How to verify compass accuracy ? - looking for "perfect" calibration

ArduCopter Copter 3.6
1

The vehicle flies well. when braking from higher speeds (poshold) it tends to go slightly to one side, and the other when flying in the other direction. , then stop and hold perfectly.

  • Primary magnetometer is far from electric components, down on a landing grear strut.
    How can I use dataflashlogs (or tlogs) to reliably confirm if there is a mismatch between magnetometer data, and reality ?
2

If the magnetometer is part of the GPS, it definitely needs to have a clear view on the sky. Please post a picture of it mounted.

3

Andre-K
How can I use dataflashlogs (or tlogs) to reliably confirm if there is a mismatch between magnetometer data, and reality ?

I also look for some more information somewhere.
At the moment the calibration was successful or failed. That it, I believe.

Regarding GPS and mag need a clear view to the sky is not always true. My Helicopters have big carbon Spin Blades (brand) and covering the GPS units as a carbon disk. No problems what I can see.
I had the unit mounted outside the disk, absolutely no difference.

4

ok, thanks for trying:
magnetometer senses magnetism, it has absolutely no need for “seeing the sky” , as long it’s far away from other magnetic parts, ferrous metals or electromagnetic interference.

Still, this is not about bad placement, just a question like - if two calibration sets fly fine, how do I quantify hpw perfect they are, and which one is “better” ?

5

I realize this doesn’t answer your question, but by your description, it sounds like there is a mechanical issue causing the steering bias (motor alignment, CG balance, etc.), since you say it only happens when it is braking from high speed. If the compass is misaligned, it should become apparent when trying to fly straight forward in Loiter at any speed, as the drone will fly a few degrees to the left or right.

Either way, the cause could be confirmed by the flight log.

6

Whatever reason, one quick re-calibration later, there was perfection. :slight_smile:
(and it’s maybe a year since last it was calibrated before that, so I do not even know when the deviation spawned, unless it always were there.

So this is why I wanted to have a “perfect” way to tell if it were a bit off. - without doing special tests. If I wanted a special test I would fly a huge circle or an octagon (auto, heading forward), and just verify heading vs bearing.
So this is why I asked for a bulletproof, simple way. (and hoped somebody pointed me to some EKF data that could tell) :slight_smile:

1 Like