Breakout Garden, Bosch BME680 sensor, high value differences!


just installed refering Pimoroni tutorial … all works fine. I want to see the raw values from the bme-sensor and the result was astonishing. My references are a DHT22 and a commerical part. Comparing the values after 24 hour burn in time:
Temperatur plus 8 degress (BME 30 / DHT22 21 / Wetherstation 21,2)
Humidity minus 20 % (BME 32 % / DHT22 55% / Wetherstation 55%)
Pressure munis 50 hPa (BME 973,06 / Commerical Wetherstation 1015)
Gas varied from 102.000 and 48.07???

Interesting thing … changing the oversampling values has only a real small effect?

Is it possible, that there are such big differences are possible?



All of these values can vary pretty widely and are difficult to validate since a sensor thermally coupled to the Pi (a really good source of heat) will always see warmer temperatures than a sensor that’s well isolated from a heat source.

I’d say it’s fairly possible to explain away the other results, too, since the sensor will be very sensitive to location and other external factors, but finding the precise reason why a particular reading varies is difficult.

The “gas” sensor measures volatile organic compounds, which is a whole bunch of random disparate stuff. I’d expect it to have a pretty high noise floor in anything but a sealed environment, and you should expect to see meaningful readings from very close readings of organic compounds (stick it over a glass of alcohol for example) and you can generally use a low-pass filter to remove the noise and get an estimate of air quality.

Running a gas sensor in a typical household is, I suspect, a lot like trying to run an earthquake detector on the back of a moving truck. I think we don’t do enough to explain how to use sensors, and manage expectations from them because we don’t really know all that much ourselves!

I made a couple of basic weather station Pi’s based around the Pi foundation Sense Hat. I quickly discovered that the heat given off by the Pi affected the temperature reading. It would read high. I got around this by mounting a Proto Hat between the Pi and the Sense Hat. That put the two farther apart and added a bit of shielding between the two. With no case the temp reading were accurate, close enough to ambient for me anyway.
My second one was a portable and even with vent holes in the case the temp read artificially high. I ended up mounting a BME680 to the outside on the case. Its connected to the Pi via jumper wires. Temps read accurate now. I suspect the heat was also affecting the Sense Hat humidity readings.


many thx of that info an the tipps … will be work on that :-)