i have tests of apm 2.5.2 with disabled compass
internal gyro only works better than "gyro + compass"
with compass + gyro if i do fast rotation at 180deg. i see in mission planner another angle, after 5-10 seconds this error slowly fixed (gyro corrected by compass)
if i starts this test with disabled compass i can see correct angle of rotation immediately
i think compass correction must to be disabled in fast rotation or gyro correction gain must to be decreased
actual compass gain may be used in startup correction only
is compass gain available in full parameters list?
tested with AC 3.1.2
This is the complete list of user-modifiable MAVLink parameters. You may want to also review this list and this one as well for clarification.
Yes, thanks for testing this.
Can I ask, how do you know the final angle is correct? Once you disable the compass you won't see the slow correction after the copter stops moving but that doesn't mean of course that the final angle is correct.
Also are you testing this with the copter armed or disarmed? the compass's updates on the heading are very weak when the copter is armed. If you test it with the copter armed, you don't need to plug in a battery though, just as long as it thinks it's armed you'll see what i'm describing.
If you're right that the compass is making the heading worse during fast rotations then I suspect it's because of a delay in the compass readings. so you can imagine, if the compass was always 0.1sec behind then as the vehicle rotates the compass would always ends up slowing down the rotation rate & heading. As you say, disabling the compass correction during fast rotations would be one solution, another would be to compare the compass heading to a delayed heading from the gyros so they line up.
thanks for the tip about arming,
in armed state really sets a heading much more precise