*Addendum* - The "tactile buzzer" is just the battery. Brain fart, not sure where my mind was when writing that out. Whoops! 😅 - Some folks were curious how the middle gap between the layers is made. I don't know for sure, but it's likely that they used a sacrificial silicon dioxide* (SiO2 aka glass) layer in between the two "functional" layers. So the process flow would have been: pattern and etch bottom layer's array of holes, deposit a thick layer of SiO2, deposit and pattern subsequent polysilicon layers (doped or undoped), then finally etch out the SiO2 layer with HF or plasma. Then flip over and DRIE etch the big cavity from the backside. That is also likely why the dimples are dimple-shaped... they are just following the curve of the sacrificial layer that was filling the holes from the very first layer. *I suspect SiO2 because there was some EDS data (not shown in the video) which showed high concentrations of SiO2 right at the broken edge between the layers, where they meet at the "bulk" of the substrate. I think that's leftover from the sacrificial etch process.
making conceptual basic 3d models is not that hard if someone has a good sense of 3d imagination. i find it crazier that he is able to break a thing many times smaller than a hair in two 🤯
@@DaveNagy1for sure hand made. The model itself would be pretty quick to make, but texturing, creating the environment, animating all that would take a decent amount of time. I'd imagine this would probably be something he modelled then sent out to an animator to render up. There's not a lot of cross over between cad modelling for engineering and pretty stuff sadly. That said I wouldn't put it past him to do it all himself, legend.
Your video making skills are off the hook. I love the CGI of the microphone and all the beautiful imagery. Your hard work to make these videos is super appreciated.
I am in awe, I have gotten into microelectronics lately after watching lots of Asianometry videos and this visual exploration of this microphone was astonishing. Seeing the small features contrasted with a human hair really put everything in perspective in a wonderful way.
It's perhaps not that surprising that you could create a capacitive microsphone from silicon, but what's mind-blowing to me is that it's such a good microphone. It doesn't seem obvious that you would be able to make a microphone that could do anything other than simply detect the presence of sound. Insane engineering to get to a useful microphone
key to it all is the perfect repeatability and precision of silicon lithography. the signal is very weak and the frequency response is terrible but it can be precisely characterised and corrected using the built in integrated circuit.
What might blow your mind too are a type of sensor, found in most smartphones, smart watches, etc., which is closely related to MEMS microphones in their construction (and which in fact came first): MEMS barometers. Rather than the membrane being moved by sound, it’s got a sealed cavity so that as external air pressure changes, the air in the cavity expands or contracts, making the membrane bow in or out, changing its distance and thus the capacitance. What’s incredible is that they’re so sensitive that this is what your phone uses to detect changes in your altitude. Yes, your phone tells how many floors of stairs you climbed by measuring the difference in air pressure. So you might think it’s sensitive enough to measure a meter or two of altitude, right? Nope, they have a resolution of a few _centimeters._ I find it truly incredible that these sensors can actually tell the difference in air pressure over literally one palm’s width of altitude.
@@tookitogo it’s truly awesome tech. I remember figuring out that one of my old android phones had a Yamaha component for an accelerometer. Been pretty hooked on micro tech since then.
@@unixuxyeah... Instead they turn to magic creatures and conspiracy theories. You know, no way the Egyptians could have built the pyramids. Because people sure were dumber thousands of years ago. 😂
The 3d model of the mic kind of blew my mind. I loooooooooooove seeing stuff under electron microscopes, thank you for making this. Fantastic all around.
Also not a seasoned audio engineer here but my trivial explanation for the cavity below the membrane is, that it provides a neutral pressure reference against the outside. Thus the microphone becomes omnidirectional. If it would be open from the back, sound waves coming from the side would not be picked up. Thank you for that brilliant deep dive of a video!
To my best knowledge, the size of the cavity vs. the diameter of the central hole define your lower cut-off frequency, otherwise this thing would be driven into saturation by low-frequency or static pressure.
@@commander-tomalak that's my thought. Vented microphone to control for VLF and barometric pressure, the cavity for resonance, the steps to tune and reflect various frequency harmonics. Impressive, given the frequencies used in narrowband telephony is around 300 - 3400 Hz, wavelengths ranging from around 45" - 4"! Yep, a quick lookup shows they're called a MEMS resonant microphone array. Here's a discussion on active noise cancellation using the technology. www.ncbi.nlm.nih.gov/pmc/articles/PMC7978172/
Absolutely crazy that something like this is 2x inside a $10 headphone, so each maybe 10cents, at most. 300mm waver gives maybe 50'000, so a whole waver with bonding and everything for less than $5000. That "buzzer" most likely is the battery, btw.
There's probably just a single mic in one of the earbuds. Lookup "digikey mems mic" and you can find them starting at 47 cents each if you buy 1000. This is a US retailer selling reputable parts. If you get them directly from China, which is probably where these earbuds originate, they'd of course be significantly cheaper.
I told Santa that I wanted an SEM for Christmas. Unfortunately, he said that I was too heavy, get the fuck off of his lap. Back during the last Ice Age and I was in school, our junior high and high school had donated TEM units, which we were allowed to use. By the time my kids went to school, the electron microscopes were long gone, as were the optical microscopes.
Dude, this is awesome to see so detailed and even broken open. And on top of that, as if that wasn't enough, you explain it all as well and even use super beautiful renders for that explanation!
Wow, excellent presentation. The SEM images and CGI blend perfectly. What an amazing piece of technology. I wonder if the dimples in the top layer are for controlling the stiffness of the disk.
Thanks! I believe the dimples are _mostly_ an artifact of the manufacturing steps to make one. There are a few ways it could have been made, but my current theory is: pattern and etch the base substrate giving nice clean holes, deposit a layer of glass on top (which will naturally form rounded dimples over the holes), deposit another layer of undoped polysilicon and then a doped polysilicon layer, then finally etch out the sacrificial glass layer (with HF or plasma) leaving the gap between the two layers. Finally flip and etch the big cavity. Just a guess but it makes sense to me. The holes in the lower layer are to help air move past it with minimal resistance, but the dimples on the surface aren't really needed. So I think it's leftover from the layered nature of fabrication.
@@BreakingTaps You mean a totally flat disk without dimples could not be fabricated? I wouldn't expect that, but I don't know anything about processes at this micro level. I'm just blown away that they actually work as well as they do. Are neodymium magnets used at this scale?
@@professordeb It's technically possible to get a flat disk on top of the hole'y layer, but it would be a lot more work. There's another process called "chemical-mechanical planarization" which is basically a super fancy sand paper for wafers 😁 It's used to flatten the top layer by grinding/polishing until all the ridges are gone. It's often used on high density microchips like computer CPUs, because you have soooo many layers that everything starts to get rounded. So they periodically flatten it with the planarization tools. So to get a flat surface for this device, you'd deposit a really thick sacrificial layer, then grind it back flat, then proceed with the next steps. But if you don't _need_ it to be flat, you can skip and save money. Magnets aren't used a lot at this scale because (I think) the magnetization process needs high temperatures and it can be difficult for the devices to survive. Although I've seen some papers about using laser-heating and such, so I'm sure it's doable. At this size, electrostatic, thermal and piezo mechanisms tend to be more common.
the way this microphone works is, there is a small hole that allows to equalize the pressure between the inside and outside of the chamber slowly later when pressure is applied to the top the fluctuations are relative to the mean pressure.
The cavity is a resonant chamber. The microphone is referred to as a MEMS resonant microphone array. Pair them up and one can have quality active noise cancellation. The cavity is a resonant chamber, the steps for different frequencies.
I work at a very old 200mm semiconductor fab as an equipment engineer. One of my processes is polysilicon deposition through LPCVD. Hearing these terms in a video about mics in earbuds is awesome.
You just spoke about almost every topic I had in my master's degree lecture "Physical Sensors in Silicon Technology", we also had the etching process RIE (Plasma etching - Reactive ion etching) explained in details in there. Thank you for making this video, I just finished my university degree and it's cool to see some practical stuff for a change!
Thank you for making these videos. They really help give perspective on this extremely tiny yet extremely impactful part of all our lives. Plus the SEM images and CGI you make are absolutely beautiful to look at.
This is great! Hah, I just bought these ONN buds on sale for ~10$ and they work great as a basic hands/wires free headset. I was marveling at the amount of tech crammed into these cheap lil guys and you've revealed their innermost secrets :D Always enjoy your microscopy.
I’d love to know more about the hole pattern. Most is a hexagonal fill which is good for maximum density and uniformity. Edges are concentric rings, and in between is a hybrid.
Why there is such a big cavity: my guess is that it is simply to do with ease of manufacture. They first make all the structures on top, then flip it over and etch through from the backside. Importantly they also intentionally leave a controlled-size hole that allows internal pressure to equalize over a controlled time (e.g. if device takes a plane ride or happens to be put in a vacuum during further processing), not too slow but not too quickly that the device wouldn't be able to pick up bass. Given the tiny size of the cavity, it can't have anything to do with acoustic resonances.
These microphones typically have a digital output using "pulse density modulation", where the rate of toggles encodes the analog signal value. The three ports coming off the control die are almost certainly power. ground, and audio out. Also +1 for DRIE video. That was the first thing I noticed when you cracked it open. The Bosch process is cool!
I counted 7 heavy etchings on one component, couldn't get a good count for thinner etchings for things like the resonant chamber. Then, I considered how many ways I could badly injure myself on the equipment that builds these devices... Ion beam, HF, yeah, gotta be a pain to maintain those machines!
I would guess the large cavity behind the membrane matters for the microphone's frequency response: Most practical microphones don't want to react to slow changes in ambient air pressure, because those can easily be much bigger than typical sound pressure levels, and could blow out the membrane. This is what the small hole in the middle of the membrane is for, to let the pressure equalize on both sides of the membrane (equivalent to a high-pass filter). Of course, if the equalization is too fast, then it can also equalize out low-frequency sounds, which would impact the frequency response of the microphone. The speed of this equalization depends on the hole size and the volume behind the hole (similar to a Helmholtz resonator with an additional loss term), so the manufacturer will tune either the hole size or the volume behind the membrane to set this frequency to a sensible value - I think 1-2Hz are typical for typical electret microphones. I would guess, then, that the hole is already as small as feasible in this process, for some reason or another. Then it would make sense for the manufacturer to make the volume larger (requiring more etching steps) in order to improve the frequency response at low frequencies.
The amplifier for the microphone looks quite interesting too. Looks like too many parts to be purely an analog amplifier. I wonder if they're driving the microphone with AC and de-modulating to get the audio?
Really cool video dude! I just hooked up a MEMS mic to my WLED display for music reactivity. Cool to see exactly how these little pieces of “fly sh!t” actually work! Merry Christmas!!!
Ordinary ceramic capacitors can also respond to sound. I've seen speculation that this effect could be used to turn ordinary electronics into surveillance devices but I haven't seen a proof of concept yet.
Interesting to see the ridges inside that microphone cavity. At first i thought they were there to help with echo and reflections, similar to how some speaker boxes deal with it. Now I'm wondering if the etching process is calibrated to make those ridges a certain size specifically for essentially tuning it.
around 3:23 that chip looks like it was cut by shearing, instead of with a cutoff wheel. a cutoff wheel leaves a clean edge, shearing leaves that sharp clean beginning of cut then it kinda just breaks
Just one correction. The balls at the ends of the bond wires are tiny solder bumps and not Au balls. Solder bumping, wire bonding, die stacking and 3D packaging in general would make a great episode!
Commonly, after making the bond to the lead frame (after first bonding to the bond pad on the chip) the wire-bond machine severs the wire with an electric arc. This arc produces a spherical blob of molten wire, which cools and solidifies. That little ball is right below the surface of the bonding foot. So when the bonding foot is pressed down onto the next bond pad on the chip, it compresses the ball onto the bond pad, creating the electrical and mechanical connection to the bond pad. So the little balls are of gold or aluminum, whatever the bond wire metal happens to be.
Now, where chips are bonded directly to each other or PCB, the balls are usually made of solder or tin or indium. Actually, they are referred to as bumps for this kind of bonding operation.
What is interesting about the bond-wire at 4:40 is that the bondwires seem to be mounted on top of a gold ball. If it was ball bonding, I would expect a clean transistion from the 'mushroomed' out ball to the wire, but this is not the case, you can clearly see that the wire itself has been smooshed (through what looks like wedge bonding) onto the gold ball.
I'm kinda surprised you didn't mention the outer layer's hundreds of dimples. They're convex parabolic cones (Dimples 3:37) that directs the sound waves around the the actual mic pickup. Being such a small device the time delay of the different waves is infinitesimal. All this multi magnifies minute sound waves much like cupping your hand around your ear. KEWL VID, THANKS
The geometry of the whole thing is interesting. The dimples probably make the upper membrane more stiff, and might have been calculated or arrived at experimentally. Stiffness is useful in this application as you don't want to respond to the inertia of the membrane itself (flopping around). The chamber below is indeed for resonance, and it should be possible to physically measure it and calculate what frequencies it is sympathetic to; my bet is that it works really well for key frequencies in "toll-quality" audio (voice). I'm also betting that the device is good at picking up low audio frequencies from whatever it's mounted to, letting the whole earbud vibrate as an extension of the sensor.
Spitballing... I imagine the dimples are to make the upper membrane less stiff. Stiff membranes resonate---like a drumskin---and you want as little resonance as possible; while still attaining high deflection for gain. (High Q, high amplitude voltage ringing, even above hearing frequency, can upset amplifiers.)
I'll add my own unqualified take into the pot: it's silicon. It probably has a really wonky breakup pattern that the dimples help with! Only joking! It's a cheap MEMS mic, guys. They probably just make the capacitance less spiky and mic a bit more sensitive since the holes are under the dimples. The only design features here are to loosen tolerances or lower cost. But the cavity under just seems like a classic Helmholtz resonator. It likely is for tuning.
It’s just FASCINATING to say the least to not only come up with such solutions but make them at scale for dirt cheapp! That work sooo well! The amount of research, knowledge, experience, and creativity of these engineers is legendary
The pattern of the dimples is interesting, it seems like a thought through pattern, I wonder how much different patterns, depths and shapes of dimples would alter the sound.
Wow, that was a truly incredible video. I was especially surprised and delighted with the model and animation you created to explain how the device works. I know that in terms of the principle of operation it is quite a simple device, but the scale of miniaturization and the way you presented it make me want to show it to my wife, children and friends. I am really impressed with your channel - keep up the good work. Regards
I guess the dimples would cause a nonlinear response to displacement? It's hard to tell what the neutral position is between the dimples and holes, but it looks like the dimples don't quite cross the holes. If I'm imagining this right, at low SPL/small deflections you'd get an initially sharp response as the dimple intrudes into the hole, and the changing size of the annular gap as the curve of the dimple passes the hole dominates the response. Once the straight(ish) part of the dimple enters the hole, the annular gap stops changing and the overall motion of the plate creates the response at higher SPLs. Pretty neat way to balance sensitivity and dynamic range if that's what they're doing.
My Pixel Buds came with a warning that they have a Class 1 laser inside. Any idea what that might be used for and are lasers of this size particularly interesting?
Laser microphones are a thing, but I'd be shocked if that was the actual use case. Seems like a more complex and lower fidelity approach. A quick search suggests it's an IR laser to detect when the bud is in an ear.
it works like a reverse electrostatic speaker. size is not audio quality related.. in fact a small membrane would be able to pickup high frequencies from all directions like a true omni direction mic. only downside of such small mic might be efficiency or noise floor
This was on my feed since it was on TH-cam.. but was scrolling down.. but after seeing the reel had to watch the full video.. Absolutely amazing. The amount of tech that goes in inside 10 dollar microphone just blows my mind..
I'm curious about the details on it's response curve and if others can be made with different geometry with different curves. It's be cool to see tiny arrays of these that have insane sound quality.
6:54 the strip appears to be continuing on the lowest level, the steps and rounding are artifacts of depositing uniform thicknesses of additional layers. The layers on top of the lowest features only appear to also have the feature. ... meaning, the strip only appears to be deposited on top, if you ignore the rounded edges from additional layers on top of the original feature. I'm not sure if you understood that, since to me it seems like you implied that... anyways really awesome to see tiny devices like that 👌
*Summary* - *0:00* A microphone is crammed inside a tiny earbud, and the video will explore its internal structure. - *0:07* Introduction to the concept of micro electromechanical microphones, which are much smaller than traditional microphones and can fit inside earbuds. - *0:24* Description of miniature microphones like electrostatic ones, which are smaller but still too large for earbuds. The need for even smaller devices fabricated from silicon wafers is highlighted. - *0:46* A $10 pair of earbuds is opened to reveal its internal components, including the speaker, buzzer, charging magnet, and a long PCB with microchips and a tiny microphone. - *1:17* Examination of the tiny microphone, which is about 3mm long, using an optical microscope. The microphone is a square block of silicon with a pattern of holes. - *1:50* Detailed analysis of the microphone's structure, including its size (under a millimeter wide), the arrangement of gold bonding wires, and the layers of the device. - *3:22* Exploration of the microphone's internal structure, revealing a second layer beneath the first, and speculation about the function of the deep cavity underneath the layers. - *4:27* Discussion of how the MEMS microphone converts sound waves into electrical signals using a silicon membrane and capacitive changes. The top dimpled surface is the vibrating membrane, and the chip amplifies these vibrations. - *6:14* The video shows the intricate etching process used to create the microphone's cavity, highlighting the deep reactive ion etching (DRIE) technique. - *8:16* Conclusion that MEMS microphones, while simpler than other MEMS devices, provide excellent audio quality and are widely used in cell phones and earbuds. The video ends with a thank you to Patreon supporters and a recommendation for another video on MEMS technology.
Thought I was subscribed here a long time ago… apparently I was wrong, so I remedied that and hit the like button while I was at it (I know I’ve done that before). Great work, concise explanations and descriptions that cater to the layman without condescending or even a hint of pretense in your tonal delivery. Kinda wanna have a beer with you, even though I hung up the drinking habit half a decade ago now. Keep up the good work and when I have expendable dollars again here in the near future, I’ll drop a few of them off over at your patreon. Until the next one, Bravo, good sir!
Great video and images! If you have a fine diamond saw or pad, you might consider grinding off material from the side, instead of breaking it open. In that way you can make reasonably clean cross sections, especially with the use of some extra resin.
MEMS is the quiet revolution. That tiny microphone, tiny sensors, gryoscopes on drones and modern airplanes, all done with MEMS. It started as engineers seeing other applications for IC processes than just electronics and making such things as working microscopic motors, but then quickly advanced to more useful concerns.
This is an amazing and very detailed video! Loved it! If I could give one suggestion though, I think it would be less distracting if you kept the same tone to the voice throughout any sentence, instead of starting each line with a high pitch and then ending them with a really deep voice. Anyways, beautiful video!
The back cavity is to allow space for the membrane to vibrate. The combination of the stiffness of the membrane and the volume of the cavity form the equivalent of an LC circuit, and you want the resonant frequency above 20khz, so the frequency response is flat in the audible range
07:18 A microphone can only detect pressure differences. If both sides of the membrane are exposed to the same preasurechange, nothing or next to nothing is detectable.
Im not sure if its a microphone from a different manufacturer but I saw the actual MEMS wafers with the holes in it. The poly etchers basically punch holes in them via bosch process to go down so far without having an profile with an angle. A lot of loops.
eally really cool, informative and educational. Thank you so much!!! I loved all the angles and even the 3D model that i have now idea how you got it. You, are, amazing!
Back of napkin calculation..: I'm eyeballing from this video that the membranes are about 0.5mm in diameter, say 0.2mm^2 surface area. Seems like the plate separation is 20 micron-ish. Plate capacitance calculation: 0.88pF. Grossly oversimplifying I'm assuming total harmonic audio distortion, say 0.1%, correlates linearly with the capacitance variation, say 0.88pF/1000=0.88fF. Less than ONE FEMTO Farad. This is just crazy. Even if I'm two orders of magnitude off, it is still crazy.
I feel like the advent of the transistor is really under appreciated Because we would never have been able to reach this small of a scale in electronics had the transistor never been invented
Back in 2003 a company called Akustika built a MEMS capacitive microphone on a CMOS chip. The chip was about 2mm on a side and the microphone was a membrane of oxide and metal over a cavity. The CMOS chip also had a preamplifier and an analog-to-digital converter. So the chip produced a digitized signal with only a connection to an external battery. Total harmonic distortion was hi-fi quality and noise was low enough for use in hearing aids. This was an early prototype. Akustika designed the microphone. I and a colleague designed the preamp and ADC. After fabrication in a CMOS foundry Akustika did the additional processing to fully develop the microphone structure. It worked. The device you show has separate microphone and preamp dice. Probably also a pretty decent performing device. What people are doing these days with MEMS is stunning. Thanks for the video!
This kind tech content is mesmerizing, I'm far from understanding how all this parts connect and talk to each other, but its exciting to see what can be done.
*Addendum*
- The "tactile buzzer" is just the battery. Brain fart, not sure where my mind was when writing that out. Whoops! 😅
- Some folks were curious how the middle gap between the layers is made. I don't know for sure, but it's likely that they used a sacrificial silicon dioxide* (SiO2 aka glass) layer in between the two "functional" layers. So the process flow would have been: pattern and etch bottom layer's array of holes, deposit a thick layer of SiO2, deposit and pattern subsequent polysilicon layers (doped or undoped), then finally etch out the SiO2 layer with HF or plasma. Then flip over and DRIE etch the big cavity from the backside. That is also likely why the dimples are dimple-shaped... they are just following the curve of the sacrificial layer that was filling the holes from the very first layer.
*I suspect SiO2 because there was some EDS data (not shown in the video) which showed high concentrations of SiO2 right at the broken edge between the layers, where they meet at the "bulk" of the substrate. I think that's leftover from the sacrificial etch process.
And here i was googling wtf a tactile buzzer is lol
Your work continues to blow my mind also. Thank you for bringing this high quality educational entertainment to me. 😊😊
I didn't catch it - I was too busy looking at that mems marvel (:
It's interesting we see no microorganisms. It's really work to keep things sterile.
So did the rapper MIMS name himself after MEMS, or did MEMS get backronym'd from MIMS?
Or is it just a coincidence?
Either way 🤯 mind blown! 😂😂
Honestly I might be most impressed by the fact that you made a 3d model of the microphone for a mere couple seconds of footage!
Yeah, that was a really impressive render! Was that model hand made, or is there some way to automagically process SEM images into 3D models?
@@DaveNagy1 I believe it was hand made as it looked different in many ways
making conceptual basic 3d models is not that hard if someone has a good sense of 3d imagination.
i find it crazier that he is able to break a thing many times smaller than a hair in two 🤯
And all that for ten bucks
@@DaveNagy1for sure hand made. The model itself would be pretty quick to make, but texturing, creating the environment, animating all that would take a decent amount of time.
I'd imagine this would probably be something he modelled then sent out to an animator to render up. There's not a lot of cross over between cad modelling for engineering and pretty stuff sadly.
That said I wouldn't put it past him to do it all himself, legend.
The fact that this sort of tech is $10 for a whole system is literally marvelous. 30 years ago this would be actual magic.
Crazy. I just made much the same comment before seeing yours.
It really is amazing how fast microelectronics has developed
@@jimurrata6785 I absolutely cannot wait to see what’s coming in another 30 years.
Magic is technology not yet discovered
@@GeraltOfRivia69exactly! Everything is possible
"Any sufficiently advanced technology is indistinguishable from magic" Arthur C. Clarke
Your video making skills are off the hook. I love the CGI of the microphone and all the beautiful imagery. Your hard work to make these videos is super appreciated.
The rendered footage was some of the best I've seen before on educational content.
I am in awe, I have gotten into microelectronics lately after watching lots of Asianometry videos and this visual exploration of this microphone was astonishing.
Seeing the small features contrasted with a human hair really put everything in perspective in a wonderful way.
Awesome video!
It's perhaps not that surprising that you could create a capacitive microsphone from silicon, but what's mind-blowing to me is that it's such a good microphone. It doesn't seem obvious that you would be able to make a microphone that could do anything other than simply detect the presence of sound. Insane engineering to get to a useful microphone
key to it all is the perfect repeatability and precision of silicon lithography. the signal is very weak and the frequency response is terrible but it can be precisely characterised and corrected using the built in integrated circuit.
@@drkastenbrot cool
@@drkastenbrotalso the ability to make an ic with that capability fit in there with such low power
What might blow your mind too are a type of sensor, found in most smartphones, smart watches, etc., which is closely related to MEMS microphones in their construction (and which in fact came first): MEMS barometers. Rather than the membrane being moved by sound, it’s got a sealed cavity so that as external air pressure changes, the air in the cavity expands or contracts, making the membrane bow in or out, changing its distance and thus the capacitance. What’s incredible is that they’re so sensitive that this is what your phone uses to detect changes in your altitude. Yes, your phone tells how many floors of stairs you climbed by measuring the difference in air pressure. So you might think it’s sensitive enough to measure a meter or two of altitude, right? Nope, they have a resolution of a few _centimeters._ I find it truly incredible that these sensors can actually tell the difference in air pressure over literally one palm’s width of altitude.
@@tookitogo it’s truly awesome tech. I remember figuring out that one of my old android phones had a Yamaha component for an accelerometer. Been pretty hooked on micro tech since then.
"Buddy I can't hear ya, think you forgot your microphone in the electron microscope again"
Sounds like vacuum in there!
@@linecraftman3907 sounds like space!
I'm always blown away by how intricate fab stuff can get! way cool investigation
you alao got access to a SEM, right? maybe investigating something like that would be a nice video idea as well :)
It's amazing that even a crappy $10 pair of earbuds has this much engineering put into it.
exatly what i was thinking, its probably pretty "plug and play" for the manufacturers but still so cool
so much fabbed silicon in a cheap throwaway device
People don’t appreciate just how far we got
Economy of scale is a crazy thing
@@unixuxyeah... Instead they turn to magic creatures and conspiracy theories. You know, no way the Egyptians could have built the pyramids. Because people sure were dumber thousands of years ago. 😂
The 3d model of the mic kind of blew my mind. I loooooooooooove seeing stuff under electron microscopes, thank you for making this. Fantastic all around.
Also not a seasoned audio engineer here but my trivial explanation for the cavity below the membrane is, that it provides a neutral pressure reference against the outside. Thus the microphone becomes omnidirectional. If it would be open from the back, sound waves coming from the side would not be picked up. Thank you for that brilliant deep dive of a video!
huh, that's a concise but insightful bit of knowledge, and goes to explain quite a bit, thanks!
Neat, TIL! Thanks for the explanation!
To my best knowledge, the size of the cavity vs. the diameter of the central hole define your lower cut-off frequency, otherwise this thing would be driven into saturation by low-frequency or static pressure.
@@commander-tomalak that's my thought. Vented microphone to control for VLF and barometric pressure, the cavity for resonance, the steps to tune and reflect various frequency harmonics. Impressive, given the frequencies used in narrowband telephony is around 300 - 3400 Hz, wavelengths ranging from around 45" - 4"!
Yep, a quick lookup shows they're called a MEMS resonant microphone array.
Here's a discussion on active noise cancellation using the technology.
www.ncbi.nlm.nih.gov/pmc/articles/PMC7978172/
Excellent explanation! Never realised there was so much complexity in there, it's certainly a lot more than just a tinier microphone!
Absolutely crazy that something like this is 2x inside a $10 headphone, so each maybe 10cents, at most.
300mm waver gives maybe 50'000, so a whole waver with bonding and everything for less than $5000.
That "buzzer" most likely is the battery, btw.
There's probably just a single mic in one of the earbuds. Lookup "digikey mems mic" and you can find them starting at 47 cents each if you buy 1000. This is a US retailer selling reputable parts. If you get them directly from China, which is probably where these earbuds originate, they'd of course be significantly cheaper.
such an underrated channel
I have so many other things to do today but your SEM experiments just have me glued, amazing stuff.
I told Santa that I wanted an SEM for Christmas. Unfortunately, he said that I was too heavy, get the fuck off of his lap.
Back during the last Ice Age and I was in school, our junior high and high school had donated TEM units, which we were allowed to use. By the time my kids went to school, the electron microscopes were long gone, as were the optical microscopes.
Dude, this is awesome to see so detailed and even broken open. And on top of that, as if that wasn't enough, you explain it all as well and even use super beautiful renders for that explanation!
Wow, excellent presentation. The SEM images and CGI blend perfectly. What an amazing piece of technology. I wonder if the dimples in the top layer are for controlling the stiffness of the disk.
They would probably also help increase the capacity by enlarging the surface area when the membrane is close to the other electrode.
i would speculate that it's mostly just a byproduct of how the thing is manufactured
Thanks! I believe the dimples are _mostly_ an artifact of the manufacturing steps to make one. There are a few ways it could have been made, but my current theory is: pattern and etch the base substrate giving nice clean holes, deposit a layer of glass on top (which will naturally form rounded dimples over the holes), deposit another layer of undoped polysilicon and then a doped polysilicon layer, then finally etch out the sacrificial glass layer (with HF or plasma) leaving the gap between the two layers. Finally flip and etch the big cavity.
Just a guess but it makes sense to me. The holes in the lower layer are to help air move past it with minimal resistance, but the dimples on the surface aren't really needed. So I think it's leftover from the layered nature of fabrication.
@@BreakingTaps You mean a totally flat disk without dimples could not be fabricated? I wouldn't expect that, but I don't know anything about processes at this micro level. I'm just blown away that they actually work as well as they do. Are neodymium magnets used at this scale?
@@professordeb It's technically possible to get a flat disk on top of the hole'y layer, but it would be a lot more work. There's another process called "chemical-mechanical planarization" which is basically a super fancy sand paper for wafers 😁 It's used to flatten the top layer by grinding/polishing until all the ridges are gone. It's often used on high density microchips like computer CPUs, because you have soooo many layers that everything starts to get rounded. So they periodically flatten it with the planarization tools.
So to get a flat surface for this device, you'd deposit a really thick sacrificial layer, then grind it back flat, then proceed with the next steps. But if you don't _need_ it to be flat, you can skip and save money.
Magnets aren't used a lot at this scale because (I think) the magnetization process needs high temperatures and it can be difficult for the devices to survive. Although I've seen some papers about using laser-heating and such, so I'm sure it's doable.
At this size, electrostatic, thermal and piezo mechanisms tend to be more common.
the way this microphone works is, there is a small hole that allows to equalize the pressure between the inside and outside of the chamber slowly later when pressure is applied to the top the fluctuations are relative to the mean pressure.
The cavity is a resonant chamber. The microphone is referred to as a MEMS resonant microphone array. Pair them up and one can have quality active noise cancellation.
The cavity is a resonant chamber, the steps for different frequencies.
I work at a very old 200mm semiconductor fab as an equipment engineer. One of my processes is polysilicon deposition through LPCVD. Hearing these terms in a video about mics in earbuds is awesome.
Love the work, will sign up for Patreon
You just spoke about almost every topic I had in my master's degree lecture "Physical Sensors in Silicon Technology", we also had the etching process RIE (Plasma etching - Reactive ion etching) explained in details in there. Thank you for making this video, I just finished my university degree and it's cool to see some practical stuff for a change!
The MEMS resonant microphone array is a fascinating technology, especially the tiny resonant chamber within the unit.
Thank you for making these videos. They really help give perspective on this extremely tiny yet extremely impactful part of all our lives. Plus the SEM images and CGI you make are absolutely beautiful to look at.
This is great! Hah, I just bought these ONN buds on sale for ~10$ and they work great as a basic hands/wires free headset. I was marveling at the amount of tech crammed into these cheap lil guys and you've revealed their innermost secrets :D Always enjoy your microscopy.
The fact people figured out how to make this stuff!!! It’s insane to think about. We don’t give ourselves enough credit as a species.
Man I am glad I found your channel. This stuff is awesome. TY
I’d love to know more about the hole pattern. Most is a hexagonal fill which is good for maximum density and uniformity. Edges are concentric rings, and in between is a hybrid.
Your animations are fantastic!
Why there is such a big cavity: my guess is that it is simply to do with ease of manufacture. They first make all the structures on top, then flip it over and etch through from the backside. Importantly they also intentionally leave a controlled-size hole that allows internal pressure to equalize over a controlled time (e.g. if device takes a plane ride or happens to be put in a vacuum during further processing), not too slow but not too quickly that the device wouldn't be able to pick up bass.
Given the tiny size of the cavity, it can't have anything to do with acoustic resonances.
Has everything to do with acoustic resonances, and there are much more ambitious ones in the public market.
These microphones typically have a digital output using "pulse density modulation", where the rate of toggles encodes the analog signal value. The three ports coming off the control die are almost certainly power. ground, and audio out.
Also +1 for DRIE video. That was the first thing I noticed when you cracked it open. The Bosch process is cool!
I counted 7 heavy etchings on one component, couldn't get a good count for thinner etchings for things like the resonant chamber.
Then, I considered how many ways I could badly injure myself on the equipment that builds these devices... Ion beam, HF, yeah, gotta be a pain to maintain those machines!
Awesome video! Thanks
I love electron microscopes and pictures they produce.
Really like to watch your content.
I always learn something new.
Thank you!
I would guess the large cavity behind the membrane matters for the microphone's frequency response: Most practical microphones don't want to react to slow changes in ambient air pressure, because those can easily be much bigger than typical sound pressure levels, and could blow out the membrane. This is what the small hole in the middle of the membrane is for, to let the pressure equalize on both sides of the membrane (equivalent to a high-pass filter). Of course, if the equalization is too fast, then it can also equalize out low-frequency sounds, which would impact the frequency response of the microphone. The speed of this equalization depends on the hole size and the volume behind the hole (similar to a Helmholtz resonator with an additional loss term), so the manufacturer will tune either the hole size or the volume behind the membrane to set this frequency to a sensible value - I think 1-2Hz are typical for typical electret microphones.
I would guess, then, that the hole is already as small as feasible in this process, for some reason or another. Then it would make sense for the manufacturer to make the volume larger (requiring more etching steps) in order to improve the frequency response at low frequencies.
can you explain what's the additional loss term?
The amplifier for the microphone looks quite interesting too. Looks like too many parts to be purely an analog amplifier. I wonder if they're driving the microphone with AC and de-modulating to get the audio?
Incredible video with stunning visual and intriguing explanation.
Keep the good work.
Sick, never knew how they fit microphones into those earbuds, thanks for showing!
Really cool video dude! I just hooked up a MEMS mic to my WLED display for music reactivity. Cool to see exactly how these little pieces of “fly sh!t” actually work! Merry Christmas!!!
What amazes me is the price of this tiny marvel being under $1.
Nice work. This looks surprisingly easy to make.
Ordinary ceramic capacitors can also respond to sound. I've seen speculation that this effect could be used to turn ordinary electronics into surveillance devices but I haven't seen a proof of concept yet.
What happened to the video "The Science of SpaceX Starship's Thermal Tiles" which was up until a few hours ago?
I came looking for it to send to a friend. I found a link to it on Hackaday, but it was listed as private. I wonder what happened?
Me too!
Same I wanted to rewatch it :(
Interesting to see the ridges inside that microphone cavity. At first i thought they were there to help with echo and reflections, similar to how some speaker boxes deal with it. Now I'm wondering if the etching process is calibrated to make those ridges a certain size specifically for essentially tuning it.
around 3:23 that chip looks like it was cut by shearing, instead of with a cutoff wheel. a cutoff wheel leaves a clean edge, shearing leaves that sharp clean beginning of cut then it kinda just breaks
Just one correction. The balls at the ends of the bond wires are tiny solder bumps and not Au balls. Solder bumping, wire bonding, die stacking and 3D packaging in general would make a great episode!
Commonly, after making the bond to the lead frame (after first bonding to the bond pad on the chip) the wire-bond machine severs the wire with an electric arc. This arc produces a spherical blob of molten wire, which cools and solidifies. That little ball is right below the surface of the bonding foot. So when the bonding foot is pressed down onto the next bond pad on the chip, it compresses the ball onto the bond pad, creating the electrical and mechanical connection to the bond pad.
So the little balls are of gold or aluminum, whatever the bond wire metal happens to be.
Now, where chips are bonded directly to each other or PCB, the balls are usually made of solder or tin or indium. Actually, they are referred to as bumps for this kind of bonding operation.
What is interesting about the bond-wire at 4:40 is that the bondwires seem to be mounted on top of a gold ball. If it was ball bonding, I would expect a clean transistion from the 'mushroomed' out ball to the wire, but this is not the case, you can clearly see that the wire itself has been smooshed (through what looks like wedge bonding) onto the gold ball.
I look forward to these posts so much. It's the highlight of an otherwise rather mundate youtube experience for me.
This was awesome - I will be subbing!
I'm kinda surprised you didn't mention the outer layer's hundreds of dimples. They're convex parabolic cones (Dimples 3:37) that directs the sound waves around the the actual mic pickup. Being such a small device the time delay of the different waves is infinitesimal. All this multi magnifies minute sound waves much like cupping your hand around your ear. KEWL VID, THANKS
The geometry of the whole thing is interesting. The dimples probably make the upper membrane more stiff, and might have been calculated or arrived at experimentally. Stiffness is useful in this application as you don't want to respond to the inertia of the membrane itself (flopping around). The chamber below is indeed for resonance, and it should be possible to physically measure it and calculate what frequencies it is sympathetic to; my bet is that it works really well for key frequencies in "toll-quality" audio (voice). I'm also betting that the device is good at picking up low audio frequencies from whatever it's mounted to, letting the whole earbud vibrate as an extension of the sensor.
Spitballing... I imagine the dimples are to make the upper membrane less stiff. Stiff membranes resonate---like a drumskin---and you want as little resonance as possible; while still attaining high deflection for gain. (High Q, high amplitude voltage ringing, even above hearing frequency, can upset amplifiers.)
I'll add my own unqualified take into the pot: it's silicon. It probably has a really wonky breakup pattern that the dimples help with!
Only joking! It's a cheap MEMS mic, guys. They probably just make the capacitance less spiky and mic a bit more sensitive since the holes are under the dimples. The only design features here are to loosen tolerances or lower cost.
But the cavity under just seems like a classic Helmholtz resonator. It likely is for tuning.
I was thinking they did it to make the surface area of the diaphragm a little bit bigger. Maybe even tune it to a specific frequency.
Thanks!
It’s just FASCINATING to say the least to not only come up with such solutions but make them at scale for dirt cheapp! That work sooo well! The amount of research, knowledge, experience, and creativity of these engineers is legendary
The pattern of the dimples is interesting, it seems like a thought through pattern, I wonder how much different patterns, depths and shapes of dimples would alter the sound.
I can't see u😂
It’s called anti-stiction bumps… to unmute the mic after an high sound event or mechanical shock
Wow, that was a truly incredible video. I was especially surprised and delighted with the model and animation you created to explain how the device works. I know that in terms of the principle of operation it is quite a simple device, but the scale of miniaturization and the way you presented it make me want to show it to my wife, children and friends. I am really impressed with your channel - keep up the good work. Regards
Im actually more amazed at the quality of this video than anything else.. and wow, SEMs have really improved over the last decade or so.
You should have recorded the audio of this video in an airbud mic
I guess the dimples would cause a nonlinear response to displacement? It's hard to tell what the neutral position is between the dimples and holes, but it looks like the dimples don't quite cross the holes. If I'm imagining this right, at low SPL/small deflections you'd get an initially sharp response as the dimple intrudes into the hole, and the changing size of the annular gap as the curve of the dimple passes the hole dominates the response. Once the straight(ish) part of the dimple enters the hole, the annular gap stops changing and the overall motion of the plate creates the response at higher SPLs. Pretty neat way to balance sensitivity and dynamic range if that's what they're doing.
I have to remind myself that this is only one component out of a $10.00 pair of earbuds.
Absolutely fantastic as always! Man I wish I had and SEM to play with!
This has answered questions I didn't know I had 👏
My Pixel Buds came with a warning that they have a Class 1 laser inside. Any idea what that might be used for and are lasers of this size particularly interesting?
Laser microphones are a thing, but I'd be shocked if that was the actual use case. Seems like a more complex and lower fidelity approach. A quick search suggests it's an IR laser to detect when the bud is in an ear.
What a perfectly paced video, I couldn't stop watching it. Thank for you the impressive images and insightful analysis!
it works like a reverse electrostatic speaker. size is not audio quality related.. in fact a small membrane would be able to pickup high frequencies from all directions like a true omni direction mic. only downside of such small mic might be efficiency or noise floor
This was on my feed since it was on TH-cam.. but was scrolling down.. but after seeing the reel had to watch the full video..
Absolutely amazing. The amount of tech that goes in inside 10 dollar microphone just blows my mind..
I'm curious about the details on it's response curve and if others can be made with different geometry with different curves. It's be cool to see tiny arrays of these that have insane sound quality.
Thats exactly my thought also!
EXCELLENT! This has been a long time coming. Thank you!
Fascinating. Thank you for the detailed explanation of how these mics work. Now I see my earbuds in a different way.
6:54 the strip appears to be continuing on the lowest level, the steps and rounding are artifacts of depositing uniform thicknesses of additional layers.
The layers on top of the lowest features only appear to also have the feature.
... meaning, the strip only appears to be deposited on top, if you ignore the rounded edges from additional layers on top of the original feature.
I'm not sure if you understood that, since to me it seems like you implied that...
anyways really awesome to see tiny devices like that 👌
*Summary*
- *0:00* A microphone is crammed inside a tiny earbud, and the video will explore its internal structure.
- *0:07* Introduction to the concept of micro electromechanical microphones, which are much smaller than traditional microphones and can fit inside earbuds.
- *0:24* Description of miniature microphones like electrostatic ones, which are smaller but still too large for earbuds. The need for even smaller devices fabricated from silicon wafers is highlighted.
- *0:46* A $10 pair of earbuds is opened to reveal its internal components, including the speaker, buzzer, charging magnet, and a long PCB with microchips and a tiny microphone.
- *1:17* Examination of the tiny microphone, which is about 3mm long, using an optical microscope. The microphone is a square block of silicon with a pattern of holes.
- *1:50* Detailed analysis of the microphone's structure, including its size (under a millimeter wide), the arrangement of gold bonding wires, and the layers of the device.
- *3:22* Exploration of the microphone's internal structure, revealing a second layer beneath the first, and speculation about the function of the deep cavity underneath the layers.
- *4:27* Discussion of how the MEMS microphone converts sound waves into electrical signals using a silicon membrane and capacitive changes. The top dimpled surface is the vibrating membrane, and the chip amplifies these vibrations.
- *6:14* The video shows the intricate etching process used to create the microphone's cavity, highlighting the deep reactive ion etching (DRIE) technique.
- *8:16* Conclusion that MEMS microphones, while simpler than other MEMS devices, provide excellent audio quality and are widely used in cell phones and earbuds. The video ends with a thank you to Patreon supporters and a recommendation for another video on MEMS technology.
wait wait wait. How do they etch the giant hole and then make the membrane on top?? Do they come from the back?
Any headphone can be a microphone.
Thought I was subscribed here a long time ago… apparently I was wrong, so I remedied that and hit the like button while I was at it (I know I’ve done that before). Great work, concise explanations and descriptions that cater to the layman without condescending or even a hint of pretense in your tonal delivery. Kinda wanna have a beer with you, even though I hung up the drinking habit half a decade ago now. Keep up the good work and when I have expendable dollars again here in the near future, I’ll drop a few of them off over at your patreon. Until the next one, Bravo, good sir!
Great video and images! If you have a fine diamond saw or pad, you might consider grinding off material from the side, instead of breaking it open. In that way you can make reasonably clean cross sections, especially with the use of some extra resin.
Will try that on the next one! I admit to being a bit lazy and just smashed it with some tweezers haha
MEMS is the quiet revolution. That tiny microphone, tiny sensors, gryoscopes on drones and modern airplanes, all done with MEMS. It started as engineers seeing other applications for IC processes than just electronics and making such things as working microscopic motors, but then quickly advanced to more useful concerns.
thank you for creating and sharing this, that was amazing and enlightening
Wow…it’s really amazing!! U can explain difficult things for us to understand brilliantly and the content used is very good.💪❤️
Oh wow this is amazing!! Thank you for making such a detailed video about it!
Crazy... Thank you!
This is an amazing and very detailed video! Loved it! If I could give one suggestion though, I think it would be less distracting if you kept the same tone to the voice throughout any sentence, instead of starting each line with a high pitch and then ending them with a really deep voice. Anyways, beautiful video!
The back cavity is to allow space for the membrane to vibrate. The combination of the stiffness of the membrane and the volume of the cavity form the equivalent of an LC circuit, and you want the resonant frequency above 20khz, so the frequency response is flat in the audible range
I hope you still get a thrill every time you say "now let's look at it under the electron microscope..."
The quality of these videos never ceases to amaze me :)
07:18 A microphone can only detect pressure differences. If both sides of the membrane are exposed to the same preasurechange, nothing or next to nothing is detectable.
Im not sure if its a microphone from a different manufacturer but I saw the actual MEMS wafers with the holes in it. The poly etchers basically punch holes in them via bosch process to go down so far without having an profile with an angle. A lot of loops.
Bosch process is absolutely wild! Especially the more refined variants that can etch with much smoother sidewalls.
Awesome! A new microscopy video!
eally really cool, informative and educational. Thank you so much!!!
I loved all the angles and even the 3D model that i have now idea how you got it.
You, are, amazing!
Your video quality blows me away
Back of napkin calculation..: I'm eyeballing from this video that the membranes are about 0.5mm in diameter, say 0.2mm^2 surface area. Seems like the plate separation is 20 micron-ish. Plate capacitance calculation: 0.88pF. Grossly oversimplifying I'm assuming total harmonic audio distortion, say 0.1%, correlates linearly with the capacitance variation, say 0.88pF/1000=0.88fF. Less than ONE FEMTO Farad. This is just crazy. Even if I'm two orders of magnitude off, it is still crazy.
Thanks
Always a pleasure to see a new video from you.
I didn’t know Thor was in my earbud 😮
Really enjoy all your videos mate, great work!
What a fascinating dive into cool microscopic engineering. Great video all round!
I feel like the advent of the transistor is really under appreciated
Because we would never have been able to reach this small of a scale in electronics had the transistor never been invented
Back in 2003 a company called Akustika built a MEMS capacitive microphone on a CMOS chip. The chip was about 2mm on a side and the microphone was a membrane of oxide and metal over a cavity. The CMOS chip also had a preamplifier and an analog-to-digital converter. So the chip produced a digitized signal with only a connection to an external battery. Total harmonic distortion was hi-fi quality and noise was low enough for use in hearing aids. This was an early prototype. Akustika designed the microphone. I and a colleague designed the preamp and ADC. After fabrication in a CMOS foundry Akustika did the additional processing to fully develop the microphone structure. It worked.
The device you show has separate microphone and preamp dice. Probably also a pretty decent performing device. What people are doing these days with MEMS is stunning.
Thanks for the video!
Correction: the company name was Akustica, rather than “Akustika”. Akustica is now part of Bosch.
When you find a topic that you are interested in or want to cover, how do you go about learning about it?
I'm amazed at how polished the second layer appears under the Electron Microscope!
This kind tech content is mesmerizing, I'm far from understanding how all this parts connect and talk to each other, but its exciting to see what can be done.
Holy moly the presentation and info is so onpoint
This was so fascinating to watch! Thank you much. I just stumbled accros your channel and you got so much more videos :O
Seriously the most impressive microscopic images I've ever seen.
I really needed a bit of marvel and wonderment today. Thank you!
ooh, your blender animations are getting really good!
As an engineer, it has been a long long time since the last time I was curios & actually had fun watching electronics under a microscope.
Thank you