I’ve seen lots of discussion on how to make better fits when calibrating OD but it seems to me as though maybe the sigmoid curve isn’t the best way to to fit the points. Why was is chosen? Has anyone tried calibrating with different curves, such as exponential, or 1/x?
Sigmoid was chosen because historically we didn’t calibrate ODs past ~.8 with 1 photodiode. If the cut off is there, the sigmoidal fit works pretty well (OD measurements with two photodiodes? - #5 by bgwong). But after adding an additional photodiode at 90 degrees, we found the range is much higher. You can feel free to add more types of fit. Just make sure to modify the code appropriately.
The easiest way is to change the evolver.py transform_data function and have it transform the incoming data with your own custom fit: