@Gamer43 is the actual designer, and the funder and parts buyer. I am the electronics tradesman that did the layout and mechanical design. His motor control algorithm is much better than VESC. VESC doesn’t do a good job estimating rotor position over 100A and you wind up with a poor lead angle pushing far more current for far less torque.
Our current test rigs are an electric go kart and the dynamometer at work, which is limited to 10kW and is a sterile lab setting so not good for real world type testing. If anyone knows of a way we could aquire an efoil deck to test, any suggestions would be appreciated.
3 Likes
Bang-bang control eh… how does that one work? I’m super curious how you’re performing better then VESC. also what particular part are you facing a shortage on?
it’s full on until you reach a certain speed. Then it gives full current intermittently at high frequency to maintain speed. You need a pretty wide hysteresis band for it to behave well.
There are a number of things we suspect that might cause vesc to perform poorly in the scenarios Nick mentioned, but not sure what. I tried to implement the cleanest possible version of FOC to my ability to try to achieve maximum torque efficiency.
Additionally, a number of people on the skateboard forum have been reporting an increasing number of bugs on recent firmware, whose symptoms I can only deduce to running out of clock cycles, (e.g. weird sputtering noises, aberrant braking behavior) and we just didn’t want to risk dealing with that.
I am facing a shortage on every single line item on the BOM, it’s terrible.
Would love to see your FOC’n Code. You’ve implemented you’re own Park and Clark Transforms? Or have you borrowed some older parts of VESC software?
I’ve developed quite a bit of VESC HW. I have a decently working but a nit noisy controller that can do significant power.
Also know a few work arounds for a few BOM items… For Eg. STMF405 off Feather development boards.
Cheers.
Mo
The motor control is unfortunately proprietary, I will have comms and interface code that will be open source eventually.
I developed a proprietary no sensors zero speed algorithm for my master’s thesis which is the proprietary stuff, it turned out it was also good for getting accurate motor parameters.
I use the STM32G4 family, which there aren’t many in the wild. It features hardware accelerators among other things that are useful for high performance motor control.
We also spent a lot of time ironing out the signal processing chain to get the most accurate measurements.
1 Like