Turn on a dime: putting tilt nav into handhelds
By Steve Nasiri, InvenSense
Motion sensing is quickly becoming the next critical technology to enable innovative applications for the handheld consumer market. Device makers are demanding motion sensing in order to develop solutions that address a range of challenges, from blur-free camera phone pictures that will spark more wireless image sharing to gesture-based commands and more intuitive user interfaces.
While motion-sensing devices have existed for some time, the current generation of products is unsuited to the mobile handset market. These devices are discrete, bulky and expensive for a high-volume industry where components are evaluated based on millimeters and cents. They are often composed of single-axis ceramic or quartz-based gyroscopes, cobbled together in a module with an add-on three-axis accelerometer. This multi-component architecture is a barrier to entry, with different mixes of technology, output signals and calibration performance. Making these devices work together is challenging and too costly for mainstream consumer applications.
The solution to motion sensing challenges for handheld devices lies in the development of miniature integrated motion-sensing devices, or IMUs (inertial measurement units), that measure all axes of motion. Such devices can integrate three axes of angular rotation and three axes of linear motion into a small silicon chip. Unlike single accelerometers, IMUs will offer greater precision, better responsiveness and easier integration into multiple applications.
Tiny inertial measurement units could add a new “twist” to handheld devices by giving them a gesture-based user interface. Photo courtesy of InvenSense
However the development of an IMU device poses significant technical and business hurdles, and to date no such products are on the market. While single-axis gyros, dual-axis gyros and three-axis accelerometers are more readily available, integration of all these sensing axes and accompanying electronics into a single die and package would be a breakthrough solution that could enable a whole new motion-sensing industry. At the same time, the solution must be manufacturable in consumer volumes - at a unit price point that meets industry requirements, currently $0.50 per axis.
InvenSense is a fabless MEMS company specializing in motion sensing technology that sought to tackle size, cost and manufacturability issues during the initial design phase. MEMS technology is clearly the correct choice for achieving miniaturization, integration and price by its inherent virtue of leveraging semiconductor-based fabrication processes. Although MEMS fabrication is cost effective, packaging and testing of MEMS devices have long been problematic. Because MEMS devices by definition are different than standard semiconductor dies, standard semiconductor packaging and assembly solutions could not be leveraged as efficiently - hence the old adage, “Eighty percent of MEMS device costs are in packaging and test.”
The main challenge is designing a MEMS solution that not only leverages standard fabrication processes, but also standard assembly and test technologies. InvenSense uses an approach called “wafer-scale integration” that combines finished bulk silicon MEMS with finished CMOS wafers at wafer level. The integration technology is completely CMOS-compatible and requires no additional material over aluminum. It allows for simultaneous electrical interconnections and mechanical hermetic sealing between the MEMS and CMOS wafers. This solution enables complete, finished MEMS wafers that can be handled and tested just like CMOS wafers, permitting packaging in standard QFN housings using standard semiconductor manufacturing lines. Special handling, custom assembly lines, cavity packages, multichip modules, etc., are eliminated.
The technology was applied to the company’s first family of products, the first integrated dual-axis gyroscope, addressing the immediate market need for image stabilization for digital still and video cameras. The silicon die measures 3.5 x 3.5 x 1 mm, and is packaged in a 6 x 6 x 1.45 mm QFN package. The second-generation silicon is 3 x 3 x 0.8 mm and will be packaged in a 5 x 5 x 1.3 mm QFN package.
The new dual-axis gyroscope chip is produced as two products with different specifications targeting two major application areas. The first is an image stabilization device for camera phones and digital still cameras (DSCs) that senses natural hand jitter in order to accurately correct for blurring due to low frequency hand motion. The second addresses the motion sensing market, with a full scale range of 400 degrees per second. It is used for applications requiring accurate tracking of hand or object motion.
Steve Nasiri is chief executive of InvenSense Inc. (www.invensense.com) in Santa Clara, Calif.