Total 30.5 hours
July 3
1.5 hours
Set up a preliminary layout. I think Iāll probably do something similar to the mistel barocco; so itāll be 60% ish, tentable, and have a puzzle piece feel. Originally thought of using two separate picoās but that would require two separate bluetooth modules for communication, increasing overall cost. I will most likely use nice!nanoās instead. Also worked a bit on the matrix today:
For space conservation Iām thinking of creating an outward protrusion down and out in a slope to house mcus. Will likely cover this section with wood for wrist support.
July 4
4 hours
Today I created a tentative starter list of materials: Nice!nano x 2 Cherry mx brown switches x 83 PBT OEM keycaps for mx style switches x 83 EC11 Rotary Encoder x 1 (might later look into a keycap set that includes the cap for the encoder) Encoder Cap x 1
Cool thought for later: a magnet in the hollow of the case for ease of use when the two halves of the keeb are connected.
Also! Schematic almost finished! I was a little worried about the difference between using the high frequency vs low frequency pins but there shouldnāt be too much of a difference in terms of firmware and performance.
July 5
3 hours
So due to some incorrect file paths I had to abandon hierarchical sheets :(( I ended up just splitting the two halves of the keeb into two separate kicad folders. It also took a little bit of time but I finalized all footprints. (links here) I also debated using a power switch, hoping to rely on ZMK for low power functions because I didnāt want to get rid of any more keys. But I ended up using one anyway because it seemed like a more direct method for power control.
July 6
3 hours
I got started on pcb finally. I finished the left side first, aligning everything from the dxf file given from kle. Iāve heard a lot of ppl mention the use of ground pours, especially for keeb pcbs. However, since they are heat sinks, itāll probably mess with me when I go to solder my tht components, so I opted to just fully route the whole thing. I ended up actually sacrificing keys for a minimalist design. So sad. I may add acrylic to make the section with mcu visible, so there isnāt any meaningless gap.
July 7
3.5 hours
Right half of keeb pcb finished! Worked on edge cuts so the two halves fit. Started on the cad imported stuff from grabcad and then cried because lol the kailh hot swap sockets were supposed to go on the other freaking side.
So thereās two things I can do atp: flip the mcu, switches, and battery connector to the back copper layer or flip every single switch and diode to the back copper layer. Even though mostly everything is through holes (aka just the diodes), I am resorting to the latter.
July 8
1.5 hour
So I finished flipping the switches and rerouting for the left side. However, I realized when I went to start the cad - again - that the offset for the edge of the pcb is off. If I try to fit the two halves of the keeb together, there would have been no additional room for the case wall thickness which should be around 1.5 - 3 mm in width.
October 8
1 hour
Reviving this since its been too long. Used kle to create an entirely different layout with more empty space needed to fit the mcu. The offset between the edge of the pcb and the end of the keycap is roughly 1.51 mm which should be enough thickness - at least on the āpuzzle pieceā side - for the keeb case thickness. Did about roughly half the traces on the left side.
October 10
2 hours
Finished the pcb for both the left and right sides. Initially I had the mcu for the right side on the back copper layer which I later had to fix by flipping it and rerouting traces. When I went to trace the routes on the other side, the hotswap sockets were too close to the edge cuts so I flipped the switches on the borders upside down. Considering that I plan to use PBT OEM keycaps, this change shouldnāt play into sound or overall functionality of the keeb.
October 11
2 hours
Got started on basic cad. Extruded the top plate for the left side. Since Iām in a time crunch I think Iāll just do a simple sandwich mount. I initially made the side thickness around 3mm, but this was odd-looking as well as flimsy. So the side thickness (except for the center split line) is around 22 mm. Similar to the mistel barocco, I think I plan on having the top frame lean at an angle on the sides.
October 17
4 hours
Just realized I did my sandwich mount wrong -_- Instead of having three separate pieces for the top frame, bottom frame, and plate, I basically merged the plate into the thickness of the keeb and had a flat non-extruded base - which proves for a not structurally sound case all in all. I finished that today, all I have to finish is drafting the edges for a smoother look, adding holes for screws, and making a blender render.
October 24
3 hours
A large issue that I had throughout this entire process, and more so today, was getting edge cuts and case thickness right. Earlier I had cut into the pcb on the sides by 1.510 mm so that the case could wrap around the entire pcb, specifically the āpuzzle pieceā section. So today when I went to put together the two āhalvesā of the keeb, there were a lot of inconsistencies. The m2 screw holes did not line up and the left hand keeb had thicker borders - these issues mainly came up because I had forgot to / had an additional offset of 1.510 mm somewhere in a sketch - so today I was cleaning up the misalignments and such.
October 26
1 hour
LAST BIT OF CLEAN UP THE CAD IS DONE! tidied up and fixed up the appearance - still not sure about the aesthetic but whatever. Later this week I'll work on the blender render!
October 27
1 hour
Finishing touches - added ZMK firmware which included a custom keycap in which a Lily58 keymap was used as a previous template. The current firmware holds both keyboards to be connected and to share a single mcu - however both halves are connected wirelessly, so the keymap would be split between two sets of config files and would be connected wirelessly when handled 'in-person' one microcontroller set to controller the other two a worker