Some years back there was talk of an atomic clock that came in a very small package (matchbox / cigarette pack size?). Iirc, saw a price indication of $1500 somewhere.
Something like that still around, and/or available? Any updated designs?
Personally I have no need for ultra-accurate timekeeping. But hey... an atomic clock is way cooler than a Nixie clock or oven-controlled Xtal oscillator. And no... huge 2nd hand atomic clock found on eBay etc doesn't cut it. Too big /heavy / power-hungry.
philipkglass · 6h ago
You're thinking of the Chip Scale Atomic Clock, first demonstrated by NIST in 2003 and commercialized in 2011:
I've heard they're struggling to get good consistent perfomance on them compared to other alternatives in similar sizes
alnwlsn · 2h ago
How big is big? You can still get used FE-5680A rubidiums on Ebay for a few hundred. They are about the size of a small book and need ~20 watts.
Not sure how good/useful/not broken they are since I've never had one, but over the last 10 years I've seen them in a good number of hobbyist projects.
Or you could just go for GPS. That's technically still atomic clocks, but in space!
ooterness · 6h ago
You're probably thinking of chip-scale atomic clocks (CSAC). There's at least two companies that make them [1][2].
Any idea what one of those would cost? (and where to buy them?)
nullc · 4h ago
SA.45s (and the eval board) has been on mouser in the past, I'm not sure if they carry it now because mouser continually blocks me these days. It was previously around $1500.
If you're interested in a small atomic clock and don't absolutely require the very low power consumption of the SA.45s and the very small package size you can get better performance, reliability, and cost in something a big larger and quite a bit higher power consumption.
The CSAC improved atomic oscillators a lot more in power than size... but its timekeeping performance is so/so as far as atomic clocks go.
In particular, the surplus market has a lot of telecom rubidium that can be had quite inexpensively.
I'm personally a fan of the PRS-10 (which also exists in a benchtop form, the SRS725). They seem to regularly sell on ebay for about $300 with a little breakout board for power. And unlike many small rubidiums they have very good phase noise. You can sync them to GPS time using a 1pps input (though I believe on the bare modules the 1pps sync is optional so if you want to use it be sure to get one that has it).
schoen · 2h ago
> SA.45s (and the eval board) has been on mouser in the past, I'm not sure if they carry it now because mouser continually blocks me these days. It was previously around $1500.
I searched on Mouser and found that part no longer has a listed price; there are fancier chip-scale atomic clocks from the same company, the only one with a listed price going for $3,528.23.
Others from a different manufacturer are just over $2,000. A kind of awesome thing is that one of those chips is marked
Frequency 10.000000 MHz
It's awesome that that's not just an estimate or some kind of exaggeration!
RetroTechie · 3h ago
>The CSAC improved atomic oscillators a lot more in power than size... but its timekeeping performance is so/so as far as atomic clocks go.
Timekeeping of atomic clocks is still waaaayy better than next-best technologies (like temp-compensated Xtal oscillators). And size/power/$ constraints matter. So CSACs for the win imho.
fsh · 2h ago
A $5 GPS receiver runs circles around a CSAC, so the range of useful applications is quite limited.
nullc · 2h ago
The timing output from a $5 gps receiver isn't particularly impressive. You have to go to pretty long time intervals for a non-timing GPS to win out. Depending on your application those intervals may be thoroughly irrelevant, e.g. using the device as a frequency source rather than a clock.
There is also, of course, the issue of infrastructure dependence. Particularly since wireless telephony has moved almost exclusively to GNSS time we're going to have a really bad time if kessler syndrome takes out the GNSS satellites.
A quick look didn't turn up any cheap non-timing receivers, but my experience is they're pretty bad (I mean relative to atomic standards, of course).
There are better timing receivers than the ones charted above, of course, but they are not $5. Their cost is now in the same general ballpark as surplus atomic clocks.
Of course, if you have both you can sync one to the other with whatever time constant maximizes the composite performance and have the best of both.
(and primary atomic clocks don't have this drift issue, but sadly the days of the occasional sub $1000 5071 showing up on auction sites seem to be over. :P )
infogulch · 6h ago
Cesium-based atomic clocks ("fountain clocks") like these use the natural resonance frequency of electron orbital energy states under a microwave laser which can be counted to measure time. Since there is a natural background noise of microwaves and many frequencies can interact with orbitals it's important to isolate the atoms from outside sources of electromagnetic radiation and heat in order to maintain accuracy.
Earlier this year there was a big leap in so-called "nuclear clocks" which uses the resonant frequency of energy states of a nucleus itself as opposed to electron orbitals around it. Besides the "more frequency = more better" factor that has always driven clock accuracy -- thorium-229 nuclei excites in ultraviolet wavelengths -- nuclear clocks are better isolated than electron orbital-based clocks because the frequency band where they interact is impossibly narrow. In fact, the reason why it was only recently demonstrated is due to the difficulty of producing the required frequency at a high enough precision to interact reliably. This could lead to more accurate and more compact and cheaper clocks.
How do you sync atomic clocks? When the error rate is 1/1e16, your error in propagating the time from one clock to another is going to be off by many orders of magnitude vs the timekeeper itself.
The modern approach is common view time transfer. Both clocks can observe signal sources such as satellites and compute their relative offsets. The signal doesn't need to be particularly accurate (though it doesn't hurt), though you do need a good model of its position and propagation. GNSS satellites are an obvious choice though there are also specialized services for CVTT.
Note that this is distinct from syncing from GPS, which is a thing people obviously do too, but CVTT can achieve much higher accuracy.
Because you're synchronizing extremely stable clocks the difference between them will primary be an offset (plus/minus a slope from relativistic effects of different altitude). Because of this you can average a large number of readings, so the only major source of error will be systematic effects in propagation/orbit/etc.
Historically, sync was obtained via traveling clocks-- e.g. you sync one atomic clock up and load it, running, in a station wagon... which is the same thing that is most often done for voltage standards today (as atomic voltage standards remain rare, compared to atomic clocks-- I think the least I've paid for one is $15 excluding the ones that were free).
But vibration isn't great for anything with a crystal oscillator in it, and the most modern atomic fountain clocks don't work if they're accelerating in any direction except the designed 'up' direction (gravity), because the little cloud of cooled atoms will fall out of the measurement channel, which makes sync by station wagon not viable.
Of course, once you talk about syncing there is always a question of what you're syncing to. UTC doesn't exist until after the fact. Laboratories measure their offsets via CVTT and UTC is calculated after the fact as past offsets to each of the contributing clocks.
algorithmsRcool · 8h ago
I recall reading that our ability to measure time accurately exceeds that of any other quantity. According to the NIST, the newest Optical Lattice clocks would drift by less than 1 second if they were started 13 billion years ago at the big bang. What else can we measure down past 1 part per 10e18?
analog31 · 8h ago
Curiously, there's also a contender for the worst, which I think at present is the gravitational constant.
teraflop · 7h ago
Yup. And an interesting detail is that we know the product G·M_e (where G is the gravitational constant, and M_e is the mass of the earth) to much higher precision than we know either of its factors. And the same goes for the sun and most of the other planets.
This is because the motion of celestial bodies and spacecraft is dominated by gravitational forces which depend only on G·M, and that motion can be measured extremely accurately with e.g. Doppler radar.
AlotOfReading · 3h ago
That's around the same sensitivity that LIGO operates at.
ipdashc · 5h ago
> our ability to measure time accurately exceeds that of any other quantity
What are 'limits' on how accurate enough clocks 'should' be?
Presumably there's diminishing returns, but as the article says we're at one part in 2.2e-16, are there practical application of going further?
0_____0 · 8h ago
With radio astronomy, where you're measuring phase of incoming radiation, I think "more is more" applies. Would be interesting to hear from someone who actually has experience in that domain though (not me!)
move-on-by · 9h ago
How many atomic clocks are in operation in Colorado now? It would be nice if they could be spread around a bit. I suppose there are logistical issues that keep them centralized?
fanf2 · 5h ago
As well as NIST there is Schriever spave force base https://en.m.wikipedia.org/wiki/Schriever_Space_Force_Base which is the ground operations centre for GPS. They have the USNO alternate master clock, which maintains a copy of the USNO time scale based on caesium beam and rubidium fountain clocks.
algorithmsRcool · 8h ago
The Naval Observatory in Washington DC has quite a few also
CamperBob2 · 6h ago
Commercial atomic clocks of various types aren't that rare. Every cell site has a rubidium standard and/or GPS timing, many data centers probably have a cesium standard, and radio astronomers use H-masers for interferometry.
Everybody with a GPS-disciplined oscillator has access to time and frequency from the Naval Observatory at the sub-100 ns level, optionally augmented to +/- 1 ns with reasonably affordable gear like https://www.sparkfun.com/sparkpnt-gnss-disciplined-oscillato... .
A fountain clock is on a whole different level than any of these. The same researchers who build fountains also work on even better optical lattice clocks, none of which you can buy from Sparkfun. These are research tools that don't have a market, at least not yet.
The SI second definition will likely move from Cs-133 at 9 GHz to Sr-87 at 400 THz before too long ( https://www.nist.gov/si-redefinition/second-future ), but that probably won't shake up the existing market too much.
Something like that still around, and/or available? Any updated designs?
Personally I have no need for ultra-accurate timekeeping. But hey... an atomic clock is way cooler than a Nixie clock or oven-controlled Xtal oscillator. And no... huge 2nd hand atomic clock found on eBay etc doesn't cut it. Too big /heavy / power-hungry.
https://www.nist.gov/noac/technology/time-and-frequency/chip...
https://spectrum.ieee.org/chipscale-atomic-clock
Microchip launched their latest version earlier this year:
https://www.electronicspecifier.com/products/frequency-contr...
Spectratime's version (mRO-50) went into a $2M piece of eye-candy that mechanically syncs a watch.
Not sure how good/useful/not broken they are since I've never had one, but over the last 10 years I've seen them in a good number of hobbyist projects.
Or you could just go for GPS. That's technically still atomic clocks, but in space!
[1] https://www.microchip.com/en-us/products/clock-and-timing/co...
[2] https://www.teledyne-si.com/en-us/Products-and-Services_/Pag...
If you're interested in a small atomic clock and don't absolutely require the very low power consumption of the SA.45s and the very small package size you can get better performance, reliability, and cost in something a big larger and quite a bit higher power consumption.
The CSAC improved atomic oscillators a lot more in power than size... but its timekeeping performance is so/so as far as atomic clocks go.
In particular, the surplus market has a lot of telecom rubidium that can be had quite inexpensively.
I'm personally a fan of the PRS-10 (which also exists in a benchtop form, the SRS725). They seem to regularly sell on ebay for about $300 with a little breakout board for power. And unlike many small rubidiums they have very good phase noise. You can sync them to GPS time using a 1pps input (though I believe on the bare modules the 1pps sync is optional so if you want to use it be sure to get one that has it).
I searched on Mouser and found that part no longer has a listed price; there are fancier chip-scale atomic clocks from the same company, the only one with a listed price going for $3,528.23.
Others from a different manufacturer are just over $2,000. A kind of awesome thing is that one of those chips is marked
Frequency 10.000000 MHz
It's awesome that that's not just an estimate or some kind of exaggeration!
Timekeeping of atomic clocks is still waaaayy better than next-best technologies (like temp-compensated Xtal oscillators). And size/power/$ constraints matter. So CSACs for the win imho.
There is also, of course, the issue of infrastructure dependence. Particularly since wireless telephony has moved almost exclusively to GNSS time we're going to have a really bad time if kessler syndrome takes out the GNSS satellites.
Edit: Here is an example adev chart for a inexpensive atomic clock vs what appears to be a pretty good timing GPS receiver: https://www.thinksrs.com/images/instr/prs10/PRS10diag2LG.gif
So in that case the GPS accuracy only beats out the free running atomic clock at intervals greater than 200,000 seconds or so.
Here is a collection of older timing receivers: http://www.leapsecond.com/pages/3gps/gps-adev-mdev.gif
A quick look didn't turn up any cheap non-timing receivers, but my experience is they're pretty bad (I mean relative to atomic standards, of course).
There are better timing receivers than the ones charted above, of course, but they are not $5. Their cost is now in the same general ballpark as surplus atomic clocks.
Of course, if you have both you can sync one to the other with whatever time constant maximizes the composite performance and have the best of both.
(and primary atomic clocks don't have this drift issue, but sadly the days of the occasional sub $1000 5071 showing up on auction sites seem to be over. :P )
Earlier this year there was a big leap in so-called "nuclear clocks" which uses the resonant frequency of energy states of a nucleus itself as opposed to electron orbitals around it. Besides the "more frequency = more better" factor that has always driven clock accuracy -- thorium-229 nuclei excites in ultraviolet wavelengths -- nuclear clocks are better isolated than electron orbital-based clocks because the frequency band where they interact is impossibly narrow. In fact, the reason why it was only recently demonstrated is due to the difficulty of producing the required frequency at a high enough precision to interact reliably. This could lead to more accurate and more compact and cheaper clocks.
Discussion 4 months ago: https://news.ycombinator.com/item?id=42362215 | Major Leap for Nuclear Clock Paves Way for Ultraprecise Timekeeping (nist.gov)
Note that this is distinct from syncing from GPS, which is a thing people obviously do too, but CVTT can achieve much higher accuracy.
Because you're synchronizing extremely stable clocks the difference between them will primary be an offset (plus/minus a slope from relativistic effects of different altitude). Because of this you can average a large number of readings, so the only major source of error will be systematic effects in propagation/orbit/etc.
Historically, sync was obtained via traveling clocks-- e.g. you sync one atomic clock up and load it, running, in a station wagon... which is the same thing that is most often done for voltage standards today (as atomic voltage standards remain rare, compared to atomic clocks-- I think the least I've paid for one is $15 excluding the ones that were free).
But vibration isn't great for anything with a crystal oscillator in it, and the most modern atomic fountain clocks don't work if they're accelerating in any direction except the designed 'up' direction (gravity), because the little cloud of cooled atoms will fall out of the measurement channel, which makes sync by station wagon not viable.
Of course, once you talk about syncing there is always a question of what you're syncing to. UTC doesn't exist until after the fact. Laboratories measure their offsets via CVTT and UTC is calculated after the fact as past offsets to each of the contributing clocks.
This is because the motion of celestial bodies and spacecraft is dominated by gravitational forces which depend only on G·M, and that motion can be measured extremely accurately with e.g. Doppler radar.
TIL. I guess maybe that explains why the second is used as the base of the SI (https://en.wikipedia.org/wiki/2019_revision_of_the_SI) post-2019, if my understanding of it is correct?
I've also been reading about nuclear clocks[1]... skipping over the uncertainty of the entire atom's chaotic oscillations entirely!
[1] https://en.wikipedia.org/wiki/Nuclear_clock
https://www.iflscience.com/this-incredible-islamic-fountain-...
https://www.nist.gov/news-events/news/1999/12/nist-f1-cesium...
Presumably there's diminishing returns, but as the article says we're at one part in 2.2e-16, are there practical application of going further?
Everybody with a GPS-disciplined oscillator has access to time and frequency from the Naval Observatory at the sub-100 ns level, optionally augmented to +/- 1 ns with reasonably affordable gear like https://www.sparkfun.com/sparkpnt-gnss-disciplined-oscillato... .
A fountain clock is on a whole different level than any of these. The same researchers who build fountains also work on even better optical lattice clocks, none of which you can buy from Sparkfun. These are research tools that don't have a market, at least not yet.
The SI second definition will likely move from Cs-133 at 9 GHz to Sr-87 at 400 THz before too long ( https://www.nist.gov/si-redefinition/second-future ), but that probably won't shake up the existing market too much.