There are many ways to add more intelligence to industrial systems, including matching analog and digital components with sensors featuring edge and cloud artificial intelligence (AI). Due to the diversity of AI methods, sensor designers need to consider multiple conflicting requirements, including decision delay, network usage, power consumption/battery life, and AI models suitable for the machine. This article will focus on introducing the application of intelligent AI wireless motor monitoring sensors and the related solutions launched by ADI.
Motor health monitoring is carried out using wireless industrial sensors
Condition monitoring (CbM) on robots and rotating machinery (such as turbines, fans, pumps and motors) can record real-time data related to the health and performance of the machines, thereby enabling targeted predictive maintenance and optimized control. Targeted predictive maintenance carried out in the early stage of a machine's life cycle can reduce the risk of production downtime, thereby enhancing reliability, significantly saving costs and improving production efficiency in the factory workshop. Condition monitoring (CbM) of industrial machines can utilize a series of sensor data, such as electrical measurements, vibration, temperature, oil quality, acoustics, magnetic fields, as well as process measurements like flow and pressure. However, vibration measurement is by far the most common because it can provide the most reliable indication of mechanical problems, such as imbalance and bearing failure.
Currently, wireless industrial sensors on the market typically operate at extremely low duty cycles. Users set the sleep duration of the sensors. After the sleep period ends, the sensors are awakened to measure temperature and vibration, and then the data is sent back to the user's data aggregator via radio signals. Commercially available sensors typically claim a battery life of five years, based on collecting data once every 24 hours or multiple times every 24 hours.
In most cases, the sensor is in sleep mode for over 90% of the time. Take ADI's Voyager4 sensor as an example. It operates in a similar way, but uses edge AI anomaly detection (with the MAX78000 AI microcontroller) to limit the use of radio. When the sensor is awakened and measures data, only when the microcontroller detects abnormal data will it send the data back to the user, thereby triggering machine diagnosis and maintenance and extending the service life of the motor. By using AI at the edge, battery life can be extended by at least 50%.
Voyager4 is a wireless condition monitoring platform developed by Analog Devices (ADI), designed to help developers quickly deploy and test wireless solutions for machines or test equipment. Motor health monitoring solutions such as Voyager4 are widely used in robots, as well as in rotating machinery like turbines, fans, pumps and motors.
The working principle of the Voyager4 sensor system
The Voyager4 sensor, in combination with the ADXL382 three-axis 8 kHz digital micro-electromechanical system (MEMS), is used to collect vibration data. Firstly, the raw vibration data is transmitted to the MAX32666 low-power Bluetooth ® (BLE) processor. The data can be sent to the user via BLE radio or USB. These raw vibration data are used to train edge AI algorithms with the MAX78000 tool.
Use the MAX78000 tool to synthesize the AI model into C code. The edge AI algorithm is sent to the Voyager4 sensor via BLE wireless (OTA) updates and stored in memory using the MAX78000 processor with an edge AI hardware accelerator. After the initial training phase of Voyager4, ADXL382 MEMS data can be transmitted along the path. The MAX78000 edge AI algorithm will predict whether the machine has malfunctioned or is operating normally based on the collected vibration data. If the vibration data is normal, there is no need to use the MAX32666 radio, and the MEMS will return to sleep mode. However, if the predicted vibration data is incorrect, an abnormal vibration alert will be sent to the user via BLE.
In the hardware system of Voyager4, the ADXL382 adopted is a 3-axis MEMS accelerometer with low noise density and low power consumption, featuring an selectable measurement range. This device supports ± 15g, ± 30g and ± 60g measurement ranges as well as a wide measurement bandwidth of 8 kHz. The ADG1634 is a single-pole double-throw (SPDT) CMOS switch, which is used to route the MEMS raw vibration data to the MAX32666 BLE radio or the MAX78000 AI microcontroller, with the BLE microcontroller used to control the SPDT switch. Several other peripherals are connected to the MAX32666, including the MAX17262 battery gauge for monitoring battery current and the ultra-low power ADXL367 MEMS accelerometer. The ADXL367 is used to wake up BLE radios from deep sleep mode during high-vibration shock events. In motion-activated wake mode, its power consumption is only 180 nA. The BLE microcontroller can transfer the raw data of ADXL382 MEMS to the host via BLE or USB of FTDI FT234XD-R.
The Voyager4 sensor adopts the MAX20335 power management integrated circuit (PMIC), which features two ultra-low quiescent current step-down regulators and three ultra-low quiescent current low-dropout (LDO) linear regulators. The output voltages of each LDO and step-down regulator can be enabled or disabled independently, and each output voltage value can be programmed via I2C with default pre-configuration. The BLE processor is used to enable or disable a single PMIC power output for different Voyager4 working modes.
In training mode, the BLE microcontroller must first notify its presence in the BLE network and then establish a BLE connection with the network manager. Then, Voyager4 transmits the raw ADXL382 MEMS data via the BLE network to train AI algorithms on the user's PC. After that, the Voyager4 sensor returns to deep sleep mode. In normal (AI) mode, BLE radio signaling, connection and transmission functions are disabled by default. The MAX78000 will wake up regularly and run AI inference. If no anomaly is detected, Voyager4 will return to deep sleep mode.
The Voyager4 evaluation kit (EV-CBM-VOYAGER4-1Z) launched by ADI includes multiple components (LED, pull-up resistor), making it convenient for customers to conduct evaluations. These components generate a deep sleep current of 0.3 mW on the LDO1OUT voltage rail. The average power consumption of the Voyager4 evaluation suite is calculated based on the time interval between events in deep sleep, training, and normal/AI modes.
The following will further introduce the functional characteristics of these related devices to you.
AI microcontrollers enable neural networks to operate at ultra-low power consumption at the edge of the Internet of Things
The MAX78000 is an AI microcontroller that adopts an ultra-low power convolutional neural network accelerometer. This new type of AI microcontroller enables neural networks to operate at ultra-low power at the edge of the Internet of Things, combining high-efficiency AI processing with the proven Maxim ultra-low power microcontroller. With this hardware-based convolutional neural network (CNN) accelerator, even battery-powered applications can perform AI inference with power consumption at the microjoule level. The MAX78000 is an advanced system-on-chip that integrates an Arm® Cortex®-M4 core with an FPU CPU and achieves efficient system control through an ultra-low power deep neural network accelerator. This device adopts an 81-pin CTBGA (8mm x 8mm, 0.8mm pitch) package.
The MAX32666 is a low-power ARM Cortex-M4 FPU-based microcontroller with Bluetooth 5, suitable for wearable applications. The design of this new-generation UB MCU is intended to meet the complex application requirements of battery-powered and wirelessly connected devices. This intelligent controller is equipped with a larger memory among similar products and adopts a memory architecture that can be expanded on a large scale. The device adopts wearable power technology, which can operate for a long time, be durable and capable of withstanding high-level cyber attacks. This device is packaged in a 109-pin WLP (0.35mm pitch) and a 121-pin CTBGA (0.65mm pitch).
The ADXL382 is a low-noise, low-power, wide-bandwidth, 3-axis MEMS accelerometer with selectable measurement ranges, supporting ± 15g, ± 30g and ± 60g measurement ranges. The ADXL382 offers industry-leading noise levels, enabling precise applications with minimal calibration. Its low noise and low power consumption characteristics allow for accurate measurement of audio signals or heart sounds even in high-vibration environments. The multi-functional pin names of ADXL382 can be referenced solely by their functions related to the Serial peripheral Interface (SPI) or I2C interface, or by their audio functions (pulse density Modulation (PDM), I2S, or Time Division multiplexing (TDM)). The ADXL382 is available in a 14-pin LGA package of 2.9mm x 2.8mm x 0.87mm.
A complete solution for wireless asset status monitoring using edge AI
Voyager4 can use edge AI for wireless asset status monitoring. It employs three-axis digital output MEMS sensors, including ADXL382 and ADXL367. This design also includes the MAX32666 BLE and MAX78000 AI microcontrollers. Flexible and PCB-space-saving PMIC power supply devices have been added as load switches to enhance the energy-saving effect of wireless sensors. Each Voyager4 kit includes a BLE 5.3 adapter with an antenna. Voyager4 uses BLE, so it is compatible with any PC with a Bluetooth radio. However, to ensure the best performance and range, it is recommended to use an adapter when communicating with Voyager4.
The ADG1633/ADG1634 is a 4.5Ω RON, three/four-channel single-pole double-throw (SPDT), ± 5V /+ 12V /+ 5V /+ 3.3V switch. Both the ADG1633 and ADG1634 are single-chip industrial CMOS (iCMOS®) analog switches. It is equipped with three or four independent and selectable single-pole double-throw switches respectively. All channels are equipped with first-open and last-close switches to prevent instantaneous short circuits when opening or closing the channels. The ADG1633 (LFCSP and TSSOP packages) and ADG1634 (LFCSP package only) provide EN inputs to enable or disable devices. The iCMOS structure can ensure extremely low power consumption, so these devices are very suitable for portable battery-powered instruments.
The ADXL367 is A MEMS accelerometer with nanometer-level power consumption, 3-axis, ± 2g /± 4g /± 8g digital output. At an output data rate of 100Hz, it only consumes 0.89 µA, and in the action-triggered wake-up mode, it only consumes 180 nA. Unlike accelerometers that achieve low power consumption by using power duty cycle, the ADXL367 does not aliasing the input signal through under-sampling but instead samples the full bandwidth of the sensor at all data rates. The ADXL367 is available in a 2.2mm x 2.3mm x 0.87mm package.
The MAX17262 is A 5.2µA, ModelGauge m5 EZ single-cell battery level gauge with built-in current detection. It is the battery level gauge with the lowest IQ in the industry, featuring an integrated current detector and the ModelGauge m5 EZ algorithm, eliminating the need for battery characteristic analysis. The MAX17262 can monitor a single battery cell, integrates an internal current detector, and can detect pulse currents up to 3.1A. The IC is optimized for battery metering with capacities ranging from 100mAhr to 6Ahr. The MAX17262 features a tiny, lead-free, 0.4mm solder pitch, 1.5mm x 1.5mm, 9-pin wafer-level package (WLP).
The MAX20335 is a small PMIC for lithium-ion systems, equipped with an ultra-low IQ voltage regulator and battery charger. It features an optimized power management solution and supports 7 x 24-hour monitoring systems for wearables and IoT. The MAX20335 battery charging management solution is ideal for low-power wearable applications. The device includes a linear battery charger with an intelligent power selector and a variety of power-optimized peripherals. The MAX20335 adopts 36 solder balls, 0.4mm solder ball pitch, and a wafer-level package (WLP) of 2.72mm x 2.47mm.
Conclusion
The microcontroller with an integrated AI hardware accelerator provides wireless sensor nodes with superior decision-making capabilities and longer battery life. By using AI at the edge, battery life can be extended by at least 50%. The modal analysis included in the vibration sensor can accelerate the sensor development cycle and ensure the capture of high-quality vibration data from the monitored assets. The Voyager4 wireless condition monitoring platform launched by ADI, combined with related component solutions, will be your best assistant to add intelligence to industrial systems.
Your message must be between 20-3,000 characters!