Atomic Clocks | Hacker News Reader

Atomic Clocks

nist.gov

141 pointsteleforce86 comments

> Today’s best atomic clocks are billions of times more accurate and precise than any other type of clock. If they had been running since the Big Bang, 13.8 billion years ago, they would have lost or gained less than a second.

Absolutely incredible, but then confused by the next assertion that is made without context:

> New kinds of clocks may soon open up capabilities that previous generations could only dream about.

What is possible by pushing past that incredibly narrow margin of 1s in 13b years? What dreamy capabilities do we get that we don’t enjoy today?

Edit: I’d assumed they were talking about new approaches to making these clocks more accurate, but I wonder if they’re referring to power and form factors? I.E. an atomic clock the size of a grain of rice: https://www.nist.gov/noac/success-story-chip-scale-atomic-cl...

There are still big leaps, like their optical lattice clock that's sensitive to the gravitational difference caused by tidal effects on the earth's crust.

It can detect gravitational differences of a few cm!

What are these atomic clocks measured against?

I think that they typically use lots of other atomic clocks [0].

[0]: https://en.wikipedia.org/wiki/International_Atomic_Time

Generally, other atomic clocks. The fact they they tick consistently compared to each other is the main thing (though at these scales, you need to account for time dilation!).

Clocks at NIST Boulder in Colorado run faster than NIST Gaithersburg in Maryland by about 1.55 microseconds per year because of the 1500 meter altitude difference. It's about 1 nanosecond per year per meter. They both have to use a fudge factor to adjust down to sea level.

Pretty sure you can rig something far better than radar if you have the ability to detect extremely tiny gravitational waves via an array of hyper-accurate clocks.

> detect extremely tiny gravitational waves via an array of hyper-accurate clocks.

I wonder how tiny is feasible. Would it get to a level of accuracy where you could detect neutrinos? We know they have some (v tiny) mass, so it should in theory be detectable that way?

Moving objects don't emit gravitational waves though. They have to be orbiting.

1. Isn't everything orbiting something?

2. If a black hole passed us by, no gravitional wave?

Is a single peak really a wave?

First, that claim is patently false. Two objects with size accelerating directly towards each other (i.e. zero orbital angular momentum hence zero "orbit") will produce waves. A moving object will produce detectable changes in gravity so arguing about "waves" in the context of atomic clocks detecting gravity is irrelevant.

I don't GR but I'm fairly certain dipole does not work for gravitational waves (i.e. two objects accelerating straight at each other). Supposedly, there's a specific definition of "gravitational wave" here. You can, of course, detect the change in gravity though, regardless of "gravitational wave" or not.

In radio astronomy, more precise clocks can directly improve the quality/resolution of observations.

> Data received at each antenna in the array include arrival times from a local atomic clock, such as a hydrogen maser. At a later time, the data are correlated with data from other antennas that recorded the same radio signal, to produce the resulting image. The resolution achievable using interferometry is proportional to the observing frequency.

https://en.wikipedia.org/wiki/Very-long-baseline_interferome...

> What is possible by pushing past that incredibly narrow margin of 1s in 13b years? What dreamy capabilities do we get that we don’t enjoy today?

You can measure speed (via special relativity) and altitude (via gravity/general relativity) with atomic clocks [0], so with a more accurate clock, you should be able to calculate your speed/altitude more accurately.

GPS is already pretty good at measuring speed/altitude, but its not always very accurate, and it only works if you can receive the satellite's signal. In theory, a more accurate atomic clock would solve both of these problems (as long as you had an equally-accurate clock at a fixed location to compare to).

[0]: https://news.ycombinator.com/item?id=16473776

> What dreamy capabilities do we get that we don’t enjoy today?

The next gen will have an alarm with snooze button.

Cost, size, weight, and power are definitely things that can be improved on. The atomic clocks you can buy for a lab are big 3U boxes but shrinking it down to something portable is the future.

I actually work with one of those chip scale atomic clocks. The goal is to put it in space. If you need precise timing in space, the current state is to lock to a GPS reference clock (just for timing, don't care about position), train an OCXO, and let it drift for a day or two before you require GPS again. OCXOs (and crystal oscillators in general) can be very accurate but the stability is easily affected by temperature swings. So as your satellite goes in and out of the sunlight or as components inside turn on and off, the surrounding temperature changes and can make the clock drift. This isn't a problem on earth since you can always have a GPS receiver running but on a power constrained satellite, this is not ideal. The CSAC gets 100x better thermal stability than even some of the best space-qualified OCXOs. This means you could potentially run for 100x longer before you need to reacquire GPS, assuming the same environment and performance requirements. The CSAC also has a brain inside so potentially you can do all of your characterization on the ground and not even need a GPS receiver onboard. It also requires much less time to become stable after powering on than a normal crystal oscillator.

You get the same benefits on the ground, super precise timing for the processors and modems without necessarily needing a GPS receiver. The modern assumption (and pretty much proven in Ukraine) is that GPS will be completely unavailable on a battlefield. Anything portable or battery-powered will likely not be able to make use of newer anti-jam features of GPS (very power hungry) so you need something small, precise, and quick to come online.

Modern waveforms often require very tight timing to properly demodulate so having a pocket sized atomic clock solves a lot of problems.

Where are you getting GPS receivers that work in space?

Probably one of the numerous companies making qualified receivers? E.g. [0]. There's even a published table comparing basically all of them [1].

[0] https://www.aac-clyde.space/what-we-do/space-products-compon...

[1] https://www.mdpi.com/1424-8220/23/17/7648#app1-sensors-23-07...

So that poster asked where to get blue carrots, and you replied 'probably one of the numerous companies making blue carrots?'. It seems they were unaware of these products, you don't have to be fake incredulous about it.

A little bit of hostility is healthy to an incredibly basic question like that. Plus I’ve encountered that user a few times, it’s always the same tier of asinine comment.

I'm always astounded by this particular activity. If you're curious about the existence of a thing, the activity required is typing some such "where can I get". But you could type that in a search bar and be gauranteed an immediate result or type it in a comment section and possibly get a definitely delayed result.

I suppose the interaction could be the point, but if the curiosity is the only factor, why choose the comment section?

It’s almost always a secondary motivation on forums, someone with an agenda trying to undermine the GP in some attempted subtle way that’s not immediately apparent to most casual browsers.

General Dynamics, for example, makes them.

https://gdmissionsystems.com/communications/spaceborne-commu...

I’m not well versed in this field but if you take GPS using time as one of its parameters I could imagine being able to pinpoint the location of an atom from space to be one avenue of an endgame in making time so precise we would lose less than a Planck second in time over that period. That being said I don’t even know if time is the bottleneck for accurate GPS coordinates.

Not many people know NIST offers a cookbook for perfectly calibrated food

https://www.nist.gov/pml/owm/metric-si/metric-kitchen/metric...

Also known as an ordinary recipe everywhere else.

Well, recipes do often use measurements like teaspoons and cups, but they have straightforward metric equivalents - a teaspoon being 5mL and a cup 250mL.

You should really measure mass rather than volume.

That depends on the state of what you are measuring. For some ingredients mass and volume will always be equivalent.

I recently bought my first not-super-cheap watch, and found out that all around the world there radio stations which broadcast the atomic clock time that a lot of clocks/watches listen to.

After setting the time zone (I guess this also changes the frequency that it listens on) it starts to sync itself using that signal. It is also solar-charging - so this basically means that I never have to touch it. Although it's an analog watch, because of this sync, it correctly handles DST, the date window (including leap year) - basically I never have to touch it again until it breaks . All without bluetooth or internet connection - pretty cool I think.

Casio Oceanus t200 for anyone that cares.

That “all around the world” leaves significant gaps: https://hagensieker.com/2019/11/17/multiband-6-what-is-it/

I had a weather station with an external antenna that could never sync, even during the night, even placed outside in the "best" side of the house on my neck of the EU.

A solar powered, low battery wristwatch that's never outside in the middle of the night will fare much worse.

It's interesting old school tech that works fine in the places where it does work.

For elsewhere, I guess synching through a phone is more practical. So far it works fine. I just hope it doesn't get deprecated (I do, unfortunately, have a feeling the Casio app will break before the radios stop transmitting).

Syncing directly from a 4G signal would be a lot more practical than VLF or HF time signals, but I haven't found a wall clock that does so.

On the one hand, yes. On the other, my country has just shut down 3G and 2G only stays because it's still used by emergency services. How long will 4G last? Longer than bluetooth and the Casio app?

It'd be nice if something like this could be maintained long term. The US, UK and Germany signals have been working since the ~1960s.

Could be, I don't have much experience with it, but so far it has worked every time without me doing anything special - I can check the status of the last sync attempt, which it does nightly on its own (or I can trigger it). The UX isn't great - all manipulation of the seconds hand, but other than that it has been flawless. This is in Germany - your mileage may clearly vary.

As I wrote: it does work great where it does work. I'd definitely have gone that route if it worked where I live. It's just that the marketing that it works "all around the world" falls pretty short, IMO.

I kinda wish the German transmitter had more range (like the US one does) but even the blog I linked to, you can see someone in the continental US who needs to hang their watch from their window blinds for syncing to work reliably.

I even looked into hacking my own transmitter (https://github.com/hzeller/txtempus) before giving up and just buying one of the models with bluetooth phone sync.

In the US, those are radio stations WWVB for clocks and watches, and WWV/WWVH for shortwave.

One thing I think is interesting is that not only are the digital formats these carry synced to an atomic clock, but the radio carrier wave is too. So if you need a very precise 10 MHz signal in the lab, you can hook up a shortwave antenna and a filter, and receive the precisely-on-frequency signal broadcast from WWV.

I'm pretty sure it's a lot more reasonable to use a fixed set of frequences and use the timezone for offset

Long before most of you were born there was a special number you could call to get the time here in the USA. It would give you a localized time every 10 seconds. It was considered an authoritative value, but I knew that normal analog clocks drifted a fair amount, and I was fascinated by the Time voice.

There was also a special number to dial that was called Information. When I was about 12, I suddenly thought that’s amazing, you can just call and get information. What I didn’t know at the time was that it only applied to phone numbers.

One day, I called Information and asked them how accurate the time number was and what mechanism was used to deliver it. Incredibly, they connected my preadolescent self to a public relations person at AT&T. That person somehow took me seriously and hooked me up with another party. And that person told me about the cesium atomic clock used to keep track of the time.

Later, I would call Information to find out the acreage occupied by the Los Angeles Zoo, and it was then that they brusquely explained to me that the number was not for general knowledge.

Here in Australia we had a number that gave you the time called 'George the talking clock'.

https://1194online.com

This has been reimplemented as a tribute by a modern phone phreak named Shaun Ewing, who also created Lenny.

https://se.au/notes/george-the-talking-clock-lives-on/

Number Region

02 8403 1194 Sydney/Melbourne/Canberra

08 7078 7794 Adelaide

07 3667 9594 Brisbane

08 7918 0808 Darwin

08 6244 3294 Perth

Thanks for sharing this with me!

In Northern California the number corresponded to POP-CORN, so if you wanted to know the time you called popcorn, and got a lovely lady's voice: "At the tone, the time will be...". The POP prefix was a shortcut to the phone company.

https://en.wikipedia.org/wiki/Point_of_presence

https://en.wikipedia.org/wiki/Telephone_exchange_names#Stand...

  76x-nnnn: POplar, POrter, ROckwell, ROger(s), SOuth(field)

411 Directory Assistance in 1978 https://youtu.be/UpsGpZ7wi9k?si=gSKP2K7WGPn-cbBj&t=60

That's brilliant. Sometimes talking to real humans is great.

It reminds me of two things. When I was a child I was given a postage stamp and wrote a letter to "Wallace and Gromit Ltd" and posted it. Remarkably, I got a reply from Aardman Animations with a card signed by Nick Park himself. Someone at Royal Mail obviously understood and looked up the real address of the studio.

Later as a teen I noticed the clock on our oven displayed "24:00" at midnight instead of the more conventional "0:00"[0]. I wrote to the manufacturer (Neff) on a support channel and it actually got forwarded through several departments before I received a reply from an engineer quoting ISO8601 and how it doesn't actually specify that midnight has to be 0:00 (although, sadly, I never received a positive reason for why someone made the oven clock like this).

For those of us who remember such interactions it's even more frustrating when you find there's a dumb robot at the end of the line. Even if it's LLM powered. I actually asked ChatGPT what it would do with the Wallace and Gromit letter and it couldn't find the same solution. We lost a lot with Google-style faceless companies.

As for the time, in the UK we had (have?) "The Talking Clock" which you could get to by dialling 123. This was well enough known in 2001 to be a key part of this scene from The Office: https://www.youtube.com/watch?v=SdBCFV6qa3A&t=146

[0] I just looked this up again and, funnily enough, John Ward showed a similar oven on his YouTube channel: https://www.youtube.com/watch?v=4_oLT7lWQ44

When I was about 10 I wrote a letter to Microsoft and they sent me back a package of Visual Studio (real licensed edition that didn't display a dialog box every time you ran a compiled program!) and, on two CDs and in HTML Help format, MSDN Library.

I like that both 0:00 and 24:00 exist. If you say "midnight, August 12", do you mean at the start or the end of the day? There might be a convention, but I don't know it so others probably don't either. With 0:00 and 24:00 and a date, you can disambiguate a time precisely.

The standard convention is that midnight is the start of the day. The instant 23:59:59 is followed by 00:00:00 of the next day.

The only instant that could exist would be 24:00:00. If you told me it was 24:00:05, I'd say poppycock - that's tomorrow, 00:00:05 instead.

My GE microwave has a whole damn button the same size as the start button for AM/PM. Has to be one of the dumbest design wastes I have seen. I have always wished the 24 hr clock had taken off in the US. I use it on any clock that supports it.

In Canada, Santa's postcode is H0H 0H0. In the 1800s a letter addressed Sromfidevi England was eventually delivered to Sir Humphry Davy.

For the kids out there, calling a number for recorded information was common in the 90s I guess? Movie times was a big one. Where I was locally, multiple surf shops did 3-4x daily surf reports. The later was a bit funny because if one of the usual update times was missed you knew the waves were good and the guy was surfing :)

777-FILM "Hello, and welcome to Moviefone!"

Also, where I live, radio transmitted the precise time in the most low tech way possible. A person said: "It is XX:XX time with the beep" and this was followed by four beeps, three short, one long, with the end of the last beep signifying the exact moment of the announced time.

We had a similar beep scheme, except it would play before the next programme stated, typically before the morning/evening news.

We didn’t have that in the Los Angeles area that I know of

Listen to "What Time Is It?" by Ken Nordine. It's a treat.

We had a similar time service in Uruguay and, whenever we had an important appointment (e.g., a medical one) we’d call the number to make sure our clocks and watches were on time.

It was a female voice that said something that could be translated to “the signal will mark it is <time>,” and, the funny thing is, between the way it had been recorded and the fact that signal (señal) and lady (señora) can sound very similar in Spanish when over the phone and followed by a vowel (“indicará” for will mark in this case), a lot of us thought it said “the lady will mark” and, as a little kid, I thought there was an old time-keeping lady on whom we all depended to be in sync. Miss NTP I guess!

https://popcornph.one

You mean popcorn?

The one we used is still active JU-55961 (area code 502), only difference is it plays an ad first. It had a little jingle used to memorize it, so I still remember it today. I searched the internet for that number, and apparently there were hundreds of these JU numbers for different services.

You actually can still call NIST and hear the signal broadcast from WWV[0], though that's probably different from the telephone time of day you're thinking of, which might have come from one of these machines [1] in a local telephone exchange.

[0] https://www.nist.gov/pml/time-and-frequency-division/time-di...

[1] https://en.wikipedia.org/wiki/Audichron

There are still a ton of time numbers kicking around. My local phone company still offers (513) 721-1700 and I have (513) 241-1010 burned into my memory from dialing it a million times as a kid playing with the phone.

Plus you can hear two signals from NIST themselves:

(303) 499-7111 - WWV (John Doyle)

(808) 335-4363 - WWVH (Jane Barbe) - in Hawaii, might cost extra to call depending on plan

And there are these two offered by US Naval Observatory in Washington

(202) 762-1069

(202) 762-1401

Plenty more out there if you do some Googling.

I built a phone accessed service for a school system some time ago, programming an off-the-shelf call handler and adding an industrial thermometer. When you called it would tell you the time and temperature and whether school would open normally, open late or be closed due to weather. School staff could call in and change the message setting.

What, Exactly, Is a Clock? To understand atomic time, we must first understand something about clocks.

A clock is, at its core, a marriage of two components: a mechanism that oscillates, or ticks with a steady beat, and a device that counts those beats and displays the time

Yeah but more deeply time is a property of the universe that the clock including atomic clocks are tracking. Kant said it is 'schema' i.e. without understanding time we wouldn't even understand causality so it is the purest possible a-priori concept that doesn't rely on sensory input to learn about

This is why I like using my Alphadec system that splits the year into 26 chunks from A-Z <https://github.com/firasd/alphadec> as my working dir; eg. right now the current Alphadec is '2026_P5M7_200286' so my working dir is Documents/work/2026_P. That way there is a folder system automatically imposed on various .js or .py stuff without having to pre-create a taxonomy since 'time is schema'--I can estimate "that ffmpeg project was in Spring so it must be around 2026_F.."

Not sure I get what is better about this than simply using the month, like Documents/work/202606 (which also sorts nicely and does not need a conversion in my head to know what time of the year that must be)

Yup as far as the folder taxonomy thing ISO YYYYMM works fine too

Alphadec is just a timezone free UTC encoding that works down the millisecond so I also use it for snapshots eg

  myscript.py.backup.2026_P5G6_328054_Aug06.txt

So then when I'm doing multiple revisions I don't have to worry about which revision of 20260806's day it was etc. Unix time in ms would work too but Alphadec is a little bit nicer to me than saying myscript.py.backup.1786086462604.txt

> Yeah but more deeply time is a property of the universe that the clock including atomic clocks are tracking.

I'll quote Einstein who said "time is what the clock says".

To fully understand time even as simply as a sequence of events is excruciatingly difficult, and we are still working out the details of our understanding. I think Einstein was on the right track with his quote, and trying to get beyond that would have simply been an obstacle in his path of knowledge.

NIST is a great example of a true Public Good.

Also, has anyone else ever used their Randomness Beacon? I once used it to settle an argument in a book club over which book to read next. It's basically a very fancy and high tech coin flip.

https://csrc.nist.gov/projects/interoperable-randomness-beac...

As long as we forget that little thing called Dual_EC_DRBG, right?

Admittedly, that was probably the NSA getting their grubby hands into NIST. A good thing being almost permanently tarnished by a bad thing.

"RT is good except when the Kremlin gets its hands on it"

Fun fact, two (very expensive) wristwatches have actual atomic timekeeping in them, instead of just receiving a radio time signal: the Bathys Hawaii Cesium 133, and a few different shaped from Hoptroff London.

All released 2013-2015. Not sure what happened then for there to be suddenly two.

Symmetricom (now owned by Microchip) released the first Chip Scale Atomic Clocks in 2011. It probably took some time for them to become available to watch designers and for them to be integrated into watch designs.

https://www.nist.gov/noac/success-story-chip-scale-atomic-cl...

I'm guessing the watches sold poorly and the chip packages are still large compared to a watch movement, so no new designs.

Sadly, no major miniaturization breakthroughs since. And those CSACs still eat up a lot of power (as do good OCXOs), making them rough for battery powered wearables.

MEMS! You should get in touch with the folks at SiTime.

Surely a good opportunity to tout your recent wristware Jeff? :D

CSACs are small and efficient enough to put in a 1U box and that is all that is required by the markets that the manufacturers are chasing.

Yeah, likewise the Breitling Emergency is big because it's an actual working 406 MHz Personal Locator Beacon. In principle that's all you need to have a rescue team come get you if you, I dunno, crashed your plane on a desert island or you broke your leg while hiking in Siberia, or whatever. Of course if you're the sort of person who wears a wristwatch which costs as much as a car you probably aren't doing anything like that, but you could be and maybe you're an elite solo who "travels light" and so you couldn't pack a conventional much cheaper PLB.

I can't imagine Breitling sells very many of these, they're huge heavy watches, if you're The Rock they probably look fine on your arm but the average tough guy can't justify owning a watch this expensive and if you're the 5'4" female CEO of a famous company you can afford the Emergency but it looks ridiculous and feels like you're carring a boulder on your wrist.

Like Concorde, sometimes we prove something was possible but it turns out it was unavoidably too expensive and annoying for it to make much sense in the market.

[Edit: Fix spelling of Concorde]

you would be surprised how large and passionate the watch community is. In the world of watches that breitling is big, but not really that big. It is quite a popular watch, and not that expensive : https://www.chrono24.com/breitling/emergency--mod21.htm?dose...

Notice that for the actual purpose of rescue only what they're calling "Gen2" actually works. The earlier devices which make up most of the linked examples are pretty, and would doubtless make interesting jewellery for a man at a society dinner or in the boardroom, but they do almost nothing if you're alone somewhere and need rescuing today.

Once upon a time, decades ago, a "beacon" is just a primitive radio transmitter which when activated produces a clearly man-made signal (it sounds like a siren basically if you're listening to it). This works pretty good if people nearby know to look for it. Then COSPAS SARSAT is a "peaceful use" satellite technology, put some satellites up above the Earth, they relay the chosen emergency frequencies and now a human operator can "hear" that beacon from hundreds of miles away. So that's all the original Emergency does. Rescue this way was sometimes a long-shot, but you did not have better options.

But obviously now we can do a lot better, so eventually COSPAS SARSAT deprecated and then eventually eliminated that relay technology in favour of 406 MHz digital beacons. So now it can tell which beacon has been triggered, and you can add stuff like contact details to a database for the rescuers. This way when they get a beacon they can see the signal is Steve Smith, solo mountaineer, here's a contact number for Steve's mum, so before we commit a mountain rescue team to 18 hours of work we can verify whether Steve is currently up a mountain [maybe in urgent need of rescue] or whether he lost that beacon last month and is at home now watching Netflix.

Your point is well taken - I just wanted to be absolutely clear for anybody who might get one of these that the original Emergency is now basically just pretty and only a modern "Gen2" Emergency "works" for its stated purpose.

It's still relative. I think...

If you could have a atomic wristwatch and you travel around the world with it, you will have a very precise time for you, but you'll end up out of sync with almost every other atomic clock.

If you sync two atomic clocks and put one at the surface of the arctic ocean and another one on top of Mount Nevado in Peru, and leave them there for one million years, the clock in Peru will be 1.88 years ahead of the clock in the arctic.

That is a difference of 0.44 nanoseconds per day.

Don't quote me on that though.. I'm not a physicist.

You are off by a factor of the speed of light.

The clocks would be off by less than a quarter second over a million years.

You mean everyone else would have the "wrong" time whereas your clock is still bang-on correct for you. So you have become a time traveler.

If you want to see how an actual application of this works, I've yet to see a description better than in CuriousMarc's HP 5061A series:

https://www.youtube.com/watch?v=eOti3kKWX-c

I wish we had books for kids at school that explains in this fashion. Even a beginner can grasp the idea behind it and dive deeper if interested. Loved it.

If I had an atomic clock, it'd probably be blinking 12:00.

Time is not a constant

Reply on news.ycombinator.com