July 12, 2015 at 1:46 am #795
I’m wondering about the best setup for a specific application:
During flight phase of an UAV – lasting about 15 minutes – I’ve to log 3D attitude with best quality (the data won’t be used for controling the UAV). For yaw an absolute azimut regarding magnetic north would be fine, but an relative value would do as well. As far as static tests on gound have shown, pitch an roll are quite quite stable whereas yaw shows a drift of about one degree per minute (NMEA output).
What I’ve done so far:
– Magnetic calibration procedure done once.
– Powering up and waiting some time for stable temperature.
– Issuing command “ZERO_GYROS”
– Issuing command “SET_MAG_REFERENCE”
(this command seems to have wrong description on page 78, I assume it’s equal to “set mag vector” in the tool)
I don’t understand exactly how the EKF combines gyro/acc-data together with mag-data. E.g. there is configurable some “mag stat update”. Is there some more information like application notes rearding this topics and might it be possible to decrease yaw drift?
July 12, 2015 at 7:21 am #797
Usually, if there is yaw drift on the UM7, it is because there are magnetometer calibration issues that prevent the angle from being measured properly. If there are time-varying magnetics fields (from the permanent magnets in a motor, for example), it can make it very difficult to get a good yaw estimate.
The magnetometer reference vector tells the UM7 what magnetometer measurement to expect when yaw = pitch = roll = 0 degrees. During operation, it takes the measured magnetic field in the sensor body frame and uses the current attitude estimate to rotate it into the inertial frame (see this library article for details about coordinate frames used on the UM7). If the attitude estimate is perfect and there are no magnetic field distortions, the rotated magnetic field measurement should match the reference vector perfectly. Distortions cause the sensor to erroneously measure attitude error.
You are doing all the right things to prevent yaw drift, but it could be that motor magnetics are causing problems. Also, be sure that you are writing everything to FLASH.
July 12, 2015 at 7:25 am #799
You must be logged in to reply to this topic.