We ran into an interesting feature of the 2CAN this weekend. The ability to be powered without a negative lead! One of the students told me that the 2CAN LED was still flashing despite the black wire (battery GND) falling out. I assured him that this was impossible and we continued working. Later he insisted on proving it to me and indeed he was correct! We proceeded to disconnect the CAN cable between the 2CAN and the Jaguar, the 2CAN turned off. We reconnected the CAN cable and disconnected the Jaguar battery ground wire, the 2CAN turned off. Based on these tests we decided the 2CAN was finding a path back to the battery through the CAN network cable and the negative lead on the Jaguar power connection. I attached an illustration for clarity. We were surprised to find out there is no isolation between the CAN network and the power electronics inside the Jaguar. Comments on this? Do we have a defective part, faulty wiring, or is this no surprise to some of you out there?
Magic2CAN.bmp (1.47 MB)
Magic2CAN.bmp (1.47 MB)
This is correct and also the case with the Jaguars. The grounds are common through the CAN cable. This should not affect performance. There are some thoughts that suggest having a ground return through the CAN cabling can cause a higher level of common mode noise on the differential pair however if you provide a proper ground to the Jaguar and the 2CAN this should not be the case. If you are concerned, you may cut the black cable inside the can cable to break the ground path.
We expected the CAN network to be optically isolated from the power network within the Jaguar.
Since CAN is a differential physical layer isolation is not required. Ground is not referenced as it is with PWM. Since ground is not referenced by the data lines having common grounds or ground noise should not affect data integrity.