MEMS Design Shrinks Speakers to Chip Scale

　　Microelectromechanical-system-based audio speakers for earbuds, smartphones, wearables and other Internet of Things (IoT) devices have proved a tough row to hoe. But USound GmbH (Graz, Austria) now says it will be first to market with a family of MEMS audio speakers, with production volumes planned for the first quarter.

　　USound calls its MEMS speaker Ganymede and says it will offer a reference design, called Magaclite, by the end of this year. The devices have been fitted to high-end sunglasses and are being developed for earbuds; smartphones; and multidriver, high-fidelity above-ear speakers.

　　“It was a challenge to make the tiny MEMS drivers sound good,” Mark Laich, USound senior business development adviser, told EE Times at SEMI’s MEMS & Sensor Executive Congress (San Jose, Calif.) earlier this month. The difficulties can be chalked up to the physics of sound, which dictates that the cone size pushing the air be proportional to the wavelength of the sounds emitted. High-fidelity home speaker systems use 12- to 15-inch bass drivers along with midrange drivers in the 3-to-6 inch range and high-frequency tweeters as small as (and sometimes smaller than) an inch.

　　For wearables, the proportions of the driver's size must be some small fraction of the wavelength of the sounds emitted plus some mechanical or electronic frequency equalization to make them sound truly high fidelity. High-end headphones, and even some expensive earbuds, use multiple drivers to achieve the highest fidelity. Most reasonably priced earbuds sacrifice fidelity for frugality by using a single driver plus a lot of electronic equalization.

　　The same can be said for USound’s MEMS speakers. The company’s low-end model uses a single driver plus electronic equalization in a chip-scale package bonded directly to the MEMS die. “Our MEMS frame uses a rectangular actuator that pushes air using piezoelectric suspension beams, with a surrounding diaphragm that seals the chamber,” Laich said. “As a result, we have very fast actuation, with microsecond response time, which will assist with noise cancellation in future models that will be built with a MEMS codec partner.” Laich declined to identify the partner.

　　Noise cancellation is accomplished by including a MEMS microphone in each earbud. The mic samples the common background noise and injects the sampled signal with 180? of phase change, thus inverting it so that the sampled signal zeroes out the ambient noise when mixed with the music signal. Of course, the response time of the speaker driver should be instantaneous to be 100 percent effective, so the microsecond response time of USound’s drivers suits them well for noise-canceling designs.

　　But how do they sound?

　　The best way to describe the speakers’ sound is “digital,” like the difference in sound between a CD and vinyl record. Of course, even with electronic equalization to boost the low frequencies, the sound from a single-driver reference design lacks the high fidelity of multidriver designs.

　　The company has already designed multidriver reference designs for an unidentified brand of designer sunglasses. By combining a conventional low-frequency driver with USound’s mid- and high-frequency MEMS speakers, the design achieves high-fidelity sound, according to Laich. It also allows the drivers to be located above the ear canal and the output to be beam steered into the ear so that the listener can also converse with others, hear traffic sounds, and receive other audible cues from the environment.

　　“Our above-the-ear designs project sound into the ear using a dipole, open-back design instead of the traditional monopole, closed-back design,” Laich told EE Times. “The dipole sacrifices a little volume in return for a narrow concentration of acoustic energy directed into ear canal, in the manner of beam forming.”

　　USound has partnerships with Austria Technologie & Systemtechnik AG, for AT&S’ tiny printed-circuit boards; the Fraunhofer Institute for Silicon Technology, for ISIT’s expertise in producing power electronics for microsystems; the Institute for Electronic Music, for IEM’s interface expertise between acoustic technologies and audio practice; and STMicroelectronics, for ST’s manufacturing expertise in robotic ears.