Closed coelner closed 3 years ago
I have a SGP30 and CCS811 at the same spot and did some graphs. The sensor is on the window board and the bay is separated with a paper curtain from the room itself. The curtail has got a gap to the board about 3cm. Measurement is every second.
Nice graphs.
I agree that values like eCO2: 8421 ppm eTVOC: 14433 ppb
make no sense.
I also cannot explain the differences between CCS811 and SGP30. The CCS811: errstat=899=--vhXmrwF--AD-iE
does not raise my trust in the values.
Please do note that these sensors are TVOC sensors not CO2 sensors (the CO2 is computed from the TVOC, assuming the sensor is indoor).
The SGP30 (I use the sparfun library) is also not very trustworthy, sometimes it keeps the eCO2 value at 400 which can't be true. (even for a MOX sensor, which should change the resistance by blowing over it).
These are relative sensors.
If you take a bottle, spray a bit of alcohol in it, hang your sensor in it, close the bottle, and then switch on the sensor, it will happily report clean air (assuming you do not use the set baseline feature).
It also means that the 'zero' value is arbitrary. what the sensors typically do is take the value at power on (again, assuming you do not use the set baseline feature), and use that as reference. Typically assuming clean air, and mapping that to 400 (SGP30), 450 (CCS811), 500 (iAQcore).
By the way, I heard once, blowing on the sensor is a bit abusive. I forgot why. Maybe you cool the heater too much and it needs to step current, maybe it was the moist in your breath. Anyhow they told me not to blow, but to use a marker pen or breath some breath freshener on it.
Good luck
It is not directly related to this library itself, but maybe the right sport to discuss this:
I put the sensor out of the window to get the baseline. The measurement is done every second but the values get summed up for a minute. First of all, there occurs some errors because of maximum resistance was reached. After this errors the sensors was placed inside. The absolute values are far beyond useful. The sensor itself is burned-in (at least for a week) and this specific print-out was done after three hours running.