Operation of the BME690 in long-term Measurement

Hello everyone,

For my project, I want to record changes in the atmosphere over a longer period of time (2 or more weeks) in order to measure the ripening process of fruit. I have therefore conducted numerous experiments on the long-term stability of the BME 690 and have not found a good operating/measurement configuration for stable long-term measurement.

I tried different sensor boards (Polverine and 8x Shuttleboard) and configurations. All measurements were taken in a sealed container with a fan to eliminate environmental influences but still have a slight air flow in the container.

I took measurements with normal air and with a banana as a test object.

The main effect that I don't understand or that concerns me is that in all experiments, the gas resistance increases in a very unpredictable way after a certain amount of time. The increase and the time at which the increase begins vary greatly from sensor to sensor and are therefore difficult to eliminate through calculations.

I would expect a flat resistance curve when measuring only air in a sealed environment, and a slight decrease in resistance over time when measuring bananas, as the ripening gases (mainly ethylene) should concentrate in a sealed container over time.

I tried the following configurations:

offical 8xShuttleboard was configured with attached file

Configuration_Fruit.bmeconfig.txt
1.88KB

general settings for the Polverine Board:
temp_prof[10] = { 100, 150, 200, 250, 300, 350, 400, 0, 0, 0 };
mul_prof[10] = { 10, 10, 10, 10, 10, 10, 10, 0, 0, 0 };

conf.filter = BME69X_FILTER_SIZE_127;
conf.odr = BME69X_ODR_NONE;
conf.os_hum = BME69X_OS_16X;
conf.os_pres = BME69X_OS_16X;
conf.os_temp = BME69X_OS_16X;

1) Contiunos Measurement:
The Polverine board continuously measures the temperature profile without pausing. I have attached some graphs to illustrate the increase. For clarity, I will only upload the graph for the 400 °C step. If you need more information, please let me know. Measurement time: 21 hours

Continuos_Bosch_Polverine.pdf
284.39KB
Continuos_Polverine_solo.pdf
101.16KB

2) Cyclical measurement:
The shuttleboard configuration was not changed, but the Polverine board was interrupted several times to allow it to cool down. The setting used was 105 minutes of measurement and 15 minutes of pause (sleep mode). Measurement time: 68 hours

2_105Meas_15break_Bosch_Polverine.pdf
577.76KB
2_105Meas_15break_Polverine_solo.pdf
156.06KB

3) Cycled Measurement:
The shuttleboard configuration was not changed, but the Polverine board was given several breaks to cool down. The setting used was 15 minutes of measurement and 15 minutes of break (sleep mode). Measurement time: 17 hours

3_15Meas_15break_Bosch_Polverine.pdf
241.05KB
3_15Meas_15break_Polverine_solo.pdf
48.67KB

4) pause after each Temperatur Cycle:
The shuttleboard configuration was not changed. Polverine was paused (sleep mode) for 3 seconds after each temperature cycle.
Measurement time: 17 hours

4_Paused_Bosch_Polverine.pdf
235.43KB
4_Paused_Polverine_solo.pdf
58.01KB


These experiments lead me to believe that the BME690 is not capable of continuous measurement over a longer period of time. Is there any experience or official statement on how best to perform long-term measurements with the BME690?

All diagrams were measured with a banana in the container.

Even if the sensor pauses for a few seconds after each cycle, there is a slight deviation.

Experiments 3 and 4 were carried out in parallel with two Polverine boards. However, these were the shortest measurements.

It is interesting to note that this effect is stronger at higher temperatures in experiment 4 and weaker or non-existent at lower temperatures. However, this could also be due to the shorter measurement period?

I only use the API for the Polverine Board and not BSEC, as the sensor is more configurable with the API.

Another interesting observation is that all sensors exhibit this effect to varying degrees. I did not configure anything on the shuttleboard except with the bmeconfig file, and yet some sensors exhibit the effect more strongly than others. What is the best way to deal with this, or is there an explanation for it?

Thank you in advance 😃
Luca

5 replies