Yes, I preheated the sensor for 5 days. The results are the same like at the beginning of burning. Looking at the documentation of the Winsen GM sensors, it looks that the sensors must give a concentration, closer to the real! OK, I agree that they are not super accurate, but for example, the GM102B sensor measures NO2 in the range 0,1-10ppm. In our case, there is nothing real with these reading of hundreds of ppm! It looks that the grove multichannel sensor is not calibrated at all! And I still cannot understand why all sensors on the multichannel sensor react together when there is some gas. Stupid think… I use a lot of different sensors, also different type of Winsen sensor for CO and the results are perfect. It is possible, that these values are not in PPM, but PPB. I will make some tests.