I’ve been researching the crap out of this, here goes:
Yesterday, around noon, CST I was doing a research flight looking for battery use data. This consisted of 3 missions. 1st mission completed without incident or errors of any kind and lasted about 7-8 minutes. I landed, it disarmed, I loaded a new mission and took off. About 2 minutes in all hell broke loose. The bird kicked out of mission mode and into DESCEND mode (which only the autopilot can command) and cycled 7 times between position and alt hold modes despite me putting it into stabilize mode. It was VERY bad.
Question: What caused the GPS Here4 module to report a loss of compass AND a loss of GPS?
Here is what I’ve found:
Flight Review Link: https://review.px4.io/plot_app?log=64b2655b-bf27-4801-8e48-694b796fba87
Param File: Dropbox
First Good Flight Log (.ugl): Dropbox
Second BAD Flight Log (.ugl): Dropbox
- Vehicle reported no errors on takeoff and had completed one successful mission prior with no reboot in-between.
- Loss of compass occurred and 15 seconds later GPS took a hit.
- We went from 32 locked satiates down to 8, then 6 then 5. No known cause.
- HDOP was at .5, then diluted to 9, no known cause.
- Prior successful mission had a rock solid full time 32 satellite lock
- I completed a mission successfully only minutes prior under full auto mode (mission mode). No issues.
- It was early in the second mission while in mission mode when uncommanded flight modes were activated. No known cause.
- These Flight modes were not possible to activate via hand held transmitter. From Mission Mode to Land then descend, mission, descend, mission, descend, land, stabilized, position, altitude, position, altitude sequence. Hand held transmitter switches were not being activated by the pilot with exception of stabilize mode one time after it was known that the drone was not being controlled properly.
- Pilot CAN NOT COMMAND DESCEND MODE. It is a failsafe mode only.
- Current flight mode settings are Hold, Stabilize and Land on 3 position switch. RTL mode on a different 2 position switch. Recommend to change STABILIZE to ALTITUDE HOLD mode.
- Flight Analysis website shows 3D pathing that is underground. Unknown cause but probably due to loss of GPS locking.
- I can’t find any proof that a loss of connection between the hand held transmitter and the drone ever occurred.
- No companion computer was connected
- Manned aircraft callsign N314VA was reported by ADSB as in the area with separation distance of 293 and another at 457 (ft? meters?). No aircraft was observed in flight though.
- Max tilt angle was 50 degrees which is a little high (45 degrees set as limit while vehicle is in air).
- Mission started with 32 satellites.
- First mode commanded by pilot was from mission mode (where it started) to stabilize mode for 6 seconds then to position mode. Flight Mode was kicked back and forth between altitude and position 7 times as the GPS positional data continued to degrade.
- HDOP climbed up to a value of 9, anything below 1 is flyable.
- No jamming was detected, this also means no EMI from onboard sources. Noise was not present when GPS was lost.
- HDOP never recovered after the troubles began, even when number of satellites climbed back to 9.
- 5V power supply to the autopilot was good and steady throughout the flights
- At approximately 20:41 UTC the GPS system took a hit that it never recovered from. The compass failed 10 seconds prior to this event.
- High Rates logging were ON by pure chance so we have better than average sampling.
- CPU load was about 51% throughout the flight
- Beginning shortly before the GPS interruption there was a raw magnetic field fluctuation that went from -.2 gauss to +.25 gauss. Unknown cause. Possibly due to yaw controls reorienting the drones heading but that correction only lasts about 3 seconds, not 30 seconds.
- Prior log 15-51-22.ulg shows a correlation between this yaw control and the magnetic spiking events.
- Kp was 2.67 with an expected geo storm of magnitude 5 coming within the next 48 hours. A class G1 CME was reported for August 17, 2026 but it was not a direct hit on earth.
- RSSI was always at 41% which indicates a fault in settings. Investigated this and found the Window time on the air side RFD was at 131 instead of 120. No other discrepancies were found. RSSI now functions as expected.
- Total flight time was 4 minutes 51 seconds.
- Resting battery voltage post flight is 23.21 volts with no power dips shown in the logs.
TWO DAYS LATER:
Sat the bird in the backyard and just let it soak. Satellites climbed to 30 or so then dropped to 5 then 4 then 0. Notice the dropping to ZERO and still a Sensor Status of Normal??!!
20 minutes later this happens, it is suddenly working!:
I caught the whole thing while logging. New log of this event is here.