Is it possible One time calibration of 9 axis sensor?

cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

Is it possible One time calibration of 9 axis sensor?

Jump to solution
738 Views
harshpatel
Contributor IV

Hello

 

I have used sensor fusion library from ST. In that we need to calibrate sensor once after that  no need to calibrate every power cycle. Such a facility required when device is fix after installation. So is there such functionality available in freescale sensor fusion library?

 

Thanks & regards

Harsh Patel

Labels (1)
1 Solution
512 Views
michaelestanley
NXP Employee
NXP Employee

Harsh,

The Version 7.00 library includes the capability to store MagCal and sensor trim parameters to flash and then reload them on startup.  However, we've never claimed a one time calibration to be sufficient.  The primary reason for this relates to the temperature coefficient of the sensors.  They are probably negligible for accelerometer and gyro for most applications.  However the Z-axis of our magnetometer has a different physical structure then the X and Y axes.  I've never characterized how well the three axes track, which would impact the quality of the stored magnetic calibration.  For this reason, I've always recommended that MagCal be run as a continuous background task, even when we've used stored coefficients from flash to speed the startup process.

Mike

View solution in original post

4 Replies
512 Views
harshpatel
Contributor IV

Hello Mike

We are using KL25 sensor fusion library in our project. We are applying  8 Zig movement on every power cycle. But in future requirement we need only one time calibration type system. well i will try it in KL25.

Thank you

Regards

Harsh

0 Kudos
512 Views
harshpatel
Contributor IV

Hello Mike

Thank you for your inputs...

So Is there any other way that  no need of 8 zig movement on every power cycle to calibrate sensor?

Regards,

Harsh

0 Kudos
512 Views
michaelestanley
NXP Employee
NXP Employee

Harsh,

Magnetic calibration requires a collection of data points taken at different orientations.  These are used (via a full least squares fit methodology) to compute the calibration coefficients.  There's no way around that, as it relates to the very nature of hard/soft iron interference - independent of the sensor itself.  Storing the previous calibration as I mentioned will probably get you in the ballpark right away.  But how close is related to the temperature differential from last to current run.  And as I've said, I've not had the opportunity to characterize the magnitude of those effects.  My best advice is to simply try it using a FRDM-K64F or FRDM-K22F with one of our sensor shields.  The ISSDK option within the SDK for those two boards include sensor fusion projects that will work out of the box and let you experiment.

Regards,

Mike

513 Views
michaelestanley
NXP Employee
NXP Employee

Harsh,

The Version 7.00 library includes the capability to store MagCal and sensor trim parameters to flash and then reload them on startup.  However, we've never claimed a one time calibration to be sufficient.  The primary reason for this relates to the temperature coefficient of the sensors.  They are probably negligible for accelerometer and gyro for most applications.  However the Z-axis of our magnetometer has a different physical structure then the X and Y axes.  I've never characterized how well the three axes track, which would impact the quality of the stored magnetic calibration.  For this reason, I've always recommended that MagCal be run as a continuous background task, even when we've used stored coefficients from flash to speed the startup process.

Mike