Dual atomic clock experiment

[iMo]

Curious Marc tries with his clocks:

[Doctorandus_P]

I find it hilarious that these clocks which were advertised as "accurate to within one second in (several) million years", are all broken after 20 to 30 or so years.

Edit:
Nah, I'm perfectly aware of the difference between accuracy and longevity (See below). Still find it hilarious.

[tggzzz]

I find it hilarious that these clocks which were advertised as "accurate to within one second in (several) million years", are all broken after 20 to 30 or so years.

You are confusing precision with longevity For the latter, see https://en.wikipedia.org/wiki/Clock_of_the_Long_Now and https://en.wikipedia.org/wiki/Long_Now_Foundation

[iMo]

Now it looks better, but still..

[SanderMTI]

I find it hilarious that these clocks which were advertised as "accurate to within one second in (several) million years", are all broken after 20 to 30 or so years.

Edit:
Nah, I'm perfectly aware of the difference between accuracy and longevity (See below). Still find it hilarious.

It is indeed hilarious.
I have old actual pictures of my parents when they were young.
If they would have been digital, I likely could not look at them today.

Over land en zee met...?

Sander

[Johnny B Good]

@iMo,

 thanks for posting that youtube link to the 2nd video which has now given me some idea of their frequency/phase stability. I don't know whether a phase drift of just 6ns in a 24 hour run ( frequency offset value of 694pHz) is typical of these secondary frequency standards or still a little short of what to expect with a new/factory refurbished unit.

 The reason for my interest in this being down to my pet project/obsession to convert an LPRO 101 into a "Poor Man's Cesium class frequency reference". At the moment, I'm seeing a stability about an order of magnitude worse with the current prototype.

 I'm in the middle of a major revamp to eliminate the issue of internal temperature gradient varying with ambient temperature. I have high hopes of limiting the phase wander to less than 10ns in 24 hours over an ambient temperaturet range that encompasses  0 to 33 degrees C (and obviously much less over the typically specified 23 +/- 5 deg C range) with the revamped version.

 It would be useful to have some idea as to when I can call this quest done and dusted.

[iMo]

..it seems to me Marc has silently resigned in his attempt to repeat the relativistic experiment with his clocks, a pity..

[dietert1]

For a cesium clock the usual number people tell you as accuracy is 1E-13. That is about 10 ns in 24 hours. Some special cesium fountain clock at the NIST reached 1E-16, but that's a different story.
https://www.mdpi.com/2218-1997/9/3/133
The introduction talks about modern atomic clocks and says: ".. near the surface of the Earth, a fractional frequency shift of 1.1E−18 corresponds to a height difference of 1 cm". In order to get a clear result comparing two cesium clocks one needs a height difference of several thousand meters. It can't be done at home, sorry.

Regards, Dieter

[Johnny B Good]

@iMo,

 That seems a pessimistic point of view. He seems to be a man after my own heart, only with more exotic (to me) bits of kit to play with.

 I've been working on my own pet project for just the past 4 1/2 years ever since I'd acquired my first LPRO 101 back in August 2020.

 Initially I took this on as a challenge to see just how far I could go in reducing the 0 to 50 deg C tempco effect to a vanishingly small value merely by shrinking this temperature range down 50mK or less. Over the following years, I kept maintaining progress and overcoming unexpected snags, every so often stepping back to admire my handiwork thus far and to contemplate my next step (a process most outside observers might well mistake for procrastination).

 I've no doubt Marc is merely taking pause to consider his options before resuming this project. "This project is not dead! It's merely resting" to misquote the pet shop proprietor in that famous Monty Python "Dead Parrot" sketch.

 I too, like you, thought it was rather a pity that his project had come to a (hopefully only temporary) stop, mainly I have to admit, due to my impatience in seeing the next installment.

[iMo]

..
The introduction talks about modern atomic clocks and says: ".. near the surface of the Earth, a fractional frequency shift of 1.1E−18 corresponds to a height difference of 1 cm". In order to get a clear result comparing two cesium clocks one needs a height difference of several thousand meters. It can't be done at home, sorry.

What Marc wanted, afaik, is to take one of his clock to a mountain top, the other will stay down in his lab.
Not sure what specific summit he meant - we have to look in his first video above..
PS: Mount Wilson (1740m, the top near LA accessible by car) perhaps ("..bringing it to an elevated altitude for couple of days..")??
He targeted 20ns diff averaged during "couple of days" (he said it is 15x smaller diff as the guys saw in the famous 1971 experiment).

[coppercone2]

if you shoot it out of a catapult can you get something else to measure from time like from rapid acceleration?

Gravity static force, but acceleration is a rapid transient one with big magnitude.

acceleration has something to do with relativity

perhaps there is truth in the statement of time flying


I don't know many things that will get a giant jump like that other then catapults. maybe a rocket sled.

[iMo]

..
The introduction talks about modern atomic clocks and says: ".. near the surface of the Earth, a fractional frequency shift of 1.1E−18 corresponds to a height difference of 1 cm". In order to get a clear result comparing two cesium clocks one needs a height difference of several thousand meters. It can't be done at home, sorry.

What Marc wanted, afaik, is to take one of his clock to a mountain top, the other will stay down in his lab.
Not sure what specific summit he meant - we have to look in his first video above..
PS: Mount Wilson (1740m, the top near LA accessible by car) perhaps ("..bringing it to an elevated altitude for couple of days..")??
He targeted 20ns diff averaged during "couple of days" (he said it is 15x smaller diff as the guys saw in the famous 1971 experiment).

AI made some math for me:

The atomic clock at Mount Wilson will be ahead of the one in Los Angeles by about 31.4 nanoseconds after 2 days.

[dietert1]

If your "noise" is 10 nsec and the effect is 15.7 nsec in 24 hours, there is some chance. But you need to bring a clock up and down like 10 times to get the noise suppression of a lock-in. Maybe each time another clock. And assume losing some of the trips due to unexpected difficulties with the "mobile lab".
How much is the power consumption of the cesium clock and the required air conditioning?

Regards, Dieter

[iMo]

There is an observatory on the Mount Wilson, so he may arrange an X day stay there in a controlled conditions.
Say a week, that is ~100ns diff, that should be easily visible against 10ns noise..
What could be an issue is the travel to the MWO and back by car, indeed..

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

[edpalmer42]

This experiment has been successfully done twice by one guy - with help from family & friends.

http://leapsecond.com/great2005/

The second time, Stephan Hawking (Yes, THAT Stephen Hawking) presented the experiment to the public on a TV show.

http://leapsecond.com/great2016a/index.htm

[dietert1]

Both experiments were using several clocks in order to reduce the measurement error, e.g. to 2 nsec in 24 hours. Probably with another even bigger set of clocks to stay for reference. And they expected an effect of 20 nsec. That makes more sense.
The videos above mention completely different numbers, like the magnetic field tuning wasn't possible to better than 6 nsec per 24 hours. And for somebody with just two clocks he will go with one clock and the other one will stay. The error of the difference measurement will be 14 instead of 10 nsec per day and the 16 nsec on top of Mt. Wilson won't serve.
Also to be a scientific experiment one needs to study how much time shift is caused by lower air pressure at high altitude. The pressurized plane cabin may be easier in that sense.

Regards, Dieter

[TimFox]

Another way to look for the effect of gravity on timing uses the Moessbauer Effect to measure the frequency shift of ⁶⁰Co gammas (roughly 1 MeV) between source and detector at different altitudes.  I believe the first successful test involved “dropping” the gammas down a tower at Harvard (22.5 m altitude).
The interested reader can find accounts of the experiment, which can detect minuscule differences in photon energy, hence frequency shift.

[Johnny B Good]

@dietert1

Excuse me, are you telling me that these caesium beam secondary atomic standards also suffer air pressure induced frequency drifts like the rubidium atomic frequency references? 

[dietert1]

@dietert1

Excuse me, are you telling me that these caesium beam secondary atomic standards also suffer air pressure induced frequency drifts like the rubidium atomic frequency references? 

I can't tell you. Just wrote down a question that any physicist would ask. And the experimenter should have some number, i mean something more than an opinion.
You mentioned rubidium clocks and i found this paper:
https://apps.dtic.mil/sti/pdfs/ADA485423.pdf
From their diagram i read about 1E-10 change after 200 days. So that could be 43 nsec in one day, or let's say a fraction of that for imperfect vacuum. And it's a dilatation effect, as the clocks are frequency aging faster at lower pressure. A pitty one can't tell the technical reason of the observed effect.

Regards, Dieter

[iMo]

Provided such helium penetration in/out the evacuated tube gets say "1000days time constant" it will have a negligible effect during the "couple of days long experiment", imho. And it has nothing to do with relativistic phenomenons, I would compare it to the humidity effect on an epoxy packaged reference ..

[Johnny B Good]

@dietert1

 Well, this experimenter did his own homework (see attached research paper and Datum's user guide for the Efratom LPRO 101 wherein they quote a rather wishy washy figure of "Altitude (operating) -200 ft. to 40,000 ft. <1E - 13/mbar").

 That Swiss Lab paper had arrived at a figure of 0.8 E - 13/mbar which, since it didn't violently disagree with Datum's rather vague figure, I took as my initial starting point to compensate for this effect until I could run my own verification tests to get an actual figure for the LPRO 101 oscillators.

 After figuring out a cost effective way to vacuum test one of my spare LPROs using a cheap 11 litre pressure cooker and vacuum pump, I purchased said items last February and ran my tests around the beginning of March. I eventually, after a few false starts, obtained Frequency shift delta figures of 0.607 E -13 and 0.587 E -13 per mBar on my final two test runs which I averaged to a value of 0.597 E -13/mBar to use as my barometric compensation factor.

 As far as I'm aware, the caesium beam tube in these secondary caesium atomic frequency standards aren't effected by this pressure effect to anywhere near what a typical rubidium vapor lamp pumped earthbound RFS has to endure, hence my response to your statement.

[/* EDIT]  Also worthy of note are the aging figures given in an entirely separate Datum datasheet (naturally!) which I've now attached which claims a 10 year figure of <1E -9 which, as best as I can figure it would by now correspond to a daily aging rate somewhere in the region of 1 to 2ns rather than your suggested figure of 43ns a day.
[*/ EDIT]

[dietert1]

In the report i linked above they measured a drift of -1,7E-13 per day in vacuum and about zero after taking the clocks out of vacuum. So the "surprise" when going to a high altitude with those rubidium clocks could amount to -1,7€-13, which is 15 nsec slowdown within one day, or a fraction of that. The 43 nsec overestimate was obtained as visual reading the difference of the slopes fitted in the diagram. The numbers are there.
One would need a similar test of cesium clocks to give a limit on that systematic error.

Regards, Dieter

[Johnny B Good]

Provided such helium penetration in/out the evacuated tube gets say "1000days time constant" it will have a negligible effect during the "couple of days long experiment", imho. And it has nothing to do with relativistic phenomenons, I would compare it to the humidity effect on an epoxy packaged reference ..

 When I looked at that document, I had a strong feeling that I'd come across this in my earlier searches on the subject of Rubidium oscillator stability issues. Since it wasn't already in my 30 strong collection of downloaded pdfs, I assumed I'd dismissed it as being not really relevant enough at the time.

 Having downloaded it to facilitate a more relaxed perusal, I've added it to my collection despite it still not being as relevant as most of the other documentation (no firm conclusions, only interesting hypotheses on the subject of the aging phenomena).

 And, as you said, has no relevance to relativistic phenomena. As for the aging effect, I'll deal with that if I ever gain enough stability to be able to discern (and quantify) this effect. Like the proverbial Boy Scout, I've already got a strategy in mind to compensate even for that, involving the use of a battery backed RTC module.

[edpalmer42]

John,

Are you also stabilizing the temperature of the baseplate?  It looks like temperature effects are larger than pressure effects.  The LPRO User's Guide shows a graph (Fig. 1.4) of temperature vs. frequency on 19 random units.  The spec (<3e-10 over 0 - 50C) suggests a value in the range of xE-12 per degree.

[Johnny B Good]

John,

Are you also stabilizing the temperature of the baseplate?  It looks like temperature effects are larger than pressure effects.  The LPRO User's Guide shows a graph (Fig. 1.4) of temperature vs. frequency on 19 random units.  The spec (<3e-10 over 0 - 50C) suggests a value in the range of xE-12 per degree.

 I certainly am (37.050 deg +/- 1mK). 

 It was that which had inspired me to improve the frequency versus base plate temperature variations by simply reducing this temperature range to less than 1% to gain (in theory, all else aside) a 2 orders of magnitude improvement over the worst example's figure of +/- 5 E -11 over that 0 to 50 deg range.

 The implied x E-12 per degree totally ignores all the other tempco effects which, as far as its intended use as a telecoms GPS disciplined oscillator needing only to limit the 72 hour holdover drift to a whopping (by metrology standards) 140μs is concerned is regarded as inconsequential "noise".

 It's not as simple as merely stabilising the base plate temperature to within a milli Kelvin of a set point against a 20 to 30 degree variation in ambient temperature. The obvious measure of thermally insulating the hell out of the other 5 sides helps but I'd woefully underestimated just how (impracticably) much insulation would be required.

 Holding the base plate to a set temperature also requires a measure to eliminate variation in the temperature gradients within the LPRO itself. I'm in the middle of building a version where the heatsink exhaust is fed to the top cover to reduce this thermal gradient variation close to a vanishing point. I suspect I'll still need to add a couple of "after burners" to the heatsink exhaust to compensate for cooler ambient conditions (I think a watt or two each should do the trick).