I couldn’t agree more! It would be nice to have a free SPI to communicate with some kind of optional module like an EtherCAT driver.
And for hobbyists, an SPI in slave mode would allow to communicate very quickly with a RaspberyPi or BeagleBone master. With a DMA on the ODrive side, it should be pretty transparent by the way.
And of course, using UART or SPI communication can only be done for very short distances…
SPI->EtherCAT is available… piggy back on this:
http://www.bausano.net/en/hardware/ethercat-e-arduino/easycat.html
I would be massively in favour of having SPI slave interface for coms to RaspPi / Arduino. It can be run really fast if the nodes are close to each other, and opens up opportunities to interface with all sorts of devices. Then can also rely on EtherCAT for the long hops
I don’t think to use a PC or raspberry because the OS is not real time (OK you have special OS with real time but my robot is not as complex as the humanoid) If I can run my code on the ODrive it would be great.
Finally I am not absolutely sure to need this type of control. I would probably implement the easiest first: speed control and if I have the time, try to implement the inverse dynamics. (this is my challenge! I really want to try because I am interested in). But there is something that I still don’t understand: If I implement the command in speed (in voltage), the motor will take the current needed to follow this speed. As long as I generate trajectory that doesn’t draw too much current, it should be the same as calculating the needed current and feed the motor with it. What am I missing?
PS: Nice job with the pinocchio and the humanoid! Thanks to madcowswe and the ODrive system too!