[M-TT43] Brand-new scope of coupling geophysics being established by infrasound and associated waves

[MTT43-03] Evaluation of prototype miniature microphones for observing infrasound

*Ryouichi Nishimura¹, Taira Itoh², Dai Adachi², Yasutaka Nakajima², Kenichi Kidokoro², Toshiyuki Hasumi², Yôiti Suzuki¹ (1.National Institute of Information and Communications Technology, 2.RION Co., Ltd.)

Keywords:Infrasound, Miniature microphone, MEMS, Parallel observation

Experimental comparison was carried out among MEMS pressure sensors: DPS310, MBL1000-8, and prototype miniature microphones from the viewpoint of performance in observing relatively high frequency infrasound such as generated at volcanic eruptions. These sensors were put in a box set up in the Kagoshima University Research Forests, which is located approximately 12 km southeast from the Minamidake summit crater. MBL1000-8 is an infrasound measurement device on which eight BOSCH BME280 modules are mounted. It records the mean value of the eight sensor outputs, thereby suppressing internal noise signals relatively large at a high frequency region. On the other hand, DPS310 is a brand new MEMS pressure sensor. Many vendors supply modules to access it through a serial bus using I2C or SPI. We employed a module supplied by Seeed Studio Co. Ltd. in the experiment. According to the data sheets, relative accuracy is ±12 Pa for BME280 and ±6 Pa for DPS310. Other features are almost the same between the both. The miniature microphones were experimentally made by RION Co. Ltd., based on MEMS microphones initially designed for hearing aids and modified so as to measure infrasound. They can capture the sound up to 1 kHz so that outputs were preliminarily processed by an RC low-pass filter with the cut-off frequency of 15.9 Hz and then sampled at the sampling rate of 32 Hz with adafruit ADS1116 to acquire the data. In case of large level input, two microphones of different sensitivities were employed; one was –39.3 dBV/Pa and the other –67.1 dBV/Pa. Because of the different DC biases, outputs of these microphones were sampled with different input ranges. As a result, the minimum quantization resolution for the low sensitivity microphone was four times coarser than that of the high sensitivity. We compared the power spectrum density of the data obtained during the night because less influence of human activity was expected. The data length was six hours from the midnight. To determine the day with the least wind influence, we compared standard deviation of the data obtained everyday since the beginning of the experiment, January 19, 2021. Although the least variance was observed on 28 January, the data on 31 January was chosen for the analysis because it showed the second least variance and odd signals with an amplitude of 5 Pa often occurred on 28 January. The data was segmented into frames of 20 minutes with 50% overlap, and the mean power spectrum density was calculated. The results were approximately –12 [dB re 1μPa²/Hz] for the high sensitivity microphone and 24 [dB re 1μPa²/Hz] for the low sensitivity microphone at 10 Hz. According to literature, the electronic noise level of DASE MB2000, which is used IS of CTBTO, is approximately –8 [dB re 1μPa²/Hz]. Therefore, it suggests that one of the tested miniature microphones could capture infrasound with as the same noise level as MB2000. Besides, the results of MEMS pressure sensors were approximately 70 [dB re 1μPa²/Hz] for DPS310 and 81 [dB re 1μPa²/Hz] for MBL1000-8. The MEMS pressure sensors showed relatively large noise level at this frequency, but they are considered effective in measuring very low frequency infrasound, for example, a duration of over two minutes, such as observed when a tsunami happens, because its frequency is too low for microphones to capture. As described before, signals of approximately 5 Pa in amplitude were sometimes observed in the experiment. These signals were come not from the electronic circuit but from the air because all the tested sensors captured them. However, such signals were not observed in the data captured by other sensors located in different directions around Sakurajima. Therefore, the origin of these signals has not been determined yet. Possible reasons are an artificial noise made in a nearby factory or a wind generated by a truck passing by. This work was partly supported by JSPS KAKENHI Grant Number 17K01351 and 19H02396.