Here is a quote from BME680 PDF:
Moreover, the software algorithms handle humidity compensation, baseline as well as long-term drift correction of the gas sensor signal.
An experiment was conducted several times and the same behavior was observed.
Additional details - humidity is increased with ultrasonic humidifier filled with clean water (passed a simple filtering process - Brita Marella Water Filter). The BSEC library is version 1.4.6.x . There is another sensors at the same location that is measuring TVOC and eCO2 and it is not impacted by the increased humidity (Sensiron SGP30, humidity data is passed to it and the SGP30 is applying proper compensation for it). The room is under uninterupted ventilation (done by an air recuperator) and there is nobody inside while the the experiment is conducted.
The PMSx003 (PMS7003) sensor is measuring dust particles and the water droplets are detected as such. This is the reason for the spike of the PM2.5 and PM10 particles. The CO2 readings are from an NDIR CO2 sensor - MH-Z19.
This should show the metrics data collected while the experiment was conducted: https://snapshot.raintank.io/dashboard/snapshot/MPpb2fGQWqZ95TtA1XvkyucWGuWWk6sB
Solved! Go to Solution.
Thank-you for your request, this has been reported to our experts.
To provide the best feedback about this output, would it be possible to share the complete datalog of raw sensor signals (timestamp, temperature, pressure, humidity and gas resistance) ideally for the whole duration of BSEC's history?
I'll try to do that, but I'll re-start a sensor with the latest BSEC (184.108.40.206) and after a few days I'll reproduce the issue. The sensor provided as example above is working for several (6+) months and I'm not sure if I can export all the requested data.
I suppose this would take about a week to be done.
Just to report some update on the data collection. With the latest BSEC library version this seems to be working just fine. I'll try for a few more days and most likely I'm going to close this as resolved.
The compensation coefficients in general depends on the age of the sensor. We provide config strings that contain the relevant coefficients for different stages during the early use of the sensor, typically at 4d and 28d. You could use either of these depending on how long the device has been used.
Hope this helps.