http://www.yoctopuce.com/EN/article/how-...co2-sensor
I know this article does not refer to the MH-Z19, nonetheless it is interesting to see how the ABC function works. It seems as if mine is the "B" version and therefore has the ABC function built in. This does not seem to be a good idea if we are trying to measure atmospheric CO2. The ABC function works on the premise that the sensor will at some stage in the day sit in an atmosphere of 400ppm. It assumes air which has been drawn in from outside to be at this level. It is this "normal" that the calibration is done against.
I think these sensors are really designed for HVAC systems in for example schools where concentrations of humans can significantly increase CO2 levels in a room. The sensor picks this increase up and triggers the ventilation system to let in fresh air. At night (for example) the sensor recalibrates itself on the assumption that the air in an empty room is at 400ppm.
If our sensor is left outside it will recalibrate itself to an assumed 400ppm at regular intervals irrespective of the actual CO2 ppm levels.
Having said all this, this does still not explain the large ranges we are experiencing. If our sensors do have the ABC function we should see more normal ranges even if the air has fooled the sensor into calibrating 400 ppm in an atmosphere that is actually at for example 420 ppm.
If the sensor does not have ABC one would think it would have been calibrated at factory in a controlled environment, and that the sensor would drift off that calibration, for the reasons outlined in the link above over time, but not day to day.
My thought is that the sensor is sensing as a proportion of the 0-2000 range and reporting it as a proportion of the 0-5000 range, ie multiplying the range by 2.5. This is real stab in the dark and I am not sure if this is even possible.
I know this article does not refer to the MH-Z19, nonetheless it is interesting to see how the ABC function works. It seems as if mine is the "B" version and therefore has the ABC function built in. This does not seem to be a good idea if we are trying to measure atmospheric CO2. The ABC function works on the premise that the sensor will at some stage in the day sit in an atmosphere of 400ppm. It assumes air which has been drawn in from outside to be at this level. It is this "normal" that the calibration is done against.
I think these sensors are really designed for HVAC systems in for example schools where concentrations of humans can significantly increase CO2 levels in a room. The sensor picks this increase up and triggers the ventilation system to let in fresh air. At night (for example) the sensor recalibrates itself on the assumption that the air in an empty room is at 400ppm.
If our sensor is left outside it will recalibrate itself to an assumed 400ppm at regular intervals irrespective of the actual CO2 ppm levels.
Having said all this, this does still not explain the large ranges we are experiencing. If our sensors do have the ABC function we should see more normal ranges even if the air has fooled the sensor into calibrating 400 ppm in an atmosphere that is actually at for example 420 ppm.
If the sensor does not have ABC one would think it would have been calibrated at factory in a controlled environment, and that the sensor would drift off that calibration, for the reasons outlined in the link above over time, but not day to day.
My thought is that the sensor is sensing as a proportion of the 0-2000 range and reporting it as a proportion of the 0-5000 range, ie multiplying the range by 2.5. This is real stab in the dark and I am not sure if this is even possible.