CAN specifies use of 'Daisy Chain' for reliability. Short branches are allowable but must be kept quite short. Tree structure as mentioned in this video can be problematic if any of the branches (Spurs) get too long. Confusingly, the problem mostly doesn't cause glitches with the device at the end of the long branch, but rather other devices on the BUS. Absolutely avoid any 'branch' longer than 30cm otherwise you will get the occasional glitch in the system. If you have a distant CAN device on your robot it is essential to run the CAN wire pair up to it and a return CAN wire pair back down. (Unless you make that device the 120 ohm terminated end of the CAN Bus) At the frequencies used in CAN, a long branch will produce reflections of the CAN signal that may/will cause the bus to glitch occasionally or continuously. Data will be lost and error correction (if enabled) will be repeatedly having to resend. As stated above - Often (In fact mostly!), it won't be the device at the end of the long branch that is having the errors. It depends how and where the reflections 'beat' with the existing signal. I totally understand the need for mechanical ruggedness of the wiring, just don't have long branches and all should be good...YMMV
CAN specifies use of 'Daisy Chain' for reliability. Short branches are allowable but must be kept quite short.
Tree structure as mentioned in this video can be problematic if any of the branches (Spurs) get too long.
Confusingly, the problem mostly doesn't cause glitches with the device at the end of the long branch, but rather other devices on the BUS.
Absolutely avoid any 'branch' longer than 30cm otherwise you will get the occasional glitch in the system.
If you have a distant CAN device on your robot it is essential to run the CAN wire pair up to it and a return CAN wire pair back down. (Unless you make that device the 120 ohm terminated end of the CAN Bus)
At the frequencies used in CAN, a long branch will produce reflections of the CAN signal that may/will cause the bus to glitch occasionally or continuously. Data will be lost and error correction (if enabled) will be repeatedly having to resend.
As stated above - Often (In fact mostly!), it won't be the device at the end of the long branch that is having the errors. It depends how and where the reflections 'beat' with the existing signal.
I totally understand the need for mechanical ruggedness of the wiring, just don't have long branches and all should be good...YMMV
Is daisy chaining better than this?
That's our question too
@@bleh1323 see my comment
Sorry cant understand anything you say with your dog muzzle on.