FXOS8700CQ x, y, z ranges not all the same?

[tedhuntington1]

I am getting data from a FXOS8700CQ and the ranges are:

X: -160 to +200 uT

Y: -160 to +190 uT

Z: +35 to +350 uT

Are these accurate values? (I am in Santa Ana, CA if that makes any difference). Since the magnetometer only measures the magnetic field of the Earth, shouldn't the Z range also be -160 to +200 uT? Is there something wrong with the Z range?

Thanks,

Ted

[anthonyduhamel]

Hello Ted,

Your values look too higher for the earth magnetic field. Did you scale the mag register values to normalized units? The magnetometer resolution is 0.1 uT/LSB.

Considering my previous note (by dividing value per 10), measurements are okay (geomagnetic field range is [+-25uT ; -+65uT])

About the Z axis, I have a question, how did you check the range? Did you rotate the magnetometer is the same plan (drawing a circle with the magnetometer ) or did you move the sensor over all direction (drawing a sphere)?

By the way, it seems to be an offset on the Z axis. Did you use the embedded  offset correction in the magnetometer?

Anthony

[tedhuntington1]

Thanks Anthony,

Yes I see that I need to divide by 10 more (*0.03662 which is 1200/32767, I had *0.3662). But shouldn't the Z have a negative reading when pointing up?

I checked the Z range just by rotating it in 360 degrees around the X and Y axes.

So you think there may be an offset on the Z axis? I have not tried to use the embedded offset correction yet, but I will try and see what happens. But should the Z be in the same range as the X and Y, or is it normal to have the Z as being positive only?

[anthonyduhamel · Accepted solution]

Hello,

No, you don't need to use those features yet.

If you rotate the device around Z axis, the Z magnetic filed should not change significantly because the geomagnetic field is horizontal.

If you rotate the sensor randomly thought all axis you get:

Z measurements are always positive, because there is offset/bias, we called it "Hard-Iron". It's why we calibrate the magnetometer.

The sensor fusion ​will correct that.

Picture below displays the extracted parameters.

Anthony

[tedhuntington1]

Thanks Anthony,

So the reason the Z readings are all positive is because the particles in the Earth's magnetic field are creating a magnetic field (a so-called "hard iron source") in the copper in the PCB under the sensor? Or perhaps the hard-iron source is coming from a magnetic field from my computers that are in the same room? I may take a look at sensor fusion, I am running Linux but have Windows too, I'm not sure how I can import my data into sensor fusion (perhaps comma or tab delimited?)- or possibly I will just subtract a hard iron offset of 19- to bring Z into -16 to +16, and X,Y would be -2 to bring it to -18 to +18.

Ted

[anthonyduhamel]

Ted,

Hard-Iron comes from permanently magnetized components located on the pcb plus magnetometer imperfections. I suggest you to read AN4247 & AN4246 (files attached) for furthers information on hard-iron distortions.

Yes you can subtract the hard-iron vector to the raw measurement vector to remove the bias but be careful about "static" calibration. Hard-iron vector can shift over time due to temperature changing, external magnetic fields disruptions coming close to the sensor and sensor position changes. 

FYI, FXOS8700CQ comes with a self magnetic calibration feature (m_acal field to set in M_CTRL_REG1 plus m_raw field to clear in M_CTRL_REG3).

Thanks,

Anthony

[tedhuntington1]

Thanks for the 2 links Anthony- it seems like a lot of calculating for my particular application. Of course I would prefer the device to not have hard-coded offsets and to calibrate itself to whatever location it is in.

One of these sensors will be on each segment of a two-leg humanoid robot- so I don't think they will be turning more than a few degrees in any dimension- hopefully that will be enough for the m_acal autocalibration to work. 

Ted