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 642 times
IMG_3325.jpg
IMG_3325.jpg (866.23 KiB) Viewed 642 times


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

IMG_3337.jpg
IMG_3337.jpg (394.14 KiB) Viewed 642 times
IMG_3327.jpg
IMG_3327.jpg (1.02 MiB) Viewed 642 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 271 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 147 times
IMG_20200527_164003.jpg
IMG_20200527_164003.jpg (3.89 MiB) Viewed 147 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

Yesterday, 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

Post Reply

Return to “Workshop”