The semiconductor story is the story of constant replacement. Early transistors increasingly replaced vacuum tubes. Integrated circuits (ICs) replaced transistors. Our large-screen, flat-panel displays and televisions, which replaced cathode ray tubes, are giant-sized, thin-film semiconductors. Semiconductor-based solid-state drives are currently replacing rotating hard-disk storage in computers and servers. Even dynamic and electret microphones have started to yield to tiny arrays of semiconductor MEMS (Micro-ElectroMechanical Systems) transducers. You’ll find at least one such MEMS microphone in many new mobile phone models.

However, one electronic technology has long resisted replacement by semiconductors: the loudspeaker. Today’s audio speaker or earbud driver continues to use century-old, moving-coil or balanced armature technology. A company named xMEMS Labs is determined to change that situation by introducing two second-generation semiconductor MEMS speakers and a device called the Skyline DynamicVent, which is simply an electrically actuated air valve used to equalize differential pressure between the ear canal and ambient air.

The three new second-generation products from xMEMS include the Cowell and Montara Plus micro speakers and the Skyline DynamicVent electrically actuated air valve. Image Credit: xMEMS Labs

These devices are all manufactured using a MEMS semiconductor process developed by TSMC (Taiwan Semiconductor Manufacturing Company), which has a resolution of 110 nanometers. That’s gigantic relative to today’s most advanced CMOS (Complementary-Symmetry Metal Oxide Semiconductor) process technologies that feature single-digit nanometer resolution. But that sort of tiny, top-of-the-Ziggurat resolution isn’t required when interacting with the real world, such as when a loudspeaker moves air to produce sound or when a microphone converts sound into electrical signals.

TSMC’s MEMS process involves a sandwich of etched silicon bonded to a layer of piezoelectric material made from PZT (lead zirconium titanate). PZT has been around for years. Bell Labs was playing with PZT ferroelectric memory as far back as 1955. It’s the material used in ferroelectric RAM to store the bits. I was writing articles about Ramtron PZT ferroelectric memory back in the mid-1980s. When electrically energized, PZT bends or flexes. That’s the piezoelectric effect. Bonded to an etched silicon membrane or diaphragm, the flexing PZT causes the membrane to move and, at the right frequencies, the moving membrane generates sound.

The two new xMEMS speakers, called Cowell and Montara Plus, join the company’s previously announced first generation speaker called Montara. The second-generation xMEMS speakers feature a redesigned driver design that allows the Montara Plus speaker to outperform the Montara speaker in loudness and enables a size reduction for the Cowell speaker, which is intended for earbuds and in-ear hearing aids and hearing instruments. xMEMS claims several advantages for these solid-state speakers including:

• Uniformity, with ± 1.5 dB of output consistency from speaker to speaker. This advantage eliminates the binning needed to match conventional voice-coil speaker pairs.
• Enhanced spatial imaging due to ± 1 degree of phase variation from speaker to speaker.
• Superior clarity, with 40 kHz audio bandwidth and fast impulse response that “opens” the music.
• Reduced speaker breakup due to silicon’s vastly superior stiffness.
• Robustness, with resistance to 10,000g of mechanical shock.

Despite these advantages, there are some disadvantages to the current MEMS speaker generation. The first is low-frequency response. For earbud designs that are all about that bass, a moving-coil woofer will likely be needed. It’s difficult to get much bass out of a small speaker, no matter the underlying technology. However, this is not quite the disadvantage it might seem to be, because some high-end audiophile headsets and in-ear monitors used by musicians already incorporate multiple drivers for improved frequency response.

Also, these xMEMS speakers do not have flat frequency response. They peak at higher frequencies, and the rise in frequency response starts at 1000 Hz. Earbud OEMs will need to apply DSP to flatten the frequency response. xMEMS prefers to frame this potential disadvantage as an opportunity for each OEM to tailor the sound for its own signature sound coloring. In addition, the xMEMS speakers present the same sort of electrical interface as do moving-coil speakers, so xMEMS has developed a companion Class-H amplifier chip named Aptos that’s needed to drive these MEMS speakers.

Perhaps the biggest disadvantage is price. According to xMEMS, these speakers currently cost more than moving-coil speakers. One reason for this situation is simply sales volume. Although these xMEMS devices are potentially less expensive to manufacture and certainly require less manual selection for sound matching, it’s always difficult for a new technology to compete with a well-established technology on price, at first.

One extremely interesting part of the xMEMS rollout presentation was a graph by Yole Developpement showing the 20-year impact that the introduction of MEMS-based microphones had on the electret microphone market. Based on the graph, the electret CAGR (compound annual growth rate) fell 2.6 percent per year over those 20 years while the MEMS microphone CAGR increased at 11.3 percent per year. However, a closer examination of the bar chart showed that the MEMS microphones didn’t so much cut into the electret market as they grew the overall microphone market and took all of that growth.

With earbuds in practically everyone’s heads seemingly all the time, the same market growth might not be expected for the MEMS speaker market, but there’s a submarket where this sort of growth might take place: hearing aids and hearing instruments. Currently, the market for these devices is artificially stunted because of the difficulty in acquiring these devices and their cost. The superior attributes of MEMS speakers could potentially grow this market significantly, thanks to an aging population that increasingly needs these devices and their ability to combine the best features of Bluetooth earbuds with active noise cancellation and hearing aids and instruments.

Finally, there’s the Skyline DynamicVent to consider. It’s a component quite unlike anything else available in the realm of sound reproduction, and it solves a problem that’s unique to in-ear devices: differential pressure buildup between ambient air pressure and the pressure in the ear canal. Earbuds that seal off the ear canal with silicone ear cups isolate the canal from ambient pressure. That ambient air pressure can often change, which is something you experience, for example, in an airplane changing altitude, on a ski lift, or even in a car when climbing a mountain or dropping into a deep valley. This differential pressure can cause ear discomfort or even pain if the air-pressure difference becomes too large.

The Skyline vents permit pressure to be equalized electronically. When activated, the vents open and allow air to flow into or out of the ear canal. xMEMS envisions that these vents would be operated by the DSP associated with the speakers and would either automatically or manually equalize the pressure when needed. Used alone, the vents could be used for another type of product – active sleep vents or “sleep buds” – which are earbuds that open a sleeper’s ears to ambient air unless the partner in the bed or bedroom starts to snore. Then the sleep bud would close the vent and block the sound. I can envision using these devices for active hearing protection in locations with high ambient noise, such as in factories, as well.

As with MEMS microphones, these three xMEMs products seem to open a range of possibilities for new sound devices. It will be interesting to see if that truly happens.