How Atomic Clocks Actually Work: From Quantum Resonance to GPS

Last Updated: 2026-05-16

Architecture at a glance

A single nanosecond of clock error in a GPS satellite translates into about 30 centimetres of position error on your phone — light travels roughly 30 cm in a nanosecond, and trilateration is a sum of travel times. That number alone explains why atomic clocks exist: no quartz crystal, no pendulum, no mechanical escapement can hold time tightly enough for the modern world. Understanding how atomic clocks work is one of those rare cases where the underlying physics is genuinely beautiful and the engineering payoff is staring at you from the corner of your phone screen every time it shows a location. This post is a friendly but technically honest tour: what a “second” actually means now, how a cloud of cesium atoms becomes the reference, why optical lattice clocks have leapfrogged cesium by a factor of 100, how GPS satellites carry four clocks each (and why relativity matters more than most people realise), and where the field is heading before the international second itself gets redefined.

What this post covers: the SI definition of the second, cesium beam and fountain physics, hydrogen masers and rubidium workhorses, the optical-lattice revolution, GPS architecture and its relativistic corrections, how the internet inherits atomic time, and the roadmap to redefining the second around an optical transition by roughly 2030.

What the Second Actually Is, Officially

For most of human history a second was a chunk of a day — specifically 1/86,400 of a mean solar day. That worked for sundials and pendulum clocks but collapsed once we could measure time to better than a part in 10^8. The Earth’s rotation is irregular: tides slow it, post-glacial rebound speeds parts of it up, and atmospheric angular momentum nudges it on monthly timescales. A “second” defined by Earth’s spin is a second that drifts.

The fix arrived in 1967 at the 13th General Conference on Weights and Measures, when the SI second was redefined as the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the cesium-133 atom, at rest at a temperature of 0 K (later refined to specify the unperturbed atom). That is not an arbitrary number. Louis Essen and Jack Parry at the UK’s National Physical Laboratory had built the first practical cesium beam clock in 1955, and the 9.192631770 GHz figure was simply the best measurement of that transition relative to the existing ephemeris-time second, rounded to the precision their instrument could justify. From 1967 onwards, the second became a count of microwave wiggles instead of a fraction of a day.

This matters for one subtle reason. After redefinition, the cesium transition does not equal 9,192,631,770 Hz because of any physical principle — it equals that number by definition, and every other physical quantity that involves time is now expressed in terms of it. The 2018 redefinition of the SI base units (the 26th CGPM resolution) tightened this further by fixing the values of fundamental constants like the Planck constant and elementary charge, with the second still anchoring the whole structure. The full text is in the BIPM SI Brochure (9th edition), which is the official source.

This is the thesis of this post, and we will keep coming back to it: a clock is just a stable, countable oscillation, and atomic clocks win because the oscillation is set by atomic structure — a property of the universe — rather than by something mechanical that can drift, wear, or feel temperature.

Cesium Beam Clocks: How a Cloud of Atoms Becomes a Reference

A cesium beam clock turns the abstract definition above into a piece of hardware roughly the size of a desktop PC. The architecture — established by Essen and Parry and refined for fifty years — sits inside almost every commercial “atomic clock” you can buy today. See ./assets/arch_01.png for the data path.

The chain has seven steps and a feedback loop.

1. Oven. A small heated reservoir at around 100 C boils off a beam of neutral cesium-133 atoms. Cesium has a single valence electron whose spin can be parallel or antiparallel to the nuclear spin, giving rise to two hyperfine ground-state sublevels (in spectroscopic notation, F = 3 and F = 4). At room temperature roughly half the atoms are in each state.

2. State selector A. A magnetic gradient (in older designs, a Stern-Gerlach magnet; in modern compact units, a hexapole or quadrupole field, and in optically-pumped versions, a laser) sorts the beam so that only atoms in one of the two hyperfine states continue toward the cavity. This is the “preparation” stage: you need a known starting state if you want to measure a transition.

3. First Ramsey microwave pulse. The atoms enter a microwave cavity tuned near 9.192 GHz. A short pulse rotates the atomic state into a superposition — quantum mechanically, the atom is now half in each hyperfine level. This is one of the conceptual moves that makes atomic clocks work: instead of asking “is this atom flipped?” we use coherent superpositions whose phase evolves at exactly the transition frequency.

4. Drift region. The atoms then fly freely for ~50 ms in a magnetically-shielded drift tube. During this drift the superposition’s phase advances at the rate set by the atomic transition itself. The longer the drift, the narrower the resulting spectroscopic line — this is the heart of Norman Ramsey’s separated oscillatory fields method, the technique that won Ramsey the 1989 Nobel Prize in Physics.

5. Second Ramsey microwave pulse. A second pulse, phase-coherent with the first, interferes with the now-evolved superposition. If the microwave frequency exactly matches the atomic transition, the second pulse completes the flip; if it is even slightly off, the interference is imperfect and fewer atoms end up flipped.

6. State selector B and detector. A second magnetic gradient sorts the post-cavity atoms by state, and a hot-wire ionizer or fluorescence detector counts the ones that ended up in the target level.

7. Servo loop. This count is the error signal. If the count is at its maximum, the microwave source is on resonance and the loop holds. If the count drops, the servo electronics adjust the microwave oscillator (typically a low-noise quartz crystal multiplied up to 9.192 GHz) to push back onto the peak. The locked oscillator’s output is then divided down to give a 1 Hz tick — the SI second.

A modern commercial cesium beam clock — the kind used in time labs, telecom networks, and some financial trading houses — achieves a fractional frequency stability of roughly 1 × 10^-13 at a day. The lab-grade primary standards do better. NIST-F2, the cesium fountain clock at NIST in Boulder commissioned in 2014, throws ultracold cesium atoms upward and probes them on the way up and back down, extending the effective drift time to nearly a second — pushing accuracy to around 3 × 10^-16, or about one second of drift in 100 million years (see the NIST primary frequency standards page and successor F2 documentation). The full physics — laser cooling, magneto-optical traps, microwave interrogation in a fountain geometry — is the same Ramsey scheme, just with cold atoms tossed against gravity instead of a horizontal beam.

Hydrogen Masers and Rubidium: The Workhorses

Cesium is the standard, but it is not the only atomic clock. Two other species cover use cases where cesium’s accuracy is overkill or its size and cost are prohibitive. See ./assets/arch_02.png for the relative stability picture.

Active and Passive Hydrogen Masers

A hydrogen maser uses the 1.420 GHz hyperfine transition of atomic hydrogen — the same line radio astronomers use to map neutral hydrogen in galaxies. Atomic hydrogen is prepared in the upper hyperfine state, focused into a quartz storage bulb whose walls are coated with PTFE to prevent collisional dephasing, and stimulated to emit microwaves coherently inside a tuned cavity. The result is a self-sustaining maser oscillation whose frequency is set by the atomic line.

Hydrogen masers come in two flavours. Active masers generate their own output power from the stimulated emission and have the best short-term stability of any practical clock — fractional frequency noise around 1 × 10^-15 at averaging times of 1,000 to 10,000 seconds. Passive masers use an external local oscillator that gets locked to a weaker maser action; they are less stable short-term but more compact, and they appear in some space applications. The European Space Agency’s Galileo satellites carry passive hydrogen masers as their primary onboard clocks.

The trade-off: masers drift on long timescales because the wall coatings degrade and the cavity tuning shifts with temperature. A maser is the best clock in the world over an hour; a cesium fountain is better over a year.

Rubidium Cell Clocks

Rubidium clocks use a different geometry. A glass cell containing rubidium-87 vapour, buffer gas, and a sealed environment is illuminated by a discharge lamp (or in modern units, a diode laser) that optically pumps the atoms into one ground hyperfine level. A microwave field at 6.834 GHz — the rubidium hyperfine frequency — drives the transition, and the amount of light absorbed by the cell tells you when you are on resonance. The whole device fits inside a 10 cm cube.

Rubidium clocks are everywhere — GPS satellites, cell tower base stations, defence radars, scientific instruments — because they hit a sweet spot. Stability is around 5 × 10^-12 at a day, which is plenty for most network synchronisation tasks, and the cost is one to two orders of magnitude less than cesium. The downside is a noticeable long-term ageing drift (the cell chemistry evolves) and a sensitivity to buffer-gas pressure, so rubidium clocks are typically disciplined by GPS or by an external cesium reference.

This is why the GPS satellite clock loadout is two cesium beam clocks plus two rubidium cell clocks per satellite from Block IIR onward, with newer Block IIIF satellites using improved digital rubidiums; the cesium clocks set the long-term accuracy and the rubidiums provide redundancy and short-term smoothness.

The Optical Lattice Revolution: 10^-18 Accuracy

For most of the last century, “atomic clock” meant “microwave atomic clock,” because the cesium and rubidium and hydrogen transitions all live in the GHz range. There is a fundamental reason you might want a higher-frequency reference: clock stability scales with the frequency you are counting. If you can count an oscillation at 429 terahertz instead of 9.2 gigahertz, you have roughly 50,000 more cycles per second to lock to, and your fractional uncertainty for a fixed cycle-counting error shrinks by the same factor. See ./assets/arch_03.png for the basic scheme.

The breakthrough came in two stages. First, the optical frequency comb — a femtosecond laser whose spectrum is a perfectly even ruler of optical frequencies, locked to a microwave reference — gave physicists a way to count optical cycles for the first time. Theodor Hansch and John Hall shared the 2005 Nobel Prize in Physics for this development. Before the comb, an optical clock was unusable because the output frequency was too high to divide down to anything you could put on a wire.

Second, Hidetoshi Katori at the University of Tokyo and RIKEN proposed in 2003 a clever trick now called the magic wavelength optical lattice. Atoms held in a laser standing-wave trap normally have their clock transition perturbed by the trap light — the AC Stark shift. Katori showed that for certain atomic species, there exists a specific trap wavelength at which both clock states are shifted by exactly the same amount, leaving the transition frequency unperturbed. For strontium-87 this magic wavelength is 813 nm; for ytterbium-171 it is 759 nm.

The architecture of an optical lattice clock looks roughly like this:

1. Source and first cooling stage. A hot atomic beam (strontium or ytterbium) is decelerated by a Zeeman slower and captured in a magneto-optical trap on a strong dipole-allowed transition — 461 nm for strontium — giving temperatures around 1 mK.
2. Second cooling stage. A narrow-line cooling transition (689 nm for Sr) on the same atoms takes them down to ~1 microkelvin, slow enough to load into the optical lattice.
3. Lattice loading. The cold atoms drop into a 1D, 2D, or 3D optical lattice at the magic wavelength. Each lattice site holds a small cloud of atoms, well isolated from collisions and from blackbody-radiation shifts (which are still the dominant systematic uncertainty in most modern lattice clocks).
4. Clock interrogation. An ultra-stable clock laser — typically a diode laser locked to a high-finesse Fabry-Perot cavity in a cryogenic, vibration-isolated chamber — probes the 698 nm clock transition of Sr-87 (or 578 nm for Yb-171). The clock laser’s linewidth is tens of milihertz in the best systems, narrower than the natural linewidth of the atomic transition itself.
5. Readout. After the probe, atoms in each clock state are imaged with state-selective fluorescence, and the ratio gives an excitation fraction that locks the clock laser to the atomic line.
6. Comb division. The clock laser is mixed with an optical frequency comb whose modes span an octave; the comb is referenced to itself and produces a microwave output that inherits the clock laser’s stability divided down to RF.

The performance is remarkable. The JILA strontium clock (Jun Ye’s group, Boulder) demonstrated systematic uncertainty below 2 × 10^-18 in 2019, and a 2024 result from the Riken Yb-171 ion clock group (Katori and collaborators) reported total uncertainties pushing 5 × 10^-19. At that level, two clocks differing in elevation by 1 cm would tick at distinguishably different rates because of general relativistic gravitational time dilation — a fact that has already been demonstrated experimentally in the same JILA lab. The clock has become a gravimeter as much as a timekeeper.

For now, optical clocks are secondary representations of the second; the SI definition is still anchored to cesium. But the BIPM Consultative Committee for Time and Frequency (CCTF) has a stated roadmap toward redefining the second around an optical transition by approximately 2030, contingent on agreement among national metrology institutes about which transition (Sr, Yb, Hg, Al+ ion, others) wins.

Inside GPS: 31 Satellites, 4 Clocks Each, and Relativity

The most consequential consumer of atomic time in the world is GPS. The Global Positioning System currently operates 31 active satellites in medium Earth orbit at 20,200 km altitude, distributed across six orbital planes inclined at 55° to the equator, each satellite completing roughly two orbits per sidereal day. See ./assets/arch_04.png for the architecture.

Each satellite carries four atomic clocks — typically two cesium beam clocks plus two rubidium cell clocks in Block IIR-M satellites, with the newest Block IIIF generation using improved digital rubidiums and (on some birds) hydrogen masers for experimental payloads. Only one clock is active at a time; the others are hot spares.

The user receiver in your phone does not measure distance directly. It measures the time difference between when each satellite says it transmitted a signal (encoded in the navigation message) and when the receiver actually saw it. Multiplied by the speed of light, that gives a pseudo-range. With four pseudo-ranges from four satellites, the receiver solves for four unknowns: three spatial coordinates plus the receiver clock’s offset from GPS time. This is why you need at least four satellites in view: the fourth equation eliminates the need for the receiver itself to have an atomic clock.

For this trilateration to work, the satellite clocks must agree to within a few nanoseconds. The control segment — the Master Control Station at Schriever Space Force Base in Colorado, with an alternate at Vandenberg, plus 16 globally distributed monitor stations and 4 ground antennas — continuously measures each satellite’s clock against the U.S. Naval Observatory’s master clock ensemble and uploads correction parameters that the satellite broadcasts in the navigation message. The full protocol is specified in IS-GPS-200, the GPS Interface Specification.

The relativity correction nobody can skip

Here is where physics turns into a 24-hour engineering problem. GPS satellites are moving at about 3.87 km/s in their orbits and sit in a weaker gravitational potential than receivers on the ground. Both effects shift the apparent rate of the onboard clocks relative to ground clocks, in opposite directions:

• Special relativistic time dilation from the satellite’s velocity slows the satellite clock by about 7 microseconds per day relative to ground.
• General relativistic gravitational time dilation speeds the satellite clock up by about 45 microseconds per day relative to ground, because the satellite is higher in Earth’s gravitational well.

The net effect is that GPS satellite clocks tick about 38 microseconds per day faster than ground clocks. If you did not correct for this, GPS positions would drift by about 11 km per day — within hours the system would be unusable. The Block II satellites were therefore engineered with a deliberate frequency offset built into the onboard clocks before launch: 10.22999999543 MHz instead of the nominal 10.23 MHz, so that after relativistic effects are accounted for, the satellite emits 10.23 MHz as seen from the ground. There is also a small residual eccentricity correction that the receiver applies in real time. This is one of the cleanest, most practically important demonstrations of general relativity in any consumer technology.

The 1971 Hafele-Keating experiment had already verified these effects to within ~10% by flying cesium clocks around the world on commercial jets in opposite directions — the original Science paper is one of the most-cited validations of relativistic time dilation outside particle physics.

For applications that demand sub-meter accuracy — surveying, aviation precision approaches, autonomous driving — there are augmentation systems. Real Time Kinematic GPS (RTK) and Precise Point Positioning (PPP) use a reference station’s known position to push civilian accuracy down to centimetres, but they still depend on the satellite atomic clocks for the underlying timing.

Why the Internet Needs Atomic Time

Outside of position fixing, atomic clocks underwrite the synchronisation that makes the internet, finance, and modern industry function. The hierarchy is layered.

NTP (Network Time Protocol). Most consumer devices sync via NTP, which was designed by David Mills in the 1980s and is now in version 4 (RFC 5905). NTP defines strata: stratum 0 is the reference clock itself — a cesium standard, hydrogen maser, or GPS receiver disciplined to UTC. Stratum 1 servers are computers directly connected to stratum 0 sources. Stratum 2 servers sync to stratum 1, and so on. Consumer machines typically operate at stratum 3 to 5, achieving synchronisation accuracy of a few milliseconds over public internet and a few hundred microseconds on a well-engineered local network.

PTP (Precision Time Protocol, IEEE 1588). When milliseconds are not enough, financial trading firms, telecom networks (5G base stations need sub-microsecond synchronisation between cells for inter-cell coordination), and broadcast TV operators run PTP. PTP exchanges timestamps at the hardware layer of network interface cards, achieving sub-microsecond and often sub-100-nanosecond accuracy on a LAN. The protocol is specified in IEEE 1588-2019. The PTP grandmaster clock is, again, typically a GPS-disciplined cesium or rubidium reference.

White Rabbit. Developed at CERN to synchronise accelerator timing, White Rabbit extends PTP with synchronous Ethernet and phase tracking to achieve sub-nanosecond accuracy. It is now used in scientific facilities, financial exchanges (some equity trading venues use White Rabbit to timestamp orders), and parts of the European power grid. The same atomic-clock backbone shows up here, just disciplined more carefully.

Time-Sensitive Networking (TSN, IEEE 802.1). Industrial control systems are migrating off proprietary fieldbuses onto deterministic Ethernet via TSN. Our detailed TSN IEEE 802.1Q time-sensitive networking architecture guide for 2026 walks through how 802.1AS (generalised PTP) provides the timing layer that makes scheduled traffic possible.

The financial reach of this is enormous. Regulation MiFID II in Europe requires equity trades to be timestamped to within 100 microseconds of UTC and traceable back to the official UTC realisation. That requirement alone has driven a multi-billion-dollar market for traceable time services and has put cesium clocks in the basements of every major investment bank.

Where Atomic Clocks Are Going: Optical Clocks, Redefining the Second

The most consequential development in the next decade is the redefinition of the SI second. The BIPM Consultative Committee for Time and Frequency has been preparing for this since at least 2015, and the working assumption is that around 2030 — possibly at the 28th or 29th CGPM — the second will be redefined in terms of an optical transition rather than cesium. See ./assets/arch_05.png for the timeline.

For this to happen, the metrology community needs to agree on five things:

1. Which optical transition wins. Candidates include Sr-87 (698 nm), Yb-171 (578 nm or the even narrower octupole transition at 467 nm in Yb+), Hg (266 nm), and Al+ ion (267 nm), plus the nuclear clock candidate Th-229 which has been the subject of recent breakthrough excitation experiments (the 2024 PTB/JILA result demonstrating laser excitation of the Th-229 nuclear isomer is a genuine landmark).
2. A demonstrated network of clocks that agree. Multiple labs running clocks of multiple species must show internal consistency at the 10^-18 level. The European Optical Clock Network (using fibre links) and the cross-continental measurements between PTB Germany, INRIM Italy, NPL UK, NIST USA, NICT Japan, and Riken have been pushing this.
3. Practical realisations. A redefinition is only useful if national labs can actually realise the new second. Compact, robust, transportable optical clocks — including the Sr lattice clocks from PTB and the Yb ion clock prototypes that have been transported across cities — are the prerequisite.
4. Traceability to existing systems. GPS, NTP, PTP, telecom timing — all the infrastructure built around microwave atomic time has to inherit the new definition smoothly. The plan is for the new second to be operationally identical to the current one within measurement uncertainty.
5. Implications for other SI units. Because the metre is defined via the speed of light and the second, redefining the second propagates into the metre. The fundamental-constants framework adopted in the 2018 SI revision is designed to handle this.

Beyond redefinition, near-term applications are pulling optical clocks out of the lab. Clock-based geodesy measures the gravitational potential difference between two distant clocks and converts that into a height difference at the centimetre level — better than satellite-based gravimetry over short baselines. Tests of fundamental physics include searches for variations in the fine-structure constant, dark-matter detection via clock-frequency modulation, and improved tests of local Lorentz invariance. And chip-scale atomic clocks (CSACs) — coin-sized rubidium clocks with sub-watt power consumption — are now in defence systems, undersea cables, and high-end IoT timing modules.

If you want to chase how this connects to a broader landscape of frontier science topics, our sibling pieces on how tokamak fusion reactors work and plasma confinement physics and the architecture of AlphaFold 3 and diffusion-based protein structure prediction cover other corners of the same modern-physics-meets-engineering space. The cryptographic implications of better timing also tie into our fact-check of viral claims that quantum computers have broken RSA-2048 and the broader quantum threat to elliptic curve cryptography in 2026 write-ups.

FAQ

Why are atomic clocks so accurate?
Atomic clocks lock an electronic oscillator to a quantum transition between two energy levels of an atom — for cesium-133, the hyperfine ground-state splitting at 9,192,631,770 Hz. That transition frequency is a property of the atom set by nuclear and electronic spins, identical in every cesium atom in the universe and immune to mechanical wear or temperature drift. As long as the atoms are well isolated from stray fields and collisions, the clock counts a reference that is fundamentally more stable than any mechanical or quartz oscillator can be.

How does GPS use atomic clocks?
Each GPS satellite carries multiple atomic clocks — typically two cesium beam clocks and two rubidium cell clocks. The clocks let satellites broadcast precisely timestamped signals; your receiver compares arrival times from at least four satellites and solves four equations for three spatial coordinates and one receiver-clock-offset unknown. The Master Control Station continuously corrects each satellite’s clock against an ensemble standard. Without atomic clocks, GPS would not be a navigation system at all.

What is the most accurate atomic clock?
As of 2026, the most accurate operating clocks are optical lattice clocks based on strontium-87 and ytterbium-171 isotopes. The JILA strontium clock (Jun Ye’s group, Boulder) and Riken’s ytterbium clock (Katori group, Japan) both demonstrate total systematic uncertainties around 10^-18 — meaning a discrepancy of about one second over the age of the universe. These remain laboratory instruments; the official SI second is still realised against cesium fountains.

Can you have an atomic clock at home?
You can buy a “radio-controlled” wall clock that picks up a longwave time signal — WWVB in the US, DCF77 in Europe, JJY in Japan — broadcast from a transmitter disciplined by a cesium standard, but those clocks themselves use ordinary quartz. True compact atomic clocks (rubidium oscillators or chip-scale atomic clocks) are commercially available for a few hundred to a few thousand dollars and show up in amateur radio, time servers, and high-end audio gear, but they are still niche outside professional use.

What happens if GPS clocks drift?
A drift of just one microsecond translates to about 300 metres of position error. The GPS control segment uploads new clock corrections to each satellite roughly once a day, and the navigation message contains correction parameters that receivers apply in real time. If a satellite’s onboard clock degrades significantly, the satellite is marked unhealthy and excluded from positioning until the clock is replaced or the satellite is decommissioned. Hot-spare clocks on each satellite let operators switch in a backup without losing service.

Further Reading

• TSN IEEE 802.1Q time-sensitive networking architecture guide for 2026 — how PTP-derived time is delivered to industrial networks.
• How tokamak fusion reactors work and plasma confinement physics — sibling science-curiosity post on another modern-physics engineering frontier.
• Quantum threat to elliptic curve cryptography in 2026 — what better quantum control means for cryptographic timing assumptions.
• Fact-check: did quantum computers break RSA-2048 in 2026? — the same lab techniques (laser cooling, ion traps) underlie both clocks and quantum processors.
• AlphaFold 3 architecture and diffusion-based protein structure prediction — companion deep-physics-meets-engineering write-up.

References

• BIPM SI Brochure (9th edition, 2019, with 2024 corrections) — official SI definition of the second. https://www.bipm.org/en/publications/si-brochure
• NIST Time and Frequency Division — primary frequency standards, NIST-F1 and NIST-F2 cesium fountains. https://www.nist.gov/pml/time-and-frequency-division
• Essen, L., and Parry, J. V. L. “An atomic standard of frequency and time interval: A caesium resonator.” Nature 176, 280-282 (1955).
• Ramsey, N. F. “A molecular beam resonance method with separated oscillating fields.” Physical Review 78, 695 (1950); Nobel Lecture (1989).
• Hafele, J. C., and Keating, R. E. “Around-the-world atomic clocks: predicted and observed relativistic time gains” and “Around-the-world atomic clocks: observed relativistic time gains.” Science 177, 166-170 (1972).
• Bothwell, T. et al. (JILA / Jun Ye group). “Resolving the gravitational redshift across a millimetre-scale atomic sample.” Nature 602, 420-424 (2022).
• Ushijima, I., Takamoto, M., Das, M., Ohkubo, T., Katori, H. (Riken). “Cryogenic optical lattice clocks.” Nature Photonics 9, 185-189 (2015); subsequent Yb-171 results 2023-2024.
• IS-GPS-200 — Interface Specification for the GPS user interface. U.S. Space Force, current revision. https://www.gps.gov/technical/icwg/
• 26th CGPM (2018) Resolution 1 — revision of the SI base units.
• BIPM CCTF Roadmap for the redefinition of the second (Consultative Committee for Time and Frequency working documents, 2020-2025).
• IEEE 1588-2019 Precision Time Protocol standard.
• RFC 5905 — Network Time Protocol v4.

---

Written by Riju — engineer and writer covering the intersection of physics, computing, and industrial systems.