[John Heden]

Hi Hugh,

With respect to my comment #3, We initially had ALL of our CRIO traffic going through the 2CAN device by connecting the CRIO to the 2CAN and then to the radio. This worked very well except for the same problem that we now discussing.

Our second topology was to run all Ethernet devices (CRIO, Camera 1, Camera 2, and the 2CAN) directly to the radio which felt like a more robust approach as the 2CAN device did not need to manage all CRIO traffic. This was an experimental change that did not seem to help or hurt but that's the way we have left our robot wired.

This approach leaves 1 of the 2 2CAN Ethernet ports unconnected and required us to disconnect 1 of our cameras when tethered (maybe the other 2CAN port could have been used for tethering but we never investigated this).

I hope this clarifies my comment #3 a bit. Perhaps our Team's next experiment would be to use port #2 to connect directly to the 2CAN device and see whether that helps. Given that this port has NO other Ethernet traffic, perhaps it would be a bit more consistent in any network influenced timing dynamics.

Thanks,

John