Traditional electret condenser microphones have long remained the most-used microphones, but micro-electromechanical systems ( MEMS) microphones are increasingly displacing them. Voice-coil speakers have long remained the most used speakers, but will MEMS speakers ever displace them? Santa Clara-based startup xMEMS, Inc., aims to change the game and eclipse voice-coil and hybrid-MEMS speakers by implementing the entire MEMS speaker — actuator and membrane — in silicon.
Tried and true?
Founded in October 2017, xMEMS is emerging from stealth mode with the introduction of what it claims is the first monolithic true MEMS speaker, delivering high fidelity, full-bandwidth sound, and low harmonic distortion for smart personal audio devices such as true wireless stereo (TWS) earbuds, in-ear monitors and headphones. Dubbed Montara, the MEMS speaker has been sampled to select customers since April. Engineering samples are now released to more OEMs and manufacturers, risk orders are expected to start at the end of this year, and mass production is planned for April 2021.
xMEMS’ Mike Housholder
If Montara is the first “true” MEMS speaker, does that mean there are “false” MEMS speakers? The answer is no, but xMEMS does not mess with the choice of words. Replacing only one component of a speaker with MEMS is not enough to make it a MEMS speaker. Montara is “the first monolithic, single-die, all-silicon speaker, so it comes out of the fab as a fully functional speaker,” Mike Housholder, VP marketing and business development at xMEMS, told EE Times. “There is no additional process and assembly step other than traditional semiconductor packaging.”
Montara has a cell-based architecture, and each square is an individual speaker. “It has got a piezoelectric MEMS actuator and, on top, there is a speaker membrane etched out of silicon.” The silicon membrane is pushed up and down by the piezoelectric MEMS actuator, generating air pressure, which moves the membrane and creates sound. “We have basically created an array of six cells for this first implementation of Montara, and those six cells move in concert to generate the sound signal.”
Voice-coil technology has been around for a century, and it seems to be working fine, said Housholder. Multi-component voice-coil speakers, however, require labor-intensive, high variability factory assembly lines. Because voice-coil speakers are semi-automated, “you will see in any speaker factory hundreds or thousands of workers manually wrapping voice-coil speakers around magnets, mounting plastic or paper membranes. And anytime humans are involved, you have variability.” When manufacturing earbuds, he continued, “workers have to go through a calibration and matching process for the speakers to make sure the left speaker and the right speaker sound are roughly the same.” This calibration step inevitably takes time and money.
What xMEMS aims to bring to the table is a full bandwidth speaker with “the benefits of a full semiconductor process: uniformity, repeatability and consistency,” said Housholder. “We have a very tight guard band, where every speaker performs very similar to the other so we can eliminate a calibration and test step on the earbud assembly line.”
Furthermore, piezoelectric materials are dust and water resistant, “so it makes it very easy to make a waterproof, dustproof speaker, without any special coating or any waterproof mesh.” Basically, the chip as it comes out of the fab is IP-57.
xMEMS’ Montara speaker
Today, every speaker on the market has a resonant frequency within the audible frequency range, but that resonance creates an unnatural sound profile that needs to be handled during the manufacturing and the design of the speaker and the earbud, Housholder explained. With voice-coil speakers, “you have to either dampen the resonance or to design some sort of filtering.” Montara’s resonant frequency is “north of the audible frequency range (20 khz), so there is no special dampening, filtering, or accommodation that need to be made for our speaker.”
And while most speakers are either glued to the earbud housing or put on some sub-assembly, Housholder said Montara can be glued, attached to a sub-assembly or soldered down on a printed circuit board, like any other semiconductor component.
Sound quality first
Media consumption has become an anywhere, anytime routine, and consumers are increasingly demanding enhanced audio quality. According to market research firm Canalys, worldwide smart device shipments will exceed 3 billion in 2023 with the smart personal audio category growing the fastest. In 2020, smart personal audio devices are expected to grow by 32.1% year-on-year, to 490 millions units.
xMEMS will position itself in the fast-growing market of near-field audio speakers, and its two first products segments will be TWS and in-ear monitors. “We have to produce an equal or better sound to voice coil to really be considered for new designs,” said Housholder. “What’s unique about piezoelectric MEMS speakers is where your traditional voice-coil speakers dip and fall off at higher frequencies.” With piezoelectric MEMS, “you gain more SPL.” It also allows OEMs to shape the sound profile, because it “introduces no artifacts into the sound if you filter it to shape either a bright sound or a warm sound.”
The Californian startup also outlined the ability to replace two to four balanced armatures with a single Montara. And “because our speaker is monolithic, all in silicon, we have a fast mechanical response time. Where that will pay the biggest dividends is in active noise cancellation.”
Though not releasing the price, Housholder said the company has benchmarked Montara with off-the-shelf, high-end solutions. “We will be competitive with today’s balanced armature speakers, and we feel we are bringing new benefits to the table for equal price.”
On the roadmap
Different configurations of Montara could be developed to address a lower price point, said Housholder. “We are looking at a two-by-two cell configuration with four speaker cells to hit a lower price and get us into the mid-range of the market.” As headphones have a bigger cup that goes over the ear, xMEMS said it is also looking at increasing the number of cells to generate a larger sound.
The startup’s roadmap goes beyond hearables, as “there is a large market for non-near-field audio, for open space, free space audio loudspeakers,” said Housholder. Loudspeakers touch smartphones, smart speakers, home entertainment soundbars, TVs, and automotive, where he sees an opportunity to take out significant weight from loudspeakers. “We are looking at different architectures to approach the variety of the loudspeaker market out there.”
In July, xMEMS raised a little more than $11 million. It has a branch office in ZhuBei, Taiwan, and employs 26 people.
>> This article was originally published on our sister site, EE Times.