Blueprint

Guilty Devboard idk

A dev board for my keyboard, because all the options kinda suck. Criteria are, high to low priority: - Can be replaced with a stock Xiao w/ shift registers or IO expander (GPIO in a TSSOP arrangement? or just along the bottom and IO expander can be elsewhere?) - Lots of GPIO - Fits on a basic TKL layout keyboard between ESC and F1, i.e. in a 1u by 1.5u space; alternatively, fits with minimal modifications - As low power draw as possible - Reasonably fast adjustable battery charging (500mA?) - Built-in chip antenna - Built-in boost converter and overdischarge protection - Internal i2c pullups

Created by autumn autumn

Tier 3

5 views

0 followers

Timeline

autumn autumn added to the journal ago

Component selection and 5V power

So a couple things.

  1. The boost converter was less powerful than I thought - would theoretically work, but barely. Managed to find an obscure, tiny, high efficiency, high current fixed voltage boost converter. Even works at high temps, has a switch (not that that switch works for my purposes but whatever), even has a power good output, it's great.
  2. Did component selection. Tiny high current diodes are also hard to find, but I did find some. Had to make my own footprint for one model diode.

I tried to document it... just a wee bit of text

image

autumn autumn added to the journal ago

Look I did stuff but I barely even know what I'm tired

IDK like, first of all I fixed the 3.3V regulator since I accidentally used an linear regulator instead of buck-boost on that. Also I adjusted the capacitors to better values - generally just bigger if they could be/it made sense. And I checked out the CM0 since apparently that has wifi/bluetooth and I'm not smart enough to figure that out lol - messaged that person for help/questions, so hopefully they're helpful.

image

autumn autumn added to the journal ago

Did power properly, added RF switch

Misunderstood how it was powered and was missing 3.3V, so fixed that by adding another PMOS and a 3.3V buck-boost converter. Added a capacitor on there just to be safe.

image

I also added an RF switch so either a chip antenna or external antenna can be used.

image

autumn autumn added to the journal ago

Finished the schematic aside from GPIO

Need to add GPIO still, but aside from that the schematic is done. I decided to upgrade it to have both a built-in 5V boost converter - so there's 5V even when on battery - as well as overdischarge protection. The charging is also variable - controlled by a digital rheostat. Everything should be done aside from adding IO/testpoints.

The hard part regarding IO/layout is figuring out how exactly it should be set up. The goal is to make it Xiao-compatible, but with extra pins, and ideally the ability to swap out this board for a Xiao plus an IO expander or something. Alternately, I could clone the Xiao's layout and extend it further down, plus pins on the back and along the bottom, maybe like some hybrid of the Xiao nRF52840 Plus, Xiao nRF54L15, and RP2040 Zero-style supermini clone?

image

autumn autumn added to the journal ago

Oops the NRF54L15 doesn't have USB

image

Apparently the nRF54L15 doesn't have USB soooo I'm redoing this with the nRF52840 again. Also turns out the QFAA variant doesn't have USB either, so it'll be very difficult to hand solder - not that I was planning to do that anyways.

No I/O, but were I to put this on a PCB, this should boot as-is.

autumn autumn added to the journal ago

Making some progress on the schematic

Nothing too interesting, just getting some schematic work done. Apparently it requires a 32 MHz crystal, and the 32.768 kHz one is optional. As for RF, just sticking with copying the reference designs, and can switch it out later. Undecided on whether I want an external antenna or not though.

Also, Figured out DNP/DNI/N.C. - same concept as 0 ohm resistors, just yk, you don't install stuff in the first place. Working on decoupling now.

image

autumn autumn added to the journal ago

Researching and figuring out antennas

The most difficult thing by far is figuring out the antenna. Need to use microstrips to match the impedance, then you're supposed to use a pi network, which is either for impedance matching or cleaning up the signal - who knows, google lies. But I've figured out microstrips at least.

image

autumn autumn started Guilty Devboard idk ago