The world's been upside down for the past few months, but I'm happy to report that I've found some time and motivation to come back to this project. After a night of whatever the reading-datasheets version of soul-searching is called, I had roughly this series of thoughts:

• My old design was far too complicated. Multiple voltages and five (?!) interconnected PCBs are overkill, and super unnecesary
• The components I'm using, it turns out, are pretty tolerant of a wide voltage range. The nrf52840 can tolerate anywhere between 1.7V and 5.5V (though it does require some different configuration depending on what you're giving it), and the M41T62 RTC module can tolerate 1.3V - 4.4V. In fact, there are only two components critical to the design that won't function all the way down to the 2.6V low-voltage cutoff of the (protected) LiPo battery: the Memory LCD, and any white LEDs I might include to illuminate it. On the upper end, the highest input voltage I could expect would be 5V from the USB bus. Ignoring the two problem components for a moment, this means that I could do one of a few things:
  • Simplest option: power everything from the battery directly, and make sure to use a charger IC that charges at or below 4.4V
  • Simple option: Use an LDO to regulate the input voltage (USB or LiPo) down to ~2.6V, then use that as my system voltage
  • Possibly more efficient option: Use an LDO to regulate down to ~3.3V, then use some MOSFET trickery to switch over to the raw battery voltage when the LDO gets into its dropout region.
• All of this is a reasonably compelling argument to find a replacement for the 5V Memory LCD which, even if it does run at 3.3V, probably won't run much lower than that. Luckily, there is another low-power display option: EPaper. EPaper displays like this one run between 2.3V and 3.6V, and only need a handful of passives and an SPI connection to drive them. Lighting EPaper is just an equally frustrating endeavor to lighting a Memory LCD, but I have some ideas on that... more on that soon. I've been inspired by this somewhat famous hackaday project. Rather than have the display wrap all the way around the band, though, I think a smaller display with a subtle curve to it more fits what I'm going for... not to mention being way more affordable. This does mean that I can't directly run the whole board from the battery voltage, though, which could be as high as 4.2V when fully charged (or 4.5V+ when charging). Luckily, the LDO options still work great.
• The only problem, then is what I had previously thought was one of the simpler things in hobby electronics: turning on an LED. I'll definitely want LEDs to illuminate the display somehow, and my ultimate goal is to have a flashlight LED built into the watch too. White LEDs typically need a higher voltage than 2.6V (and a flashlight would probably eat most of the available current from an otherwise-low-power regulator), so it looks like I may have to boost the system voltage just to make light. Luckily, there's an IC for that...

More updates to come now that I'm back at work on this project :)