The first MEMS accelerometers in iPhones and the Wii revolutionized the user interface by introducing natural motion as an input mechanism, but it's taken a while for designers to figure out how best to use these capabilities. Now the inertial sensors are starting to move into a wider range of motion control and precision location applications, helped along by lower costs, and by the generally maturing of the knowledge base and infrastructure that are making the sensor data easier to use.
Yole Développement projects these developments will drive accelerometers and magnetometers to be designed in to close to 50 percent of all mobile phones within five years, and gyroscopes to be included in some 20 percent, mostly at the higher-end smart phone part of the market. Gyroscopes are already in almost all tablets, mostly because Apple still so dominates that market with better than 90 percent share. We expect usage of inertial sensors in consumer electronics will increase by about 24 percent on average annually for the next five years, to reach some 5 billion individual sensor units by 2015.
Lower prices and better software and infrastructure help drive adoption
Major applications for inertial sensors outside of games so far have actually been somewhat limited. Accelerometers have become a must-have in mobile phones for switching between portrait and landscape mode, and seen scattered use in pedometer functions. Magnetometers hit mass adoption in phones last year, to supply correct heading for navigation. Multiaxis MEMS gyroscopes have just reached consumer price points and volumes and are showing up in first phones and essentially all tablets, although initially largely for games.
But lower prices are helping drive wider adoption. Yole sees costs of discrete inertial devices continuing their steep decline, with the ASP of a three-axis accelerometer, for example, dropping from $0.70 in 2010 to around $0.30 by 2015—or less than $0.10 per axis. Part of these cost savings will be driven by sharing the cost of one controller ASIC between two sensor devices, by packaging the accelerometer and the magnetometer, or the accelerometer and gyroscope together as one combination sensor with one ASIC. This can also improve the sensor data, directly building in the corrections of each sensor for the deviations of the other.
Also driving adoption is the fact that it's getting much easier to turn the sensor output into useful applications. The leading MEMS device makers like STMicroelectronics and InvenSense are supplying more software and libraries to make it easier for the phone and tablet makers to add basic motion functions to their systems. Dedicated motion sensor software suppliers like Movea and Hillcrest Labs are supplying device-agnostic software to allow wider applications, particularly for air mice and TV remotes to do control by gestures. And the latest version of Android operating system software supports some motion processing APIs, with more sophisticated versions expected to come.
This is an interesting trend... what began at one end of the spectrum mostly in the military world namely attitude and G-sensing (spinning mass gyro's and varieties of accelerometers) has led to MEMS versions for motion and attitude sensing. Now we are taking a page from robotics to improve gesture control.
The key to making these efficient and consume less power is really sensor fusion thru embedded software applications in an ASIC/micro.
really good post Laurent. The MEMS market is really finding new applications in the smartphones. Do you know why MEMS are finding place on smartphones only even though the price is low (30 cents or so)?
@Himanshu_Gupta: As with 3D IC's that are on track to find markets in new products (& replacements in older ones), MEMS will make it in to more products but the enabler is really the software applications. As you combine different sensing functions, processing of 'fused' data and adding value from it that the consumer / application needs is really the key. Faster & lower cost fusion of sensory data is best done at the hardware level (similar to hypervisors for a multicore architecture) and I think that is where 3D IC & MEMS in the stack will make sense.
I am working on a draft that will probably hit the EE Times editors late this week that partially answers your questions (based on IMEC's presentation at Semicon).
Good summary Laurent, I would be interested in reading something longer on this topics that compares various MEMs sensors and actuators...is MEMs growth due to standardization in this area? A few years back every MEMs manufacturer was doing things their own ways...Kris
Thanks for the kind comments!
@Himanshu_Gupta: MEMS is not limited to smartphones anymore as accelerometers are used on some feature phone platforms as well since several years for pedometer or shake control (to change music tracks...). But this is true that MEMS sensors bring a much higher added value in smartphones when combined with other components (GPS receiver, magnetometer...) and with higher level signal processing.
@iniewski: we still don't see any kind of standardization in MEMS that can be compared to the CMOS world. However it evolves step by step now that some components mature and that volumes are in the range of millions units a day. The largest MEMS IDMs or foundries have MEMS "platforms" that can be seen as internal standards. And for the test side we start to see more and more off the shelf tools or test houses involved in MEMS that are used in replacement of in-house built test equuiopments
Join our online Radio Show on Friday 11th July starting at 2:00pm Eastern, when EETimes editor of all things fun and interesting, Max Maxfield, and embedded systems expert, Jack Ganssle, will debate as to just what is, and is not, and embedded system.