Mission Planner 1.3.37 build 1.1.5917.13431
Firmware: (ArduPlane) Version 3.5.2
Yesterday I setup an auto takeoff, to auto mission on my wing.
Auto take off was set to a timer on my radio and it worked flawlessly! However as the plane reached WP-4 it tipped over and made a spiral nose dive to the ground. I took it out of auto as fast I could into stabilize and made a hard landing on the spot. No damage to the plane as this is built like a tank. But, I still want to get to the bottom of this.
So, on review of my flight I noticed that it successfully navigated WP TakeOff, 2, and 3, but when it reached WP 4, it spiraled clockwise out of the sky. It was about to make a left hand turn (sharp). I'm not sure if this had something to do with airspeed being too low. Speed was about 18 to 20. (ARSPD_RATIO calibrated to 2.726905).
The other thing I noticed, the EKF was occasionally flashing amber, but not red prior to the "Hard Landing"
The third thing it looks like a Compass Variance was announced at the point the plane went into its spiral. Now I'm not sure what to adjust.
Finally, I've noticed that when the aircraft is sitting level on the ground, or on a bench, that the nose in the GCS isn't pointing exactly in the same direction as the nose of the physical plane. There is a discrepancy of a few degrees. The compass and Pixhawk are both pointed exactly in the same direction of the nose of the plane. I have re-calibrated the compass many times in many different location (always careful to remove my cell, coins, watch, metal objects, etc) before calibration. The declination always seems to check out, and I've even tried to manually change the declination, but it always goes back to what it's reading after I click Obtain Declination Automatically and Learn Offsets.
I'm kinda stumped.
Compass Variance on takeoff is most likely cause by electro magnetic interference from the motor/ESC/batteries during high throttle. To fix twist up wires for all Pixhawk accessories, and make an effort to run the GPS and Mag wires as far away from batteries/ESC/motor and their wires as possible.
I've noticed that when the aircraft is moving the compass heading is now same as the aircraft. But when stationary there is a slight drift. It was hard to spot before as I don't have a GCS guy most days and have to go into the logs after. But this still doesn't explain why it track for a second. Maybe the turn was too sharp.