Michael, the kinds of tips and rules of thumb (based on your experience) mentioned in your last post are exactly what I'm looking for. Thanks!

It's interesting that you use rather tight tolerance to qualify the lead-in pulse and then much looser tolerance for the other intervals. Its the kind of thing that must have been determined from experimentation over time.

Pulse to pulse timing (as opposed to overall windowed sampling based on absolute time from the first pulse) intuitively looks like the best way to go, and you seem to confirm that too.

I kind of like having the option of being able to trade off processor bandwidth against code complexity, and will likely go with an interrupt scheme because there is a RS485 bus that also demands constant attention.

I'm doing this as a hobby (home automation) so time spent in doing it is not a big determining factor, but I do like to pick up the fundamental principles first and try worthwhile algorithms right from the start. Thanks again. And if anyone has other tips to add, I'm still watching the thread... :)