Just curious, and I want the opinions of others, but I thought a 'speaker' refers to something that produces sound that can be heard at a distance. And if you put it in a headset/earpiece, it was called a driver?
Speakers don't need to be heard from a distance, your headphones are still speakers. Drivers are the individual components of a speaker that create the sound. So, yes, this is more of a speaker driver than an assembled speaker.
Looking closely at the frequency response curves, I doubt these will ever be "audiophile" grade due to the horrible bass response. You can't fight physics and bass notes require greater air movement than MEMS will realistically be able to provide. But the system does seem well-suited to smartphone speakers...
If Apple puts them inside of something and touts them as an improvement, then self-proclaimed audiophiles will certainly view them as capable of offering a premium listening experience. We all know audiophiles are just victims of product marketing that are trying to categorize themselves inside of a certain identity group in order to create a self-image under the false assumption that image will matter somehow to other people.
Self proclaimed audiophiles don;t bother with Apple, they use oxygen free copper cables that are optimized to carry charge in a particular direction, suspended over the floor several centimetres on high end ceramic doorsto... I mean ceramic sound enhancers, and they'll use it all to listen to old vinyls because it's the only thing that gives that "warm", low def sound to go with their $1000 cables.
Digital is lossless, but doesn't turn itself into analog which your ears understand, and if you keep the signal for too far down the pipeline you end up with integrated DACs and amps which are crappy because of both space/power restraints and business decisions.
Audiophiles have long considered Apple to be subpar, which is why they'd spend hundreds to mod iPods to try and make them sound "good", or why they'll spend thousands on boutique DAPs from audiophile brands instead of just using an iPhone. Apple actually isn't too bad audio quality (objectively speaking), and Beats improved in audio quality after Apple purchased them (although that might be more that good quality headphone speakers became very common so you almost had to actively work to make bad sounding stuff).
With regards to the lower register, I'd be curious if they're all showing the same thing (raw or compensated). And this is a first gen product, and I'd bet they'll be able to improve that. I think balanced armatures used to be similar, but now you've got drivers intended for that. Which, with that low distortion it could also be possible to get high fidelity via filters and/or EQ. It'll definitely be interesting to see how this develops. I wouldn't be surprised if they design the drivers differently in the future (maybe more circular?), and they're comprised of much more individual cells, possibly of different sizes. There's a lot else they can do to adjust the sound (modern custom IEMs use multiple drivers of different sizes, with different tubes, and some even mix balanced armature with dynamic drivers, which could also be a method they use).
I could definitely see these being popular for the coming technology (audio is actually really important for stuff like AR and VR) and even just "smart" ear phones that seem to be slowly taking over.
Yikes, I walked into some broad-bush bashing of audiophiles. Did you know a lot of audiophiles run counter to your stereotype, and are sick of the pseudo-science marketing some of the companies push? You should check out AudioScienceReview.com. An electrical engineer objectively and quantifiably measures audio gear, much in the same vein as this site does with PC gear. Theories get debunked. Esoteric brands get exposed. High value products get uncovered. Liking good music reproduced cleanly and forcefully doesn't necessarily mean you're an idiot, does it?
Thanks for the link. I've long bought pro audio gear as it's mostly neutral-sounding, well-built, and good value for money. But I'm always a fan of quantitative measurements.
Attempting to redefine the label "audiophile" with all the idiot-baggage attached to it is rather quite futile, if you ask me. But hey, if you like to assign that label to yourself and then fight it, well, whatever floats your boat...
The audiophile community already had meltdowns when Amir showed a $1000 DAC couldn't handle 16 bit audio let alone 24 bit stuff. The community is a joke i don't why people bother with it.
That you rarely miss a chance to bash gamers, weeaboos, audiophiles, or the like, probably says more about you than you're actually saying about them. It's so bad that find it surprising when I see a post by you that isn't bashing one of these groups.
I'm not trying to be defensive, either. I don't consider myself a member of those groups, but sometimes you gotta just look at what you're putting out into the world and ask yourself if it really needs to be said.
You can get away with less air movement with a sealed IEM, as they allude to in the press release.
Otherwise, yeah... you'd want a dynamic (or better yet, a planear magnetic) driver hooked up in some kind of low pass config in headphones or speakers. And then you'd either need an amped source, a seperate amp or a coil that's happy with 30V input.
They ignore that it the same for balanced armatures, The power needed a joke when BAs need nowhere near that for 122db output. This is a portable headphone not a home desktop set up. lol
Indeed. Open bench measurements are not at all indicative of what you get out of a sealed cavity. IEMs are the gold standard for accurate audio. No chance for multipath. If you want to see how dramatic the issue is, take your fanciest pair of headphones, put it next to your nicest mic, play white noise, and observe the magnitude response. Continue to observe the response in real time as you move the headphone around. Several dB across the mid+ bands for half an inch of movement. You need a fixed geometry if you want a repeatable magnitude response. The holy grail is FIR filtering on top to minimize group delay and flatten the response. Of course, that becomes less important with breakthroughs in drivers. This is exciting stuff.
I'm not sure what frequency graph you are looking at, but the one they show is indeed very good. You don't need 105db at 20hz, but these apparently do it.
This is prototype with no filtering and it has a flatter response than others up to 1k. But the graph, more importantly is actually showing consistency, low thd, and higher spl, at high end compared to competitors. Although I don't agree with speakers reaching such high spl due to hearing damage, it does show capability.
Since they don't list the A-weighted decibels, these PowerPoint slides are pretty misleading in terms of actual output you will hear. I dunno, I look forward to testing them out, but it's not like human-scale infrasonic frequencies are suddenly possible in this form factor. There's no replacement for displacement.
You don't actually want A-weighted. A-weighting is intended for environmental noise, while C-weighting is more suitable for evaluating audio gear.
The problem is that A-weighting gives better numbers, so manufacturers started cheating and specifying those instead. And once some did, the others had to go along with it to avoid seeming artificially worse.
Just remember that this is basically a first prototype, and that they have multi-speaker designs to cover the bass range. You can't expect a single driver to produce the whole range when it's this tiny ;)
The bass require more membrane movement only applies in free-space radiation where SPL ∝ 1/f^2 for a given fixed membrane volume displacement. However, for sealed chamber, such as earbuds in a TWS, SPL ∝ dV/V, for frequency below ~1KHz, where V being total volume of each canal.
So, with proper sealing of the housing and the tip, a pair of earbud using these drivers can produce bass as high as 115dB down pretty close to 20Hz.
If their bass goes up to (or, more precisely, "down to") 10 Hz then they might be able to reproduce the deepest bass sound an earbud* can provide, even if it is weak and with a high distortion (they do not mention THD below 200 Hz for a reason). Can these be used to make a full blown home theater subwoofer though? That will never happen. *specifically an earbud, not a large headphone. These have large drivers that provide good deep bass sound.
The Etymotic ER4SR already does this with a 5 mm full rage BA, With 0.1% ~ 0.55% THD. Bass responce is poor because they need a deep fit like the ER4 these drivers aren't fit for shallow fit, It just audiophile BS. Since the ER4 can reach 17.5KHz anways & since anyone over 21 their hearing peaks at 15.5KHz so the 20+ KHz dosen't mean much at all.
I wouldn't make an assumptions about what this tech can do as it advances .This is the first MEMs speaker, I remember when the first flat panel speakers were invented and produced generally crappy sound. They are now the standard in most electronics these days.
The advantages in power savings, ideally not cancelled out entirely by the presence of the companion chip, may help wireless earbuds operate for a longer period of time without recharging. That may help (somewhat unfortunately) encourage phone manufacturers to continue omitting 3.5mm jacks as they needlessly chase reductions in thickness under the guise that a thin object is somehow more premium than a thicker object - a thing tech journalists and a fairly large segment of the public have bought into as accepted fact without putting the critical thinking portions of their brains into gear to conduct rational analysis first.
Beyond that, I can't see these things having potential to scale up to sizes that replace speakers we currently use to push audio even in speakerphone modes or say on laptops. Cool tech, but its going to exist in a niche rather than broadly replace existing speaker technologies across the board for now.
*Claimed* reductions in thickness. Most of them don't reduce thickness and some phones even simply left some empty volume where a jack (at least the 2.5mm one) could fit.
Meh, I used to care about stereo-mini jacks before phones & bluetooth headphones started supported lossless codecs.
If I'm listening to a phone, it means I'm usually somewhere that would benefit from noise cancellation. And if using NC, then a cable is actually *worse* than lossless bluetooth, because it necessitates conversion to analog and back.
The biggest benefits of a headphone jack is that you can get more battery life out of your headphones (if they support both, like my old Bose QC35 did). Also, I guess it's nice to be able to plug it into a pre-amp or A/V receiver that doesn't have a bluetooth receiver, not that I've ever done it.
Ugh, no. Get a pair of CIEMs and be done with ANC. I use mine unplugged in the shooting range, to better effect than the thick heavy earmuffs provided there.
You'd probably do fine with multiple discrete chips that aren't all on the same wafer, which would save you from needing to have an essentially perfect wafer. Plus, you could use fewer chips, at quite a cost savings.
Amazing tech! This is definitely in Anandtech's realm.
Many audiophiles balk at the idea of internal amps, but pairing a amp specifically for the load its driving is advantageous in many ways. Thats doubly true for the DAC, which can run a DSP to flatten the frequency response of that specific system.
I'd be interested to see a design with one of these paired with a Class D DAC, maybe with a dynamic driver to help with low end frequency response.
Class D seems like it *could* be nearly the perfect solution for internally-amplified speakers, if developed enough. Not sure how common it really is, though. I thought Class D would take off more than it seems like it has (last I checked).
Its definitely taking off. What really sold me on class D was a bookshelf speaker set that use an integrated class D DAC/amp, and that was a few years ago.
Any design incorporating both drivers and an amplifier (and DSP, because silicon is cheap) has no excuse for not producing a perfect flat response from every available input, limited only be the lowest peak output level across the useful frequency range. Any distortions from a flat response (e.g. the oft-desirable 'tube amp sound' can be perfectly replicated via DSP by characterising and replicating the transfer function of that system.
Yes, I wholeheartedly agree. This is revolutionary if it works as well as it is claimed. Current TWS and wireless IEMs are plagued with poor driver selection and congested acoustic chambers, and this could help alleviate that and really help bridge the gap between wired and wireless. I personally prefer IEMs over speakers and full sized headphones as for me they are more comfortable, convenient, less fragile, and easily driven. I personally also like the sound characteristics and isolation offered by most IEMs (even if that may be 'unnatural'). Modern low power hybrid chips like the ES9218 are objectively fantastic for low impedance IEMs, and as you have said comparable chips could be tailor made for the earphone.
These devices are only suitable for small sealed applications such as earbuds due to the tiny amount of moving surface. In an unsealed application the bass and lower midrange frequencies would be lost - even in a larger sealed application such as over ear headphones low frequency notes will be a problem. For room speaker applications much larger moving diaphragms are needed, the QUAD electrostatic speakers have the same ultra low mass diaphragm but moved by electrostatics instead of piezoelectrics.
Very interesting, but indeed severely limited by the need for entirely proprietary amps, even if this eventually comes wired, it'd better take a different port. They didn't say if this would fry a regular amp or vice versa if plugged in?
It wouldn't short out a 3.5mm jack, it would just be really quiet. Amps for high impedance headphones max out at like 7V or 9V at max volume, while this thing supposedly takes 30V.
Wired versions will almost certainly be USB-C anyway. Which is fine for an IEM.
"Montara can be driven up to 30V peak-to-peak signals which is well above the range of your existing amplifier designs. As such, customers wishing to deploy a MEMS speaker design such as the Montara requires an additional companion chip, such as Texas Instruments’ LM48580."
which is why I'm not holding my breath. this is exactly why electrostatic speaker designs never really took off. really good electrostatic setups need to be mono-amped and cost ~70k for a full setup.
30V to an IEM is very different than 250V+ to a loudspeaker, especially when that IEM is more efficient than a coil. The LM48580 is like $1.50 on mouser.
sure it's a entirely different power envelope, but that doesn't mean requiring 30v is a trivial difference. the point is that it necessitates a fundamentally different power supply setup. also, $1.5 for a chip isn't really the whole story. you are dealing with a whole added circuit that gets designed into it. the thought of the back and forth for debug on that alone gives me a headache. nah. not ready yet. i want to see this get incorporated by some people first.
Budgetary quote for >1M pcs is actually less than a quarter And LM48580 is not complicated to get it up and running. However, you got a point, it does require a different amp, and this made it more complicate to do a drop-in replacement.
I frankly wonder how they handle bass / low frequency sounds. These require large speakers for obvious reasons, and yet they are able to fully reproduce bass sound up to 10 Hz(!) with just a 50.9mm² die? I suppose this is intended for small speakers of earbuds, but I wonder if (and how) it can be scaled up.
For sealed application, such as all those TWS earbuds, the acoustic pressure is generated by "compression mode" via volume change. So, those rule of "Low frequency require large driver membrane movement" does not apply here. And xMEMS Montara can produce 112dB SPL within the targeted TWS application domain.
Here's hoping 2021 doesn't turn to shiet like 2020. 5nm TSMC, 7nm Intel, DDR5, potentially PCIe5, and now a big shift in consumer audio... It's looking good in the tech world from here.
"The design is said to have extremely good distortion characteristics, able to compete with planar magnetic designs and promises to have only 0.5% THD at 200Hz – 20KHz."
Wow, I didn't know electrostatic IEMs were a thing!
Looks like some use tiny transformers, which means they won't work at low frequencies. Others claim to use a "low voltage driver." I don't know anything about that, but it looks like they're paired with dynamic drivers anyway.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
58 Comments
Back to Article
TennesseeTony - Tuesday, July 7, 2020 - link
Just curious, and I want the opinions of others, but I thought a 'speaker' refers to something that produces sound that can be heard at a distance. And if you put it in a headset/earpiece, it was called a driver?MattCoz - Tuesday, July 7, 2020 - link
Speakers don't need to be heard from a distance, your headphones are still speakers. Drivers are the individual components of a speaker that create the sound. So, yes, this is more of a speaker driver than an assembled speaker.liteon163 - Tuesday, July 7, 2020 - link
Looking closely at the frequency response curves, I doubt these will ever be "audiophile" grade due to the horrible bass response. You can't fight physics and bass notes require greater air movement than MEMS will realistically be able to provide. But the system does seem well-suited to smartphone speakers...PeachNCream - Tuesday, July 7, 2020 - link
If Apple puts them inside of something and touts them as an improvement, then self-proclaimed audiophiles will certainly view them as capable of offering a premium listening experience. We all know audiophiles are just victims of product marketing that are trying to categorize themselves inside of a certain identity group in order to create a self-image under the false assumption that image will matter somehow to other people.lilkwarrior - Tuesday, July 7, 2020 - link
I'm not sure many miss wired headphones for their phones. Wireless technology has certainly done enough for most.close - Tuesday, July 7, 2020 - link
Self proclaimed audiophiles don;t bother with Apple, they use oxygen free copper cables that are optimized to carry charge in a particular direction, suspended over the floor several centimetres on high end ceramic doorsto... I mean ceramic sound enhancers, and they'll use it all to listen to old vinyls because it's the only thing that gives that "warm", low def sound to go with their $1000 cables.Oxford Guy - Wednesday, July 8, 2020 - link
Snark aside, an audiophile would try to minimize analog cabling in favor of digital cabling as that's lossless.Snark not aside, you're way too low on the pricing. Look at the silver cables out there.
s.yu - Thursday, July 9, 2020 - link
Digital is lossless, but doesn't turn itself into analog which your ears understand, and if you keep the signal for too far down the pipeline you end up with integrated DACs and amps which are crappy because of both space/power restraints and business decisions.darkswordsman17 - Tuesday, July 7, 2020 - link
Audiophiles have long considered Apple to be subpar, which is why they'd spend hundreds to mod iPods to try and make them sound "good", or why they'll spend thousands on boutique DAPs from audiophile brands instead of just using an iPhone. Apple actually isn't too bad audio quality (objectively speaking), and Beats improved in audio quality after Apple purchased them (although that might be more that good quality headphone speakers became very common so you almost had to actively work to make bad sounding stuff).With regards to the lower register, I'd be curious if they're all showing the same thing (raw or compensated). And this is a first gen product, and I'd bet they'll be able to improve that. I think balanced armatures used to be similar, but now you've got drivers intended for that. Which, with that low distortion it could also be possible to get high fidelity via filters and/or EQ. It'll definitely be interesting to see how this develops. I wouldn't be surprised if they design the drivers differently in the future (maybe more circular?), and they're comprised of much more individual cells, possibly of different sizes. There's a lot else they can do to adjust the sound (modern custom IEMs use multiple drivers of different sizes, with different tubes, and some even mix balanced armature with dynamic drivers, which could also be a method they use).
I could definitely see these being popular for the coming technology (audio is actually really important for stuff like AR and VR) and even just "smart" ear phones that seem to be slowly taking over.
Moizy - Tuesday, July 7, 2020 - link
Yikes, I walked into some broad-bush bashing of audiophiles. Did you know a lot of audiophiles run counter to your stereotype, and are sick of the pseudo-science marketing some of the companies push? You should check out AudioScienceReview.com. An electrical engineer objectively and quantifiably measures audio gear, much in the same vein as this site does with PC gear. Theories get debunked. Esoteric brands get exposed. High value products get uncovered. Liking good music reproduced cleanly and forcefully doesn't necessarily mean you're an idiot, does it?mode_13h - Tuesday, July 7, 2020 - link
Thanks for the link. I've long bought pro audio gear as it's mostly neutral-sounding, well-built, and good value for money. But I'm always a fan of quantitative measurements.MrVibrato - Wednesday, July 8, 2020 - link
Attempting to redefine the label "audiophile" with all the idiot-baggage attached to it is rather quite futile, if you ask me. But hey, if you like to assign that label to yourself and then fight it, well, whatever floats your boat...2M2B - Wednesday, July 8, 2020 - link
The audiophile community already had meltdowns when Amir showed a $1000 DAC couldn't handle 16 bit audio let alone 24 bit stuff. The community is a joke i don't why people bother with it.mode_13h - Tuesday, July 7, 2020 - link
That you rarely miss a chance to bash gamers, weeaboos, audiophiles, or the like, probably says more about you than you're actually saying about them. It's so bad that find it surprising when I see a post by you that isn't bashing one of these groups.I'm not trying to be defensive, either. I don't consider myself a member of those groups, but sometimes you gotta just look at what you're putting out into the world and ask yourself if it really needs to be said.
brucethemoose - Tuesday, July 7, 2020 - link
You can get away with less air movement with a sealed IEM, as they allude to in the press release.Otherwise, yeah... you'd want a dynamic (or better yet, a planear magnetic) driver hooked up in some kind of low pass config in headphones or speakers. And then you'd either need an amped source, a seperate amp or a coil that's happy with 30V input.
2M2B - Wednesday, July 8, 2020 - link
They ignore that it the same for balanced armatures, The power needed a joke when BAs need nowhere near that for 122db output. This is a portable headphone not a home desktop set up. lolwillis936 - Thursday, July 9, 2020 - link
Indeed. Open bench measurements are not at all indicative of what you get out of a sealed cavity. IEMs are the gold standard for accurate audio. No chance for multipath. If you want to see how dramatic the issue is, take your fanciest pair of headphones, put it next to your nicest mic, play white noise, and observe the magnitude response. Continue to observe the response in real time as you move the headphone around. Several dB across the mid+ bands for half an inch of movement. You need a fixed geometry if you want a repeatable magnitude response. The holy grail is FIR filtering on top to minimize group delay and flatten the response. Of course, that becomes less important with breakthroughs in drivers. This is exciting stuff.Dug - Tuesday, July 7, 2020 - link
I'm not sure what frequency graph you are looking at, but the one they show is indeed very good.You don't need 105db at 20hz, but these apparently do it.
This is prototype with no filtering and it has a flatter response than others up to 1k.
But the graph, more importantly is actually showing consistency, low thd, and higher spl, at high end compared to competitors. Although I don't agree with speakers reaching such high spl due to hearing damage, it does show capability.
nathanddrews - Tuesday, July 7, 2020 - link
Since they don't list the A-weighted decibels, these PowerPoint slides are pretty misleading in terms of actual output you will hear. I dunno, I look forward to testing them out, but it's not like human-scale infrasonic frequencies are suddenly possible in this form factor. There's no replacement for displacement.mode_13h - Tuesday, July 7, 2020 - link
You don't actually want A-weighted. A-weighting is intended for environmental noise, while C-weighting is more suitable for evaluating audio gear.The problem is that A-weighting gives better numbers, so manufacturers started cheating and specifying those instead. And once some did, the others had to go along with it to avoid seeming artificially worse.
Drkrieger01 - Tuesday, July 7, 2020 - link
Just remember that this is basically a first prototype, and that they have multi-speaker designs to cover the bass range. You can't expect a single driver to produce the whole range when it's this tiny ;)jyl - Tuesday, July 7, 2020 - link
The bass require more membrane movement only applies in free-space radiation where SPL ∝ 1/f^2 for a given fixed membrane volume displacement. However, for sealed chamber, such as earbuds in a TWS, SPL ∝ dV/V, for frequency below ~1KHz, where V being total volume of each canal.So, with proper sealing of the housing and the tip, a pair of earbud using these drivers can produce bass as high as 115dB down pretty close to 20Hz.
Santoval - Tuesday, July 7, 2020 - link
If their bass goes up to (or, more precisely, "down to") 10 Hz then they might be able to reproduce the deepest bass sound an earbud* can provide, even if it is weak and with a high distortion (they do not mention THD below 200 Hz for a reason). Can these be used to make a full blown home theater subwoofer though? That will never happen.*specifically an earbud, not a large headphone. These have large drivers that provide good deep bass sound.
ABR - Wednesday, July 8, 2020 - link
Not seeing what you are seeing in the headphone graphs.2M2B - Wednesday, July 8, 2020 - link
The Etymotic ER4SR already does this with a 5 mm full rage BA, With 0.1% ~ 0.55% THD. Bass responce is poor because they need a deep fit like the ER4 these drivers aren't fit for shallow fit, It just audiophile BS. Since the ER4 can reach 17.5KHz anways & since anyone over 21 their hearing peaks at 15.5KHz so the 20+ KHz dosen't mean much at all.pashhtk27 - Wednesday, July 15, 2020 - link
But if this can do so in a 'TWS' at 1/4th the price, I don't see what's wrong.rahvin - Friday, July 10, 2020 - link
I wouldn't make an assumptions about what this tech can do as it advances .This is the first MEMs speaker, I remember when the first flat panel speakers were invented and produced generally crappy sound. They are now the standard in most electronics these days.PeachNCream - Tuesday, July 7, 2020 - link
The advantages in power savings, ideally not cancelled out entirely by the presence of the companion chip, may help wireless earbuds operate for a longer period of time without recharging. That may help (somewhat unfortunately) encourage phone manufacturers to continue omitting 3.5mm jacks as they needlessly chase reductions in thickness under the guise that a thin object is somehow more premium than a thicker object - a thing tech journalists and a fairly large segment of the public have bought into as accepted fact without putting the critical thinking portions of their brains into gear to conduct rational analysis first.Beyond that, I can't see these things having potential to scale up to sizes that replace speakers we currently use to push audio even in speakerphone modes or say on laptops. Cool tech, but its going to exist in a niche rather than broadly replace existing speaker technologies across the board for now.
close - Tuesday, July 7, 2020 - link
*Claimed* reductions in thickness. Most of them don't reduce thickness and some phones even simply left some empty volume where a jack (at least the 2.5mm one) could fit.mode_13h - Tuesday, July 7, 2020 - link
Meh, I used to care about stereo-mini jacks before phones & bluetooth headphones started supported lossless codecs.If I'm listening to a phone, it means I'm usually somewhere that would benefit from noise cancellation. And if using NC, then a cable is actually *worse* than lossless bluetooth, because it necessitates conversion to analog and back.
The biggest benefits of a headphone jack is that you can get more battery life out of your headphones (if they support both, like my old Bose QC35 did). Also, I guess it's nice to be able to plug it into a pre-amp or A/V receiver that doesn't have a bluetooth receiver, not that I've ever done it.
s.yu - Thursday, July 9, 2020 - link
Ugh, no. Get a pair of CIEMs and be done with ANC. I use mine unplugged in the shooting range, to better effect than the thick heavy earmuffs provided there.willis936 - Thursday, July 9, 2020 - link
Why not both?s.yu - Saturday, July 11, 2020 - link
Currently unfeasible? There are no ANC CIEMs AFAIK and further making that wireless in his argument complicates that even further.psychobriggsy - Tuesday, July 7, 2020 - link
I await the wafer-scale MEMS speaker/driver technology for floorstanding speakers.mode_13h - Tuesday, July 7, 2020 - link
You'd probably do fine with multiple discrete chips that aren't all on the same wafer, which would save you from needing to have an essentially perfect wafer. Plus, you could use fewer chips, at quite a cost savings.brucethemoose - Tuesday, July 7, 2020 - link
Amazing tech! This is definitely in Anandtech's realm.Many audiophiles balk at the idea of internal amps, but pairing a amp specifically for the load its driving is advantageous in many ways. Thats doubly true for the DAC, which can run a DSP to flatten the frequency response of that specific system.
I'd be interested to see a design with one of these paired with a Class D DAC, maybe with a dynamic driver to help with low end frequency response.
mode_13h - Tuesday, July 7, 2020 - link
I like the way you're thinking!Class D seems like it *could* be nearly the perfect solution for internally-amplified speakers, if developed enough. Not sure how common it really is, though. I thought Class D would take off more than it seems like it has (last I checked).
brucethemoose - Wednesday, July 8, 2020 - link
Its definitely taking off. What really sold me on class D was a bookshelf speaker set that use an integrated class D DAC/amp, and that was a few years ago.edzieba - Thursday, July 9, 2020 - link
Any design incorporating both drivers and an amplifier (and DSP, because silicon is cheap) has no excuse for not producing a perfect flat response from every available input, limited only be the lowest peak output level across the useful frequency range.Any distortions from a flat response (e.g. the oft-desirable 'tube amp sound' can be perfectly replicated via DSP by characterising and replicating the transfer function of that system.
pashhtk27 - Wednesday, July 15, 2020 - link
Yes, I wholeheartedly agree. This is revolutionary if it works as well as it is claimed. Current TWS and wireless IEMs are plagued with poor driver selection and congested acoustic chambers, and this could help alleviate that and really help bridge the gap between wired and wireless.I personally prefer IEMs over speakers and full sized headphones as for me they are more comfortable, convenient, less fragile, and easily driven. I personally also like the sound characteristics and isolation offered by most IEMs (even if that may be 'unnatural'). Modern low power hybrid chips like the ES9218 are objectively fantastic for low impedance IEMs, and as you have said comparable chips could be tailor made for the earphone.
These are interesting times for audio lovers.
Duncan Macdonald - Tuesday, July 7, 2020 - link
These devices are only suitable for small sealed applications such as earbuds due to the tiny amount of moving surface. In an unsealed application the bass and lower midrange frequencies would be lost - even in a larger sealed application such as over ear headphones low frequency notes will be a problem. For room speaker applications much larger moving diaphragms are needed, the QUAD electrostatic speakers have the same ultra low mass diaphragm but moved by electrostatics instead of piezoelectrics.s.yu - Tuesday, July 7, 2020 - link
Very interesting, but indeed severely limited by the need for entirely proprietary amps, even if this eventually comes wired, it'd better take a different port. They didn't say if this would fry a regular amp or vice versa if plugged in?brucethemoose - Tuesday, July 7, 2020 - link
It wouldn't short out a 3.5mm jack, it would just be really quiet. Amps for high impedance headphones max out at like 7V or 9V at max volume, while this thing supposedly takes 30V.Wired versions will almost certainly be USB-C anyway. Which is fine for an IEM.
s.yu - Thursday, July 9, 2020 - link
I assumed more than mere voltage, or else the amps for electrostats need far higher voltage.wr3zzz - Tuesday, July 7, 2020 - link
Hasn't MEMS already being used in hearing aids for a while now and was already adapted into some IEM?vol.2 - Tuesday, July 7, 2020 - link
"Montara can be driven up to 30V peak-to-peak signals which is well above the range of your existing amplifier designs. As such, customers wishing to deploy a MEMS speaker design such as the Montara requires an additional companion chip, such as Texas Instruments’ LM48580."which is why I'm not holding my breath. this is exactly why electrostatic speaker designs never really took off. really good electrostatic setups need to be mono-amped and cost ~70k for a full setup.
brucethemoose - Tuesday, July 7, 2020 - link
30V to an IEM is very different than 250V+ to a loudspeaker, especially when that IEM is more efficient than a coil. The LM48580 is like $1.50 on mouser.vol.2 - Tuesday, July 7, 2020 - link
sure it's a entirely different power envelope, but that doesn't mean requiring 30v is a trivial difference. the point is that it necessitates a fundamentally different power supply setup. also, $1.5 for a chip isn't really the whole story. you are dealing with a whole added circuit that gets designed into it. the thought of the back and forth for debug on that alone gives me a headache. nah. not ready yet. i want to see this get incorporated by some people first.brucethemoose - Wednesday, July 8, 2020 - link
It doesnt require 30V past the LM48580 though. A quick look at the spec sheet suggests it can be driven off a 5V USB pin.So you need the amp, whatevers requred for the USB spec, and... thats pretty much it.
jyl - Wednesday, September 2, 2020 - link
You're correct. Actually LM48580 can generate 30V peak to peak from 3V, i.e. the lowest voltage output out of batter before it cut itself off.jyl - Wednesday, September 2, 2020 - link
Budgetary quote for >1M pcs is actually less than a quarterAnd LM48580 is not complicated to get it up and running.
However, you got a point, it does require a different amp, and this made it more complicate to do a drop-in replacement.
Santoval - Tuesday, July 7, 2020 - link
I frankly wonder how they handle bass / low frequency sounds. These require large speakers for obvious reasons, and yet they are able to fully reproduce bass sound up to 10 Hz(!) with just a 50.9mm² die? I suppose this is intended for small speakers of earbuds, but I wonder if (and how) it can be scaled up.jyl - Wednesday, September 2, 2020 - link
For sealed application, such as all those TWS earbuds, the acoustic pressure is generated by "compression mode" via volume change. So, those rule of "Low frequency require large driver membrane movement" does not apply here. And xMEMS Montara can produce 112dB SPL within the targeted TWS application domain.Unashamed_unoriginal_username_x86 - Tuesday, July 7, 2020 - link
Here's hoping 2021 doesn't turn to shiet like 2020. 5nm TSMC, 7nm Intel, DDR5, potentially PCIe5, and now a big shift in consumer audio... It's looking good in the tech world from here.Oxford Guy - Wednesday, July 8, 2020 - link
"The design is said to have extremely good distortion characteristics, able to compete with planar magnetic designs and promises to have only 0.5% THD at 200Hz – 20KHz."What about electrostatic?
brucethemoose - Wednesday, July 8, 2020 - link
Wow, I didn't know electrostatic IEMs were a thing!Looks like some use tiny transformers, which means they won't work at low frequencies. Others claim to use a "low voltage driver." I don't know anything about that, but it looks like they're paired with dynamic drivers anyway.
Oxford Guy - Wednesday, July 8, 2020 - link
"xMEMS is taking the fabless model: monolithic, capable of much higher voluke, very scalable, with less than 1/10 of personnel needed."I’m assuming they’re keeping the toes. Don’t know about the hands and feet, however.
zodiacfml - Thursday, July 9, 2020 - link
Interesting! Never heard of this!