In industrial applications, it’s often necessary to measure motion related quantities such as pressure, velocity, and acceleration. However, the deployment of sensors that can measure these parameters has been hampered by cost, as well as the difficulty of getting accurate and consistent measurements due to unforeseen mechanical and electronic factors.
This is unfortunate, as the actual values of these parameters, as well as relative changes in those values, can provide insight into basic system operation, impact detection, impending malfunctions, excess vibration, and unauthorized movement. As such, it’s important that designers consider taking a second look at their sensing options.
True, some solutions were viable only in limited, specialized applications where cost and complexity were secondary to necessity. However, this situation has changed dramatically with the development of low-cost, small, low-power MEMS-based ICs for motion and acceleration sensing.
This article will briefly describe some Industrial Internet of Things (IIoT)) application scenarios where designs benefit from data on motion related parameters. It will also explain the not-so-obvious difficulties of obtaining accurate and consistent measurements with these devices, which is due to both electronic and mechanical issues.
Following a brief review of the underlying physics principles, it will look at relevant products, their specifications, and application areas. It will then raise critical issues regarding mechanical placement and mounting, and look at reference designs/evaluation boards that can help developers quickly realize the end application.
The physics of sensing
Students in classical physics learn the basic equation of motion: Force = mass × acceleration, where acceleration is defined as a change in velocity (its time derivative), and velocity is a change in position (the time derivative of position). Acceleration is expressed numerically either as m/sec2 in the SI (metric) system, or very commonly in units of “g”, where “one g” is the standard acceleration value due to Earth’s gravity, 9.8 m/sec2.
Acceleration is a vector, and full characterization of acceleration requires a trio of sensors for the orthogonal x, y, and z axes. An inertial measurement unit (IMU) combines a three-axis accelerometer with a three-axis gyroscope which indicates changes in orientation, and is used for navigation and guidance. However, there are also many industrial applications for accelerometers, and these may need sensing along only one or two axes.
MEMS-based accelerometer ICs use a “proof mass” which is suspended between capacitive sensing plates via a cantilever or corner-spring arrangement; as the device accelerates, the proof mass “stays back” due to its inertia. This microscopic change in the relative position of the proof mass with respect the plates is capacitively sensed, passed to a charge amplifier, further amplified and filtered, and then presented as an analog or digitized output.
MEMS accelerometers are available with full-scale ranges as low as 1 g, and as high as 100 g and higher. The lower ranges are often a good match for many IoT applications, while the higher ones are for specialty situations such as rocket launches, vehicle crashes, and other high impact events.
Devices span wide g range, applications
The “breakout” MEMS accelerometer application was the automotive airbag sensor/trigger, which quickly replaced the mechanical “ball in tube” impact sensor. However, as acceleration is difficult to measure and quantify, many potential applications could not be addressed and/or considered. Things have now changed, as IoT applications are leveraging the availability of MEMS-based units to address situations that could not be previously, and were therefore ignored.
Vibration monitoring is the largest use of MEMS accelerometers in Industrial IoT. These include “stationary” acceleration situations where the target object remains fixed in place, such as a motor or machine, where vibration data can indicate impending bearing failure and other problems. Acceleration is also used with untethered or loosely tethered devices to provide physical security against theft and removal. It is also used to detect device drops, where an object such as a laptop PC free falls to the ground. These two basic modes determine the required g range and frequency response, and affect mounting and mechanical issues, discussed further below.
MEMS accelerometers which are well suited for IoT applications tend to be less accurate and precise than those for navigation/guidance, but as a consequence they are also smaller, lower cost, lower power devices. This makes them an especially good fit for embedded IoT applications which operate from a long-life non-rechargeable battery, or are powered by energy harvesting.
For example, the ADXL344 from Analog Devices is a 3-axis, digital output, low g MEMS accelerometer with a selectable measurement range (±2 g, ±4 g, ±8 g, and ±16 g) and bandwidth (Figure 1). It can measure the static acceleration of gravity in tilt sensing applications, as well as dynamic acceleration resulting from vibration motion or shock.
Figure 1: The ADXL344 from Analog Devices is a tiny 3-axis MEMS-based accelerometer that also includes many user settable alarms, monitoring modes, and reporting scenarios. (Image source: Analog Devices)
Among the IoT friendly features on the ADXL344 are its built-in motion detection functions based on user settable thresholds. Using these capabilities, it can determine and then report values via SPI or I2C digital interfaces if the object to which it is attached has been moved. This greatly reduces the communications burden and power demands. It can also indicate the opposite situation, such as when normal motion has stopped (machine breakdown or power failure), rather than responding to repeated queries, or sending regular but redundant updates. Despite the 3-axis functions and many internal features, it is housed in a tiny LGA package measuring just 3 x 3 x 0.93 mm.
For high g industrial applications, such as fast moving or high rate vibration of equipment, as well as high impact situations, the single-axis AIS1120SX and dual-axis AIS2120SX from STMicroelectronics offer a full-scale range of ±120 g. The 14-bit device includes the full signal conditioning and conversion chain, including a charge/voltage converter, charge amplifier, and a 2nd order sigma-delta analog-to-digital converter (ADC) (Figure 2). The digital core includes user selectable filtering (400/800/1600 Hz), compensation and interpolation, control logic, and an SPI protocol interface, all in a plastic SOIC8 package.
Figure 2: The single-axis AIS1120SX and dual-axis AIS2120SX from STMicroelectronics include multiple calibration and error compensation functions, and target high g, severe impact applications with its ±120 g range. (Image source: STMicroelectronics)
Another useful feature of the AIS1120SX/AIS2120SX is slow and fast offset cancellation. This compensates for the signal offset, which is unavoidable due to physical placement and orientation of the IC, as well as electronic effects. The fast offset cancellation is immediately used after power-on, while the slow mode is for continuously running offset cancellation.
Calibration, self-test, evaluation are key
Both the Analog Devices and STMicroelectronics devices are much more than just acceleration sensors to provide basic digitized outputs. In order to make them both power and operationally efficient at the system level (meaning that they minimize data handling and power consumption) they contain a significant amount of digital circuitry to select filters, establish thresholds, set and report alarms, and manage interface protocols to the post processor. These accelerometers are well suited for Industrial IoT applications which are remotely located, often with wireless links which must not be activated unless needed in order to minimize power consumption.
The downside of this situation is that these devices require a significant amount of planning and programming by the user to set up the many operational registers and modes, and to ensure that these accelerometers will do what is needed in the system wide context. They are not basic “plug in” sensors with a simple analog or digital output with which the host MCU must deal. Instead, they have internal processing functions and state machines which require initialization, configuration, and management in order to realize their potential by making use of their many features.
Calibration and self-test are also areas where these sensors have an advantage when compared to sensors for other physical variables such as temperature or pressure. Certainly, when using a temperature sensor such as a quality thermocouple, there is often no need for calibration if the analog front end (AFE) is well designed. On the other hand, there is no easy way to test the thermocouple element itself since that would require applying known temperatures, which is difficult at best (an “open-circuit” thermocouple is easy to detect, of course). A comparable test and calibration dilemma exists for other sensors, such as for pressure or position, since designers have to provide a known external stimulus and then assess the results.
Fortunately, there is an easy way to implement a high confidence self-test and calibration process for these MEMS accelerometers. In normal operation, the sensing plates around the proof mass measure minute changes in capacitance that are due to motion of that mass. However, complementary action is also possible: the plates can be precisely charged so that they cause the mass to move, and their resultant displacement is measured. In this way, these accelerometers are very attractive for IoT applications, especially deeply embedded ones, since they can periodically invoke a self-test and calibration mode and assess their own performance.
Mechanical issues: significant design impact
In general, where and how a sensor is placed in a product is a non-trivial issue: think of microphones, light sensors, temperature sensors, and pressure sensors. IoT accelerometers have major placement and mounting issues.
First, there is the issue of sensor alignment with the motion axis of interest. The primary axis of the accelerometer must be as closely aligned physically as possible, of course, but near perfect alignment is not always achievable. Therefore, many accelerometer systems have an initial calibration phase in which the sensor output “at rest” is assigned the zero reading. However, using the off-axis reading as zero degrades actual range span and measurement by the sine of the angular error.
More challenging is the mechanical mounting. There are many wrong and a few right ways to attach the accelerometer to the pc board, if that is where it will be mounted. If poorly or inadequately supported, the accelerometer will experience undesired, misleading vibration due to natural, unavoidable, undamped resonances of the board assembly (Figure 3).
Figure 3: Where the accelerometer IC is mounted to the pc board (green), and where that board is supported, are critical factors to ensuring meaningful and accurate data; here, some of the wrong places are shown. (Image source: Analog Devices)
For these reasons, vendors strongly suggest putting the sensor at a “hard” mounting point such as a fully supported board corner, and even using multiple such mounting points if possible. Another tactic is to use a thicker (but costlier) pc board to reduce the effect of board and system resonance. Regardless of approach, the objective is to ensure that any PCB vibration sensed by the accelerometer is above its mechanical resonant frequency and is therefore not “seen” by the accelerometer.
What about applications where the accelerometer is not mounted on the main pc board, but instead is attached to a workpiece or object of interest, and then wired to the remaining electronics? One solution is to have a small, separate pc board to which the accelerometer IC is soldered, with wire leads to the main board.
This small board, which can be encapsulated for additional protection, is then attached to the object being monitored using screw mounts, or even industrial strength double-sided tape in some cases, depending on the operating environment and expected magnitude of the acceleration. Other mounting solutions depend on the constraints of the application, the ingenuity of the design team, industry and regulatory standards, generally accepted best practices, and available enclosures.
Evaluation kits: more than just circuits
It’s now standard practice for vendors to offer evaluation kits, reference designs, and test suites for their ICs and related components. In most cases, this is because the ICs have complicated interfaces or sensitive signal paths, such as RF devices. However, for evaluating accelerometers, the need for evaluation boards is not driven by these factors, but instead by the basic challenges of setting up controlled stimuli to be measured. These allow the design team to test and modify their algorithms to meet the application requirements.
For example, the Texas Instruments DRV-ACC16-EVM evaluation module for three-axis accelerometer testing is primarily designed for haptic systems, but can also be used for vibration related tests (Figure 4).
Figure 4: The Texas Instruments DRV-ACC16-EVM evaluation module for multi-axis testing includes a board for mounting the accelerometer and connecting an oscilloscope, as well as a USB power connection. (Image source: Texas Instruments)
It not only allows vibrations to be quantified, but also provides provision for stimulating the accelerometer via user supplied vibration motors. These motors include the eccentric rotating motor (ERM) which spins around its x axis and produces rotational vibration in the yz plane, Figure 5a; a linear resonant actuator (LRA) which moves along the z axis and produces vibration only along that axis, Figure 5b; and a piezoelectric module which produces single axis motion, Figure 5c.
Figure 5: The DRV-ACC16-EVM can be used with a variety of vibration/excitation sources, each producing motion along a different axis or axes, including the eccentric rotating motor (5a), linear resonant actuator (5b), and piezoelectric effect actuator (5c). (Image source: Texas Instruments)
This module allows the designer to analyze the effects of position, mounting, and orientation of the accelerometer, and the kit guide explains how to use the scope images to calculate sensed g forces. The accelerometer board is attached to a test surface and then stimulated using one of these vibration sources to simulate actual use (Figure 6).
An interface board simplifies connecting the accelerometer output signal to an oscilloscope and allows comparison of the vibration stimulus waveform with the sensed waveform (Figure 7). The peak-to-peak voltage of the accelerometer voltage waveform (here, 155.8 mV) is divided by 2 to obtain the peak voltage, and then divided by the 57 millivolt (mV) scale factor to calculate the acceleration (every 57 mV of peak voltage corresponds to 1 g of peak acceleration for this evaluation set-up):
Acceleration (g) = Vpeak/57 mV
Here, the calculated peak acceleration is:
Acceleration = 155.8 Vpp/(2 × 57 mVpeak) = 1.367 gpeak
Figure 6: Both the accelerometer board and the vibration actuator are mounted to a common surface for driver response evaluation. The LRA is powered by the TI DRV2605 actuator driver (not shown). (Image source: Texas Instruments)
Figure 7: The blue waveform (top) is the applied stimulus consisting of a ramping up followed by a multiple pulse waveform, while the orange (lower) waveform is the accelerometer output; basic matching allows the user to calculate the peak acceleration. In this instance it is 1.367 g. (Image source: Texas Instruments)
Conclusion
A new world of IoT application possibilities is now practical due to the many low-cost, high-performance MEMS-based accelerometers now available from multiple vendors. While these are electrically simple to interface due to their internal analog and digital processing and filtering, as well as their standard I/O, they do require careful initialization and management of their many registers and user selectable options.
Further, these accelerometers bring a set of mechanical installation considerations which must be evaluated and factored into their end point use. Evaluation kits can help the process, along with an understanding of established test practices, standards, and fundamental mechanical engineering principles.
