New Part Day: MEMS Loudspeakers

MEMS, or Micro ElectroMechanical Systems, are the enabling technology that brings us smartphones, quadcopters, tire pressure monitors, and a million other devices we take for granted today. At its most basic level, MEMS is simply machining away silicon wafers to make not electronic parts, but electromechanical parts. The microphone in your cell phone isn’t an electret mic you would find in an old brick phone from the 80s — it’s a carefully crafted bit of silicon, packed in epoxy, and hanging off a serial bus.

Despite the incredible success of MEMS technology, there is still something in your smartphone that’s built on 19th-century technology. Loudspeakers haven’t changed ever, and the speaker in your newest iThing is still a coil of wire and some sort of cone.

Now there’s finally a MEMS loudspeaker A company called USound has developed the first loudspeaker that isn’t just a bunch of wire and a magnet. This is a speaker built from a silicon wafer that can be as small as 3 mm square, and as thin as 1 mm. Since these speakers are built on silicon, you can also add an amp right onto the package. This is quite literally a speaker on a chip, and we’d bet that there are already engineers at Samsung looking at stuffing this into a flagship phone.

ST and USound announced these extraordinarily small speakers would actually be made, but so far it’s been just that — an announcement. This changed at CES where ST demonstrated VR goggles with multiple MEMS speakers. Does this mean MEMS speakers are on their way to Mouser and Digikey? We eagerly await the product announcement and demo dev board kit.

47 thoughts on “New Part Day: MEMS Loudspeakers

  1. My first thouht : it has to be used in an array to make it work. Just like the radar array it would allow to shape and direct a sound beam very easily.
    There is a (in)famous company which tries to accomplish exactly that : Ubeam.
    And now USound – Ubeam. Get it?

    1. phased array
      Nice idea, but impractical for audio frequencies. To make a wavefront directional the wave front needs to be mulitple times bigger than its wavelength. For a 1kHz soundwave at 330m/s the wavelength is 0.33m and a source of this size is omnidirectional. For some directionality the array should be at least 1 meter. Impractical for mobil use. But it’s been done with ultrasonics because of shorter wavelength.

    1. Going by the datasheet on their website and looking at the datasheet for the VR reference headset, it seems to suggest that they used a more conventional woofer unit and only had the MEMS for the higher-end (although the datasheet for their MEMS units suggests that, if closely coupled, it does have a reasonable response down to ~10Hz… and all the way up to 80kHz!)

    2. Maybe they could make an avalanche configuration, something like laser (or rail gun) emulation for sound: place little MEMS loudspeakers laterally along (perpendicular to the wall of) a tube (a straw) and then drive each of them with a little lag, so that as sound wave passes from the starting end towards the output end over each little loudspeaker, that little loudspeaker adds its little boost to the wave. Now, take a matrix of such straws and place them in a box, drive them in sync so that they produce a flat wave front. It is like a directional microphone array, only backwards.

      They could thus even make strong sirens, for fixed set of resonant frequencies, by making the tubes circular with a fork for the output (like figure 6 or figure 9, or like a “paragraph” character – a circle with two forks and two ends)

          1. Yes, thanks! One “tail” would be a “source end”, where waves are induced, amplification (pumping) would be done in a ring, and the opposite tail would be a sound power outlet (horn)

        1. Well, not exactly, because for an exact analogue we would have to have some sort of metastable bubbles which would pop with a small bang when wave front reaches them, and a physical mechanism of producing them constantly, but idea is that, as the wave travels down a channel, it gains amplitude by a number of small additions, which, in broad general terms, is what lasers do.

          Only, unlike in lasers, here the amplification is not self-induced, and the loudspeaker driving circuitry must be careful to match the needed lag with the speed of sound, which could vary (and position, which could also vary, because the tube could shrink or elongate) with varying temperature and/or air pressure.

  2. U wot m8? There’re loads of loud speaker technologies that don’t involve just wire and a magnet. Plasma, electrostatic, piezo, and lots, lots more. Even 2 minutes on wikipedia would have been enough research to find that.

      1. Yeah, it doesn’t even look like a phased array of tiny elements, just a single moving element made from a flexured piezo. My day-job company was trying to put piezo speakers in cellphones in the mid 90s, but it never took off (possibly because 90s cellphones sucked and weren’t really suited for music playback yet). In that case the piezo element was bonded directly to the inner housing and used the phone’s back itself as a sounding board.

  3. These should minielectromechanical. Just moving air something bigger has to move and that’s at high frequencies only. Maybe the Heil air motion transformer squeezing in an array. The pain devices use piezo discs.

  4. except, you know… physics.
    Sound has wavelengths. So unless you’re a dolphin, you’re going to want larger speakers. Even if you mount these in an array. What would be the equivalent cone excursion when it’s only 1mm thick?

  5. Even if they’re not great, the tiny size is bound to allow some interesting use cases. Already mentioned in the comments was beam forming. Also mentioned was motors … maybe vibrator motors? How about using one as a tweeter in dual-driver earbuds. Any headphones built with an array immediately becomes surround sound. Speakers for space-limited areas (phones, laptops) that are actually strip-shaped instead small speakers in a huge decorative grill. One of the pages mentions that they can be small enough to fit in the ear canal without blocking normal sound – that tiny size would be fantastic for electronic hearing protection based on earplugs. Also such close proximity means less losses aka lower power requirements. It’s gunna be cool!

  6. Ugh. I just can’t believe that these things will give the same sound quality as an actual speaker. It’s the physics, sound is a wave with a wavelength.

    I wouldn’t care except that it seems to me that for years cellphones had crappy audio where it was much harder to understand what the person on the other end was saying. My S8 is the first cellphone that I didn’t think sucked as a phone since the days of the flip phone ended! I blame a tightening of compression rations for that. What it tells me though is that cellphone producers don’t have to consider audio quality and they can still sell their wares. When these things hit they might be awful and yet there may be no alternative available to a consumer.

      1. Simple, come up with a measure:

        Step 1: Computer generate a set of perfect waveforms. Play waveforms in near ideal conditions with several top of the line speakers and record with calibrated mic(s).
        Step 2: Rank the speakers based on how well they match the original waveforms on a 100 scale, where each count is an order of magnitude closer to perfectly reproducing the waveform.
        Step 3: Market the new Bose speakers as a “7 Maave” speaker
        Step 4: ???
        Step 5: Profit

  7. Ambient prime mover emitters by any means (MEMS, moving coil, etc.) can sound good to a Human at relatively broad frequency ranges (e.g. for music), But the problem is two-fold: 1: The prime mover requires a pre-distortion linearized driver. This is typically done in non-adaptive emitters in look-up tables (LUTs) which depend highly on the emitter’s enclosure. 2: The pre-distorition linearized driver requires large amounts of energy to compensate for the small emitter physical aperture. Everything about the pre-distortion equalizer profile depends on the specific product, and may even depend on each emitter populating the product which leads to on-of calibration of devices at the production level. Making a small physical emitter is just the tip of the design iceberg. In production, that’s where the real work begins. [Of-course we’re talking about PWM type drivers here. Pure analog drivers are out of the question, but there are hybrids emerging]

  8. When fixing an arcade machine that stuck in a reset loop, I noticed one of the 8255 PIOs giving out pulses of noise. The chip itself that is. Tried different manufacturers, made no difference, they all do that. At least Intel, Nec and Mitsubishi.
    Now to write a program that makes music using 8255 PIOs as speakers.
    Yes, it was very quiet, but producing audible sound without a speaker of whatever sort is pretty freaky.
    BTW I had the board on my bench, connected to nothing but a PSU (which incidentally makes noise of its own, but that’s the cheap primary PWM going through the cheap transformer).
    (btw. a 74LS74 that produces the NMI signal was the culprit)

  9. Devialet Phantom speaker technology just changed the world of audiophonics as I understand it , a spherical enclosure with the sound producing element sealed inside , along with some advanced electronics/drivers , apparently have made low energy high performance possible. Yes, I’m a layman, just something I read today, good hunting gentleman…

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.