[Peter Johnson]

Originally Posted by RufflesRidge

The CAN connector looks a lot like the power connector on the 2CAN. Probably just the two wires. As a differential bus CAN shouldn't need the 5V and GND being passed around.

Originally Posted by crake

Correct as well - the CAN topology in this system does not distribute power. Power would be supplied to each device separately.

Originally Posted by Meshbeard

I'd assume that those two connectors are for the CAN-high and CAN-low wires. Power is probably separate since all the power systems on an FRC robot are separate from the controls.
What will likely happen is all of CTRE's new CAN stuff will have those two CAN ports separate from the regular old 12v power inputs on all the other components.

For robustness some systems pass around the control power. In these systems, the CAN bus control logic on each component is powered from the CAN bus power lines instead of the mains supply, ensuring the component is always reachable even if the mains supply to that component is lost (with optoisolation between the CAN control logic and the rest of the circuitry). This could be useful even in the FRC world: for example, tripping a breaker would not cause that CAN bus component (e.g. motor controller) to drop off the bus (so you could still pull status, and it could report it doesn't have mains power etc).

There is a serious problem with the current CAN layer where timeouts to Jaguars could swamp other parts of the system due to error spamming. As timeouts can be caused for multiple reasons (bad termination, disconnected cable, etc), regardless of the power distribution approach, my hope is that the software layer will be updated to avoid this issue.