You can see that it took about a day from the Alpstuga to start tracking almost exactly to the MS06, despite having what is considered a more inferior sensor.
The two then stayed locked together for three days, but now, the alpstuga is drifting slightly, usual recording lower than the MS06.
Tbh, I was expecting more of a variance, given the lower quality sensor (The MS06 has a Sensirion SCD40, and the ikea an Sensirion SEN63C which is less supposedly less accurate).
I honestly don’t see any differences.
They are not laboratory or calibration sensors.
They are sensors that can differ by a few percentages from what is normal. Even any temperature sensor can vary. Each sensor can be more or less susceptible to humidity and temperature.
It is not important to have precise values but it is important that they can have trends of constant values. A value that could be 1500 or 1200 ppm is identical from a practical point of view and still indicates a high value. However, it would be totally wrong to have different values such as 1000 and 10 ppm.
even if you take two Alpstugas they will read different values and will never be totally aligned with values that will change.
Alpstuga is using a SEN63C module, which in turn is using an STCC4 sensor for CO2 sensing. The STCC4 has a specified accuracy of ±(100 + 10% of measured value) ppm, while the SCD40 of the MS06 has a specified accuracy of ±(50 + 5% of measured value) ppm.
At CO2 levels around 500 ppm the Alpstuga could be 150 ppm off in one direction, while the MS06 could be 75 ppm off in the other - in total 225 ppm apart in an extreme situation. At 1,000 ppm the accuracies are 200 ppm and 100 ppm respectively.
The accuracy specifications are likely based on a 95% confidence level and theoretically they could be further apart.
I agree that it is noteworthy that the signals are very similar Tuesday and Wednesday, but then drifts apart Thursday. Did you continue to collect data for the sensors while they were sitting close together? Did signals meet again at some stage?
If the sensors are not subjected to rapidly varying temperature and humidity, the influence from these parameters ought to be adjusted for by the on-chip compensation circuits built into both sensors.
Would you happen to have humidity and temperature data from Tuesday, Wednesday and Thursday, then an analysis could be made to see if the CO2 delta is correlated to any of these.
