Engineers tend to be traditional. Once we’ve found a way of doing something that works, we can happily keep on doing it the same way for decades. Loudspeakers are a perfect example. Whether they’re found in earbuds, smartphones, televisions, or concert halls, almost all of them still rely on the same basic principle: mechanically moving air at the frequencies we want to hear, but that’s not the only way to make sound. Intrigued? Read on…
About three years ago, I wrote a column about SonicEdge. The title of that mini-masterpiece was The Future Sounds Incredible (When Employing Ultrasonic Speaker Technology). As you may recall, the gist of that article was a new way to realize audio signals using teeny-tiny ultrasonic MEMS speakers.
Well, I just had a follow-up talk with Moti Margalit, who is CEO and Co-Founder at SonicEdge. Moti was one of the guys I spoke to for the original column. After exchanging the traditional pleasantries, including appropriate awe at my Hawaiian shirt du jour, we hurled ourselves into the fray as follows:
Me (Max): So, what’s changed since last time?
Moti: We’ve got engineering samples. It’s real!
Well, I must admit that this gave me pause for thought. Three years is a long time these days. “How on Earth am I going to spin this?” I thought to myself. Happily, it turns out that there’s more to the story than first meets the eye (or ear).
One of the things that struck me during our conversation is that this isn’t simply the story of one company bringing a product to market. Three years ago, SonicEdge looked like a fascinating outlier with an unconventional idea. Three years ago, SonicEdge appeared to be a lone voice in the wilderness championing a rather unconventional idea. Today, it seems that much of the industry has quietly arrived at the same conclusion: generating audio from modulated ultrasound may well represent the future of miniature loudspeakers.
Sound from Ultrasound
Let’s begin by reminding ourselves how all this wizardry works. Let’s start with traditional loudspeakers. Whether they’re moving-coil drivers or piezoelectric devices, they all work in fundamentally the same way. They physically push and pull the surrounding air at the audio frequencies we wish to reproduce. Want a 1 kHz tone? Move the diaphragm back and forth 1,000 times per second. Want 10 kHz? Move it ten thousand times every second.
Traditional loudspeakers versus SonicEdge’s ultrasonic MEMS architecture (Source: SonicEdge)
SonicEdge takes a completely different approach. Instead of attempting to generate audio directly, each microscopic MEMS cell acts as an ultrasonic air pump operating at around 400 kHz, which is roughly 20 times the upper limit of human hearing.
Each cell, which measures only about 50 microns across (roughly the width of a human hair), contains three conductive membranes stacked one above another. The lower membrane oscillates ultrasonically against the relatively fixed center membrane, producing an extremely rapid series of pressure pulses. The upper membrane acts like an ultrafast valve, opening and closing the aperture in the center membrane at precisely controlled moments.
Rather than allowing every ultrasonic pressure pulse to escape, the device selectively releases only the appropriate pressure pulses at precisely the right moments. The result is that the envelope formed by those pulses becomes the desired audio waveform.
Don’t worry if this all sounds a little magical. The important point isn’t the signal processing; it’s the airflow. Instead of moving a relatively large diaphragm comparatively slowly, SonicEdge moves a much smaller volume of air extraordinarily quickly.
Moti offered a delightful analogy. Think about a Dyson vacuum cleaner. Rather than relying on a large fan turning relatively slowly, Dyson uses a much smaller impeller rotating at extremely high speed. Each revolution moves less air, but because it spins so quickly, the overall airflow becomes both powerful and remarkably efficient.
SonicEdge applies much the same philosophy to loudspeakers. Instead of relying on a large diaphragm moving relatively slowly, it uses microscopic air pumps operating at ultrasonic speeds to move enough air to rival a much larger conventional speaker.
Tiny Speakers, Big Ambitions
The numbers are genuinely impressive. A conventional moving-coil speaker used in a typical earbud is usually around 12 mm in diameter and perhaps 3 or 4 mm thick. Piezoelectric MEMS speakers generally need to be larger still because their actuators don’t move very far.
By comparison, SonicEdge’s complete ultrasonic speaker module measures only around 7 × 7 × 1 mm. Inside that tiny package are approximately one thousand microscopic ultrasonic pumping cells working together. Each individual cell contributes only a tiny fraction of the total airflow, but collectively they produce performance comparable to that of the much larger moving-coil driver.
The advantages don’t stop with physical size. Traditional loudspeakers are astonishingly inefficient at converting electrical power into acoustic power. Much of the energy ultimately ends up as heat rather than sound. According to Moti, their ultrasonic pumping mechanism itself is roughly an order of magnitude more efficient than a conventional moving-coil driver. Today’s products still need to contend with system-level electronic overheads, but the underlying acoustic mechanism is already substantially more efficient—and future generations promise further improvements.
So, What Took So Long?
And so we come to the elephant in the room and the fly in the soup (I never metaphor I didn’t like). If the underlying architecture hasn’t changed very much over the past three years, why has it taken so long to reach this point?
The answer isn’t that the underlying technology didn’t work. The challenge was manufacturing. Back in 2023, SonicEdge intended to create a largely bespoke MEMS fabrication process. Unfortunately, MEMS manufacturing isn’t like digital CMOS, where designers can move relatively easily between foundries. Every MEMS process tends to evolve around its own unique sequence of fabrication steps, materials, and accumulated manufacturing know-how. Developing an entirely new process is a significant undertaking.
Eventually, the company made what, in hindsight (the one exact science), seems like an obvious decision. Rather than inventing a completely new manufacturing flow, SonicEdge redesigned its devices so they could be fabricated using essentially the same production process already employed to manufacture today’s capacitive MEMS microphones.
On the downside, that decision cost time. On the upside, it means the SonicEdge devices can now be manufactured in mature production facilities already producing billions of MEMS devices every year. The result is a much clearer path toward volume production, together with proven reliability, environmental robustness, and manufacturing scalability.
The Tide is Turning
Another interesting thing happened during those three years. Back in 2023, SonicEdge was essentially alone in promoting modulated ultrasound as the future of MEMS loudspeakers. Today, the landscape looks rather different. Companies such as xMEMS have introduced their own ultrasonic MEMS speaker architectures, while numerous researchers and semiconductor companies are actively exploring similar concepts.
Ironically, that’s probably good news for everyone involved. When multiple companies independently arrive at broadly the same conclusion, it’s usually a sign that the industry believes it’s heading in the right direction. The exact implementations differ, of course, but the fundamental principle—using ultrasound as the basis for miniature loudspeakers—is rapidly gaining acceptance.
Beyond Earbuds
Today’s engineering samples are primarily aimed at earbuds, hearing devices, and smart glasses. Numerous proof-of-concept (POC) projects are already underway with customers, although confidentiality agreements pr