TAKEOFF Mode - first attempt with odd results

I finally wanted to test TAKEOFF mode (without compass, by the way), but things didn’t go exactly as planned.

According to https://ardupilot.org/plane/docs/takeoff-mode.html, I was expecting the plane to fly in a straight line to a point located 200 meters from me, while climbing to 60 meters, following the heading I had thrown it.

Yet it went to the right soon after launch instead of keeping a straight line, then to the left after a while, as if it was already starting to enter the loiter circle, but it had reached neither the distance set nor the altitude at that point. It even paused the climb and descended for a while.

Combined with the fact that I had already thrown it too far right, the plane almost hit a treetop at 40m. Confused by all that and a new radio I’m not yet used to, I accidentally disarmed while switching to FBWA, flipping 2 switches at once. Luckily the plane was recovered intact - just in case you wonder when looking at the log. (Side note: Is there a way to safely enable re-arming in flight, without compromising prearm checks on the ground?)

Anyway, here are all the TKOFF parameters:

TKOFF_ACCEL_CNT,3
TKOFF_ALT,60
TKOFF_DIST,200
TKOFF_FLAP_PCNT,0
TKOFF_LVL_ALT,20
TKOFF_LVL_PITCH,20
TKOFF_PLIM_SEC,2
TKOFF_ROTATE_SPD,0
TKOFF_TDRAG_ELEV,0
TKOFF_TDRAG_SPD1,0
TKOFF_THR_DELAY,5
TKOFF_THR_MAX,95
TKOFF_THR_MINACC,5
TKOFF_THR_MINSPD,0
TKOFF_THR_SLEW,25
TKOFF_TIMEOUT,0

Log: https://we.tl/t-h8g7KExWKB

Thrust is definitely more than enough to make the desired climb (600mm ARWing Mini at 4S). I flew it and had it loiter many times at the same location with the same setup and no problems.

I used a triangle calculator to check the angle required to reach 60m in 200m distance - result is about 17°, so 20° should be acceptable? (Side note: I have to manually enter a new pitch angle value every time I change TKOFF_ALT and TKOFF_DIST, correct?)

I wonder if I should maybe enable the compass - but from what I read, TAKEOFF should also work without, and the lack of compass should not cause things like this?

Thanks for any hints!

quote from the instructions you linked:

“ . It is highly recommended that a compass be enabled and properly configured for auto takeoff, as takeoff with a GPS heading can lead to poor initial heading control such that heading can different from the initial heading by tens of degrees during the climb. While this may not be an issue for hand launchs, runway takeoffs require a compass for adequate heading control during the takeoff rollout”

Thanks, but as you can see from my post, I didn’t take off from the ground - it was a hand launch.

Either way, there is now another, more active thread analyzing the same issue: Takeoff magnetometer and magnets