Hi all - I'm hoping some experienced eyes can help me diagnose the root cause of unscripted spiral flying that eventually led to a crash while on a simple mission in auto flight mode.
This is the 4th mission this quad has flown, and the ~25th flight. Every other mission and flight has been perfect, and stabilize, RTL, land, alt hold, loiter, and circle modes successfully exercised. This particular mission had more steps than the others, but it was still quite simple: takeoff, down and back along ~150m course 3 times, RTL, and land, with some condition_yaw changes at the waypoints for different on-board GoPro shots. On the 2nd trip back along the course, the copter started aggressively spiraling in ~30m circles. I finally got it in to stabilize mode but made an orientation mistake and ended up crashing it into a low fluffy pine tree trying to get my bearings. Luckily the copter suffered no damage and performed wonderfully on some short stabilize mode test flights afterwards. In fact the MP and TX telemetry never lost their connection to the copter. That's great, but I obviously don't want to attempt further autonomuos flights until I understand what went wrong.
From what I have read on other posts and on ArduPilot.com this behavior during an auto-mode flight is indicative of a GPS failure. So I analyzed the tlogs and flash logs according to the Wiki and found nothing out of the ordinary. Number of sats is always 11 and the hdop never appears to climb above 1.3. When I planned the mission I changed the waypoint radius from 30m to 5m could that be a cause?(Why does it default to 30m in the first place?) Anything else I'm missing?
Tlogs and flash logs are attached, and I'm using a standard 3DR quad build, MP v1.2.70, APM firmware v3.01. The logs have a long dead space at the end where I was figuring out how to get the copter out of the tree.
Many thanks in advance, Henry
Replies
how are you powering apm, and what was your batt level. if you use jp1 and power through esc you get alot of crazy stuff if your battery voltage drops low even for a second. i had same crash as you describe resulting in full rebuild,, mine would also crash in loiter mode some times,,,,after seeing what all had in common it came back to a slightly bad cell in the batt. i tried using apm power board to correct but after ordering 4 which all had 11.1 volts on the 5 volt supply wire i gave up on that option for now
i am not pluging 11 volts into my precious apm
UPDATE - SOLVED
Some folks on Reddit pointed out that this behavior was symptomatic of magnetic interference with the compass (mine is onboard the APM). I performed a few compassmot procedures and was getting ~85% interference, which is not good according to the docs. I moved the APM and got it down to ~30%. Then I took a few test flights and graphed mag_field vs throttle before and after the APM move. Things look much better now:
Before:
After:
Thanks to Redditors for the helpful comments: http://www.reddit.com/r/Multicopter/comments/1kw8cl/unexplained_spi...