While the pilot generator was close to spec with the last changes (in terms of rise and fall time for the pilot), some component changes dialed it right in. The opening resistors are now 1 kΩ and the closing resistors are now 4.7 kΩ. With that, the rise and fall times of the pilot are now around 2 µs, which the spec demands.

I also discovered that running high current lines adjacent to the RTC and its crystal are bad for timekeeping. I've solved this in the next revision of the boards by moving the RTC to the top end of the board. I'm also going to tweak the layout of the car B pilot generator, as some of the components are too close together for easy rework.

But the most recently fabbed boards (which will arrive tomorrow) are still fine - the component changes don't affect the layout, and you just need to keep the power conductors away from the RTC (running them above the board rather than below is fine, as the ground planes provide enough isolation).