#733 – Perpetual Wearable Biosensing with Andy Kong

Intro announcer: This is The Amp Hour Podcast. Released September 9th, 2026. Episode 733. Perpetual Wearable Biosensing with Andy Kong.

Chris Gammell: Welcome to The Amp Hour. I'm Chris Gammell with Contextual Electronics.

Andy Kong: And I'm Andy Kong, founder of Chargerless.

Chris Gammell: Hey Andy, welcome. How are you doing?

Andy Kong: I'm good. Good to see you.

Chris Gammell: Yeah, same. I feel like I have to mention, I was introduced to Andy when I was talking to a past guest, friend of the show, Samy Kamkar. I'm like, Samy, who should I be talking to? And he said, Andy Kong. And I'm like, I couldn't email you fast enough. And obviously I checked out all your stuff online and you're doing really fun things. So glad you're here. What do you like working on? I mean, like when Sammy Kamkarn gives you the nod, I mean, I feel like that means you're a hacker. Do you identify as a hacker?

Andy Kong: Yeah, I think I used to identify more strongly as a hacker. I guess now I'm a startup founder, but I do miss the hacker researcher kind of vibe.

Chris Gammell: Does the hacker title make it harder to raise money maybe? Or I don't know. Sammy did it.

Andy Kong: It seems like a little hobbyist. I think that's the main problem with it. Same with researchers. They've made one of something, but they're not like...

Chris Gammell: Yeah. I mean, I think in all things, I hope to issue the issue. No, that's the wrong word. I hope to emulate the hacker mindset at least. Like that get things done, work until it works, that sort of thing. And it seems like that's your MO as well.

Andy Kong: Yeah. Getting it done with what you have, I think it's like the hacker MO. And then also, you know, being, I guess, willing to like mess with systems that maybe don't want to be messed with.

Chris Gammell: Yeah. Rules of order are there for reasons, but you know, sometimes we get a norm. So yeah. Yeah. Just one-time exceptions,

Andy Kong: you know, yeah.

Chris Gammell: So tell us about, so Chargerless. So what is that product? That's your current project. That's what you're working on. It's a funded-ish startup. That's right. I mean, like you have moving towards product in the market, that sort of thing.

Andy Kong: Right. Yeah. So we're a pre-seed funded startup. I lived in, I live in San Francisco. I moved here about a year ago to start the company. Also helped that a bunch of friends are over here. Actually, right before I moved, I stayed at Sammy's for a little bit, which is where we got to know each other. But yeah, my background has always been in like kind of biosensing and different forms of like human stimulation or like integration. Lately, I've been really into health stuff. And about a year ago, it was like, okay, I have an Apple watch, like every day or two, I have to charge it. About half the time it's like dead when I look at it. But really I just wanted the metrics out. So I started Chargerless mainly to make a really low power version of the same thing. So we have all the same sensors as the Apple watch. But if you get like a minute of sun on the device, today, you never have to charge it. And also there's a pretty big battery also. But yeah, so that's our current product. My main interest with making Chargerless is I really think devices today are, like you look at devices even 10 years ago, it's like they're limited by chip size. Like you look in the Apple, Apple two. Okay. I mean, that's not 10 years ago, but you look in the Apple two, it's like mostly logic chips, right? Like these huge things. But you look at it like stuff now, it's like mostly the battery. Um, and to me, that's just like an interesting opportunity because if you can reduce the power consumption of the whole system, you can keep the same battery life. Um, but you can now start to shrink the other large component in the device.

Chris Gammell: Then that becomes like a, is that like a hardware choice kind of activity? Is it a firmer activity? Is it all of the above activity? Like what, what actually allows that just one second a day of sunlight? Cause that is an insanely small amount of power.

Andy Kong: I hope you get to one second, but we're at like a minute right now. So it's a couple, sorry,

Chris Gammell: sorry. It's a minute, but yeah, a minute, like even, even the, the nerds in the dankest of caves will get out for a minute. Probably.

Andy Kong: Yeah. Yeah. And that 90th percentile of nerd sun exposure. Yeah. I think it's mostly on the system architecture side. Uh, it's very easy nowadays to pick a microcontroller, which is easy to develop on and has a lot of features, but maybe has like a bit too much feature, uh, for how much power draw it has. So we pick like a deliberately underpowered microcontroller. Uh, and then we have to be more clever with the firmware and the way we implement the sensing. Um, but then you get this awesome feature where it just kind of like goes as long as you have it on. Yeah.

Chris Gammell: Yeah. Yeah. And I mean, does that mean then you have to give up? So, you know, we're talking about a wearable, we're talking about a accelerometer type of, you know, wake up, take a reading, that sort of thing. Does that mean that there's fewer features? I mean, I used to have like a Fitbit, uh, what was the inspire three, like the little, the one that's like fully rubber over molded that would last like a week or two, you know, it was just for step counting. It was just step counting. Yeah. So like in terms of like features, do you have to give up on features on two or how does that work?

Andy Kong: I was surprised we don't really have to give up on features. Um, and like depending, I mean, this is a bit of a choice because I only go outside about 10 minutes a day or like 10 to 20 minutes a day. Uh, like we could do continuous monitoring if you went outside for like an, uh, maybe like half an hour to an hour. Um, so for instance, like the whoop or the Apple watch, they do this thing where like they're watching all the time. So they're constantly sampling the heart rate sensor. And that just like, you know, continuously burns like a little bit of power. Um, for us or the aura ordering, for instance, uh, we sample at a pretty often interval. So like every minute or two. Yeah. So like, that's like pretty much the only thing I think we had to give up on. But besides that, like, you know, sensing heart rate is a pretty straightforward process. And as long as you can get the timing, like to be shorter for the led, like on pulse, then you can get the same feature with much less power.

Chris Gammell: Got it. Yeah. And that, and that then is it like, if you detect that I'm like moving around a whole bunch, like an exercise type thing where you might want higher fidelity or can you move it into higher fidelity mode? Like how does, how does that work?

Andy Kong: Yeah. So, uh, I think we'll probably do something like when you trigger an exercise on the app, uh, it'll kick it into like high, like continuous mode or something. Uh, but then when you're sleeping, it's like, you know, this guy's heart rate is going to be mostly the same most of the time. So, uh, we can, we can reduce it then.

Chris Gammell: Yeah. Like intelligent. Yeah. That's smart. Okay. Yeah. And then, uh, so now you get like sensor fusion doing like a accelerometer plus heart rates. Like how, how much is it like office job? How much do you have to do, you know, kind of custom things in order to make all this work on, on such a small power budget?

Andy Kong: Yeah. It's a, it's really interesting. Like when I started, uh, it was like, Oh, you want to sense temperature, here's a temperature sensor. Uh, you want to do motion. Here's an IMU and all this stuff. Um, and these, these are all like fine and dandy or whatever. But when I went to do the heart rate detection, uh, it was like all the heart rate integrated circuits. They like, so there's a lot of wearable products obviously. So, um, TI analog, uh, they all have like these, um, integrated, like fully integrated led photodiode and, um, ADC. And so they'll do all the amplification for you. And then you just like ping it to get the heart rate out and you have to process that for peaks and stuff. Um, and I was using one of these in the beginning and it's like, I think it's like a hundred times too much power for my application. Uh, yeah. So that was the one part we had to really go custom on. Um, cause, uh, you know, you know how the heart rate detection works. Um,

Chris Gammell: I, I don't know if I, you know, I was actually thinking about just heart rate stuff. I think my only like hackery kind of like open up the other than like, as a user, a casual user wearing a, wearing a pickle pixel watch. Like I think the only one I've ever had is like the, the cardio badge. Did you ever see that from like Defcon or sorry, a chaos camp. That was the 2019 chaos camp badge. They had, uh, they had put on like a sensor on there and I, I actually got to see it working and see some of the firmware under the hood and, and things like it. That was the closest I've ever gotten. That was EKG. And I, I think so. Yeah. Yeah.

Andy Kong: Yeah. Cool. Uh, yeah, that's like lab standard, right? They hook you up to like a bunch of leads, um, like all around your chest. Uh, it's kind of interesting though. So when the heart beats, uh, it generates like a large, uh, potential across of it. Uh, and so that propagates through the body. And so most people it's like left to right ish, but, uh, the bag that surrounds the heart, the pericardium, it dulls this response a little bit and everyone's shaped a little differently. And so the vector that your heart rate, potential propagates is different. Um, this is why they have to do a 12 lead EKG to be like, like in the hospital, because some people like left to right, you don't get anything. Yeah. Uh, so that's,

Chris Gammell: I actually knew someone that once had like the, the syndrome where their whole body is flipped over the, the Y axis. What? Like, yeah. Do you know, that's a thing there's, I forget what it's called. There's a, there's a Latin term for it, but like, you know, like your heart's on the left side. Is that right? Whatever. It's like, whatever side is your heart supposed to be on? It was on the other side. Like all of his organs were, wait, how do you know that? Uh, there's a bunch of complications from it. So, okay. Okay. Well, that's, it's tough. It's really tough. Um, but yeah, it is surviving and trying. It's like, it's like, it's just amazing that like, that's a, that's a thing. I'll look up the term actually. Cause I came across it. I was like looking up some Latin term the other day. And then that came up as a search as a, as like a flip term as well. Anyways, I'll, I'll look it up here, but it's, or I'll put it in the show notes, whatever it is, but it's a thing as well.

Andy Kong: That's pretty interesting. Yeah. Body's weird. And like, you know, we're the most well studied animal. Um, and yeah, so you, you find all sorts of like funny things, uh, in the body. Anyway, so the heart rate stuff. Yeah. So if you, in the lab, it's like they hook you up to two leads. Uh, you know, you put that into a subtractor and then you amplify that signal a lot and you get like the traditional, like our peak that, um, like drawing on the, on the seismograph looking thing. Yeah. For all the wristworn. Right.

Chris Gammell: That's what I look at. You get the bleeps and the bleeps and the bleeps.

Andy Kong: Yeah. So all the wristworn stuff, uh, actually relies on the fact that your oxygenated blood is a different color from your deoxygenated blood. Yeah. So if you, you can see this on your phone camera, actually, if you take a video and the flashes on and you put your finger over the camera sensor and like you get the exposure just right, then you actually see the whole like screen, like flicker with your heart rate. And it's like the same principle in the wearables. Yeah. So they use a green light because, um, green light doesn't really go in. So most of it comes back, which is where you need it. You need it back into the device. And then you put in a photo diode, uh, you amplify that and it gives you roughly the R peak of the EKG. Uh, but it's delayed by like the propagation time out to the wrist. Yeah.

Chris Gammell: Got it. Yeah. And so I'm sorry. I do remember one other, I, I had a DIY sensor for just that where it was like, you could clip it to your ear, your, your ear lobe. And I had made, I had made my own thing where I, it was like an off the shelf ear lobe sensor. And then I had made it like, uh, I put it into a teensy and then I had it like flash, like a 3d printed heart that I had. That was like,

Andy Kong: Oh, that's cool. Yeah. Yeah. I really like that. Something like that. Yeah. So you wear it on the outside or something.

Chris Gammell: It was just like a, I mean, it was like this insanely large badge thing. It was like, it was like literally made out of red solo cups. Like that's the, the aperture side of the, the size of this thing. And then I made a cap for that. Um, yeah. And I had this real problem was that I didn't turn the flash down. So it was like, normally it was, you know, in daylight, it looked great. But then at night, everyone's just like, Oh, turn that off. You know? Yeah.

Andy Kong: Yeah. I think it's a cool idea to have like a party where like everyone has like a heart rate sensor on and it's like visualizing their heart rate. And so you can, I mean, I'm sure you can't tell like if you're paying attention to the conversation, but by looking at the, their heart rate sensor, you might be able to tell if they are enjoying the conversation or not.

Chris Gammell: Yeah. Yeah. Yeah. There's a, uh, from the badge again, I'm going back to badge life a bunch, but from the badge life community, there was, um, Carrie sharp glasses down the show a long time ago. Uh, he made a badge where it, I don't think it actually measured. I don't think it measured heart rate, but it did synchronize. So it would be like, you'd be, if you and I were talking at the party, it would like, they would communicate with each other, like with IR and they'd start to like be in sync. And so it was all based on Stevenson, Neil Stevenson's diamond age, where it was in the shape of a dragonfly. And like, like the whole party could get synchronized, which is just like, Oh, that's so cool. So fun. Yeah. Yeah. Whoa. I'll link that show into. Yeah. That's neat. Badge life. Woo. Yeah.

Andy Kong: Yeah. I don't know. Biosignals are really cool. Yeah.

Chris Gammell: Yeah. Yeah. Yeah. So I mean, and it does sound like there are, so you're saying there's like off the shelf modules, even that the, you kind of get to move up the stack. So you, you know, plebs like me can move up the sack and, and, and just get heart rate out, but that did not work for, for your use case because of current draw.

Andy Kong: Yeah. Yeah. So like it's for applications. I mean, you know, for reference are like the reason we're able to, to use a solar panel is because our system on whole uses maybe like 50 times less power than most wearables. And so most wearables, like the sensors, you can just put the integrated sensor in. And like, that's just totally fine. Like it's a small part of your total power consumption. For us, it was like, it was like going to hit like 95% of our total power consumption based doing the heart rate.

Chris Gammell: Yeah. And I mean, what are those targeted? Are those targeted like pulse oximeter style sensors, like the, the finger clip style? No, I mean, yeah,

Andy Kong: you can just put in a wearable directly. Yeah. Okay.

Chris Gammell: Just a more power hungry one. That's the, that's kind of the real target for those.

Andy Kong: Right. Yeah. And if you want to have more control. So it was mainly that I think the ADC is not super power optimized in those devices. And then the amplifier, you know, between the photo diode and the ADC is also not optimized. Uh, and then like, you can only get the pulse timings to be so short, uh, based on like how, how they implemented the heart rate sensing.

Chris Gammell: Yeah. Yeah. So it's basically like a menu system where you can choose the register set that like is the best they have, but then you get to the bottom. It's like, well, that's it.

Andy Kong: Yeah. Yeah. I mean, you know, this is like, like core, like fundamental trade-offs of like, like integration. It's like you, you lose a lot of, uh, maneuverability. So.

Chris Gammell: Yeah. Interesting. Well, I actually, I want to hear how you then deconstructed and rebuilt it yourself, but I was talking to someone today about, about just the, the challenges of like low power design and just like the, you know, the, like the error bars, like, you know, the longest pole kind of problem, right? Like, it's like, uh, yeah, maybe you have one sensor that's like 500 microamps and you're trying to get down to 10 microamps. Like just find the biggest consumer first and knock that out. But there's almost guaranteed to be that next thing down. You know, it's like, it's just this whack-a-mole game. It's so, so tough. It feels like,

Andy Kong: Oh yeah. Yeah. I mean, I, I felt that it was like not, uh, it was not like so complicated because there's only, you know, there's only three phases in our device. Uh, one is like when the Bluetooth is connected, uh, and you sync out the data. And then one is when you're sampling the heart rate. And then there's a very brief period where you're like processing the, the PPG, like the sense, like the readouts from the sensor, uh, into like an actual heart rate number. Uh, and then there's the sleeping. Yeah. And for us, the sleeping and the heart rate sensing both use the same amount of power.

Chris Gammell: Yeah. And it is interesting too, that it, it does seem like, uh, kind of somewhere between Apple watch, like you mentioned, which has kind of that, it's the superset of every, every feature you'd ever need, you know, LTE and GPS and all that, all this stuff. And then now all the way down to the single, not single use case, but the, the reduce set. It's like, you know, it's, it's basically the product decisions at that point as well. Right.

Andy Kong: Right. Yeah. I heard it was because, uh, like Apple watch, it was like, sometimes every time they announce a feature, like a, like a certain percentage of people will then buy an Apple watch. And so they have really just been like cramming, uh, features into that thing. And like, at least this is what I heard for, uh, for like why they keep doing it. Cause like sometimes they'll add stuff and you're like, I, who, who uses this one?

Chris Gammell: Yeah. Like on the ultra watches too, like they're like targeting, you know, mountaineers and it's like, like, we're going to charge $1,200 to this watch. It's like, okay. You know, that's great. Yeah. I remember you could do like, you could do like texting on it.

Andy Kong: I was like, who's going to text on this thing. And like the keyboard is like minuscule.

Chris Gammell: It is, it is kind of a lifestyle. I mean, I, I, I definitely see, you know, I, I kind of always think of like the, the, the businessman, you know, like under, you know, pulls, you know, suit coat back to check the watch time. And it's like, it was always this like dramatic action in the eighties sort of thing. And he was really trying to show off the watch. And now it's become this other, like trying to glance down at your wrists because you're trying to prove that you're not just glancing at your phone, you know, that sort of thing. It's like, yeah. Oh, I had a funny dance.

Andy Kong: I had a funny thing the other day. So a lot of people keep the display on, on their Apple watch. And like, if you look at your watch during like a good conversation, it's, it's a little rude, right? It kind of takes you out of it. But I realized that because she was like holding, her arms up, like while she was speaking, I could read the time on her watch. There's like no social penalty to, to checking the time anymore.

Chris Gammell: got it. So like, you don't need to wear a watch. You just need to have friends that like wearing watches and keep the screen on.

Andy Kong: Yeah. Yeah. And then you can see it because on the analog, there's no way I could read an analog watch. Like, you know, that's true.

Chris Gammell: All the full brightness blasting you in the face. Yeah.

Andy Kong: Yeah. Yeah. So we went with a more constrained like a feature set. So I'm really curious about the biometrics. I think most of the analysis that you're interested in can be done better, like on a server or something than on the device itself, which is quite constrained in how much power it can use for processing. And also like how much data can get out through the Bluetooth link. Yeah. But it was like, I remember I heard the app, like Google was working on a screenless tracker. And then they came out with a Fitbit Air and then Garmin has their own. And I remember being really excited about this. Like they, like these huge companies like see enough of a market and like screenless fitness tracking that they're willing to go into it. And I was just like, Oh, this is great.

Chris Gammell: Yeah. Yeah. That's nothing validates a market, like having competitors, right? Especially, I mean, your scale difference is a little different there, Andy.

Andy Kong: Yeah. Yeah. That's true. That's true. But I was very happy to see that, you know, they've gone with a similar architecture as like Hooper or a, uh, like I, I bought two Fitbit Air so I could take it apart and also use one. And I mean, yeah, these devices, once you take out the battery, you can, uh, you can put your own power profiler on it and you can just pretty much see each of the phases that they're going through extremely clearly. Cause there's nothing else using power, right? It's just idling or they're doing Bluetooth or they're doing some kind of sensing or they're doing some kind of process. Right. Right. Right.

Chris Gammell: And actually, can you run us through the, you know, the kind of the, the works, sorry, the, the looks like lookalike kind of things. You said the Google, Google has a new one without a screen. Whoop is one that I think we've maybe mentioned on the show before. I have, I knew some of that work there. Um, yeah. And, uh, so like what, where else is in the space? Like, and where are they targeting as well? They all like athletics or are they targeting more passive?

Andy Kong: So like whoop, uh, an aura were like for a long time, the main like screenless trackers, the aura is like a ring. Uh, the whoop is like kind of like an Apple watch, but it has no screen on it. So it's just fabric covering. Um, then I think Garmin released a screenless one and they also have like a big line of watches. Um, and then all the big tech companies have like their own like smartwatch. I think, uh, from Garmin is mainly targeting like running athletes. So they do a lot of GPS stuff and their, their data is known to be pretty good. Uh, I think like Apple watches, uh, in, I think in head to head comparisons, the Apple watch is the best, like in terms of, uh, accuracy versus like lab standard testing. Uh, and then the, the whoop has chosen to specialize as like extreme athletes. And then aura is like just a nicer looking like tracker. So,

Chris Gammell: yeah. And aura has a bunch of like contemporaries as well, like in the smart ring space, but actually, you know, you know, one of them in a slightly different one, right? You know, Eric Mijakovsky, who's been on the show as well with, Oh, really? Oh, it's been on, yeah, I think 10, 20 episodes ago. And, uh, you said you knew him, right?

Andy Kong: Yeah. Yeah. Yeah. Eric's been super helpful with, uh, like, yeah, he, he lives just down in Palo Alto. So, um, yeah. Yeah. Interesting. Yeah. Eric, Eric, uh, I was like, I tested one of their, their rings early on. Um, yeah. I mean, that thing's also charger, chargerless. So,

Chris Gammell: yeah, exactly. Right. Exactly. That's that, that, that, that one just kind of like one time use and then send it in for a refurb or replacement, I suppose. That's a, an alternative method. I do think from like a ceiling kind of perspective, like seal, you know, uh, environmental ceiling, that, that, that can make sense. Uh, yeah. Amazing. Any of the ring makers can put any batteries around, you know, a ring. It's just nuts to me, you know?

Andy Kong: Yeah. Have you seen the, yeah, they're, they're kind of interesting. Like the ring shaped batteries, uh, the, the characteristics are actually really bad. I don't really know why, but yeah. And then also of course, like different sizes, like you're getting a natural, it's unnatural.

Chris Gammell: Why, why would you put your finger through a battery? You know, like it's just, no, you shouldn't do that. You know? Yeah. Yeah.

Andy Kong: Yeah. I remember the Samsung blowing up, like you remember the Samsung ring on that guy on the airplane.

Chris Gammell: Yeah. Oh, I do actually, uh, my college and high school buddy, uh, also been on the show, Steve cruiser. Uh, this is just a callback show basically at this point, uh, Steve cruiser worked on that and he had to actually do puncture testing for, uh, exponent. And he was doing like testing for the Samsung batteries when that was happening. It was wild. Yeah.

Andy Kong: Yeah. So like Eric, Eric's ring is like the index has like a steel inner and outer. And so like I, the swelling can't really, can't really get you from, from that. That's good. But most, most of the ones are like resin or like plastic or something. So as soon as they get hot, like it starts deforming like, like crazy.

Chris Gammell: Yeah. Hmm. Yeah. Even just metal on my finger, you know, I get a wedding band and, um, I'm happy to have that, but that's like gold. And gold is very pliable. And like, I, you know, they can cut it off if they need to sort of thing, you know, versus like the tungsten rings. It's like, yeah. Yeah. Yeah.

Andy Kong: I'm totally scared about this. Like my first, uh, super con I met, I think Mike heard, he was telling me the story about like, Bill heard. Oh yeah. Bill's missing. Yeah. Bill's missing. Yeah. Yeah. Right. Yeah. Right. And it was cause his wedding band got caught on like something and it like deep gloved the finger. Yeah. Gruesome. Yep. Yep. So I only wear split rings currently.

Chris Gammell: Ah, that's good. That's good. Yes. Yes. Uh, yeah. Good old bill. We've, it's been a while since Bill's been on, but he's been working on the Commodore thing. Uh, you know, the Commodore reboot. So, Oh, really? Okay. Cool. Uh, okay. So let's go back to the, um, the deconstructing of like this, you know, this analog front end type chip that could measure, um, the heart rate. Now you had to deconstruct and build it yourself. So what did it take to do that?

Andy Kong: Yeah. So, uh, I think I've done this before. It's one of my earlier electronics projects was just making like a, um, like PPG sensor. And if you just leave the, like, for instance, if you just put a green led next to a photo diode and then you amplify the photo diode, um, you'll see like if you, and you scope the output of the amplifier, then on the oscilloscope screen, you'll just see your pulse, like very small, like 10 millivolts or something. Uh, and in like, this obviously leaves the led on continuously, which is like super expensive. But if you, you know, you only like the heart rate doesn't change that fast. So you can actually just turn it on at like 50 times a second, get a sample and then turn back off. And there's actually no limit to how small this pulse can get. Cause you're just checking the color of the skin essentially. Uh, and for, for me, I was like, we can do the controlling. We can control the timing with the microcontroller. Uh, and then you can just take each aspect that you need, like the led and the amplifier and just turn them on exactly when you need it. So the, for us, the photo diode, uh, amplifier comes on and then like a few microseconds later, the, the, uh, the led blinks, we acquire the sample, we turn off the led, we turn off the amplifier. And so, I mean, it's actually quite simple in, in terms of like, like, you know, describing it, but yeah, getting the timings to be smaller than, um, than like the commercial health trackers is, is pretty tricky.

Chris Gammell: Yeah. I mean, what are the relative to, you said microseconds in order to do this? Like, aren't there like ramp times and things like it, just like thinking about the detector collecting photons sort of thing. Like you gotta shine the photons, you gotta bounce off. Yeah. Obviously it's the speed of light, but like, you still gotta like charge that thing up. There's capacitance and all that other stuff too. Right. Right. Yeah.

Andy Kong: Yeah. It's a pretty small, uh, like photo diode capacitance is like, you know, the whole thing's like a capacitor, I think. Um, but it's like a very small amount of silicon and it takes a, I don't know, like after the amplifier, it takes maybe like five microseconds to amp up. Um, and the led comes on instantly and turns off instantly. Um, the main tricky bit is acquiring the sample fast enough and getting the, um, what is it? Getting the amplifier like on and off fast enough. Cause if you, yeah, actually kind of funny. So you know how they tell you to put like a decoupling caps on everything so that they're like, like the signal output is smoother.

Chris Gammell: Yeah.

Andy Kong: So if you remove the decoupling capacitors, then like the device comes on much faster. Like, you know, as basically as soon as a wire hits that voltage, like, uh, like the device is on. So, uh, got it. Yeah.

Chris Gammell: Yeah. I mean the, uh, that's interesting. So what kind of like, uh, so it's like a transimpedance amplifier to get, to go from current to voltage, that sort of thing. Exactly. What kind of gain are your, what kind of gain stages in there? Yeah.

Andy Kong: Yeah. So ideally you have two gain stages so that, um, like if someone has a different, uh, skin like color, then you can increase the led brightness or you can increase the gain on the second stage. And you can't really put like a digi pot on a, on a transimpedance amplifier because the capacitance of the digi pot is so high, uh, that it slows down like your overall rise time. And so ideally you would do like a, a basic transimpedance amplifier and then put a digi pot in the gain path of a second amplifier, which gets it, uh, to maximal range. And then you can get like good signal out of it.

Chris Gammell: Okay. And I mean, like what are the relative, I don't actually know what the requirements are. Like how accurate does it even need to be? Cause like you're doing like these samples as, as, as the bleep bloop does its, uh, various stages. Right. But like, like what is, uh, what is the expected? I guess I don't even know. Like what does the signal chain like need, need to have in order to have like a cogent signal?

Andy Kong: Yeah. I mean, this is something nice about, uh, I think health tracking is that you can kind of like, like the sensor that does health tracking cannot really be improved that much once it does, uh, the tracking, uh, like for instance, like your temperature sensor at a certain, if you can get a temperature sensor down to, uh, let's say like on average, it averages like under a microamp, uh, for you to acquire like a continuous temperature reading, then like, you kind of don't really need to do much more, at least for the human, because the human doesn't change. Right. Like if we need temperature sensing on a machine, like the machines will, the requirements will always change, but on a human for health sensing, it's kind of a fixed problem. Like we need to measure the system. System doesn't really change. Uh, and as long as you measure like the same set of characteristics, you can kind of always, uh, like you can, you can finish the task and like, yeah.

Chris Gammell: It's like precision versus accuracy. As long as you have precision, you're probably in a good space because you're consistent across population versus like absolute accuracy of like, like you're not measuring, you're not like, uh, uh, what's it called? Calibrating against like absolute zero or something like that. Right. You don't need to know. Right. It's exactly 22, 73.1 Kelvin or, you know,

Andy Kong: right. Yeah. So yeah, for us, the heart rate, uh, is like, yeah, it's kind of a choppy. It's kind of just a, you know, whenever your heart rate goes, uh, the, the color will lighten a little bit. And then, so you'll get a bit more light coming back. And then that signals like pretty small on the overall, like headroom of the, of the led. And you can just sample, uh, the pulsing and then do your own like peak detection on it. Uh, actually most sensors in the heart rate trackers today are, are just, they're just like peak detectors. Like step counting is just peak detection on absolute magnitude. Right. And then the heart rate sensing is peak detection. Uh, sleep tracking is like, not really,

Chris Gammell: but it's in with like a step tracking. They were doing like algorithms on top of the accelerometer to like detect actual motion. Is that not the case?

Andy Kong: Yeah. I think for probably activity tracking, um, you may see that like, you know, swimming profile looks different from a running profile. Um, looks different from biking, but it's, I mean, you know, it's just peak detection on the different axes, like the different motions that you do when you're doing these, like you're just looking for the repeats.

Chris Gammell: What, what made you want to do, you know, the health, health tracking in the first place. I mean, like you said you had kind of like a bio bio sensing background, that sort of thing.

Andy Kong: Yeah. Yeah. So I, I, I did a lot of like research in, in college on like biosensing and also in, uh, biostimulation actually. So like electrical muscle activation with like pulsing electricity through like, you know, like the tens unit that people use for like some kind of, uh, physical therapy. Um, and I started, yeah, so I was, I was really interested in biosensing, like actually in the body via the machine. Uh, but I was really interested in like getting, you know, our intent out through like some, some form of like biosensing. Uh, at some point I realized that like, it's actually very difficult to get like a human intent, which is like targeted. So like I was, I was looking at some EEG stuff. So like brain computer interfaces, um, like there are some non-invasive like kits for it basically, but they're very, they're very limited in how good the signal is. Uh, and I realized that it kind of isn't going anywhere. Like most of the EEG headsets generate just noise data. Uh, and so that, that I started pivoting into like health stuff.

Chris Gammell: And that would be like the, oh, what was the, there was one where like the kit you're talking about is like the one where you could actually have sensors in your head with like a plastic, like skull cap. on top of it and you get screwed. Yeah. Yeah. Yeah.

Andy Kong: The open BCI has like a, it's probably like 10 or 20 now.

Chris Gammell: Got it. And the noise is due to what? Like scalp movement versus the actual like sensing.

Andy Kong: It's so funny. Uh, like everything, uh, everything causes noise. Yeah. The, the cable sway, uh, so, uh, for reference, like a non-invasive, uh, brain computer interface measurement, you have about one to 10 microvolts of signal. And so to get that into where you can measure it, you need to like, you know, like a hundred thousand gain or something. Uh, yeah. Yeah. Right. And so small, small variations, you know,

Chris Gammell: someone walks across the carpet, you're toast.

Andy Kong: Yeah. Yeah. Uh, and like the cable moves a little bit and that induces a volt, like, you know, it's just, it's just awful. And you can't even shield because, uh, you can't shield on the head. Like the head doesn't have a ground, like your ground references at your ear or something, but you're still the cable. Like if it's not terminated, then it kind of doesn't really do that much. Uh,

Chris Gammell: what if you required your users to like lay on a piece of tinfoil or something like that? That's, you know, got a ground spike. That's a joke. For the listener. Andy is like, you're seriously considering this. That was just a joke. I didn't know.

Andy Kong: You get the tinfoil hat. The. Yeah. Yeah. Yeah. Yeah.

Chris Gammell: Yeah. That's interesting. I mean, yeah. And I mean the, I know that there's a lot of stuff, there's a lot of stuff in like the implant space right now as well. And, uh, I'm very squeamish. So, uh, yeah, yeah, I don't, yeah, I, I could see. And also just like how invasive it is from, uh, you know, controlling a computer for, uh, someone who doesn't have any health issues versus someone who's like, actually like dealing with paralysis and things like it. It's like, okay, well, maybe there's other reasons.

Andy Kong: All the nearly people. Right. Yeah. Yeah. Right.

Chris Gammell: Right. Yeah.

Andy Kong: But then I started like, yeah, it was like, if you want to get a good signal out of the head, like you have to go to the source. And that was a bit too much for me in my undergrad. Uh, so I started being interested in health stuff. Um, yeah, actually I was, I was really interested in, uh, do you know quantified self or like the personal informatics communities?

Chris Gammell: Um, maybe I, the term sounds familiar, but I don't know if I know any of the communities.

Andy Kong: It's usually people who have like some chronic health problem that like all the doctors have no clue what's, what's happening. Um, and then at some point that people are like, all right, this is really annoying. I'm just going to try, start tracking everything in my life that's pertains to like the thing that's affecting me. And then like figure out like what actually causes it. Uh, yeah. And so today it's easier than ever to sense like everything in your environment. And so these people will capture like, you know, like, uh, the noise level in their car or like their location constantly. And then put like air quality sensors everywhere. And you have to go back and like sink all the data up and then you can start doing analysis on it. And actually today it's like, it would be, it's, it's so easy to do, uh, analysis across like, you know, tons of different.

Chris Gammell: Throw the data sets in the clanker and just be like, Hey, you find any trends here? Yeah. Yeah. Yeah. It's incredible.

Andy Kong: Especially have any idea of like what you think might be causing it. Uh, it's very good at like figuring out hypotheses and, and like all the analysis is just like, you know, uh, is there a relationship between two, these two things and calculate the P value, uh, like of the effect and everything. So, yeah. So that,

Chris Gammell: that's what really got me into it. Shades of my, my old like statistical process control. And I, I always remembered like, like doing like a design of experiment. And then you find that one thing that like had the biggest knob, like that you could just turn, um, you know,

Speaker ?: yeah.

Andy Kong: That's funny. Yeah. It reminds me of like when we were doing the, I mean, now when we're doing, uh, optimization on the power consumption of our heart rate sensing, like it is kind of just figuring out where the knobs are, uh, like how, what can you reduce so that we can still get the same signal? Um, but like we keep turn the led off faster or something. Yeah.

Chris Gammell: Yeah. So, so, so then to go back to the, to the, um, the church list as well. So the two other, elements that are really interesting to me are one, just the harvesting on a relatively small cell. And then the other being the Bluetooth, like to actually have low consumption, you know, Bluetooth is lower consumption than I think of, but like how, how low can you go?

Andy Kong: Yeah. Uh, for us, the Bluetooth is, I would say like, I think under 5% of the total power consumption. Um, people are always really surprised by this. I'm, I'm not really sure why I think it's cause they're used to using ESPs or something, which, you know, kind of just burn like, I think like 60 million or something when it's transmitting.

Chris Gammell: Well, I mean, depends on how much data you're sending to, I suppose. I think it's not like, are you formulating, like, are you really bit packing things down? Are you sending Jason? Are you, what are you, what are you sending?

Andy Kong: I'm sending raw sample currently. Yeah. Most of the processing currently happens on the phone. Um, just because I wanted to be, uh, I wanted to make it easier to experiment with different algorithms. Like while we're still developing.

Chris Gammell: Yeah. And so like, like what is the, what does the payload look like then? So like if I, if I put this thing on my wrist and it's monitoring me for 10 minutes, how much data will actually go across the gap to the phone?

Andy Kong: Yeah. So you can think like we sample at 50 Hertz, let's say our, uh, window is like 10 seconds. So you get 500 samples of whatever your ADC is outputting. Let's say it's two bytes. And then you get about a kilobyte. And then like your temperature is like a couple of bytes. Uh, you know, you, you get your emotion in there, maybe like, like 300 bytes on that. So it'd be like 1.5 kilobytes per, per second, per sample. And then we do that like once a minute. So it'd be like 10, 10 kilobytes. But then you imagine like that gets off the, like once you've established the Bluetooth connection, uh, you can pretty reliably transmit like a hundred kilobits per second. And so you can get that off in like a second or something. Um,

Chris Gammell: yeah, got it. So like radio on time is still relatively low. You're not cranking the power output. You're not, you'd be, you're expecting like close proximity. You're not trying to like reach across the room for,

Andy Kong: we don't have to transmit like continuous audio,

Chris Gammell: like across like 10 meters or something. Got it. You're not doing like anything like fancy, like coded files or anything like that either. It's just straight. I haven't had to yet.

Andy Kong: Yeah. I mean, we, I want to, uh, I could, it's very easy, like, um, kind of like stream encryption you can do where like, you know, all the heart rate sensors, like numbers will be centered around like approximately like a range, which probably not your full range. So you can reduce the, uh, the, the bit rate of the sample there. Um, it's not, and there's no like,

Chris Gammell: there's no like real time stream as well. So you, you can batch stuff too. Like you can take a 10 kilobyte chunk or whatever it is and batch it up. And yeah,

Andy Kong: yeah. Yeah. You just blit out like, like every 10 minutes or so we sync with the phone. Uh, or we, we try to like the phones make it actually kind of hard to do background Bluetooth syncing. Uh, cause, cause it uses a bit of power. Yeah. Right, right, right, right. Yeah. Actually, there's, as a, as a medical device. So the glucose monitors have like, like guaranteed, like Bluetooth listening for their device. Really?

Chris Gammell: Yeah.

Andy Kong: Yeah. Yeah. But you can't get that as a, as a mere wearable.

Chris Gammell: Well, all you got to do is just go through the FDA and just, you know, do all the, all the terrible, terrible paperwork and maybe you too can get this handy, you know, but then you also get those Medicare bucks. So, you know, it'll come in, it'll come in, you know? Yeah. Yeah.

Andy Kong: We'll have to do the math. Yeah.

Chris Gammell: Yeah. I don't think, I don't think it's worth it. I think it's better to find, uh, yeah, find, find the right audience, make a, make a good product and yeah, go from there. Right. Right. Yeah.

Andy Kong: Well, well, bigger FDA ambitions in the future. Hopefully.

Chris Gammell: Yeah. Yeah. Right. Yeah. I mean, I do think that low power passive, I think passive monitoring and just like the ability to hand off data like that is, is important. Right. And especially like thinking about like, like I, I think about like elder care type stuff as well. Like asking, asking my parents to plug something in. It's just like, you know, they're, they're not, they're not like, yeah, exactly. Tech unsavvy, but it's not, it's just like, it's any kind of friction in the process. So like if, if it's a minute of sunshine a day, they're, they're going to be.

Andy Kong: So I heard this neat thing about whoop is like they, they discovered there's like doing market research, but basically if you have to take off the device to charge it, there's like always some very like small percentage, you know, maybe like under 1% of a chance that the user just never puts it back on. And so for whoop actually they constructed like a, their battery pack charger is like always plugged into the wall and then you slide it over your whoop. So you actually never take off your whoop while you're charging it. So of course this means like very inefficiently charged. It's like, it gets very hot. Sorry.

Chris Gammell: I was about to say it gets really hot all the time. Three people have died. No, I don't know.

Andy Kong: I was going to say my ankle gets hot, but like, then like I, that's where I wear it. So the whoop is kind of,

Chris Gammell: how many trackers do you, Oh, you had the thing about this on X, right? I think when I was looking up your stuff on X, you said you had like a bunch of trackers on it. Yeah.

Andy Kong: Yeah. So I used to have like three, three or four on, cause we were comparing the data. Um, but yeah, the, the whoop is, is very tall. And anyway, so it's nice what they do with the charger. Um, and yeah, I mean, we hope to avoid this problem entirely.

Chris Gammell: Yeah. Yeah. That'd be great. That'd be great. Yeah. Okay. So now solar panels. So you have a solar panel on there. Uh, it's, I mean, like reminiscent of, uh, you know, solar, solar calculator, uh, like level.

Andy Kong: It's the same shape. It's approximately. Yeah.

Chris Gammell: Right. Exactly. It's like, yeah, it's a thin band and it's more flexible. It seems like. So how does that, how does that work? I mean, like, like energy harvesting chip, what's actually happening to the hood?

Andy Kong: Yeah. So I was, I was really interested in this. So there's a lot of, uh, like big harvester chips, which are meant for like, you know, solar farms and stuff. Um, and, but like for us, it's really important that the, if while you're harvesting, the harvester doesn't use that much power. So the harvester is just a boost converter, right? So it takes like, you know, the half a volt or so that comes off a single panel solar cell and then boost it to like the battery voltage. Um, but it's really important that that thing doesn't use like 10 microamps because, uh, you may only be getting 10 microamps out of like the actual thing. So there's a lot of chips, which do the harvesting and you just have kind of have to look for them. They're actually a little difficult to source, I would say. Uh, and they're still, they're not small yet. There's only a couple of like chip scale packages, which would be small enough for us to, to fit in our device. Um,

Chris Gammell: yeah, I've looked at the linear tech one before and, and I mean, linear tech, I mean, obviously it's analog devices now, but like just the, the price alone, I'm just like, I'm out of here. I don't need it that bad. Just put a bigger battery in there. How much was it? Six bucks. I mean like six bucks, the digi key price. It was, it was not, it was not meant for, uh, you know, consumer level goods, right? It means packaging, everything was, yeah, but it was also like, that was like, like you could do like vibrate. It was multi-sensor too. You know, you can do small solar cell, you can do vibration. You can do all the things where it's just the, the, the voltage, I think it was like 38 mil micro millivolts rather. And you know, it's just like diode level, you know, small, small.

Andy Kong: Yeah. I would hate to harvest a thermal. Like I really, I'm interested in making like a thermal powered thing. Cause like, then you don't even need the sun. You just have to be like alive. Uh, yeah. It's like the Peltiers you can get on like a wrist one thing are only going to generate like, I don't know, like like 20 to 50 millivolts or something. Uh, like harvesting that, like imagine. Um, yeah.

Chris Gammell: But I feel like that's also like every, uh, every like charlatan startup be like, Oh, we were, we're using Peltiers and we're going to harvest, you know, ton, you know, kilowatts of energy from people. Yeah.

Andy Kong: Okay. The harder part I should say for the solar is actually getting the solar cells.

Chris Gammell: Yeah.

Andy Kong: Yeah. Cause like, I mean, they're just, uh, like if you want an amorphous solar cell, like the ones in, um, solar panel, but like in solar calculators, uh, they kind of like, I think they sputter it onto glass or something. Uh, then they, they lean it out, but the efficiency is quite bad. And I was really interested in optimizing our device for outdoor charging. I wish I had a minute of a day outside. And for us, like no, since no application uses like comparable power to us, like there's nobody who makes this small of a solar cell. And so we had to go like find people who can custom cut like solar cells. And like, actually when I was first starting this, I was like at MIT and, I had a friend in a lab, they had a laser cutter and I bought like the, you know, single panel wafers. And then I started dicing them on the laser. And like, you, you learn so much, you learn so much about solar panels. Yeah.

Chris Gammell: Solar panels don't like that at all.

Andy Kong: Yeah. They do not like being cut. Yeah. So it's neat. Whenever you buy them on Amazon, they're like, you know, they're mounted onto like a PCB substrate and then you just solder two wires onto it. Right.

Chris Gammell: Yeah.

Andy Kong: Yeah. But if you actually go on the panel itself, it's just like metal deposited onto silicon. And if you touch it, it like shatters. Uh, it's quite tricky. Yeah.

Chris Gammell: Yeah. Yeah. So this actually is a good, good segue into like, how are you sourcing this stuff? So like, where are you in the production life cycle of this thing? I think I saw hundreds of units you're at. Like where, yeah.

Andy Kong: So we're making like a few hundred, we're making a few hundred units right now. Uh, most of them are going out to customers and we'll send out a couple of promo stuff. So if anyone's interested, they can, uh, they can ask me for one. And, um, yeah, I would say like production, I'm not super concerned about capacity of parts in the near future, but mainly it's, um, like the next step will probably be to go to a contract manufacturer, uh, and like transfer our process knowledge over. Yeah.

Chris Gammell: Yeah. Okay. And is there anything on the, on the microcontroller side? Is there anything kind of like unique required from your, from your micro or is it kind of just off the shelf Bluetooth? Like what, what, what, what actually off the shelf.

Andy Kong: So we're using like a renaissance microcontroller. Um, I think, I think, uh, Eric put me onto this or something, but yeah, it's pretty, pretty standard. Yeah.

Chris Gammell: Yeah. Yeah. I mean, they're in a lot of consumer goods, especially like they, they, they know the space and how to do that sort of thing. So that's, that's good. On the coding side, do you have to, do you have to like be digging down and registers or? Yes. Yeah.

Andy Kong: Yeah. Yeah. Yeah. They, there's like not, I, I don't, I've never, you know, like when I, when I was doing electronics in college, I, I didn't really know, uh, like super much in detail and you're kind of just trying to do like a cool, uh, gadget kind of project. And so these are a lot of ESPs, a lot of NRA stuff. And I, I only use dev dev boards. Actually when I started, before I started this company, I'd never put a microcontroller on my own PCB. It was always a dev board, uh, it goes on a PCB because I don't even know how the programming stuff works. But yeah, now it's like, like once you leave like that nice ecosystem of like stuff that hobbyists have developed on for a long time, you, you kind of have to figure out like, you know, the SDK is just a wrapper on like the ROM functions or like there's a wrapper on writing stuff to a register. And yeah, you just do that. Yeah. Yeah. It's crazy. I think a very, very new to me.

Chris Gammell: Yeah. Yeah. Yeah. Check out the CH32. There's no other options really. CH32 fun library is just like, all right, you're going to be pouring, uh, commands into all your registers now. And, uh, you know, you write little macros and you're good. Oh, really?

Andy Kong: Yeah. Yeah. I, I, I thought it was so janky. And then I realized like, this is kind of just how you're supposed to do it.

Chris Gammell: You know, it is, it's, it is, it is absolutely. So like, if you don't, if you're not going to do Renaissance anymore, you are hosed, you're rewriting everything. Right. That's the reason, you know, that's like the story a lot of people tell themselves. It's like, I love, I love Zephyr. Right. But, and like, that is, that is the use case of like switching between parts and having cross compatibility, whatever. But you pay for that in layers of abstraction. And if you go the other direction and you're like, no, CH32 V003 or whatever. And I'm using the same register set all the time. Like who cares? You know, it's faster.

Andy Kong: It's yeah. I looked at Zephyr. Yeah. So I, I, I'm surprised you like it actually, because when I was using the NRF, you know, they're, they're migrating, at least they were migrating to Zephyr. And like, but there was still a lot of. The transformation is complete. They're,

Chris Gammell: they're, they're, they're all in. Well, not fully actually. They, they have the Air Metal NRF5 for just one part. It was very obviously made for one customer. And then they're like, we could, we could release this for other people too. They got a lot of pushback.

Andy Kong: Yeah. I see. Yeah. But even then like the Zephyr stuff, it's like integrated, but you know, you still need to do, you still need to know how the SDK works. Cause there, there's some like parts of it, which have not been abstracted. Right.

Chris Gammell: Depends on the, yeah. I mean, I think Zephyr is such a large surface area, really, you know, you and I think about it probably from the microcontroller perspective and the, you know, Bluetooth stacks and things like that. And it's like, that seems standard enough if you look across like portfolios, but like, actually I've been, I've been looking at parts that are running Zephyr on cortex a, and it's just like, Oh, wait a second. Like the same core that's running like on a B, like Linux on a B, the bone people are running Zephyr on that too. And it's just like, that is, that is an insane difference up and down the stack. Very interesting. Right. Yeah. Yeah. I think you're going to have a lot, a lot of trade-offs where you can cover both, you know?

Andy Kong: But yeah, so this is my first time. Yeah. I was like crazy also figuring out make files and stuff for the first time. Cause this is like pre LM, like they couldn't figure it out for you yet. And like, you know, it just wouldn't compile or something or, or it would like map your program into the wrong memory space. And then like, you just like nothing would run. And I remember like that, that took like, that took like three weeks or something. Only three weeks. Come on, man. You get the dev board.

Chris Gammell: This is, we're treading upon the path that many have, have gone before. And it's, yeah, it sucks. Well,

Andy Kong: I was doing it full time. And so I was like, I, Oh, actually it might've been impacted. Cause I was running like a weird sleep study at the time. You were subject to sleep study. I subjected myself to a sleep study. Yeah. That is quite, I was, I was really curious because you know, like, you know, the day is 24 hours, right? Most people know this. Depends, depends on what planet I'm on, I suppose. Yeah. Depends on what planet, right? Yeah. But if you put, if you imagine each day is 28 hours, uh, then every week. And so you, you sleep eight hours, but you are awake for 20 and every week you will completely cycle back to, uh, being awake in the morning again. So every week you can decide to stop. Don't you lose a day? And at the time, right? Uh, yeah, you lose a day and it all comes out of your sleep. Yeah. Yeah. Okay. All right. All right. Yeah. So I, this is the first time I was like unemployed and like, uh, calendar. Oh yeah. That was pretty hard. Yeah. You, you go to text your friend, it's like literally 4am and they're like, like obviously not going to respond. Yeah. Um, but that was a fun, I, I, I tinfoil all my windows and then I was, I was locked in with the, uh, the dev board and like,

Chris Gammell: okay. Yeah. I saw some friends. Good, good, good. Yeah. That was, did they come and visit you and they're like, Andy, are you okay?

Andy Kong: But yeah, the tinfoil on the, on the windows was a little concerning to some of them, but it was, it was for the light. Yeah. It wasn't about the signals. The signals were getting out. The whole room, right? You're fine. Yeah. Yeah. The, the mic controller.

Chris Gammell: That's great. And so, uh, you have not taken this product to CMs in China yet, but that's the, that's the goal maybe at some point you think?

Andy Kong: That's the next step. Yeah. I, I visited a couple of last time I went to pick up some components and like talk to other manufacturers. Um, yeah, I mean, it's pretty neat stuff. They, they have these like stickers they put on all your phone cameras and your laptop cameras. Uh, like, so you can't take pictures of their lion. Uh, but then they'll just show you everything. Yeah. Um, like they're very, they're very willing. That's great.

Chris Gammell: Yeah. It's a, it's an interesting, interesting time. I'm sure. Yeah. Yeah. That's great. I'm sure you're going to have a lot of great stories after you have to go through that.

Andy Kong: Have you, have you been, I've not been, no. Like have you done manufacturing? Okay.

Chris Gammell: I've done manufacturing in different places, but not been over to Shenzhen or anything like that. So it's on the list. It's on the list, man. Uh, we have a guest coming up who I will be a expert in that space. So no, no spoilers looking forward. Okay. So what, uh, what price point are you targeting for your device and stuff like that too? Cause I feel like this is, this is also an interesting, like, you know, it's, it's a consumer level device, but it does seem like higher end. And it also seems like it's targeting maybe, uh, it's not the bottom of the barrel kind of like cheapest. So like where, how are you kind of positioning in the market too?

Andy Kong: Yeah. The Chinese, like, uh, so like aura is like, I think like $400 now or something. And like the Chinese rings are priced at, I don't know how they're making money. The Chinese, the cheapest health tracking ring you can get is like $11. Wow. So that's probably close to the floor of like, you know, how much the hardware costs, uh, and like the, the manufacturing process, you know, whatever. Um, and so when we started, it was like, it was only like three or $400 trackers. And, and so we priced our, ours at one 99. Uh, and there's no subscription cause I want to be able to get it like continuous data from a lot of people. Um, and then sell it. Is that what you're doing? Like, I don't even know where I would sell. I don't think you can sell. I don't know, man. Well, I want to be able to have, so like, for instance, um, I did some, I used to play Tetris a lot and I would analyze my Tetris score after I had slept like poorly or well. Yeah. And you could see like very, like big differences in the performance. Yeah. Yeah. And, but like, that's, I think, you know, people are always kind of throwing off like data on this stuff without being like capturing it. But if you have a baseline of like health data to coordinate everything with, um, then you can do your own analysis. So what I want to offer is like, you know, some nerd is going to do like, Oh, here's my analysis of my Strava with my health, with my heart rate data. Uh, and then they can sort of publish that and then people will download it and run it on their data. Like with, from their services. Uh, and so this is kind of what I'm going for with the, from the health angle. Got it.

Chris Gammell: So like kind of open data format and the ability to kind of build on top as well. Right. That, that feels like that's. Yeah, exactly. Yeah.

Andy Kong: I think like, you know, our app is going to be like good or whatever, but I, the, it's most important to me that it's easy to get the data out, uh, onto like somewhere you can analyze it. And ideally like, you know, I, I want to, uh, build more like analysis tools into the device itself. Cause I think that's like kind of the bulkiest stuff. Um, and then we can just run it. But if you'd like to, you know, export it, like for instance on whoop, if you try to export your data, half the time, they don't even give it to you. The last time I export it is two months. Yeah. Yeah. Yeah. And like, welcome to whoop. You want your data? We say, like the EU, we say,

Chris Gammell: noop, noop, no data from you from where you whoop.

Andy Kong: Yeah. I don't, I don't really understand. Yeah. Yeah. But yeah, I would like to make it really easy.

Chris Gammell: That's great. Yeah. Yeah. And especially phone side is already processing all that. Then that feels like that is a easier way to kind of get it into the, there's no like cloud backend or anything like that either. That's.

Andy Kong: Uh, I mean, like we, we, we, we back up the data. Yeah. Like I, I've lost my phone before and like, it would suck to like lose all your, all your data. Yeah. That's true. But.

Chris Gammell: But device side, device side, uh, as the phone, the phone is the processing, that sort of thing.

Andy Kong: Device side, right? Yeah. You don't want to do, you don't want anyone to do something. Like, I don't want to do something to a user that I would not want done to me. Right.

Chris Gammell: Yeah.

Andy Kong: Uh, so like, I just want to make it very easy to export. Like we're not selling it. Um, yeah. I think like, if you have like someone who cares, like, yeah, like personally, like I not going to, you know, screw over someone like it could be you. Like, um, so yeah.

Chris Gammell: Empathy for your users,

Andy Kong: that sort of thing.

Chris Gammell: Okay, cool. Um, all right. Well, I am very excited to see how this project goes, the product rather. Um, I would love to talk about the wireless skin thing before we finish. So you have a product project called power over skin. So let's talk about this real quick because this is very novel. And, um, I didn't, I didn't see this coming. I gotta say, tell me about it. Oh,

Andy Kong: did you, did you read the paper? Like the talk or anything? Like how much context do you have?

Chris Gammell: I just a little bit. Yeah.

Andy Kong: Okay. Yeah. So it's my pin tweet. I did this project maybe like two years ago. So I came back from, uh, my time in Zurich and then I went back to CMU and my old prof was like, Oh yeah, you can just come hang out in the lab. And at the time I was really interested in making devices that didn't need charging. Um, like that wasn't how I was describing it, but I was like, I want to make devices that work for free, uh, like mentally and like electrically. Yeah. Yeah. And so I was excited.

Chris Gammell: Energy harvesting. Passive. That's right. Yeah, exactly. Yeah.

Andy Kong: Yeah. And same time, you know, I was starting to get like more devices. I had my like wireless earbuds. I, you have your fitness tracker, your phone, you have your laptop. Um, you know, there's a couple other things in there, but my point was like, whenever you're interacting with a device, you're touching it. Right. And what if a big battery that you wear on you, uh, is somehow able to push power through your body contact, uh, to power the device. And then like, you know, everything is dead until you pick it up. And as soon as you start using it, it's alive and you kind of energize it in that way. Yeah. And so what we're, there, there's this thing called, um, human body communication or like HBC or body channel communication, where you use your body as a antenna for sending data. And I was like, Oh, well you, you must be able to just crank that and like send power, you know, like data is just a structured power. And so, you know, whatever frequency that you're sending the data through, you can just put like a fat, uh, RF signal in, into the body. Uh, like not, not that fat, right. I mean, it's like, it's like only a few milliwatts. I mean,

Chris Gammell: in my mind, I am imagining like Wiley coyote, like getting shocked and like, just like the yellow outline, you know, that sort of thing. But it does seem like it might be a little safer than that.

Andy Kong: Yeah. Well, I had, I had to read like, you know, the, the, the, there's a council on non-ionizing radiation and they put out like a report of like all the dangers of like, um, like what the limits are based on your body weight and stuff. So I had to read that, which is fascinating. It's really interesting stuff. Yeah. And the only problem is that like whatever power goes through you is heating you.

Chris Gammell: I was going to say, do you, do you want people criticizing you for like, you know, controlling their mind with 5g? Cause I think you're right on that path. I gotta say, I mean, yeah, and this is like, this is like that model where you have the, the jelly filled head with like the GSM phone, right? Like that's the body heating thing for non-ionizing radiation. Yes,

Andy Kong: I think so. Yeah. I think so.

Chris Gammell: Galitzkaya talked about that. Cause she had like a head filled with goo and it was like a test model. And she had told me about that, but like, yeah, yeah. The goo head. Yeah.

Andy Kong: So what's neat was like, we, we want to maximize a signals transmission to the body, right? As you go higher frequency, uh, since you're like a little conductive, it becomes easier to send power through, uh, like longer distances. But as you keep increasing the frequency, it starts radiating. And so this, there's naturally, it just evolves this like peak point around like 40 to a hundred megahertz. And that just like fills the body, um, like fairly uniformly. And yeah, so it just does bulk heating on you. It's like warms you like by a milliwatt, which is very little compared to how much power, you know, your body is generating.

Chris Gammell: Got it.

Andy Kong: Um, but no cancer, like no mind control, I think.

Chris Gammell: Got it. Well, you say that, but of course you'd say that. Right.

Andy Kong: So then, okay.

Chris Gammell: So say you're just transmitting at 40 megahertz then. And how do you actually harvest that on the other side?

Andy Kong: It's really neat. Yeah. You, you know, you're putting a milliwatt through, you may only get like a, you'd be lucky to get a hundred microwatts on the other end.

Chris Gammell: Okay.

Andy Kong: Um, and so like no device is really like, is like super prepared to like harvest such tiny amounts of, uh, voltage and not to mention like, you know, we put in like, maybe it was like a five volt or 10 volt signal and you only get like a few hundred millivolts on the other side. Like you can't, you can barely even rectify it. And so there's like a very small, like there's a set of filters you can put on DigiQ or anything. Uh, you need the reverse capacitance to be really low. So it doesn't connect the other way. And then you just make a rectifier out of it. Oh, sorry. And you need the four drop to be extremely low. And so there's one chip which does this. I think it's from, uh, sky sky works. Is that what it's called?

Chris Gammell: Oh, then they just rebranded as like something silly, like one of the power companies, right?

Andy Kong: Yes. Yeah. So they have one diode, which is the perfect candidate for this. And you just do like, you know, two diodes, two capacitors. And that's, that's the harvesting. Um, we actually, we borrowed a technique from this Chinese paper, uh, on this topic. So not many people have done research on this. I think like there's five or six papers total about this.

Chris Gammell: Let me say a couple of people have, but they, they burned up in atmosphere. Yeah.

Andy Kong: Yeah. They didn't make it. They didn't read the safety guidelines. Yeah. Yeah. So there's one more neat thing you can do since you know, the frequency you're transmitting at, you can produce a tank circuit on the input, even before the rectifier. And that will trade off current for voltage. And then you can get more power going through your device. And then after that little unit, uh, and the rectifier, you can just harvest it. You just get like 300 millivolt or something. And you start harvesting that.

Speaker ?: Yeah.

Chris Gammell: This is pretty wacky. I gotta say. Yeah. Yeah. Yeah. Yeah.

Andy Kong: So, yeah. Yeah.

Chris Gammell: Uh, you built it. You built this thing.

Andy Kong: Yeah. Yeah. Yeah. We made like, uh, maybe like 10 or 20 like little units. It's really small stuff, right? Cause it's RF diodes and like tiny capacitors, like tiny power. Yeah. Um, I think our storage unit was like a hundred microfarad tantalum capacitor. Um, and yeah, our first prototype was called the snail. Cause it was like, you need, yeah, obviously like radios, uh, one leg of the path is through your body. The other leg is through the air or the environment or something. And, uh, you need like some large ground plane. And so we used a small copper heat sink, uh, for just for the surface area. But then everyone thought it was thermal harvesting, uh, when I posted it on Twitter. Yeah. But so yeah, it was like kind of snail shaped. And then whenever the capacitor filled to a certain voltage, it would discharge it to its lower bound. And then we start charging again. And so we use this to, um, measure like how fast we're transmitting power. You actually can't scope the receiver unit because when you attach the scope, you couple it directly to ground and it massively increases your, your power. Wow. Okay. So it's tricky to even measure. Yeah. Yeah. But yeah. And were you able to actually power a final device with it or like what? Yeah. Yeah. So my, my lab is really interested in applications. Like for all our videos, I think we do like five applications at least, um, like demos of it doing stuff. So we had one that was just the blinking and then we had a, an led earring, uh, which I had a decorative heat sink, like made of copper. Um, then we did, I think it was like a skin patch that did temperature sensing. Uh, and then we did a joystick, like a Bluetooth joystick, uh, that you wear like as a ring. Yeah. So varying levels of power from like, I don't know, like 200 microwatt up to, up to like a few milliwatt was what we got, depending on how far away it is and everything like this.

Chris Gammell: And the, uh, would, would there need to be some kind of like testing to figure out kind of the optimal 40 to a hundred megahertz? Like how did you, how did you figure that, you know, 43.75 meters? Yeah.

Andy Kong: It's a little different for everybody else. Like for each person, right? It, it depends a little bit, but it actually doesn't depend on the skin conductivity and it doesn't really depend on how, how much fat you have. Um, so it will change the range a little bit. What we did was, uh, I just had it on a couple of people and I just manually turned the knob on the frequency, uh, or like the frequency generator. Um, and then you just see how much voltage is coming out on the other side. And I just took a continuous video of it. And afterwards I labeled what the, what the reading was. And then that was how we approximated. Yeah. But one other thing we found was like, if you do a 40 megahertz, like sine wave, right? You only get 40, but if you do a 40 megahertz square wave, it's actually easier to generate. Uh, you get all the harmonics. And so you can actually start to harvest from the harmonics. Uh, and that boosted our power, power delivery by, by a bit.

Chris Gammell: Huh? Yeah. This is, uh, that's pretty wild, man. And so you said,

Andy Kong: if you only need a couple hundred microwatts, yeah, it's like, it's like totally great for your application. And you can think if you put it in your laptop, which is most of the time wired and you're like, you got your wrists on it. So you're got good contact. Like you could charge your AirPods from that. Like no problem. Hmm. Uh, I mean,

Chris Gammell: my old Mac, like aluminum body, it used to shock me every time I touched it when it was plugged in.

Andy Kong: Oh, they must've developed this already. Yeah. Yeah. No, Apple, they're so,

Chris Gammell: their hardware is so ahead of everything. They didn't even tell anybody. That's why they charge so much. Yeah.

Andy Kong: I imagine the approvals for this, like going consumer would be a little tricky, but it is actually completely safe. And I think you could totally get it approved. I know. Actually, I talked to this guy who's working at a startup where they're trying to make, uh, this technique work in hospitals. So they want to put it in the hospital bed and have like patients where like something which is powered through the bed. I do think perception,

Chris Gammell: perception might be the harder, the harder piece of all this. Um, you know, like you're doing what now?

Andy Kong: Yeah. Yeah. Yeah.

Chris Gammell: Yeah. Also, um, that's true. If you get into the, you know, just, just below a hundred megahertz, you're going to start interfering with, uh, radio stations.

Andy Kong: Oh yeah. Yeah. We didn't even think about the FCC. Those guys. Yeah. Other concerns. Yeah. They're a bunch.

Chris Gammell: Yeah.

Andy Kong: But when I was, because I worked on this and then I started, uh, chargerless, like everyone was like, Oh, chargerless must be powered like through the skin or something. Yeah. But no, chargerless is just solar. So,

Speaker ?: yeah.

Andy Kong: You know, I learned a lot of low power.

Chris Gammell: If you have a big enough power supply, you could charge just about anything through the skin. You just, you know, eventually the skin starts to char. Yeah. You don't need to do DC. Yeah, exactly. Yeah. Right. Right. Yeah. Yeah. I would touch the bus bar on my, uh, my panel. You know, I, I was charging all kinds of things on my, on my body.

Andy Kong: I've actually never gotten shocked by the wall. Uh, I think it's coming, but yes, you know, in my future, I see it.

Chris Gammell: Yeah. Be careful. One, one hand behind your back. That's, that's the key thing. it's okay to get shocked. Not through the heart.

Andy Kong: Do you know, there's a similar, uh, adage for laser technicians.

Chris Gammell: What's that?

Andy Kong: So like, you know, you keep one hand behind your back, so it can't pass through your heart. Cause if you have two contacts on it, you're, you're done for, um, for laser technicians is if you turn on a system, you should have one eye close.

Chris Gammell: Oh, interesting. Hmm.

Andy Kong: Cause in case it hits something that you set up and it blinds you, you still have one. Yeah. Yeah. It's when you do, it's when you do that second. I take an eye, morbid fun tips. Yeah.

Chris Gammell: That's great. Uh, where can people find out more about charger lists? Where can people sign up for pre-order testing? All that other stuff.

Andy Kong: Yeah. So we're currently selling our, for like our second batch. And then this is the one right before the skill production. So I will actually touch every unit that goes out. Currently, uh, our websites, getcharger list.com, um, like G E T charger list, or you can just look up charger lists. Yeah. Cool. And then I guess also, if you want to read more about stuff, I worked on in the past, like it's all on my personal website. It's a Andy Kong.org.

Chris Gammell: Yep. And you're on the socials. We'll put, we'll put links into all of Andy stuff, uh, GitHub.

Andy Kong: And yeah, I'm, I'm milking my, my Twitter. Yeah. Yeah. My socials for it. That's great. I got monetized on there right as everyone was leaving. I was like, why is everyone leaving? Like what are we even doing? I make money from posting on here. Yeah. Yeah. Yeah. That's too bad.

Chris Gammell: All right. Well, Hey, I really appreciate you being here and, uh, yeah, I'm looking forward to keeping following, following all your projects. It seems like you're doing a lot of interesting things.

Andy Kong: Yeah. Great chatting. Good to meet you. Yeah. Yeah.

Speaker ?: Yeah. Yeah. Yeah. Yeah. Yeah. Yeah.