Robotis Dynamixel Controller Limitations

I love Dynamixels, but have long been frustrated that all of the controllers that Robotis produces for them have very limited computing power. Because of this I have for quite a while been experimenting with different configurations to give my robots substantial onboard computing capability. One of these is using an OpenCM9 with a Pi Zero 2 W. It seems to work well, but I wonder if I am pushing the limits of the OpenCM9, given that I am using 15 AX-12s with it. I have a CM-5 and CM-530. I’ve opened the CM-5 and noticed it has big capacitors. Was the OpenCM9 engineered to handle that many AX-12s on an ongoing basis, assuming the power to the board is sufficient? I have also run my setup directly from a Beaglebone Black, powering the Dynamixels straight from my 12V power supply and using the UART from the Beaglebone through a circuit I put together (which I copied from the schematics from the CM9). It works, but not dependably (sometimes misses enumerating a servo or 2), and I’ve always wondered if it is my code, a limitation of the Beaglebone, or the quality of my circuit. It would be great if Robotis made a cape for the Beaglebone or a Hat for the Pi to control Dynamixels. Rather than make closed system controllers with limited microcontrollers, it would be nice if instead Robotis made an OS capable controller (like having a Pi onboard), or at least an interface between an SBC and the Dynamixels (like the aforementioned capes or hats).

1 Like