Offset and Slope factor issue

[octane321]

hi there, and first of all many thanks to your work - i love it!

But i´m struggling getting 10 sensors equalized in temp and humidity reading.
Trying to use only offset, only slope or both combined gives me worst equalized sensors than before..
So i dived a bit into the calculations and found out, that "calibration" with offset and slope is really annoying and need a high effort if you want to do it the right way.

My personal opinion for temp. "normalization" and i´m still on research (same again for humidity):

Step A: Correct deviation on linear behavior

• One will need measurements from all n sensors at a defined low temp (let say 5°C) e.g. in the fridge
• One will need measurements from all n sensors at a defined high temp (let say 40°C) e.g. in the oven
  (for best results log the measurements for 15-30min after the readings have been stabilized and calculate the average)
• Pick one sensor as reference (same sensor for temp & humidity!)
• Calculate the range of all sensors
• Calculate the deviation of all n sensors range-difference in relation to reference sensor
• Calculate the new slope factor and set it in all sensors

Result
All sensors should have quiet the same changes when temp and hum changes
But the measured temp. are still all different

Step B: Correct the offset

• One will need again need all sensors measurements at a defined low OR high temp like above
• Calculate the temp deviation of all n sensors range-difference in relation to the reference sensor
• Add the deviation to the default offset (-45°) and set up all sensors

Request / Idea
It would be much more easier to take only a high and low measurement per sensor and put that into a calibrate linear function with a Set Point value for low and high value
Do you see any chance getting this implemented into the firmware?

Thanks in advance and have a great time!

[octane321]

should be calculated like this:

Tcali =
((T_high_meas - T_low) / (T_high - T_low)) * x + (T_high_meas - (((T_high_meas - T_low_meas) / (T_high - T_high_meas)) * T_high))