qmk-keychron-q3-colemak-dh/keyboards/handwired/d48

D48

A Proton C based handwired 48 key keyboard with 2 rotary encoders, I2C OLED, WS2812 strip, buzzer & clock!

• Keyboard Maintainer: Andrew Dunai
• Hardware Supported: Proton C handwired

Make example for this keyboard (after setting up your build environment):

make handwired/d48:default

Details

• Proton C based handwired keyboard
• 2x custom 1.25mm stainless steel plates
• Kailh Choc White (clicky)
• 2x rotary encoders
• 0.91" 128x32 I²C OLED
• Small buzzer mounted inside (still waiting for the AST1109MLTRQ boys)
• WS2812 strip (14 LEDs)
• DS1307 I²C module real-time clock

Build process: album

Pinout

Challenges

I'm very happy with the result, but at some point Proton C was driving me nuts.

I did a lot of trial and error during assembly & programming. There were a lot of yet undocumented caveats, so I'll outline them here so that you guys can avoid the same issues I had.

Matrix & encoders

Although this is a 48-key board with a 12-col & 4-row matrix, I've decided to add an extra row above the first one to make my matrix 12x5 and wire encoders' push buttons as 2 extra keys, thus making it a total of 50 (12x4 + 2 encoders). I used 2 columns (9 & 12) for those buttons.

So, a first row actually has 2 buttons on columns 9 & 12 (because encoders are located near those columns). Encoders' push buttons are also configured via QMK's keymap.

Check out the d48.h & config.h for pins used & keymap macro definition.

I²C/OLED

Most of the stuff worked out of the box, except me choosing the right pins for my OLED.

On the Proton C pinout, there are 3 labels for I²C and for some reason there are 2 pairs of SDA/SCL for I²C¹ channel: B8/B9 (rear left side) and B6/B7 (rear right side). I'm not sure if this is a mistake or if I was doing something wrong. So initially I picked B8/B9 which were not working. When I switched to B6/B7, things worked like a charm. Later I used B9 for matrix row. No issues so far.

Oh, and by the way, while using B8/B9, keyboard was sometimes swallowing quick keypresses. I believe this was due to I(2)C timeouts (because incorrect pins were used for OLED).

Buzzer

It turns out once you switch on AUDIO_ENABLE, you cannot use A4 & A5 because they interfere with the buzzer. My guess is that buzzer uses DAC channels (not sure why both).

I couldn't find this in documentation. Honestly, Proton C has almost zero documentation and this was the biggest challenge. Anyway, apart from almost going crazy from those challenges, I really liked it!

RGB

I used pin A15 for my WS28128 RGB strip.

D1307 real-time clock

Connecting DS1307 RTC was a piece of cake: same I² pins as OLED (SDA/SCL), GND to GND and power to Proton C VUSB pin (5v).

Other issues

• B5 could not be used for matrix.
• TAP_CODE_DELAY had to be increased to 10 to fix tap_code(KC_VOLU/KC_VOLD) calls being swallowed in encoder callback.
• Be extremely attentive about the pinout: keep in mind that the official Proton C pinout displays the rear of the board, not the front. Being used to front pinouts, I ended up soldering entire matrix to the wrong side, so I had to desolder every wire and connect it to the opposite side.

Conclusion

I had a lot of fun. The layout was inspired by the Planck THK. Feel free to ask any questions!