Accurate laboratory testing of low-frequency triaxial vibration sensors under various environmental conditions

Triaxial vibration sensor are widely used used in various application. Recently, low-cost sensors based on micro electro mechanical system (MEMS) technology are also becoming more widely adopted. However, their measurement accuracy can be affected by environmental factors such as temperature. In this study, we developed an environmental testing system integrated with a triaxial vibration exciter. The system can reproduce long-stroke, low-frequency triaxial vibrations – such as those caused by huge earthquakes – under temperatures ranging from $-30~^\circ\mathrm{C}$ to $+80~^\circ\mathrm{C}$. Using this system, the measurement accuracy of vibration sensors can be evaluated under different environmental conditions. The system provides highly accurate reference measurements using a laser interferometer and reference accelerometers that are primarily calibrated within the system. The overall accuracy of the reference vibration measurement is estimated to be approximately 0.23~%. Based on these reference measurements, we investigated the accuracy of earthquake observations using a MEMS accelerometer as a demonstration. The system configuration and testing procedures are presented in this paper.

💡 Research Summary

This paper presents the development and demonstration of a novel Triaxial Environmental Test System (TETS) designed to accurately evaluate the measurement performance of low-frequency triaxial vibration sensors, such as accelerometers and seismometers, under controlled environmental conditions. The core motivation stems from the fact that the mechanical and electrical properties of these sensors, including increasingly popular low-cost MEMS-based types, can be influenced by environmental factors like temperature, potentially compromising measurement accuracy in field applications.

The developed system ingeniously integrates a large thermostatic chamber (capable of -30°C to +80°C) mounted directly onto the table of a long-stroke triaxial vibration exciter (400 mm horizontal, 100 mm vertical stroke). This configuration, unlike the method prescribed in ISO 16063-34 where the exciter armature is inserted into a fixed chamber, allows the entire chamber and the Sensor Under Test (SUT) inside to be vibrated. This key design innovation enables the reproduction of long-stroke, low-frequency triaxial vibrations—essential for simulating large earthquake motions—while simultaneously subjecting the SUT to precise temperature control.

To provide a highly accurate reference for evaluation, the system employs a two-step measurement approach. First, a primary calibration of three reference servo accelerometers is performed traceably to national standards using a laser interferometer. This calibration is conducted at various temperatures to meticulously characterize the temperature dependence of each accelerometer’s complex sensitivity (magnitude and phase). A first-order low-pass response model is fitted to this data, with its parameters (low-frequency sensitivity ratio and cutoff frequency) expressed as cubic functions of temperature, allowing for accurate interpolation within the tested range (-20°C to +75°C). Subsequently, these temperature-calibrated reference accelerometers are placed inside the chamber alongside the SUT. During a triaxial test, they provide the reference vibration waveform, with their pre-determined temperature-dependent sensitivities applied for correction.

The estimated uncertainty of a single-axis reference vibration measurement is approximately 0.11%, stemming from primary calibration uncertainty, fitting residuals, and temperature measurement errors. When considering triaxial measurements, the cross-axis sensitivity of the reference accelerometers (about 0.2%) contributes additional uncertainty, leading to an overall estimated uncertainty of about 0.23% for the triaxial reference vibration provided by the system.

As a practical demonstration, the system was used to evaluate a commercial MEMS accelerometer (ADXL355). The mainshock waveform of the 2016 Kumamoto earthquake was applied to the SUT at three different temperatures: -15°C, +23°C, and +56°C. The output waveform from the MEMS sensor was compared against the high-fidelity reference waveform generated by the calibrated servo accelerometers. The results showed waveform deviations of up to about 3%, which significantly exceeded the reference uncertainty, indicating that these errors originated from the characteristics of the MEMS sensor itself (e.g., sensitivity error, nonlinearity). Notably, the deviation pattern was consistent across all tested temperatures. Furthermore, the triaxial waveforms were used to calculate the instrumental seismic intensity scale (I_JMA), yielding nearly identical values at different temperatures, suggesting the MEMS sensor’s intensity calculation performance was robust against temperature variations for this test.

In conclusion, the research successfully establishes a comprehensive methodology and a capable hardware system for the metrologically traceable evaluation of triaxial vibration sensors under combined environmental and dynamic stress. The TETS overcomes the limitations of existing methods by enabling accurate, long-stroke, triaxial testing within a controlled climate, providing a valuable tool for validating and improving the reliability of sensors used in critical applications like seismic monitoring.