Calibration of noise sensor

[marteinnm]

We are looking into starting a noise sensor project in Iceland. Our idea is still in the general conceptualization phase.
One of the problems we can currently identify is the calibration of the microphone. A noise engineer working on our team is worried about the detoriation of the quality of the data with time. Has this been addressed elsewhere here?
We would like to use the IM72D128 due to the waterproofing of the microphone on the DNMS - T4 V1.7 with the NodeMCU ESP8266.
I have posted the same topic on the sensor.community forum as that is where we plan on publishing the data. https://forum.sensor.community/t/calibration-of-noise-sensor/4638/1

[hbitter]

Hi marteinnm,

In our experience it is good to start with the IM72D128 microphone. We had a few failures with the ICS-43434, mostly due to humidity.
Calibrate the microphone:

• We recommend mounting the microphone in a 1/2" tube so that it can be checked with a standard 1/2" calibrator. The IM72D128 microphone from Watterott is also mounted in a 1/2" tube. https://shop.watterott.com/Microphon-IM72D128V01-with-1-2-tube
• The datasheet for the IM72D128 microphone gives a range of 0f ± 1dB, so there is no urgent need for calibration to measure ambient noise.
• If you wish, you can enter a correction value from the measurement with a calibrator in both the NodeMCU and Raspberry Pi versions.

Longer term detoriation:
This is a good point and has not been addressed systematically. It would require some statistical analysis to track the readings over time. We haven't had the time or resources to do this yet. However, we do have some experience of field measurements over several years in all weather conditions. These show that the microphones do not normally have significant deviations. However, it is important that the microphones are mounted freely in the air and protected by a foam cap. Water must not get directly into the small microphone opening. If this happens, the readings will immediately become so implausible that it will be noticed, or the unit will fail completely. We have microphones in use that have been running for over 2 years without any problems. It would be interesting to see if this would be the case in Iceland's sometimes extreme weather conditions.

But it is an open system and you can add to it yourself. There are a couple of additional APIs for communicating the readings besides Sensor.Community that are specifically designed to do this sort of thing. In this case, I would recommend using a Raspberry Pi Zero 2W instead of the NodeMCU. The Raspberry Pi can even log the data to the SD card.

I hope this answer helps you with your project.
Feel free to contact me or Ewald directly: helmut.bitter@t-online.de or ewald.thoma@t-online.de

Regards,

Helmut