Multiple BME680 Sensor Mounted On Same Location & While Comparing there is offset in the values?

[msalmankhan]

Hello Everyone , I am conducting some tests on the BME680 sensors  and their respose to the enviroment, So  for the test purpose I have placed two of the sensor exactly on the same location , and others are placed at some distance from them, I have let the sensors to run for 3 days,after three days , I have burnt some plastic material, in order to create smoky enviroment, the enviromentwas quite smoky, 
What I found some strange things , the two sensors placed next to each other  performs differently , Which I will  show  you on the graph , The two sensors placed almostly on the same location , then it should responeded to the enviroment in exactly the same way, but from the graph it behaves differently I just want to know the reason why this different behaviour is shown by the sensor that are placed exactly on the same location.

Note: I am using BSEC Library , and the sensors are operating in LP Mode, IAQ Configuration is : generic_33v_300s_4d, but because I am using LP mode of the sensor , the interval between readings is : 3 Seconds.

Two sensor placed next to each other , marked with white Color pencile, The shape of the graph are different

As you can see the sensor marked with white colour , are placed next to each other , but the graph shows different behaviour, This should be of same shape.

NOTE: This is the Static IAQ values reporting by the sensors.

Any help in regarding this will much be appreciated Thanks.

[NickXML]

Could we get some suggestions/answers here?

To recap, we're running multiple sensors with BME680s and using BSEC for calibration. IAQ accuracy is typically 3, with brief "blips" to 2.

And yet, as we have shown, we are getting significantly different values from sensors physically located next to each other.

Isn't this precisely what BSEC is supposed to prevent? Some reasonable margin of error is fine, but a difference of 50-100% or more between adjacent sensors is not okay.

One example captured earlier today registered 213 IAQ on one sensor and 576 IAQ on the IC less than 3" away.

Another shows one sensor at 70.5, a " Good" rating based on the chart in the 680 datasheet. The neighboring sensor less than 3" away reads 222, or " Heavily Polluted" at exactly the same time.

Either the air is good or it is polluted. It cannot be both.

1/ Is this expected behavior?
2/ if yes, what is the point of BSEC calibration?
3/ if no, what are possible causes that could lead to such wild variations?

The BME680 is promoted as a low cost but reliable sensor for indoor air quality. Our data currently suggests otherwise. Can we please get some meaningful feedback to resolve this???

I've attached yet more examples showing big discrepancies between two sensors located next to each other.

[BSTRobin · Accepted solution]

Hi msalmankhan,

Suggest to burn-in and run-in activity before use BME sensor.
For burn-in process, quickest way to use DC mode running 12hrs for the band new sensor. For run-in process, suggest 2hrs for gas measurement stabilization.

[BSTRobin]

Hi msalmankhan,

Did you have further test and feedback?

[BSTRobin]

Since there was no feedback for a long time, we close this topic first. You could update feedback in the future if you have.