Hi,
We’ve always done the test with the same parameters, by uploading the same set. Not only that, but we’ve tested over 6 pixhawk black, and 3 pixhawk orange, which makes it a conclusive set of tests.
After some debugging, we believe we find the problem and a possible solution.
The Jetson Nano (or Jetson TX NX) does not produce a perfect square signal wave. After sending this signal to the level-shifter (1.8V to 3.3V) the defects of the Jetson waves are highlighted. Cube Black might have some additional resilience to this, whereas Cube Orange does not. I know that Cube Orange is a close source, but can you try to have any idea what sort of modifications were done on the Cube Orange connection to Telem 1 that is causing such issues? I guess this will affect most people that will use Jetson with the cube Orange.
To fix it, we will try 2 different approaches:
- A simple circuit integrated on the board (adding anything to a highspeed connection is usually a bad idea, but we’ve done a simple PoC and it solved the issue.
- A STM32 doing the wave correction and level shifting.
On the other hand, both Pixhawks send an almost perfect signal, which is correctly interpreted by Jetson Nano. No issues with this connection.
So in reality, what was happening was that the Pixhawk was sending all the information, but missing almost all acknowledges, making it to fail in parameter and mission waypoints readings.