A geophone-based and low-cost data acquisition and analysis system designed for microtremor measurements

MicDAC consists of a developed graphical user interface and external hardware that includes amplifiers, low-pass and notch filter circuits, clipper circuits, voltage converter circuit, external analog-to-digital converter, and Arduino Uno board. The graphical abstract of MicDAC is shown in Fig. 1. The designed external hardware is connected to the computer via a USB. It utilizes the USB port as a power supply. Thus, no external battery is required.

3.1

 Hardware implementation

The vertical and horizontal geophones manufactured by the EGL Company were used to measure three-component microtremor data. They have some characteristic features such as a natural frequency of 4.5±10 % Hz, typical spurious frequency greater than 150 Hz, damping of 0.6±5 %, and open circuit sensitivity of 28.8±5 % V m−1 s−1. The external view, amplitude response, and phase response of these geophones are shown in Fig. 2. There are also geophones with lower frequency on the market, but the price of these instruments increases exponentially with decreasing frequencies. Moreover, the cost of high-quality broadband seismometers can reach a mid-level car price. For this reason, in this paper a low-cost hardware–software device is presented to both measure and interpret three-component microtremor data. The estimated costs of the electronic components of the designed external hardware in this study are presented in Table 1. The total cost of the hardware, including the sensors, is approximately EUR 255.

Download

The output of a geophone consists of two poles, and its output voltage is too weak to be recorded without amplification. The voltage difference between these poles can be measured by using various operational amplifiers. The first stage of this signal conditioning circuit consists of differential-input and single-output INA122 amplifiers from Texas Instruments (1997). These amplifiers are instrumentation amplifiers with very important properties such as low-noise (60 nV ∕ Hz), high-quality, and rail-to-rail output, wide power supply range (single supply: 2.2 to 36 V, dual supply: -0.9/+1.3 to -18/+18 V), low offset voltage (250 µA max), and low quiescent current (60 µA).

In this study, symmetrical power supplies (±5 V) were used to supply these amplifiers. The Arduino Uno board can only provide 3.3 and 5 V positive outputs. For this reason, the ICL7660 integrated circuit was used to obtain the negative power supply (−5 V) from the positive power supply (+5 V). The pin connections of the ICL7660 voltage converter and INA122 amplifier are shown in Fig. 3. It can be set to four different gain levels (46, 56, 60, and 66 dB) for each channel thanks to the DIP switch and resistors connected to pins 1 and 8 of the INA122 amplifiers.

Download

The next step in the external hardware is a passive RC (resistor–capacitor) low-pass circuit, which is used to attenuate the high-frequency components in the amplified signal and to avoid the aliasing phenomenon. The cutoff frequency of the low-pass filter was set at approximately 22 Hz because the corresponding frequency range in microtremor studies is 0–20 Hz. In the last stage of the signal conditioning circuits, the Twin-T notch filters and clipper circuits were used to remove the 50 Hz interferences and to clip the negative voltage in the output signal, respectively (Fig. 3). Instead of the internal 10 bit analog-to-digital converter on the Arduino board, the MCP3208 12 bit external analog-to-digital converter was preferred to ensure higher-resolution signals.

Download

3.2

 Software implementation

The MicDAC sketch (Arduino uses the term sketch for a program) was compiled with Arduino version 1.8.8 and stored as a file called MicDAC.ino. This sketch digitizes seismic data with a 200 Hz sampling frequency and transfers the digitized data to the computer via USB. The second program (MicDAC-GUI), developed by using .NET Framework 4.5.2 in C# language, is a user-friendly and Windows operating-system-based software. The MicDAC-GUI consists of data monitoring mode, data recording mode, and analysis tools. It detects the available COM ports automatically. The data monitoring mode is used to display three-component microtremor data in real time and to test the geophones before starting the recording operation. In addition, this mode is also used to adjust the offset needed to see signals symmetrically through the 10 K potentiometer, a component of the external hardware (Fig. 3). The data recording mode is used to record three-component microtremor data during the desired time.

The MicDAC-GUI allows for performing records with a maximum of 180 min. When the recording operation ends, the user is alerted with an alarm and the temporary data are saved into a file called “datam.txt”. The microtremor data are stored in a text file. This data file consists of three columns: V (vertical), NS (north–south), and EW (east–west) components. The recorded signal values vary in the range between 0 and 4095 because the analog-to-digital converter is 12 bit. The coordinates of the survey area and descriptions are stored in a file with info extension.

The MicDAC-GUI allows many operations such as low-pass filtering, tapering, windowing, and smoothing on the recorded data. The desired signals can be selected thanks to the windowing feature of the MicDAC-GUI. The window length (10.24, 20.48, 40.96, 81.92, and 163.83 s) is a user-defined parameter. The time duration of the analysis process increases or decreases depending on the number of enabled time windows and data length. The frequency distribution of each window is displayed separately. This feature gives an idea about the windows that will be used in the analysis. Finally, the horizontal-to-vertical spectral ratios for each selected time window are calculated. The calculated H ∕ V spectral ratios are also numerically displayed using a grid component. After the analysis process ends, the parameters (cutoff frequency of low-pass filter, taper ratio, length of time window, bandwidth of Konno–Ohmachi smoothing function), raw data, low-pass-filtered data, Fourier amplitude spectra data, smoothed Fourier amplitude spectra data, and H ∕ V spectra data are saved into a text file with the “soln” extension.

Download

3.3

 The reliability and accuracy tests of MicDAC

Three tests were performed in order to demonstrate the accuracy and precision of MicDAC: (1) channel consistency tests using synthetic and real data; (2) internal noise measurement test; and (3) comparison of the characteristics and frequency contents of recorded signals using the proposed system and a commercial microtremor measurement device.

Download Print Version | Download XLSX

The channel consistency test was performed to evaluate the time–amplitude differences for each channel. Firstly, a 1 Hz sinusoidal signal of 15 mV amplitude generated by the model FG-8002 function generator manufactured by EZ Digital was connected to each input channel, and these signals were recorded over a period of 60 s with a 46 dB gain. After that, the differences of the recorded signals for each channel were calculated. As shown in Fig. 4a, the error percentages of the difference signals (channel 1–channel 2, channel 1–channel 3, and channel 2–channel 3) are lower than 1 %. The H ∕ V ratios were calculated using the same sinusoidal signals recorded by three channels and presented in Fig. 4b. The H ∕ V ratio will be equal to 1 because the same sinusoidal signals were applied to each channel input.

Secondly, a channel consistency test was conducted with real sensors. For this purpose, the same Vertical Geophone, placed on a table, was connected to the inputs of each channel and the ambient noise was recorded over a period of 180 s (Fig. 5). In this test, the calculated Fourier spectra were utilized to demonstrate the frequency contents of three channels. The calculated Fourier spectra and H ∕ V ratio are shown in Fig. 6a and b, respectively.

Download

Download

In the next test, the internal noise of MicDAC was recorded during a time period of 60 s with a 200 Hz sampling frequency (Fig. 7). The noise levels of three channels were observed as approximately ±3 counts (7.3242 mV, 9.39×10-5 m s−1).

Finally, MicDAC was compared with a triaxial digital seismograph called GeoBox manufactured by SARA Electronic Instruments. GeoBox is an instrument designed especially for recording ambient seismic noise, and its different versions with sensors of 2 and 4.5 Hz are available on the market. In this study, the SR04HS model with 4.5 Hz sensors of was used to make a comparison with MicDAC. The signals recorded simultaneously by MicDAC and GeoBox are shown in Fig. 8a and b. Log-MT software was used to monitor and record the signals with GeoBox. Two different gain levels, 46 and 60 dB, were used in the signals recorded with MicDAC. As a result of this comparison in the time domain, a good correlation between the recorded signals using GeoBox and MicDAC was observed. This similarity was also observed in their frequency spectra (Fig. 8c). In addition to the comparisons in the time and frequency domains, their H ∕ V spectral ratios were also obtained using the Geopsy software (Fig. 9a and b). The obtained H ∕ V peak frequencies and amplitudes are given in Table 2. A good correlation was observed between the calculated H ∕ V peak frequencies and amplitudes. The main reason for the differences in the amplitudes of the H ∕ V curves at frequencies below 2 Hz is related to the electronic design of GeoBox. GeoBox has an electronic architecture that can obtain a flat band wider than the natural band of the geophone embedded in the instrument. Therefore, it obtains better sensitivity at low frequencies.

Download