We are seeing some differences in the communication error rates between the ttyTHS1 serial port nano, and the ttyTHS2 on the eMMC nano. Are there any signification differences between the 2 ports that we should be aware of? Specifically, when we communicate with ttyTHS1 on either an SD or eMMC nano, we can run at 921600 baud, sending 2048 byte messages - no problem!
When we send from one nano (ttyTHS1) to an eMMC nano ttyTHS2, we’re seeing errors if we go above 57600 baud, 200 byte messages.
I’m trying to do due diligence here - there are other potential culprits for this problem (e.g. a non-NVidia carrier card), but I want to make sure that the eMMC2 ttyTHS2 is not a path that needs further research on our part
The first suggestion would always be to set to performance mode before testing:
sudo jetson_clocks
This may or may not be enough to help. If this does not do the job, then can you set your UARTs to use two stop bits? This would allow correct function over a slightly larger tolerance (the clocks normally are within tolerance all the way up to and including speed 115200, but above this the clocks will deviate from standards…two stop bits increases the available working speeds when clocks are the issue).
hello surfingsoccerdad,
are you able to reproduce the same errors on the Nano DevKits?
thanks
linuxdev - apologies, I should have mentioned that we’re using 1 start bit and 2 stop bits in all of our communications. “jetson_clocks” … I’ll have to research that one. We *are" using the lowest power setting because it helps us thermally.
JerryChang - in the DevKits we use ttyTHS1 on both sides. We’ve been focused on delivering our system, and … well, I haven’t had time to do the test that you suggest. I’ll look up the pins for ttyTHS2 on the DevKits and give it a test
Thank you both -