Hi all, can someone confirm what acceptable compass offsets and mag_field are?
I've been trying to improve my compass accuracy lately (3DR Quad C), and had a bit of a win today by raising the APM platform higher to get away from the battery wires connecting to the power distribution board immediately under it. This seems to have improved things a lot, the COMPAS_OFS_Z value halved:
And here's a graph of mag_field versus throttle from a flight today. I can't find the info about what is acceptable, I had the figure of 150 in the back of my mind, so I'm guessing that's a delta of 150 from the baseline (i.e, with throttle off)? In which case the graph below is within that threshold - how does it look?
BTW, I did some pretty accurate auto flights today, there was a slight deviation after each waypoint but once it was established en-route between waypoints it's very close to the tracks.
Another auto mission was to stay in one location and use CONDITION_YAW to point to each point of the compass in turn with delays in between, and it did a great job of that too.
However, I notice when I'm on the ground connected to the Mission Planner, the direction it shows is up to 90 degrees off from the heading I know it to be pointing in. Is this because it has the compass calibrated to flight conditions with the motors running, so when they're idle it's out of whack?
I had the same problems and solved it by moving my isolation platform up so it is now about 2.5 cm above the ESCs and power distribution board. Initially my offsets were very similar to yours with the Z around -390.xxx. The platform move brought it back to those you see here.
compass ofs x 26.665997
compass ofs y 1.67235
compass ofs z -10.118506
I'm continuing to tune it, and (most importantly) learning to fly it since I'm quite new, but haven't tried any missions yet. It will regularly do a RTL from 150 to 200 meters without any heading variance. While on the ground there is no deviation from true compass headings. My next step will be to move it back down about 8 or 9 mm and see what that will do to the offsets.
Here's a picture of the platform, which is pretty rough, but effective. My plan is to replace it with a two level platform which can hold a camera, and telemetry gear. It is isolated with sorbathane grommets. I believe the key to the isolation dampening, is having mass of the entire platform isolated which is much easier than trying to dampen the APM alone. Here are the accelerometer graphs from a recent flight.
I put these rubber spacers at the bottom of the stack:
This raised the APM further away from the power distro board to help the compass as well as decreasing accel vibration. My accelZ was +/- 3 before, now +/- 1 (i.e. -9 to -11 in a hover).
While things can always be improved, I'm curious what acceptable values are for mag_field and the compass offsets, and whether this accounts for the slight inaccuracies in waypoint nav.
Great looking accel vibration numbers...plus the spacers are clean, much better than the squishy sorbathane grommets I'm using. Thanks for the link.
I did see a forum thread a couple of weeks back in which Randy was involved, speaking about the mag_field and compass offsets and how they affect the nav. I've just searched but can't find them again...they're out there somewhere.
Just tying this up with some info on max mag offsets. Apparently -150 to 150 is good (from Randy on the dev call)