MEMs Revolutionizes Sensor Landscape
Until fairly recently—roughly 15 to 20 years ago—sensors for ubiquitous physical parameters such as pressure, motion and magnetic fields were relatively large, costly, power-hungry and electrically incompatible with standard circuitry. All that changed dramatically with the development of MEMS-based sensors (microelectromechanical systems) which etch a sensor into silicon by expanding on the well-developed fabrication processes and techniques already used for conventional electronic integrated circuits.
Beginning with the mass-market airbag sensor which quickly made the spring-loaded, ball-in-tube sensor obsolete and validated MEMS technology, these MEMS-based sensors have completely changed both how engineers think about sensing in general and, more importantly, what they can actually accomplish in their products. Products such as smart phones, drones, autonomous vehicles and tactile feedback “fingers” now have one and often more tiny, low-cost, low-power accelerometers, gyroscopes and other sensors on their crowded internal circuit boards, almost effortlessly providing high-performance sensing capabilities. Even microphones, which are acoustic-vibration transducers, have adopted MEMS technology for many situations including smartphones.
The direct and indirect market for high-value MEMS devices related IoT applications is expected to grow dramatically to 2018, according to market research firm IHS. (Source: IHS)
MEMS-based sensor applications span small, low-cost products to ones which are much larger and costlier. At one end, they are embedded in wearables to track use motion and pulse rates, while at the other end, today’s cars are loaded with multiple MEMS sensors to trigger airbags, sense impending rollover, adjust shock-absorber settings to match on load and road conditions, oversee brake performance and monitor many other performance and safety-related factors.
“All that changed dramatically with the development of MEMS-based sensors (microelectromechanical systems) which etch a sensor into silicon by expanding on the well-developed fabrication processes and techniques already used for conventional electronic integrated circuits.”
Unlike their predecessor, larger “assembled” sensors which produced a simple, basic analog or crude digital output, MEMS devices have many outstanding virtues. They are small, rugged, low-power, suitable for monolithic multi-axis devices often designed with calibration and compensation for non-linearities or temperature drift, as well as self-test.
MEMs Sensor basics
Nearly all MEMS sensors use silicon, the most-common semiconductor material. However, unlike electronic circuits which use the element primarily for its electrical properties when it is “doped” to create a semiconductor, MEMS devices focus on its physical properties. It is an easily handled and etched crystalline material with consistent and rugged mechanical properties. It can simultaneously function as a mechanical and as an electronic component.
Consider the MEMS accelerometer, the first high-volume MEMS sensor and thus a major learning experience for both vendors and users. There are many ways to build this sensor: in one common approach, a “proof mass” with a side plate is etched into the silicon between fixed plates, and this mass is anchored by tethers acting as silicon springs on the side of its free-motion direction. Electrically, this structure is a micro-capacitor, with capacitance on the order of picofarads (one trillionth or 10−12 F) and even femtofarads (10−15 F).
The basic MEMS-based accelerometer uses a tethered but movable mass surrounded by fixed plates to form an ultra-sensitive motion-responding capacitor; the actual silicon design is much more subtle and complex.
As the die accelerates along its active axis, the proof mass moves and the capacitance between the plates changes by a tiny amount, and this minute change is sensed and amplified by associated on-chip circuitry. After amplification, filtering and buffering, the output corresponding to the proof mass’s displacement is available as an analog signal, or can be digitized within the MEMS device to provide a format compatible with the associated microcontroller.
Of course, this all sounds easy, but development of the first viable MEMS-based accelerometer was a long and difficult effort taking roughly a decade. Although it is based on well-known silicon-IC design and production techniques, those were developed for electronic circuit on silicon, not mechanical ones. Therefore, new design tools had to be developed which focused on modeling and production related to mechanical aspects (i.e. strength of materials, thermal issues), as well as the active circuitry.
Further, the standard silicon masking and etching techniques, which were mostly two-dimensional or had modest 3-D aspects, had to be greatly expanded to support complex masking and etching in a third dimension. This was necessary to allow for undercutting and allowing the proof mass to move. Packaging was also a challenge, as any minute particles which got trapped during the package cycle, or came from within the package itself, would interfere with motion of the proof mass. Standard electronic-production test systems and lines were also an issue, as they were designed to verify electronic performance. Now, however, there was a difficult mechanical aspect to the test as well, because the existing test systems for electromechanical-type sensors were not designed to handle IC packages.
While some sophisticated and quite successful systems were obviously built before MEMS devices became available, each sensing aspect was a challenge and major accomplishment. Many systems could only implement sensing along one or two dimensions due to cost, weight, power and performance limits. A few MEMS-sensor examples show how this situation has changed; note that many MEMS sensors are related to motion and orientation, which are among the most difficult parameters to effectively sense, and yet are critical enablers for drones and IoT applications.
A step above the triaxial accelerometer is a complete inertial measurement unit (IMU) which adds a gyroscope for assessing orientation in an absolute (inertial) frame of reference. This is needed where GPS is not available (underwater, indoors, or due to interference or spoofing), for implementing image stabilization, or if the system needs to cross-check the GPS results. Highest-performance IMUs use spinning-wheel gyros to stabilize the reference platform and are the extreme realization of mechanical and electronic design and implementation. These spherical units cost millions of dollars, consume hundreds of watts, range in size from about 10 cm to about 30 cm diameter, and weigh tens of kilograms.
In contrast, MEMS has made IMUs available to applications which cannot afford any of these specifications, admittedly with performance which, like that of the MEMS versus PIGA accelerometers, is not quite as good but certainly good enough for many applications. A six-axis IMU offers acceleration sensing along X, Y and Z axes plus rotation sensing around those same axes.
9-Axis Inertial Motion Sensor
There can, however, be a problem with the 6-axis sensor. While the accelerometer and gyroscope do provide information about accelerations and rotations, small errors which build up over time can cause the general direction to drift.
This problem can be resolved through the addition of one more absolute directional sensor – a 3-axis magnetometer. This has led to the emergence of a 9-axis inertial motion sensor which comprises a 3-axis gyroscope, a 3-axis accelerometer and a 3-axis magnetometer.
With this extra magnetic field, the sensing algorithms will be able to compensate for any drifts over time. Consequently, the measurement and tracking of location are more accurate and precise even during very complex movements.
A whole slew of new applications can be enabled as a result. One such application is the full control of 3D cursor in CAD packages, flight simulators, head-mounted mice and remote controlled vehicles. Another involves major improvements in navigation through virtual environments such as CAD models and virtual gaming environments. Short-term full-space head movements and activities over longer distances can also be easily tracked. Small gestures, respiration and heart rates as well as gaits and tremors can be captured.
The impressive opportunities in mass applications such as the Internet of Things and autonomous/semiautonomous vehicles for land, water and air are largely due to dramatic, nearly simultaneous developments across all aspects of technology. These include high-efficiency electric motors and controllers; lightweight, high-capacity batteries; tiny GPS receivers and processors; powerful central processors; effective RF links for control (whether smartphone or dedicated); and tiny high-resolution video cameras, among others.
Also critical have been the multiple sensors which have allowed remote guidance and navigation (with or without GPS), such as electronic compasses, and MEMS-based motion, acceleration, inertial measurement and magnetic field units. In broad context, the availability of these high-performance devices has completely changed the design tradeoff balance, transforming sensors from large, costly, heavy, power-hungry transducers with incompatible electrical interfaces to their complete opposite.
At Avnet we have partnered with InvenSense, Molex, Murata, NXP, ST Microelectronics and Instruments to support the research phase of your next sensor based design. We have assembled a library of product information, development kits and reference designs to jump start your next project. Download our new Sensor Solutions Guide which features some of the latest MEMs sensors products available from our supplier partners.
“Everything about STMicroelectronics’ 3-axis digital MEMS Gyroscopes,” http://www.st.com/st-web-ui/static/active/en/resource/technical/document/technical_article/DM00034730.pdf
Written By: Bill Schweber
Bill Schweber is an electronics engineer who has written three textbooks on electronic communications systems, as well as hundreds of technical articles, opinion columns, and product features. In past roles, he worked as a technical web-site manager for multiple topic-specific sites for EE Times, as well as both the Executive Editor and Analog Editor at EDN. He has an MSEE (Univ. of Mass) and BSEE (Columbia Univ.), is a Registered Professional Engineer, and holds an Advanced Class amateur radio license. Bill has also planned, written, and presented on-line courses on a variety of engineering topics, including MOSFET basics, ADC selection, and driving LEDs.