Feedback wanted: Rigel96, a mid-ergonomic, split, columnar keyboard

[ncw33]

Hello everyone,

I'm new to geekhack, and indeed, haven't designed a keyboard before. But, I'm hoping for some feedback on my thoughts about a keyboard design.

I'm written a long brain-dump on GitHub here: https://keyboards.nicholaswilson.me.uk/rigel96/

To summarize, firstly, I'm puzzled by the lack of cross-over between the two keyboard communities (mainstream 65% boards vs ergonomic boards).

It looks like there's a gap in the community (or in the market) for a keyboard that's more ergonomic than a standard 65%, but wouldn't feel like a step down or compromise for for the people who are currently enjoying their 65%, worrying about lubing their switches, trying out different plate mounting systems, etc.

Secondly, I have a suggestion for some design goals that try and address that gap:
1. To make an ergonomic board that doesn't re-arrange the standard key assignments, since people seem very attached to that.
2. No compromise on typing feel. Ergonomic, but with the best features of a premium 65%.
3. No trade-offs or decisions bad for ergonomics.

Finally, I have a proposal for a keyboard:

Rigel96, a 96% ergonomic keyboard

Columnar, but with standard key assignments. Basically what Alice should have been.

Split, because that's the most flexible form factor.

So far, I have a CAD design for the bent plate, and I plan of how to manufacture it (laser-cut, then bent with a 3D-printed or steel die in a press).

Questions for feedback:

1. Is this crazy? Am I crazy? Why has this not been done before, as far as I can tell?

2. Any feedback on the design? Is the plate design something that any of you would be interested in actually typing on?

3. I plan to go ahead and make some of these plates, for prototyping. Is there anyone here who'd be interested in joining in, and making/ordering a prototype with me?

[vvp]

Well, I'm curious whether there will be an interest in half standard / half ergonomic keyboard

[fpazos]

Is this crazy? Somehow.
Am I crazy? I don't think so, only a bit naive.
Why has this not been done before, as far as I can tell? Some factors as to why this hasn't been done:

• The ergo community is a niche inside a niche. And people go very strong on their preferences, starting with no one really needs / goes over a 50% layout and most of the users only like the 40% keyboards. I only use 50% orthos that are the biggests ones and I could count the new keyboards of the last two years with the fingers of a hand and I
   would have spare fingers.
• Doing a tenting plate is a manufacturing craziness, also, when someone enters a GB, even a small one. A big factor is having a PCB. A tenting PCB is one of the biggest manufacturing craziness.
• Whoever who is willing to buy a case to do a handwired keyboard is usually willing to design and manufacture its own case or is in such a sort budget that isn't going to join any GB. One of the advantages of the dactyl/manuform is that it is almost the definitive ergonomic keyboard as you can customize it to your likes / needs, that's something worth going for handwiring and despite that you'll also see very few users with one.
• High end orthos are usually crazy expensive, I would love to have buy one but I can't afford it.

Any feedback on the design? Is the plate design something that any of you would be interested in actually typing on?
I'm not interested on any 96% ortho, most of the ortho community neither nor most of the non-ortho community.
Despite that, it is an interesting design? Of course it is.
I don't try to discourage you with this (despite my comments).
If you can, try to build and use an ortho kit to understand how QMK allows you to have very few keys using layers and such things.
I highly recommend you to build your design, it is a very satisfying journey to design and build your own design, don't think at the moment about using CNC or anything complex, just buy a cheap 3D printer (even a second hand one) and just make your keyboard (a good 3d printed design, while not the same as aluminium, can be almost as good and buying a 3D printer is going to be far cheaper than CNC'ing anything). If you have the tools and knowledge to produce an aluminium plate it would be nice (instead of producing the metal tools for it you can also print them, just use 100% infill hahaha). And when you are done with it, you'll have a completely unique and exclusive keyboard.
Just please, don't forget to send the project to kbd.news I would love to see it finished.

[Leslieann]

Fpazos has you pretty covered on a lot of this so I'll tackle the manufacturing.

Why has it not been done, because this would be EXPENSIVE to do right.
Based on others who have tried, printed dies will last for only a few uses. You have the right idea with your method but longevity of your dies and consistency of your bends will be a problem, not just in manufacture but also feel. Steel dies will not be cheap and to really do it right means doing it all in on go, not lots of small bends. But there is a way...


Have you rendered this with switches and caps?
With the way your switches are angled it's very likely that your positions are very wrong (based on a glance at the numbers). Some caps/switches will be too tight (inward bends), others too far apart (outward bends), unlike a normal keyboard you need to design the plate based on the keycap location not the switch location. I'd highly recommend rendering EVERYTHING, screws holes and fasteners, caps, switches, even the tools... and virtually assemble it to make sure it can even be assembled. It's very easy to miss this sort of thing until you have it in hand by which point it's too late and you may have far too much invested.


How much are you planning on spending on this, for yourself or in a group buy?
How much CAD experience do you have? How much manufacturing experience do you have? I don't mean what you've built in your garage, I mean sending a cad file you made out for manufacturing... The less experience you have the more the price (and time required) goes up, FAST. In my experience, this design just doesn't lend itself to being easily prototyped or manufactured in small quantities. While people think we CNC cases because it's "the best", the truth is CNC is one of the cheapest ways to do low volume manufacturing. Manufacturing exists on a sliding scale, you can have very low buy-in/high item cost (like CNC) or very high buy-in/low volume cost (injection molding) and while there is lots in between, you never get low volume low buy-in/low item cost. If you're only making 10, it may be cheaper to just cnc the plate and top frame as one part than trying to do a cnc top frame and stamp/bend the plate. The benefit being you can adjust the spacing and mill a plate for less than making dies over and over till you get it right and then stamping it (stamping like this can be tricky). One consideration to make it easier (and a lot cheaper)  would be to just use 2 or 4 horizontal bends and no various levels in the middle, then it becomes pretty simple for a shop to laser cut/water jet then add the bends and only needs 3-5 pcb per side. It would be a fraction of the cost to make and wire.

Don't expect customers to hand wire if you decide to make more.
Soldering a keyboard is relatively simple in terms of soldering but soldering itself is not that simple for everyone, I've seen some people really just muck things up in ways I didn't even think possible. Hot swap really did open up the deeper aspects of this hobby to a lot more people than you can imagine. While you could sell it as a project, it's going to be one heck of an expensive project, people also hate reading instructions and all it takes is one wrong wire and it no longer complies with your firmware. As for a pcb, you're looking at about 30(?) different pcb's, technically you could order them as 1 or 2 with breakaway parts but you're still looking at quite a bit for those prototypes though nothing abnormal, people just under-estimate this. You also need a way to connect all of them, no matter how you do it it's going to be expensive or very time consuming and one mistake and it will not line up with the plate properly. And don't even consider hot swapping this (without simplification). While doable it would really up the complexity of the case and pcb design, especially if the case is cnc'd.


Realistic ways to do this if you just want one for yourself and maybe pass it on to others who want to do similar...
Honestly, especially on such a small keyboard, a 3d printed plate is plenty stiff and strong if designed properly. Remember, you have several mm between the top of the plate and the pcb, use it to create webbing, look at other 3d printed keyboards (my PF65 and the Dactyl use this method). Use Petg (or T-glase), 3 perimeters @ 30-40% infill, it works/feels/sounds really good. It's a softer plastic that better absorbs vibration and sound and will not compress/reshape itself in time like PLA does. It's also much less harsh on your hands. On the bad side, it doesn't glue or take paint as well as PLA but for me, it's much better. While this should be small enough to do in a single print on a typical home printer, it won't be an easy print due to size/warpage, probably one of the more complicated prints someone can do. Once done though you could still contract it out to be 3d printed in metal (or better done in plastic). It's still not going to be cheap but much more realistic than trying to make the plate the way you want and you can directly transfer your tested STL file right to the metal printer, unlike cnc milling. Better still, if it works, you could post the files for others if they want one with a pretty good estimate on cost for them.

Probably missing bunch more but this should get you thinking.

[Rhienfo]

This is interesting even though it's not for me, I really like how you laid out your thought process there. I find especially interesting how you have decided to optimize the board for the new gmk profile, I like the experimentation there, but if you are doing that, I feel that you shouldn't make protos (or if you are, do cheap 3d printed ones) until the caps come which will delay the project, based on gmk lead times but will lead to a better project.

Love seeing more unique and different projects come to fruition, good luck with the rest of the project.

[ncw33]

Thank you such much everyone for your feedback!

It's really encouraging to hear your interest, even if this an unusual design that you wouldn't use yourselves.

Replying to fpazos:

The ergo community is a niche inside a niche. And people go very strong on their preferences, starting with no one really needs / goes over a 50% layout and most of the users only like the 40% keyboards.

My core idea really is to try and improve on where the Alice currently is aiming. Several companies are making Alice boards, so there must be some demand there - and yet it's just a franken design (in my opinion) that tries to improve ergonomics but doesn't really hit the spot.

Imagine a keyboard you'd give to your mother-in-law, or put on the desk at home for your spouse/partner and children. I feel I can't replace the keyboard at home with a 40% Dactyl!

Replying to fpazos and Leslieann:

I'm quite interested in the idea of using 3D-printing to produce the bending die, to make the bent metal plate. That's not something I'd considered.

I think that with steel dies, getting extremely accurate bends would be possible: at the end of the day, you can empirically measure the k-factor for each bend, and the springback, and just keep tweaking the plate until you achieve accuracy. The dies will then ensure consistency. If you assume an initial k-factor of 0.45, then the actual k-factor is going to be around 0.425-0.475, which gives a variation for the bend allowance of less than 0.1mm. So basically fine-tuning the bend allowance won't buy you much additional accuracy over your process error anyway.

I _have_ rendered this with switches and caps, and designed carefully accordingly. The FreeCAD design is fully-parameterized, with all the parameters (stagger/recess/splay, per column), and there's a big spreadsheet with a ton of trigonometry to get all the positions accurate. The base of the keycaps is 5mm above the plate, and the keycap bases are laid out on a 19.05mm grid, then the plate is offset below that.

The concave bends (such as between Z and A, and Q and 1) work perfectly. The convex bends (such as between Ctrl and left-shift) don't quite work, there will be a gap there. That's because you need to allow both keys to be pressed at once, meaning the keys will travel apart as they come up.

How much are you planning on spending on this, for yourself or in a group buy?

Very good questions! I have fairly good experience of woodwork and metalwork myself (in my shed), but very little experience of sending stuff out to be manufactured. I admit this is a bit of a problem.

I hadn't considered milling the plate at all - because I was planning to build it myself at home, and then move to CNC only if there was lots of interest. As you say, there are big fixed costs (of time and money) sending something off to be manufactured, so if it's just 1 or 5 units I was going to avoid CNC at all, and just hand-build the case and plate.

Don't expect customers to hand wire if you decide to make more.

Yes, agreed! I've made a couple of PCBs by hand (in an acid bath, printing off a mask using an inkjet printer etc). But never sent one off to be made.

Have you seen the amazing PCBs for the Charybdis (Bastardkb)? Very, very neat. He describes in a post how he did it: using a thin PCB, and using experimentation. Offsetting the plate dimensions to get an initial PCB, but then fine-tuning the allowances for each bend by experimentation (sending off for another tweaked PCB...) until the pins lined up exactly with each hole in the bent PCB. Sounds pretty hard work.

So - I think for a Dactyl it's either going to be: handwiring, single-switch PCBs (Amoeba), or something like Bastardkb's designs.

Honestly, especially on such a small keyboard, a 3d printed plate is plenty stiff and strong if designed properly

That's a fascinating idea. I hadn't considered printing the plate either. I've never 3D-printed anything before, and I certainly don't have budget to get one myself, but it looks like there are services where I can send an STL file and have it printed for reasonable cost. My metalwork experience is from >10 years ago, and predates 3D printing and CNC, and was mostly on hand-lathes and mills! I would gladly buy a hand-mill over a 3D printer if I had budget for either. ("Budget" here refers to spouse approval as well as absolute cost!)

I found your post on the PF65, it looks an awesome DIY project!


Thank you again everyone for your thoughts.

[Leslieann]

That's a fascinating idea. I hadn't considered printing the plate either. I've never 3D-printed anything before, and I certainly don't have budget to get one myself, but it looks like there are services where I can send an STL file and have it printed for reasonable cost. My metalwork experience is from >10 years ago, and predates 3D printing and CNC, and was mostly on hand-lathes and mills! I would gladly buy a hand-mill over a 3D printer if I had budget for either. ("Budget" here refers to spouse approval as well as absolute cost!)

I found your post on the PF65, it looks an awesome DIY project!

Thank you again everyone for your thoughts.

You're welcome and thanks.

I trimmed your reply for clarity and less need for multi quotes as it gets messy.

I'm sure  the bends could be done with some experimentation but I'm not sure all of this is going to be very budget friendly by the time you get it right. Depends on your skill, supply of scrap metal and tools I guess.

You can get functional 3d printers for under $300 these days, though those ones are a project in themselves (pretty much all of them are really it's just a matter of degree), you really want to spend a little over double that. However, many libraries offer them up for nearly free + plastic (which is cheap), just beware it can take hours but you could get each half done for just a few dollars and a few hours of your time... If you do that, print it with PLA It's more forgiving), 2 perimeters and like 10%-15% infill and use it mostly for verification and test fitting, not use. This makes it cheap and fast, then you can redo it with more perimeters and infill or send it off for a pro to do it for you knowing it should work as you already have tested it. Beware, print service quotes tend to be 1 or 2 perimeter @10 or 20%, so the bill and time will be higher than the quote. If you go this route the PF65 thread and Thingiverse page probably has a wealth of info you can use.


Couple important things for designing for a 3d printer...
You want to try and design with one side flat and as few overhangs and bridges as possible. Best to think of it like building a brick wall as that's how it prints is layer by layer or brick by brick, if you do this, you won't have problems. They can do overhangs, they can jump gaps (bridges) but some do it better than others so keep it to a minimum where you can, I've seen some do 1/4in long bridges, others defy gravity and print almost horizontal for 8inches, but they too sag if there's no support at the other end, there's ways to go even more extreme but that's getting into some advanced design work there. Stick to the basics. You can also use breakway supports, which the software can generate, it just tends to be a bit ugly there. If you're concerned ask someone on here to take a look at the file and give you some tips. Again the PF65 can offer clues. Drop the files into a slicing program like Cura or Prusaslicer and have a look, Thingiverse also has an STL file viewer. On the slicing program (those break files into the layers and generate machine code), use the default printer and do a preview, you can see tool path and support structures.

Generally,
Pass through holes are oversized by .4 or 0.5mm, you can always ream them with a drill bit.
Brass inserts I undersize by .2mm.
I often use #2, #3 wood screws and 3mm screws (will work for anything up to about 5mm), For threads I tend to make them exact size (for example a 3mm screw = 3mm hole) as the printer over feeds a tad but you also want plastic to fill threads not simply cut. Pre-thread them just enough to hold themself in place then heat it (the screw head!) with a cigarette lighter (it will start to sag) and drive it in, wait for them to cool then undo them and assemble. Works way better than people think. Not as good as a brass insert but in many cases that is overkill and costs more money. Inserts also need to be pushed just a tad below the surface for best results.

Try and leave .4mm gap when things fit together, again, this is not CNC and many printers (especially home/hobby grade)have a bit of skew so leave some room in your design to accommodate. This is not a big deal if everything prints the same way, but often some parts are printed right side up or on their side while others are on the other side or upside down which can multiply the skew.

Oh...
If you go the library route (local schools, hackerspace, makerspaces also work too), don't do the whole thing at once, print a cube with switch hole, a screw hole, threaded hole, and brass insert, just a small part to get and understand your tolerances. Do the same for overhangs and such, could even wrap it all into a single small part and just get some basics down without spending hours waiting on a failed print (not a bad idea for contracting out either). Also do one with 2 switches to test clearances side by side. Take advantage of failures to test other things, strength, feel, how it sands and paints if you want it... This will save you a TON of time, then armed with that info finish your design and come back and do a full test print. I've seen tons of people who just go at it and waste tons of plastic and time. Even with tests, failures, multi versions, etc... I think the PF65 only cost me like $30 in plastic.

Sorry, I know this is a lot but refer back as you reach each step.
Good luck and keep us posted.

[ncw33]

Wow, that's super-detailed help, thank you!

Given that I have limited time (with kids...) I'm probably going to stick with what I know for the prototyping phase, and use wood and metal to start off.

I'll be making wooden dies for pressing the metal keyboard plate. I have decided to allow myself one extravagance: I'll send off to an online service to have the sheet metal laser-cut. I would do it by hand... but making ~100 times 14×14mm holes by hand doesn't particularly thrill me. Last time I did something like that, it took ages just to scribe them out on the metal, then the cutting with a jigsaw takes time, then hand filing and finishing with a dremel takes yet more time. And I still struggle to consistently get the holes square within ±0.1mm tolerance. Alas for no shed room for a milling machine, or I'd be able to get them dead square and accurate! Still time-consuming though to scribe and cut them all individually.

So, I'll pay to have the plate CNC'd with a laser. My experiments so far show that my plywood bending dies will be more than strong enough to do the bending. The case can be just hand-cut from wood - to me that still seems simpler (and cheaper) than using a CNC machine to 3D print it from plastic.

It'll take me a long time to work on it in stages (maybe an hour or two a week), but I'll report back when I have some progress.