3 replies to this topic

[odtu_ee]

Has anyone checked the effect of quad motors on the mag outputs?

if those motors cause hard-iron type errors then they can still be (somehow) compensated. Does anyone have any idea about those effects? (e.g. how many Gauss is that? Does it depend on rotor rate?)

[D-Lite]

odtu_ee, on 21 July 2011 - 01:19 AM, said:

if those motors cause hard-iron type errors then they can still be (somehow) compensated.

No, it's not a static magnetic field at all. If you have an outrunner, the magnets are located on the rotating cover of the motor. That means the magnetic field changes with the rotation of the motor depending on the number of magnetic poles. For a 14 pole motor rotating at 5000 rpm, I would assume 873Hz. The magnetic field from the (static) coils rotates with a slightliy higher frequency. Both fields overlap and the resulting field would most likely be somthing low frequency.

odtu_ee, on 21 July 2011 - 01:19 AM, said:

Does anyone have any idea about those effects? (e.g. how many Gauss is that? Does it depend on rotor rate?)

Depends on many factors. For a high quality motor, you would expect most of the magnetic flow to run inside the iron. A big factor (of course) is the distance between the motor and the sensor. I did a quick test with the motor mounted 25cm away from the sensor and could hardly see an impact on the output of the sensor when turning the motor by hand. If I mount the motors about 2cm from the sensor, it's effect is at about the same magnitude as the earth's magnetic field. I don't have absolute calibrated numbers, but this rough estimation shines some light on the subject.
Another issue could be the magnetic field induced by motor and battery wires but as long as both (back and forth) wires run in parallel and close to each other, the effect should cancel out.

[peabody124]

I know Bill Nesbitt with his AutoQuad project claims to have a good calibration for his accounting for some of this.  However, it's not something I've really pursued as I suspect these complicated calibrations will end up causing more problems than they solve for the average user.  Same for cross axis misalignment which typically ends up < 1 degree.

[D-Lite]

I did another test with the motors running. I cannot see significant increase of noise on the magnetometer data. But there's a shift of x/y data related to the total current drained from the battery. This is most likely caused by the "power distribution" board located only 2 cm below the mangnetometer. My conclusion until now: the motors doesn't do much harm and can most likely be ignored for this subject. The layout of the "high amp" wires seems to be much more critical and can easyly produce errors that make the measurement totally useless without correction. This should be kept in mind when doing a frame design that is intented to include a magnetometer.