i have a standard jdrones hexa with the 30A motors and 12x4.5 props.
stabilized flight works well, as does altitude hold. but the two automatic modes i have tried so far have a stabilization problem: on entering LOITER, for example the hexa simply "slips" off to one side, it looks just as if the stabilizing correction is being applied in exact opposite of what is needed. the hexa slides off and follows a smoth arc towards the vertical, switching back to STABILIZE does not instantly stop the dive and it seems i have to send some command for it to begin to fly again - just a moment of throttle and roll away from the dive brings things back under control.
LOITER and CIRCLE behave exactly the same, i have waited several seconds (letting the dive get way beyond 30º before switching back to manual control) but it simply doesnt stabilize in auto flight.
what have i missed? as manual flight is fine.
from CLI: "ArduCopter V2.1.0 Alpha" - im not too keen about running alpha tbh, but this is what mission planner installed a couple of days ago?
im currently waiting for props to fly and test this, but believe i may have found the cause in incorrect magnetic declination setup.
magnetic declination here is 5º55' west, so according to setup instructions i entered "-5.55" and didnt think much more of it. it was only the other day that i realised the heading was roughly 180º out, on checking the declination setup parameter it read "-555", so i entered it again as "-5.55" and mission planner parsed it to "-555".
finally i entered "-5,55" (comma, not point) and it was parsed to "-5.48..." much better - seems like a regional settings issue with windows, although i really think mission planner should warn that the parameter is outside reasonable bounds.
555 - 360 = 195, which to me goes a long way to explaining why it seemed to be applying correction in exactly the opposite direction, but begs the question "is this the only parameter taken into account when loitering?" are there no limits to the angle it will apply to attempt to return to the loiter point?
I had exactly the same issue, I had to use a comma to enter the declination...
That kind of detail should be added to the arducopter wiki!