Calibrating measurement system by offsetting readings with a mathematical function?

[davey31415]

Hi,

I am sensing an angle from two linear potentiometers.  Several factors associated with the pot mounting are adding non-linear errors to the system over time.  In the past bolts and screws have been tightened and loosened to reduce these errors but the process is rather time consuming. 

I have been considering teaching LabVIEW the correct angles with a sinebar and then performing a non-linear regression on acquired data using the taught points.  Has anyone else here tried something like this in the past? 

I'm looking for some general advice.  If there are any reasons not to pursue such a method I'd like to learn about them before investing time in the idea.

Thanks,

Dave

[davey31415]

Maybe the DAQ forum would be a better place for me to ask the calibration-specific question.

Can anyone recommend functions/examples that I should inspect to perform this?

I haven't performed a regression in LV before but I'm hoping this can be accomplished easily.  I'm thinking maybe store the calibrated values in a file and load them into global variables when the program starts?

Dave

[NAKeel]

Hi Davey,

You calibration idea sounds like a good one.  You could load calibration data from a file, and that should work. You could possibly use the non-linear curve fit VI to do your non-linear regression.  Take a look at the VI and see if that is something you can work with.

Nick Keel 

Applications Engineering 

National Instruments

[davey31415]

Thanks NAKeel.  Where might I find the non-linear curve fit VI?  I've tried searching examples on both 7.1 and 8.5 installations but have not come accross it.

Regards,

Dave

[NAKeel]

Hi Dave,

The VI is located in the Functions Palette»Mathematics»Fitting»Non-Linear Curve Fit.vi.  It comes as part of the Full Development System, so you would not see it if you have the Base Development System.   That might explain why you are having trouble finding it.  If you don't have that VI, you would need to create the regression with basic math VIs, which could be a difficult task.

Nick Keel 

Applications Engineering 

National Instruments