Ok, still hopeful, but unconvinced, I want to break it down simply and think about how these things can seem to defy the laws of physics, starting with the most obvious and move to some more complex ways. I may be wrong in some math, and their may be some discussion of that, but what I hope to achieve is a framework for how to think about the ideas, and refine them together and drop in some more highly refined equations. Please I ask that the thread is gentle, builds upon the theories, or critiques individual parts, not ablates the whole. This thread hopes to understand, just don't post, "you don't understand, it's magic", because there are other threads for that.
I've read the technical specs that are available. The pixel diaphragms are the same as normal speakers and work in the same way, ie. they need to displace volume of of air which radiates outwards, pressure dropping according to the inverse square law, and hits our eardrums with fractions of the input energy. The only difference is the scale and of course that they are driven, electrostatically rather than electromagnetically.
1. So what we know so far, is that the displacement is tiny, so crudely, the max volume of displacement of any speaker is equal to the width of the diaphragm multiplied by the throw (forwards and backwards movement). These pixels, are not going to be loud individually, and even 1024 of them on a wafer are going to be low volume. What we need is a way to think of this sanely, so take a container the width of your loudspeaker, work out how far you can push the speaker in, and double it (because speakers move outwards too), then cut the height of the container down to that size, now fill the container with liquid, and pour it over the floor until the puddle is wafer thin, yes that's a big puddle, and that's how huge a wafer thin speaker will be in order to make displace the same volume and send that energy to your ear resulting in same sound pressure. So far it doesn't make sense that these chips will make the same volume and take up less footprint than normal speakers, you decrease the depth, you must increase the footprint. Audio pixels are promising smaller footprints, not larger, so far it doesn't check out.
2. Second level of complexity in our thought experiment: the resonant / fundamental of these chips from memory is 4khz - and would sound like tinnitus. Our container of spilled liquid is only optimally efficient at 4khz, if you drop it down to 2khz, you'll half the energy, 1khz is a quarter, 500hz is an eighth, 250hz (close to middle c) is going to be a sixteenth of the energy, down at the lower end of our hearing at 20hz This is pretty low. And still, this isn't a nice sine wave, it would begin to look pretty square at those lower frequencies due to the latent time, not at all able to produce an arbitrary (non square) wave and be musically useful. So the case for producing great, loud sound is drastically reduced. At bottom range of our hearing, say a generous 31.25 hz we can only transmit 128th of the power into sound waves. To summarise, the already low sound output potential of the chip is being wasted because the potential is mostly in high frequencies; when made to produce low frequencies the chips are mostly idle. A traditional speaker is comparatively in its element producing these long slow movements of air.
3. Okay so what is obvious at this point, is that the above wouldn't work to produce low frequencies acceptably loud, and acceptably round enough to reproduce an arbitrary waveform (musically useful), so what is possibly being done, and is alluded to is the chips themselves are orchestrating the pixels to produce a kind of Mexican wave, over long periods of time where each pixel fired is added to the pressure front in sequence. It must be noted however that the low volume in #2 is not increased by this action, instead it trades off even more volume to convert a square wave (produced by 1024 individual pixels oscillating simultaneously) into a more rounded wave (1024 pixels asynchronously) in order to be musically useful. remember in #2 low frequencies are low volume because the potential of the chip isn't being used, in fact from the charts I've seen low frequencies aren't even (or are barely) audible until multiple chips are stacked together, giving the Mexican wave effectively a bigger stadium. At this stage in our thought experiment, the promise of having audio pixel footprint anything but larger than traditional speakers (especially at truthful musical frequencies) is going to be highly unlikely bordering on ludicrous.
4. Placeholder possibility - Audio Pixels are being used as high frequency pumps to create pressure which is stored and undergoes delayed release at high volumes / low frequencies to create waves
This is way too many steps ahead of the cutting edge of current MEMS technology.
5. Placeholder possibility - Coandă effect - using air inducement
The problem with this is that objects that do this require an input air stream, audio pixels don't seem to have a design that allows to collect extra air.
6. Placeholder possibility - Beats
When two high frequencies are created together, but differing by a small amount, the differences can be audible heard as frequencies in their own right, these are called beats, potentially we could create two hypersonic (too high to be audible) waves, and redraw them with useful signal overlaid in their differences, and we should be able to hear only the signal; also because the two carrier waves are high frequency, they carry a higher volume which can be manifest in different ways. With miniature loudspeakers such that MEMS chips could provide, and circuitry to calculate precise driving, this could work. Something I'd considered when I was tuning my guitar once, but I've never done the maths. To be honest I think this is the most likely route. A quick Google search for "hypersonic tones create beats" leads us to this little gem
https://acoustics.org/pressroom/httpdocs/133rd/2pea.html and of course
https://en.wikipedia.org/wiki/Sound_from_ultrasound. I'm convinced this is the only potentially viable way of producing music from small packages. Now reading further, many of the claims Audio Pixels claim, eg. beam directionality, seem to fit with sound from ultrasound concept. So at this stage, just today, I've gone from a skeptic to thinking it's highly possible this will be done at some stage, and I'm a lot more interested, but the lack of public information thus far is quite frustrating.
Continues...
יום טוב
Laurence
(20min delay)