We use pixhawk 2.1 and Ublox NEO M8N GPS module.
In this simple configuration GPS reports lots of satellites and EKF2 gets GLOBAL_POSITION within ~15 seconds from boot.
But things go downhill as we add more stuff to the drone. We connect Intel NUC as on-board computer, RPLIDAR, Intel RealSense ZR300. As more devices are added (or even wires re-arranged) we see all sorts of problems with GPS.
First problem is when GPS driver report zero satellites. This seem to be caused by simple RF interference. So moving GPS away from all the main electronics helps with this.
But there seem to be another, more insidious problem with conductive noise.
The issue was discussed in 2012 here:
and at more length in 2014 here:
We managed to trace some interference to serial communication on TELEM1/2 causing noise in the power cables going close to pixhawk. After moving the cables away from pixhawk we see less issues. Also played with putting ferrites on various wires with some success.
BUT seemingly trivial changes to any part of the system tend to throw GPS out of order now. Or cause EKF2 to take minutes to get GLOBAL_POSITION.
For example swapping cameras from Logitech to RealSense to MyntEye gives unpredictable GPS performance.
What is the current state of the game with GPS self-jamming on pixhawk hardware? Any advice?
I would imagine lots of people use Pixhawk/PX4 in progressively more sophisticated configurations and must encounter these EMI issues.