# How Chip-Scale Millimeter-Wave Radars Are Sensing the World

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 5 Aug 2026, 11 min.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0146.mp3

Lucas and Luna explore the quiet revolution happening inside millimeter-wave radar chips. From automotive safety systems that see through fog and rain to smart-home devices that detect presence without cameras, these tiny sensors are becoming the eyes of the Internet of Things. The episode unpacks how frequency-modulated continuous wave radar works, how on-chip antennas shrink the entire system to a few millimeters, and why the 60-gigahertz band is such a sweet spot for consumer gear. Lucas digs into a specific example: a single-chip radar used in a popular smart thermostat to sense occupancy, explaining the signal processing tricks that turn raw reflections into useful data. Luna questions the trade-offs between radar and other sensing modalities, and they discuss the privacy advantages of radar over cameras. The conversation also touches on the cost curves that are making radar chips as cheap as accelerometers, and what that might mean for everything from elderly care to drone navigation. If you've ever wondered how your car avoids a pedestrian in the dark or why your lights know you're in the room, this episode gives you the concrete story. #MillimeterWave #RadarChip #FMCW #60GHz #AutomotiveRadar #SmartHome #PresenceDetection #OnChipAntenna #IoT #SensorTechnology #PrivacyTech #Semiconductor #ChipDesign #TechInnovation #Engineering #Electronics #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: So I was reading a teardown of a popular smart thermostat the other day, and there it was — a tiny radar chip, not much bigger than a grain of rice, sitting right next to the temperature sensor. And I thought, this is the quiet takeover nobody's talking about.
Luna: Radar in a thermostat? I thought that was just for military jets or autonomous cars.
Lucas: Right, but that's exactly the story. Millimeter-wave radar has been shrinking for a decade, and now it's cheap enough and small enough to end up in consumer gear. The thermostat uses it to detect whether anyone's in the room — so it can adjust heating or cooling based on presence, not just a schedule.
Luna: How does a tiny chip actually 'see' someone? It's not like it has a lens.
Lucas: It's all about radio waves. The chip transmits a signal in the 60-gigahertz band — that's millimeter-wave, because the wavelength is around five millimeters. It bounces off a person, and the chip measures the time and frequency shift of the echo. Do that millions of times a second, and you can tell not just that someone's there, but whether they're breathing, or even their heart rate.
Luna: Okay, that's impressive. But why 60 gigahertz specifically? Why not use lower frequencies like Wi-Fi?
Lucas: Good question. 60 gigahertz is a sweet spot because the wavelength is short enough that the antenna can be tiny — integrated right onto the chip. And the band has a few gigahertz of unlicensed spectrum, so you can use wide bandwidths. Wider bandwidth means finer range resolution, so you can distinguish a person from a wall, or even two people standing close together.
Luna: But doesn't higher frequency mean more attenuation? Like, it can't see through walls?
Lucas: Exactly. 60 gigahertz is great for short-range, a few meters, and it doesn't penetrate walls — which is actually a feature for privacy. You only sense what's in the room, not the neighbor's apartment. And there's a huge advantage over cameras: it's completely anonymous. You're just a moving blob of reflected radio waves, not a face.
Luna: That privacy angle is a big deal for smart home stuff. I know some people are creeped out by always-on cameras.
Lucas: Right. And that's why companies are putting radar in everything from smart speakers to bathroom mirrors. But the technology itself is fascinating. The core is a frequency-modulated continuous wave radar — FMCW. The chip sweeps the frequency up and down in a linear ramp, and by comparing the transmitted and received signals, it can calculate both distance and velocity simultaneously.
Luna: So it's like a bat, but with math.
Lucas: Pretty much. And the signal processing has gotten so good that a single chip can track multiple objects, estimate their angle, and even classify micro-motions like a hand gesture. That's how you get radar-based gesture control in some headphones — you wave your hand, and it pauses the music.
Luna: Hold on, headphones? I thought radar was for cars. Let's talk about automotive too, because that's where the big money is.
Lucas: Yeah, automotive is the volume driver. Every new car now has at least a few radar sensors — for adaptive cruise control, blind-spot monitoring, automatic emergency braking. And they're all millimeter-wave, typically in the 76 to 81 gigahertz band. That's a different part of the spectrum, but the same principle.
Luna: And radar works in rain and fog and darkness, which cameras don't do well.
Lucas: Exactly. That's why radar is non-negotiable for autonomous driving. Tesla famously tried to do it with cameras only, but most of the industry uses a sensor fusion approach — radar for distance and velocity, cameras for color and detail, and LiDAR for high-resolution 3D mapping. Radar's advantage is that it's cheap and solid.
Luna: How cheap are we talking? I imagine the cost has dropped a lot.
Lucas: A decade ago, a radar chipset might cost a few hundred dollars. Now you can get a complete single-chip radar with integrated antennas for under ten dollars in volume. Some are even cheaper. That's what's unlocked the smart home market.
Luna: And that cost drop is because of integration, right? Everything on one die.
Lucas: Yes. The big breakthrough was putting the antennas on the chip itself. Antennas are normally off-chip, which adds cost and size. But at 60 gigahertz, the antenna is just a few millimeters, so you can etch it into the metal layers of the chip package. That's what makes the whole thing fit on a single piece of silicon.
Luna: Speaking of chips, this feels like a good moment to mention something that keeps our little show running. We're independent, we don't run ads, and that's only possible because some listeners choose to support us over at buy me a coffee dot com slash fexingo. If today's tech deep-dive gave you something useful, that's a way to keep it going.
Lucas: Yeah, and honestly, every little bit helps us keep the lights on and the microphones warm. No pressure, but if you're in a position to, we'd appreciate it. Now, back to radar — there's a specific design challenge that's really interesting.
Luna: What's that?
Lucas: Power consumption. A radar chip is transmitting and sampling constantly, and that can drain a battery quickly. In a thermostat, you're plugged into mains, so it's fine. But for battery-powered sensors — like a motion detector that's supposed to last a year on a coin cell — you need to be smart about duty cycling.
Luna: So they turn it on for a fraction of a second, take a measurement, then go back to sleep?
Lucas: Exactly. The chip has a low-power mode where it wakes up every few hundred milliseconds, sends a quick pulse, and if it detects motion, it boots up the full signal chain. That way, the average current draw can be under a few hundred microamps. That's how you get a radar that runs for years on a small battery.
Luna: And that's the same trick they use in occupancy sensors for smart lighting.
Lucas: Right. They're everywhere — in offices, in warehouses, even in some public restrooms. But the next frontier is using radar to monitor vital signs. The chip can detect the tiny chest movements from breathing, and even the heartbeat. There are prototypes for in-home elderly care that can detect falls and alert a caregiver.
Luna: Fall detection is huge. My grandmother lives alone, and a camera in her bedroom feels invasive. Radar could be a privacy-respecting alternative.
Lucas: Absolutely. And that's the social benefit — you get safety without surveillance. But there are technical challenges. The reflected signal from a person is tiny compared to the background clutter. You have to filter out static reflections from walls and furniture, and that takes some clever digital signal processing.
Luna: And what about interference? If every device in the house is transmitting on 60 gigahertz, aren't they going to jam each other?
Lucas: That's a real concern. The unlicensed band is shared with WiGig and other devices. But radar chips use spread-spectrum techniques and randomize their sweeps, so the probability of two devices colliding is low. And if they do interfere, the processing can often reject the spurious signals. It's not perfect, but it works well enough for most applications.
Luna: I also wonder about the range. How far can these tiny chips 'see'?
Lucas: For a standard consumer chip, maybe ten to twenty meters at best. But for automotive, with larger antennas and more power, you can get up to two hundred meters. The difference is mainly in antenna gain and output power. Regulators limit how much power you can transmit, especially in unlicensed bands.
Luna: So the physics sets the boundaries, but the engineering is what makes it useful.
Lucas: Exactly. And there's a really elegant example of that: some smart speakers now use radar to detect not just presence, but the direction of the user. That lets the speaker beam the audio toward you, so you can hear it better in a noisy room. It's a simple idea, but it takes a lot of signal processing to get the angle right.
Luna: That's like a party trick. But I'm curious — how does the chip actually estimate angle? Is it using multiple antennas?
Lucas: Yes, that's the key. A single antenna gives you distance and velocity, but not angle. To get angle, you need an array of antennas. The chip has multiple receive antennas, and by comparing the phase differences between them, you can compute the bearing of the target. That's called angle of arrival estimation.
Luna: And that's all happening on-chip, right? No external processing?
Lucas: Right. Modern radar chips have a built-in digital signal processor that does all the FFTs and beamforming. You just stream out a detection list over SPI or I2C. It's remarkably simple to integrate.
Luna: It really feels like the sensor itself has become a commodity, and the value is in the algorithms.
Lucas: That's the trend across the industry. The hardware is getting so good that differentiation comes from software. Companies are building machine learning models on top of the raw radar data to recognize gestures, classify activities, even detect emotions from breathing patterns. It's a whole new sensing modality.
Luna: Emotions? That's getting into sci-fi territory.
Lucas: It's early research, but the idea is that your breathing and heart rate variability correlate with stress. Radar can measure those non-invasively. So a device could sense when you're stressed and play calming music. That's a long way from a thermostat.
Luna: I guess we've come a long way from just turning on the heat.
Lucas: And the next step is even more integration. We're seeing radar chips that combine with other sensors — like MEMS accelerometers or microphones — on the same package. That allows sensor fusion at the hardware level, which improves accuracy and reduces cost further.
Luna: So eventually, every smart device will have a tiny radar doing something invisible.
Lucas: I think that's the direction. Radar is the only sensor that gives you presence, motion, and vital signs while respecting privacy. That combination is just too valuable to ignore. It'll be everywhere, but you won't notice it.
Luna: As long as it doesn't end up reading my mind, I'm okay with that.
Lucas: Give it a decade.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.