http://letsmakerobots.com/node/30202

Uses tri-state capability of Arduino digital I/O pins to build a 1-pin H-Bridge controller.

Cannot use standard PWM. To PWM this, you need to switch between
input and output mode on the pin. High is for foward. Low for reverse.

I have been reading up on BLDC motor control, and am wondering if a novel approach might be more simple, and require a few less parts. Almost all solutions I see involve PWM. The PWM throws up a ton of noise and makes sensing the zero crossings tricky. It seems to me like I could use a single pulse, of short duration on each commutation step. This would reduce noise, and at lower duty cycles (where sensing the back EMF is trickiest) there would be a "quiet" period in which we could estimate zero crossing times.

The main idea is to drive the motor from open-loop style timings, but asycnronously monitor and update these open-loop timings. Estimations of zero crossing times and frequency would get updated, and used to tune FUTURE commutation steps. Control would always be one commutation step behind, but can be predictive and use some more complicated math, like least-squares fits to estimate zero-crossing times.

I wonder if most solutions out there are inspired as digitizations of traditional analog designs, instead of re-thinking the problem in a more digital manner given the availability of cheap, fast microcontrollers.

To do the "sensorless" BLDC driver, I need to sense motor back EMF at voltage ranges outside of the Arduino inputs.
I also need to LPF this signal, since it will contain switching noise.
I needed to recall how to solve RC circuits, the following link covers what I needed to recall to get the specs for the voltage dividing LPF.
http://boim.com/misc/LPFVoltageDivider.html