Work on the watch mk2 is underway! Based on the musings in my last update, the new watch has the following notable features:

Electronics

• A buck converter to knock the input voltage (the higher of USB power or the LiPo battery) down to 2.8V
• A flexible epaper display
• Five side-mounted buttons (3 on one side, 2 on the other)
• Drive circuitry for an LED backlight, LED flashlight, and buzzer motor
• USB-C charging (or, borrowing from this project, a magnetic charge connector semi-permanently inserted into the USB port)
• The same nrf52840 SoC, M41T62 RTC chip, and W25X20 SPI flash

All of this is on one PCB, which is now ordered and should hopefully be here in time to assemble and test next weekend.

Notably unsolved is the actual mechanism to light the epaper; I have some ideas, but I want to test them out on prototypes before I go too far into fleshing out the design.

Mechanics

This watch is a bit bigger than the last one (in order to fit the larger display). That's both good and bad; good because I have much more space to work on the PCB, but bad because it's... well.. big. It's about 15mm thick (which is actually on par with my current watch). It should be able to take any standard 22mm watch band, though it remains to be seen if I'll be able to 3d print the 1mm-diameter holes required for standard spring bars. 

I haven't included the flashlight LED or caps for the buttons in the design yet, but here's how it looks so far (ignore the weird render artifacts that come from Fusion360 trying to render glass).

Power

The E-Paper display uses effectively no power to maintain its image, and around 26mW (which, at 2.8V, is around 9.5mA) when updating. A partial update, which I hope to do once each second, takes about 250ms, which gives a total average power usage of around 2.4mA. This is higher than I wanted; if we account for about 500uA for the rest of the device (which, somehow, is reasonable), we end up with an average power draw of about 3mA, which gives about a week of runtime on the 500mAh battery. If we drop this down to a full refresh (which takes 2s, but provides better display quality) every 1 minute, the epaper display uses 300uA on average, which brings us under 1mA average power draw (or a runtime of 500h, or just shy of 3 weeks).

There are a lot of assumptions built in here, especially around my ability to get the power draw of the nrf52840 down close to its advertised abilities. It's probably good to set expectations low and assume about 1 week of battery life, corresponding to around 4mA average draw, then be pleasantly surprised if I can do better.