What’s that rumbling? It’s the ever-loudening boom expected to stand the silicon microelectromechanical systems (MEMS) microphone market on its collective ears. Forecasters say the market explosion will really unfold after 2009.
Two years ago, only three or four dominant silicon MEMS microphone manufacturers existed. Knowles Acoustics had the lion’s share, followed by Akustica, Pulse Engineering’s Sonion MEMS Division, and Infineon Technologies. Now the list includes at least a dozen others (see “Representative Silicon MEMS Microphones” at www.electronicdesign.com, ED Online 20667).
They’re all preparing design and marketing plans for a major push into this field. In fact, many see it as the next “commodity” MEMS IC market with rapid growth, after accelerometers (see “Those Elusive MEMS Market Figures,” ED Online 20665).
In addition to their wide use in mobile phones and notebooks, scads of new applications for these microphones will soon emerge. These include handsets, headsets, voice recorders, camcorders, laptops for Voice-over-IP (VoIP) uses, digital cameras, MP3 players, and interactive games. Voice activation for a wide range of consumer electronic products is seen as a lucrative market. The automotive field offers potential growth in hands-free communication and navigation devices, too.
Established companies like Analog Devices, Freescale Semiconductor, NTT Dokomo, NXP Semiconductors, Omron Semiconductors, Panasonic, and STMicroelectronics—particularly those with expertise in both audio and MEMS technologies—are poised to grab a good share of the silicon MEMS microphone market. Others like GoerTek Acoustics are typical of the rising interest by China-based companies to compete in this market. Many of these firms have pre-announced some design details in anticipation of new product market introduction.
Electret condenser microphones (ECMs) are a much less expensive option than silicon MEMS types like those sold by Japan’s Hosiden Corp. The well-established ECMs use a simple structure consisting of a capacitive sensing plate and a field-effect transistor (FET) (Fig. 1).
The average selling price of silicon MEMS microphones is $1.50 to $2.00, depending on order volume and performance requirements. This is about three times greater than ECM prices. But silicon MEMS microphones offer several advantages, such as smaller size and greater integration capabilities, which enables greater design innovation. They also consume much less power (about one-half the current drains of ECMs).
Silicon MEMS microphones also offer greater immunity to radio-frequency interference (RFI) and electromagnetic interference (EMI), and they can withstand the high temperatures of a surface-mount technology (SMT) process. Although ECM manufacturers claim their designs can withstand the high temperatures of an SMT reflow process, silicon microphone makers contend that such products suffer from a shift in their performance characteristics due to the organic materials used in their construction.
Because they can be made on a batch-fabrication CMOS process, it’s possible to make silicon MEMS microphones much less expensively (even cheaper than ECMs) once the market opens up and mass production goes full throttle.
SINGLE CHIP OR MULTICHIP?
Most manufacturers use two chips in their microphone design— one for the transducer structure and one (usually an ASIC) for signal- conditioning electronics. Exceptions are Akustica, which uses a single-chip design for its CMOS microphone, and the three-chip approach taken by Pulse Engineering’s Sonion MEMS Division. Each has its advantages and disadvantages.
The single-chip approach makes it easier to integrate more functions on the same die at less cost. It’s also more compatible with a standard CMOS process in terms of process flow, packaging, and testing. It has a higher level of reliability due to the absence of wire bonds that interconnect two or more chips as well. On the other hand, it limits design flexibility for meeting different application requirements.
The smallest silicon MEMS chip comes from Akustica. Measuring 1 by 1 mm, it forms the basis of both the analog output AKU1126 Napoli and the AKU2202C Shadyside microphones (Fig. 2). Its size makes it possible to use multiple microphones, rather than just one, to provide ambient noise cancellation in products such as Bluetooth headsets or cellular handsets.
Continue on Page 2
Small packages also allow manufacturers to supply microphones with closely matched key parameters such as sensitivity and phase response, thanks to the silicon fabrication process used. Such devices have a much smaller spread of production tolerances than other microphones.
Arrays of two or more microphones provide directional sensitivity. Devices can then isolate a voice input from a user in noisy surroundings. Example applications include hands-free operation of communication or navigation devices—as an OEM module or an aftermarket addition—in the automotive environment.
In the meantime, breakthroughs continue for silicon MEMS microphones. Analog Devices’ ADMP421/401 digital/analog output microphones include a particle filter constructed by perforating a silicon cover over the microphone’s diaphragm. These 10-µm deep and 6-µm wide holes are very effective at keeping dust particles and other contaminants from reaching the delicate MEMS diaphragm.
This design has led to high-performance specifications such as a 62-dB (typical) signal-to-noise ratio (SNR), 20-kg and 160-dB mechanical and sound-pressure shocks, respectively, and powersupply rejection ratios (PSRRs) of 80 dB (digital version) and 50 dB (analog version). According to Analog Devices, these figures represent the highest performance specifications now available in the industry (Fig. 3).
In some cases, experience in designing a MEMS package has allowed companies like Memstech to rapidly introduce to the market rugged and robust silicon MEMS microphones for mobile phones. Memstech’s silicon MEMS microphones, which feature a fixed diaphragm, are similar to the original design employed by Memstech’s founders while working at the Ford Motor Co. developing MEMS pressure sensors for cars. (It should be noted that a MEMS microphone is essentially a MEMS pressure sensor. While not all pressure sensors can be called microphones, some can be if they’re designed to sense low dynamic audio pressure waves). The original design at Ford, which was later used to make disposable blood-pressure MEMS sensors, employed a silicon capacitive absolute pressure (SCAP) sensor that has no holes in the diaphragm, enabling robust performance when exposed to dust particles.
Wolfson Microelectronics uses silicon nitride for the transducer element in its WM7110/7120 silicon MEMS microphone instead of polysilicon. According to the company, silicon nitride delivers lower stress and strain coefficients, leading to higher reliability. Wolfson also offers a silicon MEMS microphone with ±1-dB sensitivity tolerance. “We already have the solution for noise cancellation,” says Nigel Burgess, a member of Wolfson’s marketing team.
“We’re working on a higher-precision silicon MEMS microphone with ±1-dB sensitivity tolerance, an important parameter for multi-microphone arrays,” says Jacob Philipsen, VP and general manager of the Sonion MEMS Division of Pulse Engineering Corp. “Most other silicon MEMS microphones offer ±3-dB sensitivity tolerance. A dual-microphone arrangement is important for noise cancellation in handsets and mobile phones. Good acoustic amplitude and phase response over time is crucial.”
There are other approaches for dealing with background noise besides offering tighter sensitivity tolerances. Pulse Engineering’s Sonion MEMS Division, for example, offers a low-profile (l.5-mm) rubber boot to house the silicon MEMS microphone and acoustically seal it. The boot offers full mechanical and electrostatic- discharge (ESD) protection, as well as protection from dust particles. The company says that the construction of its MEMS microphone is optimized for acoustic protection (Fig. 4).
Many companies with IC production experience are joining forces with firms that have audio electronics experience to strengthen their market hold. For instance, Wolfson Microelectronics is beginning to reap the benefits of its January 2007 acquisition of MEMS design company Oligon Ltd.
The Oligon team was made part of Wolfson’s AudioPlus strategy to add additional hardware and software technologies to its basic audio-processing and analog-to-digital and digital-to-analog conversion technology. “Our audio chain expertise in codecs and DSPs gives us a leg up in the silicon MEMS microphone area,” says Burgess.
In 2007, China’s GoerTek Acoustics, armed with its experience in ECMs, teamed with China’s Forte Media to strengthen GoerTek’s silicon MEMS microphone business. GoerTek has expertise in fabless designs of advanced voice-processing ICs and small-array microphones.
Two years ago, Infineon Technologies AG produced a silicon MEMS microphone (Fig. 5). Since then, it joined up with Japan’s Hosiden Corp. to further develop and manufacture the microphone. Hosiden adds its strength in acoustics technology to Infineon’s silicon MEMS expertise. And, of course, Pulse Engineering Co. acquired pioneer Danish MEMS silicon microphone manufacturer Sonion and made it one of its divisions.
Furthermore, NXP Semiconductors, which was formed by Philips with expertise in semiconductor IC manufacturing last year, developed a silicon MEMS microphone. Since then, it has partnered with Switzerland’s Phonek to develop state-of-the-art low-power wireless medical hearing systems that use silicon MEMS microphones. Phonek has expertise in acoustics and hearing technology.
Continue on Page 3
A MATTER OF IP PROTECTION
As more companies try to enter the field, the issue of intellectual property (IP) protection looms larger, forcing established silicon MEMS microphone manufacturers to aggressively protect their patents with litigation. Court suits and rulings have been forthcoming. For example, the International Trade Commission (ITC) is in the process of making a final determination as to whether MEMSTech has infringed on Knowles Acoustics’ packaging patents.
“IP protection is an important issue that companies like us and Knowles Acoustics are trying to uphold,” says Marcie Weinstein, Akustica’s director of strategic marketing. “We have not licensed our CMOS MEMS or MEMS microphone IP to anyone.”
A few years ago, Akustica and Knowles Acoustics entered into a cross-licensing agreement. It gave each company the freedom to pursue the development, manufacturing, and delivery of MEMS microphones.
Knowles also filed an IP complaint with the ITC against AAC Acoustic Technologies in 2006. According to AAC’s Web site, the ITC denied Knowles’ motion for a preliminary injunction that bars AAC Acoustics from producing silicon MEMS microphones. Subsequently, the two companies resolved their differences and entered into a worldwide cross-licensing agreement. Cross-licensing agreements have also been reached between Pulse Engineering’s Sonion MEMS Division and Akustica.
One thing should be clear for engineers looking to incorporate a silicon MEMS microphone in their designs: Most of these device manufacturers aren’t very forthcoming about providing “complete” specifications. Some offer product brief sheets. Others provide a few “highlight” performance specifications. Many don’t even have complete datasheets, if they have any datasheets at all. There’s a game of “specsmanship” going on that can use a lot of clarification.
Once many of these strategic relationships settle down, things should become clearer, and full and explicit datasheets will become the norm. Looking forward, it’s conceivable that more than one sensor function can be combined on the same MEMS chip holding the microphone transducer element, providing reduced packaging, testing, and overall processing costs.
A silicon MEMS microphone may yet be combined with an accelerometer in, say, a mobile phone. Or, multiple chips can be combined within one package. Smart 3D packaging approaches that use through silicon vias (TSVs) for interconnects may facilitate all of these developments.
Akustica demonstrated that MEMS inertial and RF functions can be manufactured on the same process being used to massproduce silicon MEMS microphones. The company has shown that a dual-axis accelerometer (a three-axis version is in development) can be made, as well as a capacitive membrane switch for tunable components and filters.