Non-Linear characteristic calibration

which is the best method to straighten non linear characteristics of analog sensors. For example some temperature sensors are not linear, so therefore they need to be calibrated. Supposing you had true physical temperatures measured at 10, 20, 30, 40, 50 and 60 degrees, but the analog input is 10.5, 21, 33, 45, 52 and 59 degrees. (Here we can see that the sensor value inititally exeeds the acutal value slightly and then the error gradually increases only to decrease again towards the higher range). How could this be corrected/calibrated (to straighten the line out if represented graphically). Any suggestions would be highly appreciated.

Maybe have a look at the spline-curve:
You can add various points to fit a curve...
But you have to be careful: the "normal" version puts for every calculation one entry in a log file on your node-red device. Maybe a problem in long-term-usage.
In github is an issue on that: I deleted some (one?) line in one file, to switch this of.

1 Like

Thanks, that may be just the node I need, will try it out tomorrow. The challenge is to try to get these sensors to give the same result:

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.