The market for MEMS microphones has been growing steadily and, according to our research, is estimated to reach approximately $1.2 billion in 2015. Of the total market, almost $900 million is coming from smartphone, tablets, and wearable platforms. Additional applications include hearing aids, automotive, virtual reality, headsets, smart home, and IoT. As the quality of MEMS microphones continues to improve, new applications such as far field and directional audio are emerging. To get the latest scoop on this market segment, we recently spoke with Matt Crowley, CEO of MEMS microphone start-up company Vesper. In this detailed interview, Matt provides his perspective on the latest price and technology trends, emerging applications, competitive dynamics, and commercialization challenges. In addition to Matt’s comments, we also reached out to Jeremie Bouchaud at IHS to get his input and data as well.
MEMS Journal: What are the main trends with MEMS microphones today?
Matt Crowley: The main trend continues to be a push for higher signal-to-noise ratio (SNR) and higher reliability. New applications such as far field audio and directional audio need better SNR. There is also pressure for smaller size. Further in the future there will likely be more digital content in microphones for some IoT devices.
Figure 1. MEMS microphone forecast for mobile applications (handsets, tablets, wearables) by SNR level (source: IHS).
MEMS Journal: On pricing and performance, what are the key numbers to keep in mind?
Matt Crowley: There are a couple of factors that impact pricing. The main driver is higher SNR at small size. We’ve found a clear correlation between pricing and SNR, and within that, there is also a clear price distinction between analog and digital, with digital commanding higher prices. Microphone vendors can maintain attractive pricing if they can continue to push the performance envelope with high reliability. To sum up, pricing depends on a few key parameters -- SNR, size, digital-vs-analog, and volume. Our strategy is to release higher SNR products every year and maintain the price and technology lead.
MEMS Journal: What are the actual numbers for high-end, middle, and low-end SNRs? And what are the corresponding prices and recent trends?
Matt Crowley: For very high volumes (i.e., greater than 100 million devices), I would estimate a 60dB microphone at $0.20, 63dB at $0.30, 65dB at $0.45, and 67dB at $0.60. There is also some pricing variability in the marketplace, but these are reasonable estimates. I have also found that the market seems to have a barbell distribution -- in other words, people either want the cheapest or the best microphone, and the middle segment is less robust.
Also, the above estimates are for minimum, not typical, specs. For SNR specs you must carefully differentiate between minimum and typical specs. Only the minimum spec is guaranteed, but most of the marketing materials cite typical specs. For example, a 67dB capacitive microphone will have an SNR spread of +/- 2dB, which means that the minimum spec is 65db.
MEMS Journal: How much ASP erosion are you seeing with MEMS microphone prices?
Matt Crowley: Prices are eroding fastest at the low end, so microphone companies must continually release new products with better performance to improve or maintain ASPs. For example, there was a big premium for digital mics when they first emerged because few vendors could supply them. As the number of suppliers increased, the price erosion accelerated. The industry has been struggling in the past few years to increase SNR with capacitive technology which has led to price erosion for capacitive microphones.
MEMS Journal: Have you seen some recent devices on the market which incorporate far field audio and directional audio capabilities? If so, which devices and from which manufacturers?
Matt Crowley: Any application that records sound or noise at a distance is far field so that is a pretty broad category. Amazon’s Echo recently released to very positive reviews and it is basically a voice controlled computer. This is the first mass market device that heavily leverages both directional and far field audio using an array of seven microphones. Based on how successful the launch has been, I expect there will be a surge of interest in voice user interface devices that leverage far field and directional audio. Some advanced security cameras use thirteen microphones in array. The customer response to our VM101 announcement has been overwhelming and the scale and diversity of devices being developed is remarkable.
At the high end, a company called Norsonic has developed a directional audio product that is now on the market. Norsonic’s acoustic camera uses 128 to 384 mics depending on the performance needs. The product’s size ranges from 0.4 meters to 1.6 meters in diameter, which is far larger than something that can be used in a consumer device. The range and accuracy of their devices is really astounding. For Vesper the challenge and opportunity is to bring that extreme performance into mainstream consumer devices.
MEMS Journal: What’s going with MEMS microphone arrays these days? What are the applications and use cases?
Matt Crowley: Despite the efforts of a few pioneers, we’re not yet seeing mass deployment of MEMS microphone arrays, primarily because it is too difficult to match larger numbers of microphones. To have high performance microphone arrays, the individual microphones must be closely matched in sensitivity and phase. The sensitivity and phase must also be stable over time, immune to changes in environmental conditions, and resistant to contaminants. With capacitive MEMS microphones, the individual components often have to be manually matched and, as the number of microphones in the array increases, this manual matching becomes more difficult. Because the piezoelectric process is more tightly controlled and stable, it will enable device manufacturers to take any two microphones from two different reels or lots and know they will be matched.
MEMS Journal: What's different about Vesper's MEMS microphone technology as compared to the traditional mics?
Matt Crowley: Vesper uses a fundamentally different transduction method. The first capacitive, or condenser, microphones were developed at Bell Labs in 1916. Traditional condenser microphones are air gap capacitors with a backplate and a flexible diaphragm. Modern capacitive MEMS microphones operate on the same principle and are basically miniaturized condenser microphones manufactured in a silicon wafer process. Capacitive microphones use two plates, a diaphragm and a backplate, that form an air gap capacitor that has a high bias charge. When the diaphragm moves in response to sound, the capacitance changes and the resulting voltage is amplified.
Unlike the capacitive approach, we are using piezoelectric technology to build our microphones. For piezoelectric, the structure is much simpler. It is a cantilever that moves in response to sound and directly generates a voltage that is amplified. So our technology is a major departure from the way that microphones have operated for almost 100 years.
Figure 2. Vesper's MEMS microphone with lid removed to reveal the piezoelectric MEMS structure and custom ASIC inside. The small 3.35 X 2.5mm device is mounted on top of a US dime coin to show scale. (source: Vesper MEMS).
MEMS Journal: What’s the biggest technological challenge and risk with your approach?
Matt Crowley: Making piezoelectric MEMS requires lots of experience and expertise in aluminum nitride (AlN) films. If done correctly, the films are consistent and well controlled, as vendors like Avago and Qorvo have demonstrated, but there is a lot of proprietary know-how involved. This expertise takes many years to develop.
The amplifier ASIC for piezoelectric MEMS is actually simpler because our MEMS device does not require a high bias voltage or gain trimming. This eliminates the charge pump and gain trim circuit blocks which in turn makes our ASIC much smaller and simpler. The lack of a charge pump also makes our startup time nearly instantaneous and improves the power supply rejection ratio (PSRR).
MEMS Journal: What are the main technical challenges that you've had to solve in the past 2-3 years? What are the remaining technical challenges that you are working through right now?
Matt Crowley: The past few years were about perfecting the design to achieve manufacturability. We also spent a lot of time perfecting the AlN deposition process. As of today, we don’t know of any major technical challenges. Once our first product goes into volume production, we will concentrate on improving performance even further.
MEMS Journal: Regarding manufacturability, can you be bit more specific? For example, can you discuss a bit more about some of the specific issues that you guys have had to work through?
Matt Crowley: We were fortunate not to encounter too many problems so far. Getting the recipe for AlN films with the proper characteristics to make good microphones has probably been the area where we’ve expended the most effort.
MEMS Journal: How are you doing with your corporate development and funding? How much money have your raised?
Matt Crowley: We are well financed with strategic and financial investors. The financial investors are Atlas Ventures, a Boston based VC with extensive experience in MEMS, and the University of Michigan, which is where our co-founders Bobby Littrell and Karl Grosh invented this microphone technology. Our other major investor is a very large Asian company that is already in the microphone industry, and they will also become our strategic customer and partner. We’ve raised $3.5 million so far.
MEMS Journal: Large companies such as Apple and Samsung don't typically deal with smaller companies and startups directly. What's your approach for business development and market penetration?
Matt Crowley: Our business model is to leverage partners so that we can focus on design and product marketing. We will sell MEMS and ASIC wafers to companies in Asia that will package, test and distribute fully packaged microphones. Vesper will focus on design and marketing and let our partners handle assembly, sales and support.
MEMS Journal: Do you think that the InvenSense acquisition of the MEMS microphone business was a good idea? Why or why not?
Matt Crowley: I can’t comment on them specifically because I don’t really know how they are doing. Their office is only a few blocks away from ours which is a funny coincidence. It is a big market.
MEMS Journal: Vesper's core technology was initially developed at the University of Michigan at least 5-6 years ago. Why has it taken so long to commercialize it and bring it to market?
Matt Crowley: In terms of a product lifecycle in the MEMS world, we’ve actually moved quite quickly from design to production. The first working microphones were designed by our founders at the University of Michigan back in 2009. From 2009 to 2014 the company was funded through research grants, and our founders worked on perfecting the technology and process at academic fabs to develop a commercially viable design. In 2014 the company raised its first external funding, hired an engineering team, and engaged with a high-volume commercial foundry. Vesper actually assembled an engineering team and engaged a foundry to launch the product in less than one year. I think this model of patiently perfecting the technology without raising funds and only raising money when the design was ready to be rapidly commercialized makes a lot of sense, and will allow Vesper to get to market with less than $4 million in investor capital versus a typical $50-$100 million.
MEMS Journal: Which platforms are you focusing on first? For example, wearables, smartphones, or hearing aids?
Matt Crowley: Our first product is a single-ended analog microphone that has broad applicability in consumer electronics. Target applications include smartphones and wearables. With a product roadmap that features differential and digital microphones and a partnership strategy that gives us great distribution leverage, we will be able to supply a broader range of target applications moving forward.
MEMS Journal: Why do you guys think the world needs another microphone company? What are the main unsolved problems from the OEMs' perspective?
Matt Crowley: Two reasons. First, existing capacitive microphones still have major reliability challenges, but piezoelectric designs will make microphones far more reliable and rugged. Vesper will not only improve baseline reliability, we will also allow consumers to use microphone-based products at the beach or at the pool without worrying about breaking their device. Second, OEMs want to develop some incredibly creative and amazingly cool audio-enabled applications, but they are stymied by a lack of microphone performance. Examples are far field audio capture for voice-controlled smart homes, directional audio through beam forming for audio zoom, and next-generation advanced noise cancellation. These applications require small, stable, well-matched microphones with higher SNRs than are currently available on the market.
MEMS Journal: When are you planning to start shipping in volume? And what is that volume expected to be?
Matt Crowley: We plan to begin shipping small volumes of wafers to our packaging partners in Q4 2015. We expect our microphones to show up in devices in 2016. I can’t offer a volume forecast at this time.
In addition to Matt Crowley’s comments, we also reached out to Jeremie Bouchaud at IHS to get his perspective on the topic of MEMS microphones.
MEMS Journal: What are you seeing in terms of MEMS microphone ASPs and price erosion trends?
Jeremie Bouchaud: Compared to other MEMS for consumer applications, such as accelerometers, the price erosion of MEMS microphones has been rather modest in the past years. The main reasons are that, unlike the motion sensors, the MEMS microphone market leader Knowles has not dropped prices to maintain its market share. Also, there has been a demand for high performance MEMS microphones since the introduction of virtual assistant Siri in the iPhone 4. For MEMS microphones, higher performance comes with a price premium (again, unlike with MEMS motion sensors), and this has slowed down the overall price erosion.
MEMS Journal: Going forward, do you think that MEMS microphones will be able to continue to resist price erosion pressures?
Jeremie Bouchaud: There is a risk that this “exception” for MEMS microphone pricing is coming to an end. For example, after Knowles lost its MEMS microphone spot in the iPhone, it has left the field open for new players which are much more aggressive in terms of pricing. Also, at least for the moment, MEMS microphones seem to have reached a plateau in terms of performance improvement around 65-66dB SNR. If the performance does not keep growing as it has in the 2010-2014 time period, then the “performance premium” will no longer be able to slow down the overall price erosion.
MEMS Journal: What are the specific price trends that you are seeing right now?
Jeremie Bouchaud: I can provide a price range. For low performance (59dB SNR) analog MEMS microphones, in high volumes for handsets, in 2015 prices have been less than $0.20. For very high performance (65-66dB SNR) analog MEMS microphones, in high volumes for handsets, in 2015 prices have been greater than $0.30.
*********************************************
This article is a part of MEMS Journal's ongoing market research project in the area of MEMS microphones. If you would like to receive our comprehensive market research report on this topic, please contact Dr. Mike Pinelis at [email protected] for more information about rates and report contents.
Copyright 2015 MEMS Journal, Inc.
Comments