@nanocastro thats a nice find! This is also cool (on the same site) http://www.hackair.eu/hackair-cardboard/ because everything is better with cardboard. This uses:
A clean, dry, empty drinks carton (e.g. Tetrapak)
Some petroleum jelly (e.g. Vaseline)
Some household string
Some tape
A counter weight (e.g. a battery, coin, paper clips)
A macro lens for your smartphone
After putting out the Vaseline coated cardboard for 24 hours and then take a picture which s uploaded to their app.
Iām not sure that he is working fairly. Rather, it shows a tendency. I will try calibration. But, interestingly, nitrogen dioxide, carbon monoxide are negatively correlated. I think why the increase in nitrogen dioxide leads to a decrease in all other indicatorsā¦?
I wrote about this on https://github.com/Seeed-Studio/Mutichannel_Gas_Sensor/issues/8
Hi Andriy
Temperature and humidity interfere with the response of metal oxide sensors as MICS. You can check this references (1, 2)
They donāt use the same MICS models but I think the interference problems remain. They use a multivariate linear regression or machine learning with data from high-cost/reference stations to calibrate the sensors in the field. We are now trying to do something like that (co-location with a reference stations)
Hope it helps.