower than the palm rest; plus some small reserve (about 1.5 mm). So there is not much space there to move it lower.
Case redesign could make it about another (maybe) 5 mm lower but not more since there would be a problem to fit electronics below the palm rest.

I do not know about Maltron but from the pictures it does not seem to be lower. It seems to be a bit higher or the same height at best. Maybe somebody can provide the number?

Edits in strikethrough and italics.

[vvp]

I was wondering how you decided on the locations of the MX switches in regards to the curved keywell. Is that just a copy of the Kinesis? There's another project that I've been following that has been trying to figure out how to make a curve keywell like that.

No it is not the same as Kinesis, but Kinesis keywell and Ergodox were the initial inspiration.

• I removed some tilting in kinesis keywell which I think does not make sense (e.g. the slight rotation of 7UJM column along y axe).
• I measured my hand in a relaxed position when palm is horizontal and fingers are cuved down. From that I decided on column stager along Y and Z axe.
• I did some estimates (simple trigonometry) how pointing finger may move to the two centre columns and set their tilt along y axe. Their position along Y axe was as low as possible but without forcing thumb cluster father away from keywell.
• Thumb cluster was designed so that it has 8 keys, is vertically staggered so that two keys cannot be pressed by accident when hand is on palm rest and is as near to keywell and as low as possible.
• Outer pinkie column was made lower in the middle so that the movement to ' key is diagonally down and the press of the key can be a bit of continuation of that move. The column is more curved so that the bottom and top keys are not that far away.
• Based on this The first mock-up was generated.
• Each mock-up was 3dPrinted populated with switches/keycaps and discussed with friends. The proposed changes were incorporated to the next mock-up.
• The 6th version of the mock-up was extended to the full case ... and that is the current state.

Here is an example of 5th mock-up (or is it 4th? ... I do not recall).

[vivalarevolución]

An important part of the Maltron design is making the thumb keys lower than the main key cluster.  Oobly also did a nice variation on the thumb cluster by making it into a small curved thumb bowl.  Making the thumb cluster the same height as the rest of the keys is not ideal.  At least I have developed that opinion from using the Maltron, Kinesis and Ergodox.

Height difference between K key and the middle of thumb cluster on Kinesis is about 28 mm. Thumb cluster is higher.

On Katy, it is about 18 mm 12 mm. You can make it smaller by almost 3 mm very easily by using row2 keycaps instead of row4 for the thumb cluster. Using some custom keycaps it can be at most 4.5 mm lower. If you would try to make it even lower there would be a problem: when the two bottom right thumb cluster keys would be pressed then their keycap top would be lower than the top surface of the palm rest. Which is not good since you would lose option to press the key with your hand completely resting on the palm plate. This was the design limitation then setting the thumb cluster height. As low as possible but the bottom two keycaps (when pressed) cannot be lower than the palm rest; plus some small reserve (about 1.5 mm). So there is not much space there to move it lower.
Case redesign could make it about another (maybe) 5 mm lower but not more since there would be a problem to fit electronics below the palm rest.

I do not know about Maltron but from the pictures it does not seem to be lower. It seems to be a bit higher or the same height at best. Maybe somebody can provide the number?

Edits in strikethrough and italics.

Perhaps I should clarify.  When I speak of it being lower than the main key wells, I notice that the large keys in the thumb wells closest to the main key wells are a centimeter or so lower than the B and N keys.  Because the thumb in a resting position is naturally lower than the rest of the fingers, this a comfortable position.  Then for the thumb keys closer towards the middle, the height of the keys increases, making it easier for the thumb to reach the distant keys. 

Now the design is not perfect, for example, the top row of keys on the thumb cluster cannot be reached with slightly moving the hands from the home row.  But I have found the slightly sloping upwards thumb cluster to be the most comfortable thumb cluster on any ergonomic keyboard that I have tried.   The other ergonomic keyboards do not have a thumb cluster with such a design.

[vivalarevolución]

On another note, have you considered modifying the thumb cluster to make it more compatible with the natural curvature of thumb movement, such as the Keyboardio effort, or maybe make little curved thumb bowls, like Oobly's keyboard custom keyboards?

http://geekhack.org/index.php?topic=59722.0

http://geekhack.org/index.php?topic=49721.0

[vvp]

Perhaps I should clarify.  When I speak of it being lower than the main key wells, I notice that the large keys in the thumb wells closest to the main key wells are a centimeter or so lower than the B and N keys.  Because the thumb in a resting position is naturally lower than the rest of the fingers, this a comfortable position.  Then for the thumb keys closer towards the middle, the height of the keys increases, making it easier for the thumb to reach the distant keys. 

Ok, I see now. When I rest my hand palm down and bend fingers more so that they are at the same Y coordinate as end of the thumb then my thumb is about 1 cm lower too. But also near the edge of its movement down (and there is a lot of space for easy movement up). One does not want to get too near to the thumb movement limit when pressing a key. So I would say Maltron thumb cluster may be at the edge or too low for me. But I'm not sure. I would know only after doing a mock-up.
On Katy, the middle thumb cluster (row4) keycap is about 4 mm higher than the middle of the N keycap.

[vvp]

On another note, have you considered modifying the thumb cluster to make it more compatible with the natural curvature of thumb movement, such as the Keyboardio effort, or maybe make little curved thumb bowls, like Oobly's keyboard custom keyboards?

I did not know about Oobly's design before he pointed it out in this thread. So it was not considered. But I probably would not go for it even if I knew. The primary reason being that it has only 4 keys and I do not know how to extend it with more keys and not making the keyboard too high. Oobly's design looks pretty good if you are OK with 4 thumb cluster buttons only and do not mind somewhat higher keyboard.

I considered King's Assembly but I did not like it. I did not make a mock-up but I believe that if the joystick would be well accessible by thumb then (from the 5 thumb buttons) only two from the bottom group would be accessible. Maybe the two top thumb buttons can be modified so that they are pressed from bottom up but then the keyborad would be too high. And it is still only 4 buttons. And it is already too hight anyway.
King's Assembly looks like a bulky bad keyboard (not enough good thumb modifiers) and a bad gaming device too (the small joystick is worse than mouse for FPS or role playing games and it is worse than big joystick for flight simulators).

I considered keyboardio's butterfly keyboard with 4 (or 5 if we count that almost palm button) thumb cluster buttons. I dismissed it because I wanted at least 8. Even kinesis thumb cluster has 6 buttons which is more than on keyboardio. Moreover it looks to me like the keyboard design is done for visual appeal but not for functionality. Why the non-zero angle between pinkie columns and the other columns? It is not like our pinkies naturally stick out ... moreover it is harder to press the top button in the outer column. Why all the empty space around keys? It is missing the bottom row too, which makes sense since it is not contoured and without the bottom row being tilted ... it as well as does not need to be there at all.

Anyway back to thumb cluster on keyboardio butterfly keyboard. The reason not to use it was that it does not have enough keys. But now I realize that the reason may have been wrong. I could add one more row to the radial thumb cluster. The nearer row would be lower so that user cannot press the nearer key by mistake when the pressing of outer key was intended. Maybe worth a mock-up to check it out. Still this thumb cluster would take more space so even if the mock-up test would turn out well I may not go for it. The current thumb cluster on Katy has easily usable from 5 to 6 keys without moving palm when the hand rests on the palm plate (depends how exactly you place the hand). All of them are easily usable when hand is in proper posture above the palm plate. At least with my hand.

[vivalarevolución]

Thanks for the thorough explanations.  What I meant with something similar to the Keyboardio thumb cluster was maintaining a thumb cluster similar to yours with 1x keys, but introducing a slight fan to it.  Imagine if you thumb cluster fans slightly downwards and towards the center, to mimic that natural fanning motion of the thumbs, rather than a boxy shape as most thumb clusters are.  It would expand the gaps between each column on the thumb cluster, but it might be slightly more comfortable.

Also, I just notice how you made the farthest keys on the thumb cluster slightly raised.  Nice touch.

Also, also, based on your design, are any the keys on the bottom row closest to the thumb cluster (the arrow keys in your photos) comfortably accessible with the thumbs?  Sometimes I find those keys can be stealth thumb keys on an ergonomic keyboard; easily accessible with the thumbs, but not exactly on the thumb cluster.

[vvp]

Thanks for the thorough explanations.  What I meant with something similar to the Keyboardio thumb cluster was maintaining a thumb cluster similar to yours with 1x keys, but introducing a slight fan to it.  Imagine if you thumb cluster fans slightly downwards and towards the center, to mimic that natural fanning motion of the thumbs, rather than a boxy shape as most thumb clusters are.  It would expand the gaps between each column on the thumb cluster, but it might be slightly more comfortable.

Ok, I see. That may be done but does not make much difference. The angle difference between keys is not big when there are only 3 of them in a row.

Also, also, based on your design, are any the keys on the bottom row closest to the thumb cluster (the arrow keys in your photos) comfortably accessible with the thumbs?  Sometimes I find those keys can be stealth thumb keys on an ergonomic keyboard; easily accessible with the thumbs, but not exactly on the thumb cluster.

Yes, by turning the kecaps by 180°. Quite usable but not perfect. Look at the picture:

The bottom key in the column 7UJM should be about 3 mm lower for a better result. At least, it looks so without doing a new mock-up. When bottom row is tilted it is easy to use it with fingers (instead of a thumb). But if you use Kinesis Advantage or Maltron you already know that.

[vivalarevolución]

You think this case could be manufactured using CNC, similar to what czarek at Deskthority is doing with his Ergodox case?  Is CNC cheaper and more accessible than 3D printing a case?

http://falbatech.pl/prestashop/index.php?id_product=16&controller=product

http://deskthority.net/vendors-f52/ergodox-pcbs-cases-electronics-and-assembly-services-t6972.html

[jacobolus]

You think this case could be manufactured using CNC, similar to what czarek at Deskthority is doing with his Ergodox case?  Is CNC cheaper and more accessible than 3D printing a case?

To make this precise case you need a 5-axis CNC mill, and those are very expensive. You’ll get a much better quality result cutting a case out of wood or metal than you will from extruded plastic though.

[vivalarevolución]

You think this case could be manufactured using CNC, similar to what czarek at Deskthority is doing with his Ergodox case?  Is CNC cheaper and more accessible than 3D printing a case?

To make this precise case you need a 5-axis CNC mill, and those are very expensive. You’ll get a much better quality result cutting a case out of wood or metal than you will from extruded plastic though.

I was thinking in terms of hiring someone else with the machine to make the case, rather than buying your own machine.  A few people have mentioned that 3D printing this case would cost a few hundred dollars.  But I doubt a machine shop would be interested in doing a one-off case for you.

[Oobly]

You think this case could be manufactured using CNC, similar to what czarek at Deskthority is doing with his Ergodox case?  Is CNC cheaper and more accessible than 3D printing a case?

To make this precise case you need a 5-axis CNC mill, and those are very expensive. You’ll get a much better quality result cutting a case out of wood or metal than you will from extruded plastic though.

Yup, due to the angled cutaways, it would have to be done with a 5-axis. Expensive.

For prototyping it may be better to get one of these (assuming the design can be made in parts that fit the printable area): https://www.kickstarter.com/projects/362246155/littlerp-affordable-flexible-open-3d-resin-printer

Quite affordable, especially if you can source a cheap DLP projector. You could probably make your own fairly easily, actually. All it really is is an accurate single axis stepper and ballscrew, with mounts, table and vat.

Another way to do it is to use parts from a laser printer, but you'd have to design the y-axis deflector carefully to be able to keep the focus.

You could also design it in parts that can be milled on a 3-axis CNC and then joined. Or if you're crazy like me, you can try to build your own 5-axis CNC machine. Then you can build anything for the cost of the materials, power and wear and tear on your mill ends.

These are a few of the reasons I have used a flat plate for the main area of my design. Easier to machine and with staggered columns and contoured keycaps, still very comfortable to use. I don't think the curved "wells" are worth the trouble, especially if you then want to also use PCB's....

[vvp]

The current design has minimum print area of 166 mm x 150 mm x 29 mm.
It is better if the minimum print area is at least 166 mm x 188 mm x 29 mm. Then the bottom part can be printed in one piece. And the top plate too but it would require too much support.

If FFF (i.e. a simple home made RepRap printer) is used then the printer with these minimum requirements would need to have a heated chamber or you would need to try to print in PLA instead of ABS. Otherwise the build area must be bigger since adhesion pads need to be added to corners to avoid part peeling off the build plate. Without a heated chamber the built area should be at least 186 mm x 172 mm x 29 mm.

That is easily within a reach of home made RepRap printers. The RepRap printers will not produce a very precise product. But it will be good enough to build the keyboard. Hole re-drilling and some filing is needed here and there. It will not look perfectly (see the pictures I posted to get an idea) but it is usable. As for as RepRap precision, a Rostock printer here can be off by about 0.2 mm for each surface. Mostly it is within 0.1 mm. Always the surface is offset out (i.e. making the part volume bigger). I believe it is because of extruder calibration and maybe slicer. I may try to improve that.

As for as price from commercial services, it was my guess that it would be a few hundred (probably in the range of 200 - 400). I do not really know. And I do not have any experience with them. I guessed it based on example prices they had on the web.

A commercially printed case would have much better precision (especially since you could use SLA or SLS process). If there is only a limited demand then I can arrange a few cases for somewhere in the range 140 - 200 €. The quality would be as on the pictures I posted. If there is more demand, something else can be arranged.

Anyway it is a bit premature to talk about this since not even the first development prototype is done. I already know I will want some changes to the case which will make it easier to assemble (it will not change positions of the Cherry MS switches).

[vivalarevolución]

Ok, thank you for your responses.  I have to disagree with you Oobly about curved key wells.  I find them absolutely essential on an ergonomic keyboard!  They reduces the distance that a finger has to travel. for the top and bottom rows.  Perhaps I am spoiled by my Maltron.

[vvp]

Status update: http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html#p181178

[yasuo]

the switches glued on plate?

[vvp]

If you want to be sure that the switches cannot be pulled out when you are pulling out a keycap then they should be glued. When being carefull with glue you can pry them out even when glued.
If you think you will not forget to be careful when pulling out keycaps then you can leave them not glued. They will not fell out by themselves or due to gravity only

Here it is discussed more:
http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html#p181273

I do not think that the flimsy cherry mx clips (which are on the switches) can hold the switch securely in mounting plate (even in the metal ones - at least that is an experience with one metal plate I have here (sometimes it holds well sometimes not)). So if the keyboard does not have PCB and you want to be sure the switches hold even when pulling out keycaps then you should use some glue or something else.

[jacobolus]

I do not think that the flimsy cherry mx clips (which are on the switches) can hold the switch securely in mounting plate (even in the metal ones - at least that is an experience with one metal plate I have here (sometimes it holds well sometimes not)).

If you have a 1.5mm thick metal plate, and the switch holes are the right size, then they work fine for holding the switches against quite a bit of force, more than enough to pull off MX-mount keycaps. At least, the simple square cutouts work fine; I’m not so sure about the various funny-shaped cutouts.

[vvp]

You seem to be right. Good point. Should be good enough with new switches and a good metal plate.

I tried it again and it depends on switch and how many times I did pull out the same switch in and out. Probably the clips can wear out a bit on some switches and then it does not hold well enough. Some switches did withstand more pulls than others. I did not bother to release clips with a screwdriver when pulling the switches out. If I would then they should not get any damage. My current situation is that some of my switches hold in my metal plate and some do not.
Yeah and some caps are tighter than others.
Square holes 13.9mm x 13.9mm, 1.2 mm thick plate.

One could use small switch "enabler" PCBs instead of glue, but I personally would not bother. Glue is easier and cheaper too

[vvp]

Status update in the last post of
http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html

Executive summary for the lazy ones: Nothing really important. Left hand side HW is done but not tested. We will probably need to replace controller with something more powerful. That is a bit unfortunate ... unfortunate that it is needed so soon.

[andyboy]

great job! if have group buy in future, i want buy it!

[Tiramisuu]

I love the project but I suspect it will take the better part of a 1000 dollars to build one, once you have developed a spec.   Those highly beneficial curves are expensive.

[vvp]

$1000, may be if you ask somebody to build it for you, I guess he may charge about $500 for work.

If you are doing it yourself and have access to a reprap (maybe you or your friends own one) then it will be around 160€ and a lot of work (significantly more complicated than Ergodox). A group buy can make it lower (especially the PCBs and electronics). If you are willing to use reprap printed keycaps (or have some really cheap source) you can save about 27€

Here is a very approximate breakdown:
case on a reprap: 25
switches: 80*0.5
2 small PCBs: 20 maybe 30
electronic parts: 20 maybe 30
keycaps: 30 (105 ansi keyboard set - not perfect but will do (or about 4 if you accept reprap printed))

If you do not have access to a reprap or you require a prettier SLS printed case then add about 250€ to the price (that was the last estimate I got when I uploaded part of the case to Shapeaways and extrapolated the price).

[yasuo]

very similar to the original maltron would be great with SA

[jeffgran]

This is awesome. Nice work so far. How did you make the 3D printable design? What software did you use? How did you model the curves like the Kinesis? I was thinking of trying something similar but not sure how to go about that so I decided to stick with flat for now, just to keep making forward progress.

[vvp]

The model was done in FreeCAD. It is a bit buggy a rather limited. But I'm not a student so getting a good commercial CAD is too expensive (the options I like cost about $7000 which is too much for me).
http://geekhack.org/index.php?topic=58104.msg1422317#msg1422317
The other free option is OpenSCAD. Looks like OldDataHands uses it for his project.

Modelling curves: I'm not sure what you are asking about. It is rather obvious. The keywell switch sockets were built with a python script:
* build one switch socket
* to build the next switch socket, I defined angle and change from one socket to the next one,
* fused the switch sockets to one column,
* defined position and angle changes between columns
* fused columns to keywell
* draw the rest with the standard FreeCAD tools (no scripting)

I'm sure somebody announced an intention to make a program which would generate the whole keyboard from some definition of the switch locations. If he/she does it, maybe it will be reported somewhere on geekhack.

Just play with some CAD and you will quickly figure out how to use it. It is not a rocket science. Then you will have a keyboard which suits you best

If you can get a commercial CAD for a reasonable price (or do not mind to pay the full price) then get one and just use it. It is not hard to model a keybaord case with good tools.

[dorkvader]

I'm sure somebody announced an intention to make a program which would generate the whole keyboard from some definition of the switch locations. If he/she does it, maybe it will be reported somewhere on geekhack.

ne0phyte's kb layout tool now includes an option to download the dxf for a plate, but it's currently limited to 2D or flat keyboards.

http://ne0.cc/laygenV2/

[vvp]

It was in the context of a contoured keyboard. But maybe it was him and it is only the first stage. I do not recall who it was, nor where it was. I remember it because it is an interesting idea.

[jeffgran]

Thanks, yeah that's what I was wondering. I made my design using OpenJsCAD, and while it was nice at first, as my design got more complicated, it got very cumbersome having to do everything with CSG techniques, and without having access to object-local coordinate systems. You can rotate around a non-global axis using matrix multiplication but it takes like 10 lines of code to do so. So I was just wondering how other people were approaching this problem. I had not heard of FreeCAD, so I will check that out.

Edit: OpenJsCAD, not OpenSCAD... same idea though.

The model was done in FreeCAD. It is a bit buggy a rather limited. But I'm not a student so getting a good commercial CAD is too expensive (the options I like cost about $7000 which is too much for me).
http://geekhack.org/index.php?topic=58104.msg1422317#msg1422317
The other free option is OpenSCAD. Looks like OldDataHands uses it for his project.

Modelling curves: I'm not sure what you are asking about. It is rather obvious. The keywell switch sockets were built with a python script:
* build one switch socket
* to build the next switch socket, I defined angle and change from one socket to the next one,
* fused the switch sockets to one column,
* defined position and angle changes between columns
* fused columns to keywell
* draw the rest with the standard FreeCAD tools (no scripting)

I'm sure somebody announced an intention to make a program which would generate the whole keyboard from some definition of the switch locations. If he/she does it, maybe it will be reported somewhere on geekhack.

Just play with some CAD and you will quickly figure out how to use it. It is not a rocket science. Then you will have a keyboard which suits you best

If you can get a commercial CAD for a reasonable price (or do not mind to pay the full price) then get one and just use it. It is not hard to model a keybaord case with good tools.

[vvp]

If you like the scripting approach then you can script FreeCAD using python.
The python API maps very well to OpenCascade back-end.
Or if you do not like python you can use OpenCascade natively from C++.
The only problem is that OpenCascade misses some features (*) and has bugs here and there.

FreeCAD gets much more usable when you know python, since you can more easily detect why something does not work or work around possible bugs. Of course, it will not help you if the bug is in OpenCascade itself. Personally I think it is better to script OpenCascade from python than just write it in OpenScad. Despite its ocasional bugs, OpenCascade is just more powerful than OpenScad.

(*) The OpenCascade interface is more powerful than the implementation behind it. So form the interface it may look that something works but it is not implemented yet.

[vvp]

Status update: first steps to try another controller. More info at the link below:
http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html#p191612

[vvp]

Prototype v0.6 is working OK.
More information and pictures are here:
http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html#p195210

If you would like to build Katy v0.6 or v0.7 then post a message here or send me a PM. Just an interest check. But it may help to keep me going quicker
I do not recommend to build v0.6. I recommend to wait for v0.7.

[vvp]

Just to let you know that work is still being done on this.
Switch plate changes are done and it is fixed now.

Next is to draw the rest of the case and built it.
If you are interested then a bit more info is here:
http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html#p216090

[kiltannen]

The model was done in FreeCAD. It is a bit buggy a rather limited. But I'm not a student so getting a good commercial CAD is too expensive

Have you looked at: 123D Design  ?

Last time I tried it seemed pretty full featured, but I've never invested enough time to get decent at modelling...

[vvp]

No, I did not know about 123D Design. And I cannot find a feature list quickly. For work like this, one needs a parametric CAD which can replay history properly and preferably has support for assemblies so that you can match parts together easily. Google Sketchup and Blender will not do. FreeCAD would do just fine if it could replay history properly.

I bough Cubify Design, which is a limited version of Geomagic Design, which was called Alibre (or something like that) before.

Edit: To be clear: I started with FreeCAD and later switched to Cubify Design since FreeCAD has problems to replay history. If FreeCAD developers fix the history replay (in the issue tracker, they call the ticket naming) then it will be better than Cubify Design. Definitely more features and freedom than Cubify.

[vvp]

The most important information to build this is available now:
http://deskthority.net/workshop-f7/katy-keyboard-or-k80cs-key80-contoured-split-t8524.html#p243722

[Heliobb]

What you did with the old prototype ?