# How Chips Are Using On-Chip MEMS Mirrors for LiDAR

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 7 Jul 2026, 10 min.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0097.mp3

Episode 97 of The Hardware Podcast dives into a surprising trend in chip design: putting tiny MEMS mirrors directly onto silicon to create solid-state LiDAR. Lucas opens with the 2025 market projection of $2.5 billion for automotive LiDAR and explains how on-chip scanning mirrors eliminate bulky rotating assemblies. Luna asks about the manufacturing challenge of aligning mirrors to micron precision, and Lucas details how companies like MicroVision and STMicroelectronics are using wafer-level processing to solve it. They discuss applications beyond cars—drones, robotics, smartphone depth sensing—and the potential for LiDAR to hit sub-$50 per unit by 2028. The conversation stays grounded in engineering trade-offs: mirror size vs. field of view, resonance vs. durability. No hype, just how these millimeter-scale mirrors are bending lasers with the same physics as a macro mirror, but etched in silicon. #MEMS #LiDAR #OnChipMirrors #SolidStateLiDAR #MicroVision #STMicroelectronics #AutomotiveSensors #WaferLevelProcessing #SiliconPhotonics #LaserScanning #DepthSensing #AutonomousVehicles #Drones #Robotics #ChipDesign #Technology #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: Welcome back — Lucas here with Luna, and we are in the lab again for episode 97 of The Hardware Podcast. Today I want to talk about a component that is no bigger than a grain of sand but is quietly reshaping how machines see the world: the on-chip MEMS mirror.
Luna: MEMS as in micro-electromechanical systems, right — the same technology that's in your phone's accelerometer and gyroscope.
Lucas: Exactly. Same basic idea — tiny mechanical structures etched onto silicon. But instead of sensing motion, these mirrors reflect laser light. They are the core of what the industry calls solid-state LiDAR, and they are starting to appear in everything from autonomous cars to delivery drones.
Luna: So no more spinning laser turrets on top of cars? That's the image most people have when they hear 'LiDAR'.
Lucas: Right. The spinning mechanical LiDAR — like the ones Velodyne made famous — they work, but they are expensive, fragile, and bulky. A Velodyne HDL-64E from a decade ago cost about $75,000. Today you can buy a solid-state LiDAR module with a MEMS mirror for under $500. The trade-off is range and field of view, but the cost curve is dramatic.
Luna: And the size — I've seen these modules about the size of a deck of cards. How do you actually get a mirror on a chip?
Lucas: It's a multi-layer silicon process. You start with a silicon-on-insulator wafer, then use deep reactive-ion etching to carve out a mirror plate — typically a few millimeters across — suspended by tiny torsion bars. Underneath the mirror, you deposit electrodes. Apply a voltage, and the mirror tilts electrostatically. You can get a few tens of degrees of optical scan angle, and you drive it at resonance — somewhere between 1 and 30 kilohertz — to sweep a laser beam back and forth.
Luna: Resonance is key then — you're not holding the mirror at a static angle, you're letting it oscillate naturally.
Lucas: Exactly. That's how you get the speed. A single mirror scanning on one axis can generate a 2D line scan. Add a second mirror — or a second axis on the same mirror — and you get a full raster pattern. The laser fires pulses at precise intervals, the mirror sweeps the beam across the scene, and a photodetector measures the time of flight for each pulse. That gives you a point cloud.
Luna: So the mirror is doing what the spinning motor used to do, but with no moving parts except the mirror itself.
Lucas: Right — and that's the reliability win. A MEMS mirror has no bearings, no gears, no motor. It's just a sliver of silicon flexing a few microns. Companies like MicroVision have been shipping MEMS mirror modules for pico projectors for years. Now they are adapting those designs for LiDAR. STMicroelectronics has a MEMS mirror in their FlightSense line, aimed at time of flight ranging. And there are startups like Hamamatsu Photonics and Ouster — well, Ouster used to do spinning LiDAR, but they acquired a MEMS mirror company a couple years back.
Luna: I want to get into the manufacturing challenge, because aligning a laser to a mirror that's a couple of millimeters across, on a chip, with micron precision — that sounds like a packaging nightmare.
Lucas: It is, and that's where the real engineering lies. The mirror itself is fabricated on a cmos compatible process, so you can integrate drive electronics on the same die. But the laser diode and the photodetector — those are separate chips. You have to align them optically, then hermetically seal the whole module. Some companies are doing wafer-level packaging — bonding a glass lid over the MEMS wafer, then dicing it into individual units. That brings the cost down because you align once at the wafer scale.
Luna: So it's like what you see with camera modules in phones — the assembly is automated, but the tolerances are really tight.
Lucas: Exactly. And the same trends are driving cost down. A MEMS mirror LiDAR module that cost $1,000 in 2020 might be $200 today, and some analysts project it could hit $50 by 2028. That opens up markets beyond automotive — think warehouse robots, agricultural drones, even smartphone depth sensing. Apple already uses a VCSEL laser for Face ID, but that's a flood illuminator, not a scanning mirror. Add a MEMS mirror and you could get a full 3D map from your phone.
Luna: Before we go deeper into applications — and I know you want to talk about drones — I want to ask about the physics. A mirror this small — does it have the same optical quality as a macro mirror?
Lucas: Surprisingly, yes — if you do it right. The mirror surface is typically coated with aluminum or gold, and it can achieve a surface roughness of a few nanometers. That's comparable to a conventional mirror. The limiting factor is aperture size. A MEMS mirror is maybe 3 millimeters in diameter. That limits the diffraction-limited spot size, which in turn limits the angular resolution of the LiDAR. For long-range detection — say 200 meters — you want a larger aperture to keep the beam collimated. That's why some systems use multiple MEMS mirrors in an array, or combine a MEMS mirror with a larger fixed mirror.
Luna: So there's a trade-off between range and field of view. The bigger the mirror, the better the range, but the harder it is to scan fast.
Lucas: Right. And that's why you see different architectures. Some use a single mirror that scans both axes — that gives you a small field of view, maybe 30 degrees. Others use a mirror that scans in one axis and a second mirror or a galvanometer for the other axis, giving a wider field but more complexity. And there's a third approach where you use a MEMS mirror to steer a beam from a fiber laser, like a phased array, but that's more experimental.
Luna: So where does this technology stand in 2026? I know automotive LiDAR has been promised for years, but we still don't see it in consumer cars.
Lucas: We're close. The latest generation of mems based LiDAR from companies like Valeo and Hesai is now in production for Level 3 autonomous systems. The Mercedes Drive Pilot system uses a Valeo LiDAR that is a hybrid — it has a rotating mirror but also some MEMS elements. But true solid-state — no moving parts at all — is still a few years away for automotive because of the reliability requirements. MEMS mirrors have been tested to billions of cycles, but auto grade means surviving vibration, temperature extremes, and dirt. That's a hard qualification.
Luna: But for drones, the requirements are less stringent, and we're already seeing products. DJI has a drone — the Matrice 350 — that uses a mems based LiDAR for obstacle avoidance.
Lucas: Exactly. And that's where the volume is today. Drones, robots, even mapping for construction sites. A MEMS LiDAR module is light enough to fly on a small drone, and cheap enough that if it breaks, you replace it.
Luna: That brings up a broader point — the economics. If the module cost drops to $50, you could put it in home appliances. Imagine a robot vacuum with LiDAR instead of bump sensors.
Lucas: It's already happening. iRobot — or rather, the brands that succeeded iRobot — they use LiDAR for simultaneous localization and mapping. The Neato Robotics vacuums used a spinning laser, but newer ones use solid-state. The MEMS mirror gives you a 360-degree scan in a flat package. It's a natural fit.
Luna: And before we move on — I want to bring up an important point. We talk about all these free episodes — and they are free — but they take research, scripting, and production. If you've gotten something out of this conversation, a couple of dollars a month is genuinely what keeps these going. You can find us at buy me a coffee dot com slash fexingo. It makes a real difference — literally helps us cover the cost of the microphones and the hosting. And we appreciate every single person who chips in.
Lucas: Yeah, it's true. And we don't have ads, so listener support is what keeps this ad-free. So if you can, that link is buy me a coffee dot com slash fexingo. Now back to the tech — I want to talk about one more application that I think is going to be huge: gesture recognition.
Luna: Oh, like the Leap Motion or the Xbox Kinect?
Lucas: Exactly. But with MEMS LiDAR, you get much higher resolution and you can work in bright sunlight, which infrared camera-based systems struggle with. Google's Project Soli used radar, but LiDAR gives you sub-millimeter accuracy. Imagine controlling your car's infotainment with a wave of your hand, or your smart home lights. The MEMS mirror can scan a small field — like your hand — at very high frame rates, so it can track finger movements.
Luna: And that's all happening on a chip that fits under a fingernail. That's the part that still amazes me.
Lucas: It is remarkable. And the key takeaway from today is that MEMS mirrors are not a lab curiosity — they are in production, they are getting cheaper, and they are enabling a whole generation of sensing that will make machines more aware of their environment. Whether it's a car, a drone, or a light switch, the ability to see in 3D is becoming a commodity.
Luna: And it's happening because of a tiny piece of silicon that tilts with a voltage. No spinning parts, no lasers that need aligning — just a chip.
Lucas: Right. And that's the hardware revolution we track every week. Next episode we're going to look at something completely different — on-chip gas sensors that can detect disease from your breath. So stay tuned, and thanks for listening.

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.