Failsafe no global position

Hello, we have experienced an error a few times when flying our large scale drone (~1.4m motor to motor) where we supposedly the GNSS receiver (F9P with RTCM corrections) stops sending data, according to the onboard logs. We flew the same pattern in the same area for an extended period of time, then a sudden apparent loss of GNSS data from the receiver. It reports at 21:51 in the log. Then no GNSS measurement is logged for 17 seconds.

As I experienced this error before I set GPS_DUMP_COMM to 1 to have more data to debug the communication between FC and GNSS. Using pyulog I attempt to extract all data from the ulog file however, no gps_dump messages are seen, and ulog_extract_gps_dump says no gps_dump messages there.

Log can be seen here:
https://review.px4.io/plot_app?log=f02b6e93-8fa0-414c-8e44-0c40a9d408f8

Note:
I am fully aware that there are high vibration metrics on the drone, however it is the fact that there are no gps_dump messages in addition to the GNSS receiver apparently stops sending data i’m looking for here

At first glance a few things come to mind. Without knowing the details of your setup I would suspect a connector-glitch related to high vibrations or something related to the strong thrust-magnetic field correlations you have there. In any case the GNSS might be resetting/rebooting during those 17secs of silence.

But those are just wild guesses.

From what I see the thrust-magnetic field comes from the drone switching from GNSS height to barometric height and stopping its velocity to start a landing as set in the failsafe. This can be confirmed looking change in estimated Z height and attempting to stop that rapid descent before going controlled downwards as well as the velocity in x-y axis

That’s not what I meant by correlation. Check Thrust and Magnetic Field in the docs.

To put it bluntly: The red line is not supposed to consistently follow the green one. This means that you have some strong, measurable electromagnetic fields in your setup. Couple that to high vibrations and suddenly electromagnetic inductions are very likely to cause all kinds of problems.

But again, just a guess without knowing the construction details.