# How Chips Are Using On-Chip Atomic Clocks for Precision Timing

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 12 Jul 2026, 9:42.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0106.mp3

Lucas and Luna explore the cutting edge of chip-scale atomic clocks, focusing on how CSACs are shrinking atomic timing into a chip form factor. They discuss the breakthrough at NIST in 2024 that achieved a fractional frequency instability of 10^-15 at one hour of averaging, and how companies like Microsemi (now part of Microchip Technology) are commercializing these for GPS-denied environments. The episode drills into the physics of coherent population trapping and the practical trade-offs between power consumption and long-term stability. Listeners learn why a chip that loses one second every 300 million years is now a real product. #ChipScaleAtomicClock #CSAC #AtomicClock #PrecisionTiming #NIST #Microsemi #MicrochipTechnology #CoherentPopulationTrapping #GPSDenied #QuantumSensors #MEMS #Timekeeping #Technology #Hardware #ElectronicsEngineering #Semiconductors #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: So you know how your phone gets the precise time from GPS satellites, each of which carries an atomic clock? Well, what if you could shrink that entire atomic clock — the physics package, the laser, the control electronics — onto a chip small enough to fit inside a drone or a network switch?
Luna: That sounds like science fiction, but I know it's real. These are called chip-scale atomic clocks, or CSACs. We've had them for about a decade now, but the latest versions are genuinely impressive.
Lucas: Exactly. The key breakthrough came out of NIST around 2024. They demonstrated a microfabricated atomic clock that achieved a fractional frequency instability of ten to the minus fifteen at one hour of averaging. To put that in human terms — that clock would lose less than one second every 300 million years.
Luna: And that's on a chip that consumes just tens of milliwatts of power, right? Not the room-sized caesium fountains at national labs.
Lucas: Right. The traditional caesium fountain clock at NIST is about the size of a small elevator shaft. The chip-scale version is literally a few cubic centimeters. And here's the thing — if today's conversation gave you something usable, something that makes you think differently about timing technology, we want to keep these episodes free and accessible to everyone who's curious. We deliberately don't run ads on these. If you want to support that choice, the link is buy me a coffee dot com slash fexingo.
Luna: Yeah, it's a small way to keep this kind of deep-dive engineering conversation available without any sponsors interfering. So, back to the physics — how do they actually trap atoms on a chip?
Lucas: Great question. The technique is called coherent population trapping, or CPT. Instead of using a hot beam of caesium atoms like in a traditional atomic clock, they use a tiny vapor cell — about the size of a grain of rice — filled with caesium or rubidium atoms.
Luna: And they shine two laser frequencies at those atoms simultaneously, right? The frequencies are tuned so that the atoms get trapped in a 'dark state' — they stop absorbing light.
Lucas: Exactly. The width of that dark resonance is what gives you the frequency reference. And because the atoms are in a vapor cell, you don't need a high-vacuum chamber or a complex cooling setup. It's all microfabricated.
Luna: So the main components are a vertical-cavity surface-emitting laser, a caesium vapor cell, a photodetector, and some control electronics. All integrated into a package that's maybe 1.6 by 1.6 by 0.4 centimeters.
Lucas: Right. That's the commercial CSAC from Microsemi — now part of Microchip Technology. It draws about 120 milliwatts in steady state, which is remarkably low for an atomic clock. And it's been used in everything from undersea seismic sensors to military radios.
Luna: But here's the trade-off — the short-term stability isn't as good as a quartz oscillator. For the first few seconds, a good quartz crystal actually outperforms a CSAC. It's only after about a hundred seconds that the atomic clock's long-term stability starts to win.
Lucas: That's exactly right. So most real systems use a hybrid approach: a quartz oscillator disciplined by the CSAC. The quartz handles the short-term jitter, and the atomic clock corrects the drift over minutes and hours. It's the best of both worlds.
Luna: And that's critical for applications where GPS is unavailable. If you're in a deep mine, underwater, in a tunnel, or in a conflict zone where GPS is jammed, you need a local time reference that stays accurate for days or weeks.
Lucas: The military has been a huge driver. The US Army's Dismounted Assured Positioning, Navigation, and Timing system, or DAPNT, uses CSACs to keep soldiers' navigation accurate even when GPS is denied. But there are also commercial uses.
Luna: Like in financial trading. When you're doing high-frequency trading, every nanosecond matters. If your timestamp is off by a microsecond, you could lose millions. So trading firms are starting to put CSACs in their network switches to synchronize timestamps with extreme precision.
Lucas: And data centers too. The IEEE 1588 Precision Time Protocol requires accurate clocks to synchronize network traffic. A CSAC can maintain sub-microsecond accuracy even if the network connection to the master clock is lost.
Luna: So what are the latest developments? I know NIST has been working on a next-generation chip-scale clock using rubidium instead of caesium, and they've improved the long-term stability even further.
Lucas: Yes, in 2025 they reported a chip-scale clock with a stability of two times ten to the minus fifteen at one day of averaging. That's approaching the performance of much larger passive hydrogen masers. And they did it by using a pulsed optical pumping technique instead of continuous CPT.
Luna: Pulsed pumping — so you turn the laser on and off, and measure the atomic precession during the dark time. That reduces some systematic errors from the light itself.
Lucas: Exactly. It's a clever trade-off: you use more power during the pulse, but you get better long-term stability. And the total power is still under 200 milliwatts, which is fine for many applications.
Luna: And there's also work on cold-atom chip-scale clocks. Instead of using a vapor cell at room temperature, you trap and laser-cool the atoms to microkelvin temperatures. The coherence time is much longer, so the stability could be orders of magnitude better.
Lucas: That's the holy grail, but it's still in the lab. The challenge is that cooling atoms requires more complex optics and vacuum. But there have been promising demonstrations using grating magneto-optical traps — those are MOTs that can be fabricated on a chip.
Luna: Like the work at MIT Lincoln Laboratory where they built a cold-atom clock on a chip that fits in a shoebox? That was about two years ago, I think.
Lucas: Right. Their Cold Atom Micro Primary Standard, or CAMPS, achieved a stability of one times ten to the minus thirteen at one second, and it's projected to reach ten to the minus sixteen at a day. That's competitive with some primary frequency standards.
Luna: So we're moving toward a world where every base station, every network node, every drone has its own atomic clock. That's a big shift from relying on a few centralized GPS satellites.
Lucas: And it raises interesting security implications. If you can't jam a local atomic clock, then GPS jamming becomes less effective. The military is very interested in that. But also, think about the power grid — synchronized phasor measurement units use precise timing to detect grid instability. A CSAC could keep those measurements accurate even if GPS goes down.
Luna: There is a challenge, though: cost. A CSAC module still costs around $1,500 in volume. That's too expensive for consumer electronics. But the trend lines are good — the first CSACs in 2011 cost over $10,000.
Lucas: And as the volume increases, driven by 5G and IoT timing requirements, the price will come down. Some analysts project that by 2030, chip-scale atomic clocks could be under $100. At that point, they start appearing in things like smart grids and autonomous vehicle fleets.
Luna: Also, there's a push to integrate the entire CSAC on a single chip using silicon photonics. The laser, the vapor cell, the detector — all on the same die. That would dramatically reduce size and cost.
Lucas: That's what DARPA's ACES program is funding. They want to get the entire atomic clock — physics package and all — onto a chip that's less than 1 cubic centimeter and consumes under 50 milliwatts. If they succeed, you could literally put an atomic clock in a smartphone.
Luna: Imagine your phone's time being accurate to a nanosecond even when you're underground. That would change how we do location-based services, synchronization, even blockchain timestamps.
Lucas: Exactly. And it's not just about timekeeping. The same physics used for atomic clocks can also be used for magnetometers and accelerometers. You could build a complete inertial navigation system on a chip that never drifts.
Luna: Like a quantum compass. That's the long-term vision — chip-scale quantum sensors that measure time, magnetic fields, and acceleration with atomic precision.
Lucas: But we're still a few years away from that. For now, the CSAC is the most mature chip-scale quantum device in production. It's quietly reshaping how we think about timing in critical infrastructure.
Luna: So next time you see a cell tower, there might be a tiny atomic clock inside, ticking away with the stability of an entire galaxy. That's a pretty amazing piece of engineering.
Lucas: It really is. And it's a reminder that the best technology isn't always the most visible — it's the one that makes everything else work more reliably.

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.