ST (STMicroelectronics) has officially launched the IIS3DWB10IS industrial-grade MEMS vibration sensor, which integrates AI computing power on the chip. It is designed for equipment condition monitoring scenarios. This device combines the wide bandwidth vibration acquisition capability with ST's self-developed second-generation intelligent sensing processing unit (ISPU 2.0), aiming to provide equipment manufacturers with a digital alternative to piezoelectric sensors. 

Of great reference value: Predictive maintenance solutions are gradually becoming smaller in size and moving towards edge-end sensing nodes. For industrial and automation R&D engineers in Europe, this sensor can not only simplify the hardware design for vibration monitoring, but also perform data processing locally on the sensor itself. 

MEMS sensors are calibrated and achieve equivalent performance to piezoelectric sensors. 

The IIS3DWB10IS can detect vibration and shock signals up to 200g, with a sampling frequency of up to 10kHz or higher. STMicroelectronics stated that the noise floor of this device is as low as 35µg/√Hz. In scenarios of high-end equipment condition monitoring, the sensing performance has been very close to the indicators of traditional piezoelectric sensors. 

The operating temperature limit of this device can reach up to 125℃, and it can be adapted to harsh industrial conditions such as motors, pump bodies, drivers, and rotating machinery. Leveraging the inherent advantages of the MEMS architecture, this product comes with a digital interface, is smaller in size, has lower power consumption, and the difficulty of integrating the mechanical structure and the circuit has also significantly decreased. 

Simonne Ferri, Executive Vice President of the MEMS Division at STMicroelectronics, said: "This industrial MEMS vibration sensor features the dynamic range and bandwidth indicators required for high-end applications, further extending STMicroelectronics' technical advantage of 'sensor-native digital processing'. With the built-in ISPU 2.0 and the new hardware accelerator, it can perform signal calculations and AI inference at high speed, accurately identifying equipment wear characteristics while reducing processing latency and overall power consumption. This new generation of condition monitoring sensor has been born. It is also the first solution with actual competitiveness and capable of replacing piezoelectric sensors: lightweight, easy to install and design, with extremely high measurement accuracy, and with a power consumption level sufficient to support long-term battery-powered operation." 

Edge AI on-chip computing power for condition monitoring 

The on-chip integrated ISPU 2.0 processing unit is positioned closely to the sensing element to perform signal processing and AI inference operations. According to official data from STMicroelectronics, this unit can achieve a computing power of 40 MIPS and 40 MFLOPS, with processing performance being four times that of the previous generation; the data transmission rate between the sensor and the MEMS core has also increased to six times the original speed. 

It is equipped with a complete software ecosystem. It encapsulates the commonly used function libraries for vibration monitoring, including standard algorithms such as fast Fourier transform (FFT), digital filtering, envelope analysis, vibration intensity determination, and anomaly detection. Designers can delegate more computing tasks to the sensors for execution, effectively reducing transmission delay and the overall system power consumption. 

Andrea Torcelli, the Chief Technology Officer of Bonfiglioli S.P.A., commented: "The IIS3DWB10IS perfectly meets the requirements of the target industry and harsh working conditions. With its extremely large dynamic range, wide bandwidth, wide temperature operating characteristics, as well as advantages such as easy mass production integration, streamlined circuitry, and controllable costs, we were able to replace the traditional piezoelectric sensing solution that had been in use for many years. In addition, the integrated ISPU 2.0 processor enables complex signal operations and high-speed AI inference to be executed locally at the sensing end, allowing the system to make more intelligent coordinated responses." 

Device hardware packaging configuration 

The IIS3DWB10IS is equipped with a 2048×80-bit FIFO (First-In-First-Out) register and an on-chip integrated temperature sensor. It adopts a 4.5mm×4.5mm×1.5mm 16-pin LGA package with a side that can be wettable for soldering, and is suitable for automated optical inspection (AOI) production lines for quality control.