Accelerometers play the leading role in vibration monitors designed to identify failing equipment or structures well before mechanical breakdown actually occurs. By integrating key functional components, microelectromechanical systems (MEMS) sensors offer engineers a range of options needed to accommodate the broadest range of monitoring requirements. Engineers can select from an extensive range of high G MEMS sensors designed for vibration monitoring from Analog Devices, Freescale, and STMicroelectronics to implement vibration detection and health-monitoring systems.
Vibration analysis offers insight into equipment performance, enabling predictive maintenance programs that can identify and correct problems well before they require more expensive repairs and extensive downtime. Normal machine operation that results in characteristic vibration patterns and abnormal operation tends to exhibit even more distinctive vibration signatures. At or near failure, machine vibration can become pronounced. For this level of equipment failure detection, traditional piezoelectric sensors such as the MiniSense 100 Vibration Sensor from Measurement Specialties (see Figure 1) can cost-effectively serve to trigger alarms and auto-shutdown procedures. These devices are low-cost cantilever-type vibration sensors where deflections in the beam generate a piezoelectric response – offering high sensitivity but are limited to low frequency vibrations.
Figure 1: Traditional piezoelectric sensors such as this Measurement Specialties MiniSense 100 vibration sensor offer cost-effective devices for detecting vibrations associated with serious equipment problems. (Source: Measurement Specialties)
Well before actual mechanical or structural failure, however, subtle changes in higher-frequency vibration patterns can signal potentially serious problems such as bearing wear, mechanical misalignment, and more. In turn, high G accelerometers are required to detect changes in higher frequency vibration signatures, providing data needed to predict eventual failure modes and enable directed preventative maintenance. By identifying potential failures well before operations become impaired, this type of predictive health monitoring helps companies avoid damage to costly equipment and mitigate, or ideally eliminate, downtime costs.
For vibration monitoring in equipment or structures, accelerometers are carefully positioned in critical areas, translating the linear motion of vibration into a proportional voltage. Manufacturers offer families of sensors that detect vibration along single or multiple axes on the x, y, and z planes (see Figure 2) – enabling engineers to match the sensor to specific vibration modalities of interest.
Figure 2: Available accelerometers detect vibrations along one or more axis, enabling engineers to build vibration monitors that can precisely identify vibration location in monitored equipment. (Source: STMicroelectronics)
Available parts include on-chip signal processing capabilities or offer simple interfaces with microcontrollers for further processing (see Figure 3). After downstream sampling and time domain analysis provides initial results, data can be passed to more sophisticated frequency-domain analysis subsystems required to detect failure-related changes in vibration. In turn, these results can drive application-specific responses such as notifying service personnel, limiting motor speed, or initiating safety shutdown procedures.
Figure 3: Designed to simplify integration with microcontrollers, advanced accelerometers require few external components. (Source: Freescale Semiconductor)
The requirements inherent in this type of health monitoring application drive the need for MEMS devices that combine both higher-frequency response and integrated functionality. Freescale addresses these requirements with its MMA series of silicon capacitive, micromachined accelerometers. Offering comprehensive combinations of acceleration range, sensitivity and x, y, z-axis detection, each member of this family integrates on-chip signal conditioning circuitry, a four-pole low pass filter, and temperature compensation.
Although manufacturers use slightly different methods for implementing the acceleration-sensing cell within the sensor, Freescale’s approach illustrates the general approach. MMA family devices combine a surface micromachined capacitive-sensing cell and a CMOS signal conditioning ASIC within a single integrated circuit package. Formed through conventional semiconductor mask and etch fabrication processes, the sensing cell is a mechanical device that can be functionally modeled as a set of beams where a movable central mass, or proof mass, moves between fixed beams in response to changes in inertia (see Figure 4).
Figure 4: A typical accelerometer measures vibrations by sensing movement of a central proof mass relative to fixed beams. (Source: Freescale)
Electrically, the sensing cell plates form a pair of capacitors and as the central mass deflects from its rest position in response to acceleration, the distance between the beams change, resulting in a change in capacitance. In Freescale’s MMA devices, the ASIC generates acceleration data by comparing the difference between the sensing cell capacitors measured using switched-capacitor techniques.
Noise sources abound in vibration monitoring applications, necessitating the use of sharp cutoff filtering methods. Freescale’s MMA family devices include an on-chip, 4-pole switched-capacitor Bessel filter designed to preserve pulse shape. Along with its filtering advantages, the use of this switched-capacitor technique also eliminates the need for external resistors and capacitors to set the cutoff frequency.
The ability to integrate semiconductor logic with the sensor permits implementation of self-test features particularly critical in embedded mission-critical applications. Freescale’s MMA devices include a fourth plate in the sensing cell for self-test purposes. During self-test mode, the device applies a calibrated voltage across the self-test plate, causing the moveable plate to deflect, resulting in an expected output voltage.
For its LIS331HH sensor, STMicroelectronics provides a self-test function that is accessed through the self-test bit in a control register. When the self-test bit is set, the device applies a force to the sensor to simulate a specific acceleration, allowing engineers to confirm that the sensing cell and sampling circuitry are functioning as expected.
Analog Device’s ADXL001 single-axis accelerometer, among other members of Analog’s accelerometer families, offers a self-test function that moves the sensing cell plates to mimic an external acceleration force, resulting in an expected change in VOUT. As with other devices in this class, this approach to self-test ensures operation along the entire signal change from sensing cell to output pin.
The ADXL001 also features a differential sensor and a fully differential signal path to help reduce noise sources that complicate vibration monitoring (see Figure 5). For example, without careful attention to noise management, engineers can find that spikes in the power supply can be interpreted as vibrations. In this case, Analog Device’s differential signal processing helps eliminate power supply noise before it can affect actual vibration measurements.
Figure 5: The ADXL001 uses a differential sensor and a fully differential signal path to reduce the effect of noise in vibration monitoring. (Source: Analog Devices)
As engineers look to extract more subtle information from vibration analysis, issues such as noise management, self-test, temperature compensation, and more can significantly complicate design efforts. The availability of highly integrated specialized devices such as Analog Devices’ ADIS16227 iSensor, help ease these concerns. The ADIS16227 combines full tri-axis acceleration measurement with signal processing capabilities able to perform both time and frequency domain analysis.
Real-time frequency analysis is essential for timely notification of the kinds of equipment defects, imbalances, and misalignments revealed through vibration data. Typical vibration monitoring designs use external frequency-analysis subsystems – an approach that reduces or even removes options for real-time response to vibration events. By immediately performing frequency analysis at the source, integrated vibration-monitoring devices such as the ADIS16227 can more quickly apply further analysis to gain further detail on the source and nature of the vibration.
ADI’s ADIS16227 performs a 512-point fast-Fourier transform (FFT) for each axis and provides FFT averaging to reduce noise. This kind of per-axis analysis is critical for isolating the specific source of vibration energy. The device includes a 16-record FFT store to allow users to observe frequency changes over time and provides an alarm function across six programmable spectral bands, each with programmable upper and lower limits.