v2 Choc RGB variant

[brianlow]

I am working on variant that supports Kailh choc switches, hotswap sockets and per-key RGB. Based on @DaneEvans's Sofle RGB (thanks).

Pull Request:
#104

Kits:
http://Choc.brianlow.com

Welcome feedback and/or collaboration with anyone interested.

TODO

• Verify hardware
  • switches
  • OLED
  • LEDs
  • encoders
  • split
• Firmware
  • basic setup
  • split config
  • keymap
  • LED config
  • LED physical positions
  • cleanup
• PCB v2.1
  • fix traces too close to pads
  • bump version to 2.1
  • better led orientation marks
  • change all markings to silkscreen
  • vertically enlarge encoder cutout on top plate
  • regenerate gerbers
• build guide
  • controller orientation
  • LED orientation, order, soldering tips
  • BOM - no spacers needed
  • top plate is not compatible with original/rgb
  • photos
  • firmware
• publish
  • PR to this repo
  • PR for qmk firmware

[DaneEvans]

I'm happy to do reviews if you need before releasing gerbers

[brianlow]

Update: one half is built and working. Everything is tested except for the encoder and TRRS connector. One trace is a bit too close to a pad and requires careful soldering but otherwise good

[josefadamcik]

Well done, people keep asking about low profile Sofle V2 and this is even RGB. Once you have the final design, feel free to open a PR to this repository and we could talk about merging it. Of course only if you want. There is nothing wrong with having your own version / fork.

[dattanchu]

@brianlow This is fantastic work. From the commit log, it looks like this support both mini-e and choc switch? Is the pcb files ready for physical production?

Can you elaborate on the hard-to-solder spot you mentioned please? My experience is limited to a non-led sofle RGB and would like to try my hands on a choc board for personal use.

[brianlow]

@dattanchu

it looks like this support both mini-e and choc switch?

Yeah!

Is the pcb files ready for physical production?

Short answer is not yet. Long answer: I am feeling pretty good about it. Most of the 2nd half is going. I've got a few keys/leds that probably need some soldering touch ups. I haven't tried connecting the halves. If you are adventurous I suspect it is ok to build. If you'd like I can make a few changes, like fix the hard to solder spot.

My experience is limited to a non-led sofle RGB

You should have no trouble soldering this. I am a hobbist too. The SK6818 mini-e LEDs have been fairly easy. I've only over-heated or mis-soldered a couple leds.

[brianlow]

Changes I was thinking

• inner thumb key, right side has a trace too close to the pad for the socket. I am surprised DRC didn't flag this. Any suggested settings so this problem, and any others, are found automatically?

• top plate: cutout for the encoder is just a 1-2mm too small vertically
• some of the text done with solder mask (?) isn't visible at least on white PCBs, change to silk screen

[DaneEvans]

Did you run DRC? For some reason kicad makes this easy to skip. Check your rules too, but yeah, that should have been picked up I would have thought. Kicads DRC isn't the most intuitive, and a lot of things are turned off by default. Maybe increase the number too. I'm a little surprised that the PCB house ran it. Maybe it's not as bad as it looks on the screen cap. Text is usually either done in silk, copper, or mask over copper, mask alone isn't common, bare FR4 isn't very attractive, and it only works with coloured masks

…

On Wed, 28 Jul 2021, 3:11 pm Brian Low, ***@***.***> wrote: Changes I was thinking - inner thumb key, right side has a trace too close to the pad for the socket. I am surprised DRC didn't flag this. *Any suggested settings so this problem, and any others, are found automatically?* [image: image] <https://user-images.githubusercontent.com/938138/127267031-177e037a-b29e-457d-a906-eabadab0268d.png> - top plate: cutout for the encoder is just a 1-2mm too small vertically - some of the text done with solder mask (?) isn't visible at least on white PCBs, change to silk screen — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#40 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AFIRN4K4ZNGZLILDLSXQU7DTZ6GOZANCNFSM4YHXR6MA> .

[brianlow]

@DaneEvans thanks!

I was just experimenting with Footprint Editor -> select pad -> Properties -> Pad Clearance = 0.30 (default was 0). With this, the DRC finds this problem and a few others.

I'm a little surprised that the PCB house ran it. Maybe it's not as bad as it looks on the screen cap.

Here is a screen cap with measurement:

I ordered from JLCPCB. Maybe this is normal clearance and I was just too sloppy / too hot with soldering and lifted the soldermask or something.

[DaneEvans]

That still sounds like a trace clearance number is too low, a pad clearance of 0 can't be normal, it has to be added to a trace one that's .3 or similar I would think.

I do know I had similar issues where it allowed auto route to do things that seemed completely unreasonable. Thankfully I caught them.

That one also looks to be inside the solder mask cutout!
I think JLC charges more for them to check it, some of the others have a DRC tool on their upload site.

For a board like this, hand soldered by hobbyists, I aim for 2 layers,
1+mm clearance, and more around pads.
Good gaps between diodes etc
And fairly wide traces, at least one to 2 sizes up from the default.
Obviously that depends on density, but there is heaps of space on these