current is supposed to be constant during measurement. Current scheme is just like a DMM. a current measuring resistor produces a voltage drop that is buffered and read by ADC.

Yes the measurement is not linear and it was a great surprise to me. I first measure voltage at open circuit and set offset and then measure a current near maximum range capacity and set gain value but surprisingly it is not linear and when I get from a lower range to upper range I read different values though I had just calibrated both resistors with the same DMM and with the same current source. I have no idea why.the difference is small but not zero.

Yes the problem is expanding, how should I get a reference current or reference resistor for auto calibration? If I can have a reference resistor then why bother with auto calibration I use the precise resistor for current measurement.

but if I can solve auto calibration problem maybe I can solve drift problem, because I can auto calibrate at power on and then temperature remains constant during measurement and also I can solve long term drift.