Another single-switch circuit board

Reinhold

08 Apr 2020, 08:25

mtl wrote:
07 Apr 2020, 04:25
  • To make use of the Kaihl switch sockets, one really needs to use the screw mounts on the left and right sides to stabilize the PCB when a switch is inserted or removed. I don't think I've seen anyone use them before with the 1.x series.
  • A 2u version needs to be created to support stabilized keys, as the screw mounts on the 1u would interfere with the switch plate stabilizer cutouts on 2u (non-space bar) stabilized keys.
On a regular, flat keyboard this could be an issue, yes.
But where the Amoebas really shine, and where I've seen multiple people use them, is in the ergonomic keyboard community, where the keys are not flat next to each other. This means that full-keyboard PCBs are impossible, but these single-switch PCBs are perfect for the scenario. In most of these builds I see people glueing the PCBs to the case and not use the screws, because the cases are 3D-printed, and it's near impossible to 3D print a case in a fidelity that supports those small screws.
I think the best and most well-known example of such a build is the Dactyl Manuform.

I guess what I'm trying to say is, that for non-flat keyboards (where the Amoeba really shines IMHO), the stability isn't an issue, because people tend to glue the PCB or the switch anyway.

mtl

09 Apr 2020, 23:17

Ah, that makes sense. Thank you for that context. I'll put together the files for release, try to call attention to the milling issue, and create the 2u version at a later date (as time permits).

For now, here are a few pictures of prototypes that have been made. The first pictures show the current design, which suffer from the milling error and can't fit the LED without filing away at the PCB:

IMG_3322.jpg
IMG_3322.jpg (1.15 MiB) Viewed 2638 times
IMG_3325.jpg
IMG_3325.jpg (866.23 KiB) Viewed 2638 times


Here are a couple of shots of an earlier prototype with the LED:

IMG_3337.jpg
IMG_3337.jpg (394.14 KiB) Viewed 2638 times
IMG_3327.jpg
IMG_3327.jpg (1.02 MiB) Viewed 2638 times

pcaro

10 Apr 2020, 09:53

Great version! Waiting next version to build an ergonomic keyboard!
Last edited by pcaro on 10 Apr 2020, 10:36, edited 1 time in total.

Reinhold

10 Apr 2020, 10:01

That looks cool!

The milling issue for the LED is unfortunate. If I understand correctly, they use a too big of a mill head, which results in the corners being too rounded for the LED to fit in? Is it possible to make the cutout longer, such that it's not completely square, but instead extends a bit to the top and bottom, allowing for rounder corners? I don't know if that would make it impossible to fit the circuit in though.
Also, do we know if any other manufacturers supports finer mills, such as JLCPC?

Also, you don't have to be in a hurry for my sake, I won't have time to play around with them for the next couple of months anyway. I would much rather wait a few month if that was what it took to figure out how to fix the LED issue.

mtl

10 Apr 2020, 15:20

It may be possible to dogbone the corners. Will give it an attempt. (PCB space is at a premium.)

I don't have experience with other manufacturers, but hopefully someone can offer insights. I can also produce panelized gerbers if someone can offer guidelines (optimal panel size, ability to do V scoring/grooves, etc.). These parameters would possibly vary by manufacturer.

wolfwood

10 Apr 2020, 23:17

I'm very excited by this as well. I'm handwiring a Dactyl with the Kailh hotswap sockets and this would be a dramatic improvement.

mtl

11 Apr 2020, 22:46

I've shared the project on OSH Park here, and posted the gerber files here. :) I'll clean up the sources and post those soon as well.

Regarding OSH Park's milling issue, I looked at what would need to be done to mitigate and it is a significant enough change that it would warrant a version 2.1, if the idea works at all. Basically, it sort of requires the LED to be rotated 90 degrees, which will necessitate re-routing many of the traces.

For now, I would either work with OSH Park to ensure they use the finer mill bit (as they did with my first prototype run) or go with a different fab that can ensure precise cut-outs.

SuperAmi

13 May 2020, 14:08

I was looking to build a new Dactyl style keyboard, but I wanted my next keyboard to support both hot-swap and RGB. Mostly I want to more easily test drive different key switches and make use of the RGB's to better indicate active layers. I started work on my own layout, when I stumbled on this.

I must say it looks awesome! I can't wait to get my hands on some and would really appreciate a 2U version. I've done some work before with Eagle, but haven't tried KiCad yet. I may take a more concentrated crack at it soon, as I was planning to work on a small interface board to hold the connector for between the boards and a socket for the controller sub-board. So, maybe in a few weeks I can take a crack at the 2U version if you haven't gotten to it by then.

Razerban

16 May 2020, 23:49

The version 2 looks awesome!
Did you have time to check with OSH Park about the miling issue ?
I've been a long time JLCPCB customer, and I never got a problem with the PCB I ordered.
Has anyone ordered the v2 from JLCPCB and share his feedback ? If no one did, I can place an order later today.
Also I don't know if you had time to work in the v2.1 but it could be interesting to place an order for both versions to save time.

SuperAmi

18 May 2020, 17:14

So after taking a good look at the Amoeba Royale and doing some 3D prints to see how it would fit in my planned dactyl keyboard, I decided I could use some changes. It started, because I needed a little bit of space around the corners to allow it to fit into a 3D print. Then I went down the rabbit hole. I'm calling it the King Amoeba, since that's my last name and still a fitting homage to the inspiration. The changes are pretty substantial though, with only the hotswap socket kept from the original.

  • Cut back corners to allow at least 1mm edge, ie width and height at the corners is 17mm instead of ~19mm.
  • The design was optimized for ortholinear, but should work with a standard layout.
  • I went to a 1.27mm / 0.1" pitch on the board interconnects. The plan is to just use standard ribbon cable to wire up the rows.
  • The top has a left and right 5 pin connector, of which either the top 4 or bottom 4 pins are used depending on the direction of the serial bus for the LED's.
  • This same pattern has been repeated on the columns as well as an extra column pad in the opposite corners.
  • I had to drop the through hole diode for space, and I swapped the SMD diode to a SOD-323, because JLPCB has these in stock, so I might be able to avoid soldering those at all.
  • I used the SK6812MINI-E and redrew a footprint based on the datasheet I found.
All in all I think it came out pretty well. I plan on doing a double wide layout next and then hopefully ordering some boards.

I will try to get files into a Git Hub fork when I have time. I will also add my OpenSCAD files that I will be using as a base for the 3D print. I hope to also have a modular variant of that to enable quick prototyping there as well.
amoeba-king.jpg
amoeba-king.jpg (62.63 KiB) Viewed 2267 times

pcaro

19 May 2020, 21:58

SuperAmi wrote:
18 May 2020, 17:14
So after taking a good look at the Amoeba Royale and doing some 3D prints to see how it would fit in my planned dactyl keyboard, I decided I could use some changes. It started, because I needed a little bit of space around the corners to allow it to fit into a 3D print. Then I went down the rabbit hole. I'm calling it the King Amoeba, since that's my last name and still a fitting homage to the inspiration. The changes are pretty substantial though, with only the hotswap socket kept from the original.

  • Cut back corners to allow at least 1mm edge, ie width and height at the corners is 17mm instead of ~19mm.
  • The design was optimized for ortholinear, but should work with a standard layout.
  • I went to a 1.27mm / 0.1" pitch on the board interconnects. The plan is to just use standard ribbon cable to wire up the rows.
  • The top has a left and right 5 pin connector, of which either the top 4 or bottom 4 pins are used depending on the direction of the serial bus for the LED's.
  • This same pattern has been repeated on the columns as well as an extra column pad in the opposite corners.
  • I had to drop the through hole diode for space, and I swapped the SMD diode to a SOD-323, because JLPCB has these in stock, so I might be able to avoid soldering those at all.
  • I used the SK6812MINI-E and redrew a footprint based on the datasheet I found.
All in all I think it came out pretty well. I plan on doing a double wide layout next and then hopefully ordering some boards.

I will try to get files into a Git Hub fork when I have time. I will also add my OpenSCAD files that I will be using as a base for the 3D print. I hope to also have a modular variant of that to enable quick prototyping there as well.
amoeba-king.jpg
Great work!

Please don't forget upload to github and share here your files and photos!!

SuperAmi

27 May 2020, 17:16

So the boards came today,
IMG_20200527_164013.jpg
IMG_20200527_164013.jpg (3.84 MiB) Viewed 2143 times
IMG_20200527_164003.jpg
IMG_20200527_164003.jpg (3.89 MiB) Viewed 2143 times
but upon it looks like I made several mistakes. :? I don't have the rest of the parts yet, so I'm not too sure how well they will work, but the continuity testing looks good.

Here are the issues I found.

I used the SO-323 instead of the bigger SO-123 or SO-123F diode normally used.
JLCPCB also put them into a 5x18 panel instead of 18x5 panel, so the panel is not as useful as I'd like.
The biggest issue seems to be that I believe I put the pads for the LED's upside down. They are still functional, but wiring up the LED's will be a big pain in the behind.
I'm also going to make the cutout for the LED a little bigger, its a bit tight right now.

I'm going to put in an order for a corrected variant 1.1, but if anyone would like some of these, I have five 5x18 panels. They will work well without the LED's and the SMD's can work, but wiring them will be tougher. If anyone is interested in some and willing to pay shipping and maybe some beer money, please let me know.The panels are perfectly spaced for ortholinear layouts.

User avatar
kralcifer

02 Jun 2020, 06:02

mtl wrote:
07 Apr 2020, 04:25
  • To make use of the Kaihl switch sockets, one really needs to use the screw mounts on the left and right sides to stabilize the PCB when a switch is inserted or removed. I don't think I've seen anyone use them before with the 1.x series.
  • A 2u version needs to be created to support stabilized keys, as the screw mounts on the 1u would interfere with the switch plate stabilizer cutouts on 2u (non-space bar) stabilized keys.
Nice work mtl. I'd like to try Royale for a hand wire I'm doing with a "flat board" LK201. I ordered from OSH Park to check it out.

I was thinking that I would go for long rows from a sheet, i.e. break the sheet at the row boundaries. And then screw the row to the top plate every 3rd or 4th switch via riser. I like the idea of less wires for my hand wire. I like hot swap. I don't care about LEDs. The LK201 had these strange plastic blocks for their giant membrane plungers. I'm printing my own with modifications to have cherry mx holes and greater width in the vertical direction to accomodate the Royale but yet allow the plastic blocks to have their feet go into the bottom plate holes. Kind of fun, I hope it works.

Image

plastic block upside down, feet that go into the metal base plate pointing up:

Image

here is orig plastic block side by side with a previous rev of my printed mod:

Image

and here's a rough cut-away of what I'm trying to do:

Image

Razerban

05 Jun 2020, 13:40

@SuperAmi: Did you order the corrected version from JLCPCB ? Have you received them yet ?
Also, did you have time to post the gerbers on Github ?

Thanks for your efforts!

SlientClick

05 Jun 2020, 16:46

Razerban wrote:
05 Jun 2020, 13:40
@SuperAmi: Did you order the corrected version from JLCPCB ? Have you received them yet ?
Also, did you have time to post the gerbers on Github ?

Thanks for your efforts!
This, in the middle of build a tiny tight dactyl prototype right now and this board would be perfect for the next iteration!

Razerban

05 Jun 2020, 17:04

deleted

SuperAmi

09 Jun 2020, 10:17

SlientClick wrote:
05 Jun 2020, 16:46
Razerban wrote:
05 Jun 2020, 13:40
@SuperAmi: Did you order the corrected version from JLCPCB ? Have you received them yet ?
Also, did you have time to post the gerbers on Github ?

Thanks for your efforts!
This, in the middle of build a tiny tight dactyl prototype right now and this board would be perfect for the next iteration!

I got the boards back on Thursday. That makes two revisions, and I'm still waiting on the sockets, LEDs, diodes and other parts from AliExpress. I guess they are more Ali than Express... :?

I was waiting to post until I had a chance to actually verify the design and clean up the files. Its my first KiCAD project and I don't think I have the libraries and other aspects setup cleanly, but if it will help, you are free to try it out. I have 5 16x5 panels and don't need them all. I can send you a panel, but unless you are in Europe the shipping is currently prohibitive due to COVID-19.

The Fork is here: https://github.com/JKing-B16/keyboard-p ... moeba-king

I basically added my work under the amoeba-king folder. I've also added some initial work on an OpenSCAD library for a 3D design to hold the boards. I'm currently working on an ErgoDox-Dactyl layout. I've also started in on a Numpad keyboard as an initial prototype of the 3D design and to practice soldering techniques, but you can find a rough layout for my Dactyl design in the columns.scad file as well. The SCAD is based around a single key module that holds the PCB fairly well. There is a grip version for single boards that grips the bottom of the board as well.
IMG_20200606_231043.jpg
IMG_20200606_231043.jpg (4.39 MiB) Viewed 1807 times
My plan is to use standard ribbon cable to connect up the LED's/rows/columns. This hasn't been as easy as I'd hoped. A lot has to do with the short connections. They transfer heat to easily. Combine that with the stranded cable in ribbon cables, and you get bus wire that sometimes becomes a bus bush. If anyone knows a decent source for solid-core ribbon cable, that would be great.

Switching to a leaded solder allowed me to lower the heat and improved my success rate. The best technique so far was to take 4 wires of ribbon cable, strip them together and insert them carefully into the board, solder them to one board, trim the cable to length, and then pull off the insulation and solder the other side. I think for longer runs, like a dactyl layout, you could get back in with the strippers and actually leave the sheathing on, but for short runs there is not enough meat to grab.

Here is a short preview of soldering. https://youtu.be/Nhtp69AZE4M
Here is an image of the bus wire soldered. I'm still waiting on the solid core wire delivery to finish up the row and LED soldering.
IMG_20200606_231043.jpg
IMG_20200606_231043.jpg (4.39 MiB) Viewed 1807 times
As you can see in the video, when it works, its really nice, but I had to resolder every third try.

I am glad I increased the pad size on the connectors. It makes for easy surface soldering as well.

I'll try to post more, once the rest of the parts arrive in the next 2-3 weeks.
Attachments
IMG_20200609_083804.jpg
IMG_20200609_083804.jpg (4.28 MiB) Viewed 1807 times

Seibar

09 Jun 2020, 10:41

I found this thread after I had already discovered the royale in the github files by accident. I odered 100 pieces which just arrived today. They look good. Once the parts arrive I'm going to make a Dactyl Manuform keyboard with them, wish me luck!

Thank you mtl for your hard work and for making these accessible for everyone to use!
Attachments
20200609_165815.jpg
20200609_165815.jpg (1.9 MiB) Viewed 1802 times
20200609_165809.jpg
20200609_165809.jpg (1.24 MiB) Viewed 1802 times
20200609_165715.jpg
20200609_165715.jpg (1.46 MiB) Viewed 1802 times

SlientClick

11 Jun 2020, 03:58

SuperAmi wrote:
09 Jun 2020, 10:17
SlientClick wrote:
05 Jun 2020, 16:46
Razerban wrote:
05 Jun 2020, 13:40
@SuperAmi: Did you order the corrected version from JLCPCB ? Have you received them yet ?
Also, did you have time to post the gerbers on Github ?

Thanks for your efforts!
This, in the middle of build a tiny tight dactyl prototype right now and this board would be perfect for the next iteration!

I got the boards back on Thursday. That makes two revisions, and I'm still waiting on the sockets, LEDs, diodes and other parts from AliExpress. I guess they are more Ali than Express... :?

I was waiting to post until I had a chance to actually verify the design and clean up the files. Its my first KiCAD project and I don't think I have the libraries and other aspects setup cleanly, but if it will help, you are free to try it out. I have 5 16x5 panels and don't need them all. I can send you a panel, but unless you are in Europe the shipping is currently prohibitive due to COVID-19.

The Fork is here: https://github.com/JKing-B16/keyboard-p ... moeba-king

I basically added my work under the amoeba-king folder. I've also added some initial work on an OpenSCAD library for a 3D design to hold the boards. I'm currently working on an ErgoDox-Dactyl layout. I've also started in on a Numpad keyboard as an initial prototype of the 3D design and to practice soldering techniques, but you can find a rough layout for my Dactyl design in the columns.scad file as well. The SCAD is based around a single key module that holds the PCB fairly well. There is a grip version for single boards that grips the bottom of the board as well.

IMG_20200606_231043.jpg

My plan is to use standard ribbon cable to connect up the LED's/rows/columns. This hasn't been as easy as I'd hoped. A lot has to do with the short connections. They transfer heat to easily. Combine that with the stranded cable in ribbon cables, and you get bus wire that sometimes becomes a bus bush. If anyone knows a decent source for solid-core ribbon cable, that would be great.

Switching to a leaded solder allowed me to lower the heat and improved my success rate. The best technique so far was to take 4 wires of ribbon cable, strip them together and insert them carefully into the board, solder them to one board, trim the cable to length, and then pull off the insulation and solder the other side. I think for longer runs, like a dactyl layout, you could get back in with the strippers and actually leave the sheathing on, but for short runs there is not enough meat to grab.

Here is a short preview of soldering. https://youtu.be/Nhtp69AZE4M
Here is an image of the bus wire soldered. I'm still waiting on the solid core wire delivery to finish up the row and LED soldering.

IMG_20200606_231043.jpg

As you can see in the video, when it works, its really nice, but I had to resolder every third try.

I am glad I increased the pad size on the connectors. It makes for easy surface soldering as well.

I'll try to post more, once the rest of the parts arrive in the next 2-3 weeks.
Finished my hand-wired version just yesterday. Will be watching your results. Thanks for you work on this btw!!!

mtl

21 Jun 2020, 23:06

Wow.. lots of cool projects and ideas, everyone. I didn't know the SK6812MINI-E existed. That seems like the better part to use, as it should be easier to solder and the solder points should hold up longer.

I'm ordering some new prototypes and LEDs to see how well they can be incorporated into the 2.0 design. If it works out, I'll publish the updates. :)

bstiq

07 Jul 2020, 17:03

Hey guys,

I've been working on a similar project for quite some time, and only just discovered the amoeba royale - I've been (re)-creating the same PCB... I feel a bit dumb, but hey.

Disclaimer: I sell those, this is my main source of income. If this is a problem I can delete/modify the post.

So here's the different PCBs I'm making:
- a single-switch PCB (already being sold), with hotswap for MX and Choc low-profile (no RGB). I use this one in my Dactyl Manuforms
- a 2u PCB with stabs and sk6812 mini-e RGB. This has been asked to me by multiple people, I've taken the edge cuts of the amoeba 2u as basis when I found out about it.
- a 6x6 plate with 19.05mm spaced single-switch pcbs, with MX hotswap and sk6812 mini-e RGB

Now the last one is the most interesting, as everything is already connected on the 6x6 board - rows, cols, Vcc, Gnd and Din/Dout for the rgbs per-line.
So in practice it's a 6x6 orthogonal board, which can be "cut" in smaller pieces to make an otho board of any size... all the while having vcc, gnd, the rows and cols, Din/Dout already wired. This is the one I'm most excited for.

Let me know what you think, I can also post some screenshots.

Cheers!
Bstiq

Takeko

22 Jul 2020, 13:14

Also made an account just to follow this thread.
I'm planning my first build that I want as flexible as possible and I was looking for the best option for handwiring + hotswapping + backlight.
First stumbled on the amoeba, then the last rev of bitboard : https://www.reddit.com/r/MechanicalKeyb ... urce=share

Which seemed to be the best single key pcb option but the github is originally in japanese and it's a bit hard for a beginner like me to assimilate the information in this format.

The amoeba royale project seems very similar afaiu.

mtl

23 Jul 2020, 23:07

Bstiq, there is nothing in the Amoeba license to prevent you from selling it or derivatives. I'm happy if the sources and designs are useful and am interested to see your work.

Takeko, yes the bitboard does look very similar in purpose. Thanks for sharing the link!

Small update on the changes to support the SK6812MINI-E. The revised design is good, but the LEDs I received are smaller than the spec sheet indicated. The difference is less than 1 mm and I'm not sure if the manufacturing is off or the spec sheet is off. Still, I may make a small change to the design before posting it as I think the solder pads could be a little bigger.

yoshi

25 Jul 2020, 01:28

I'm jumping in here just to say I'm also really excited for this to come out! I've purchased a whole lot of Amoeba v1.5s from keeb.io for my dactyl build, and have been wondering whether I can get away with using through-hole RGB LEDs, and just hand-wiring the other two leads from each LED.
The goal would be to drive the whole thing with an IS31FL3731. Is there any reason why the same structure intended to support SK6812 minis wouldn't work for control-circuit-less RGB LEDs?

Also, for those considering the SK6812, how can you drive ~60 of these from a USB power line? It seems like 20mA * 60 = 3 A >>> 900mA of even USB-C.
Thanks for making this!

bstiq

27 Jul 2020, 14:49

Hello,

Thanks mtl for the update.
I don't plan on selling the amoebas 1u, as I find the wiring a bit too confusing.
However, my 2us are based on the amoeba 2us.

I received my new boards. They require a few modifications, but they're working as intended.
For the single-switch, I decided to go only with MX (no choc).

Here are pictures of the PCBs: https://imgur.com/a/XNvgzsK
You will notice that the ortho board has holes for chocs as well. But the sk6812 is very difficult to solder on the choc side, and for chocs the spacing is different anyway - so the new version will have only MX. Also re-worked the breakaway tabs, you can see in the pictures those just don't work at all.

All in all I'm very happy with them, and now waiting for the new version from jlcpcb.

Hey yoshi and Takeko,
Here are my 2 cents on a per-key rgb dactyl:
The ISF might work, but you might not need it if you have around 60 keys or less. The wiring will be a hassle. Also consider that to use an ISF you need RGB leds with one pin per R, G, B + common anode or cathode.
This means that for each single-key pcb, you will need 4 wires that go to the ISF.

If you chose to go without ISF, then you need to use SK or WS rgb leds. In this case, you will also need to connect GND, VCC and Din/Dout inbetween each PCB.

Both solutions require a lot of soldering, and imo is not for the faint of heart.
I have wired dozens of Dactyl Manuforms at this point, and even now I'm not considering per-key RGB.

The PCB mentioned on the reddit thread is something akin to what I'm working on, albeit mine will be made for orthogonal boards - but I don't see a reason why you wouldn't be able to break it apart to 1us.

Please, don't hesitate to let me know if you have any questions about the Dactyls, I have some experience building them...

Cheers

Post Reply

Return to “Workshop”