90% of this project, I've gone ahead with hand tools or a soldering iron and just built stuff. Very little drawing out of stuff beforehand. That's my dayjob, planning ahead isn't fun.
So far the damage on this project has been two blown caps that I've over-volted, a bunch of wasted time building some motor drivers that ended up being too noisy (they would cause the microcontroller to brownout - I may fix this later, it's not too hard), and now a few more hours of wasted time.
I was getting some weird results with my power circuit (previous log), then I ran out of time and had to leave it for the day. Later in the shower, it hit me - duh. I was trying to drive the MOSFETs from the wrong voltage reference.
N-channel MOSFETs need a positive threshold voltage from the source to the gate. Usually (or often, I should say) they're going right to 0V, so my natural inclination is the drive the gate voltage using ground as the reference.
That's totally not what's going on in this circuit! Obviously! Here's a P-channel solution that I should have started with, given, like, 5 minutes of critical thought:
An incomplete survey of my hackspace tells me that I may have issues finding appropriate P-channel MOSFETs, though. They're much less common than N-channel. One of the local brick-and-mortar places has an old old list of transistors they used to stock in the 2000s, and that has one or two potential hits, both between $4-$6 per transistor :/