Congrats on trip completion Justin ! I presume you took a quick dip in Halifax, but left the bike on shore....
justin_le said:
One of the main reasons I wasn't exactly volunteering the bike to ride is that the firmware for the sensorless operation was only at a most basic functional level at this stage.
Good call. I stupidly convinced my wife to try my bike and accidentally left power on, with a throttle with spring removed.
Fortunately no major injury/damage...
justin_le said:
With a sensorless controller there is a speed below which the controller can't stay synchronized to the wheel because the back-emf voltage is too low to give accurate position readings. When the motor slows down to this speed, then the controller should either give up trying to power the bike if it is throttling, or if it is doing regen, then it should simply short out all 3 phase windings for a smooth resistance right to a stop. I hadn't implement either of these behaviours yet since I was planning to run with the hall sensors, and the sensorless code was only there as a backup and once I had it more or less working -- I was off! --.
Anyways, the consequence of that is that the controller would attempt to stay synchronized with the motor and do regen well below the speed at which this is even possible, and that's where the heavy chugging came from. In sensored mode, then the regen and all was naturally smooth right down to zero rpm but since the rain in winnipeg took out one of my hall signals I couldn't let people test it this way.
Justin
I see. Couldn't you have code/circuitry that still worked even if 2 of the 3 halls were dead or intermittent ? I imagine it'd be enough resolution to interpolate and create virtual hall signals for the missing ones. Might even work better than Back EMF, although some responsiveness would probably disappear. (???)
So your regen was somewhere around 2.3% overall then. Seems to match or exceed the 1-2% figures I've heard elsewhere, including for hybrid cars.
When I get regen, it will be interesting to see what could be done in the steep, never-ending Gatineau Hills or similar. Those hills KILL my batteries in short order, but they're much fun to go down; will be less fun/fast with regen though... I'm speculating as much as 7-15% regen recapture in highly hilly areas with some decent DC-DC conversion efficiency (or perhaps FETs to switch battery pack to 5s4p instead of 20s1p).
I guess you had no substantial up and down steep large hill areas to cross in your trek ? If you did, I'd be interested to know the regen efficiency over that portion of your trip.
Cheers, take a nice break and get back designing/building soon...
Mike.