Test results of compass_learn=3

[jstroup1986 (Joseph G Stroup Jr)]

I tested the compass_learn=3 today.

Attached are two MagFit charts. The first is MagFit of my copter flying with compass calibration settings from before installing 4.2.0-RC1.

The second is of MagFit of the copter flying after calibrating the compass using compass_learn=3 to calibrate the compass.

There’s a difference - but even after running compass_learn=3, MagFit shows improvements can be made.

When I run MagFit I select both elliptical and MOT options. Just a guess, but I’m thinking that maybe compass_learn=3 doesn’t account for motor interference.

I’ve floated the question before - with both options available, should one be chosen over the other?

Before compass_learn=3:

After compass_learn=3:

[Yuri_Rage]

There is probably not a one-size-fits-all solution. You should use the one that yields better results for your application. That’s what tuning achieves.

COMPASS_LEARN will generally be optimized for the specific conditions during which it was executed, so it may not account for motor/magnetic field variation throughout the entire operating range.

While I don’t have much experience with the MagFit utility, it should yield excellent results so long as the components remain consistent with the range of conditions from the analyzed flight.