NIST Builds A 100,000 Times Better Atomic Clock

J Shumate writes: " NIST has built a better clock, which, no doubt, will lead to a better mouse trap." According to the article, "The new all-optical atomic clock -- so named because of its reliance on laser technology -- measures the shortest intervals of time ever recorded. In fact, those intervals are 100,000 times shorter than those observed by the best current clocks."

Re:Electronics?

[Anonymous Coward]

The effect to which you are referring (clock rates differ between ground and up in the air) is not a special relativity problem (i.e. relative speeds), but a general relativity problem. The clock rates differ because the one on the ground is deeper into the gravity well created by the earth's gravitational field. Such measurements were carried out in the early 1960's using Mossbauer effect driven clocks (that can resolve time durations down to about a part in 10^13 or so).

One would think that a claim to high precision would take this effect into account since gravitational wells do effect photons (light) - an effect first seen in the eclipse of 1919 (Einstein's great vindication).

Fun with relativistic effects

[K-Man]

Even motion at 1 m/s should yield time dilation of about 1 part in 10^17, which is only 100 times smaller than the ticks/s (10^15 = 1 quadrillion) of the clock. So walking the thing around for a about 100 sec. will yield a difference of one tick.

Anybody want to calculate the GR effects at different altitudes?

Re:Fun with relativistic effects

[K-Man]

OK, I got lazy and looked up some figures. At the Earth's surface, the gravitational time dilation for each km of altitude is 1 part in 10^-13 (source [sussex.ac.uk]), or 100 ticks/s. This is a linear approximation, so that translates into 1 tick/s gained for each 10m of altitude.

Electronics?

[awa]

So the thing seems to be pretty accurate, based on the "tick" frequency, and they talk about a lot of uses for it. However, nothing is said about the size of the monster. This is important because, if it could be made small (and cheap) enough and the "ticks" could be used to draw a sync. signal from, PetaHz computers do not seem too far away.
I know this might sound like wishful thinking, but they do talk about using the technology for many menial tasks, so, why not?

Short Term vs. Long Term Stability

[Detritus]

Anyone know what the short term stability of this clock is? Last time I checked, masers beat cesium beam clocks for short term stability, which is important in some applications.

Re:accuracy

[thogard]

Pi seconds is about a nanocentury.

Re:Electronics?

[EasyTarget]

I dont get it, I had a quick look at the PARCS stuff and I cannot find any mention of relativistic effects..

However, they seem to be talking of comparing clocks which are moving at different speeds relative to each other. Presumably, given the sort of accuracy they are trying to achieve, these clocks will suffer signifigent relativistic drift from each other. I know that relativistic effects have been observed with a pair of clocks, one on the ground, and another in a Jet doing a large number of long trips. IIRC They actually confirmed the predicted relative time dilation effects by experiment that way.

Anyone know how PARCS addresses this..?

EZ

Terry Pratchett...

[dr bacardi]

Just warned me about this in Thief of Time.

Re:accuracy

[Old Wolf]

Well, we can check that out: it ticks 10^18 times per second. If we assume that it is therefore accurate to one part in 10^18 (a reasonable assumption), then that's one second in 31.6 billion years (according to dc :)

So perhaps BBC were unexaggerating, or perhaps they halved my accuracy estimate.

BTW something very interesting happened here. This was my full calculation:
seconds/400 yrs = 86400 * (365 * 400 + 397) = 12648700800
==> seconds/yr = 31621752 (exactly)
10^18 secs / 31621752 = 31623810236 years
(ie. the number of seconds in a year is very close to the squareroot of ten! (316227766...)

Re:Electronics?

[Old Wolf]

The problem with computers is not being able to generate the signal; it's i) making components that can respond to the signal fast enough, and ii) making components that don't overheat when doing (i).

The real test

[psykocrime]

is whether or not this thing is accurate enough to fix the timing problem in the TARDIS.

Re:Millimeter GPS?

[Doctor Fishboy]

I'm an astronomer, but not a satellite expert :)

I would imagine that it's possible in theory, but the major systematic error would be in the accurate location of the satellite itself. Satellites AFAIK are known to a few meters, but they need constant orbital updating.

Any GPS experts care to comment?

M.

Re:Millimeter GPS?

[yzquxnet]

Would it be possible to triangulate the position of a satelite based on signals from other atomic based clock satelites? If this could be done you could narrow the error margins down from both sides. Where you are, and where the satelites are. Or maybe they just broadcast the time down to earth and not into space.

Also, If the two satelites, or maybe three, knew where each other where and also broadcast that information to a gps unit, would it be possible to triangulate a position with fewer satelites?

Maybe I should sharpen my pencils and see.

The clock at work.

[dankjones]

So when I arrive at work ON TIME and they say "Hey, You're 3 minutes late." and I tell them my watch is set to NIST time and their clock is 3 minutes fast. They'll actually give a shit now?

Re:accuracy

[eclectro]

NPR [npr.org] did a story on the clock today (the links are not up yet) They talked with one of the researchers who said that the clock is accurate to 10^15 - so the BBC numbers probably work out.

Re:Electronics?

[eclectro]

The Reuters report mentions that it takes up a couple of rooms. I would think that it takes pretty exotic equipment to constantly suspend the single ion of mecury so they could bounce lasers of from it.

I'm sure someday they will get all the laser/optical processing smaller, but it is probably isn't a priority right now. The most important use of accurate clocks will come from the PARCS project [nist.gov].

How long .....

[schlam]

Now we can accurately time how long it take to cum it some ridiculous significant digit...

Re:Fun with relativistic effects

[-douggy]

I would but my pupils dilated when I saw the equations

accuracy

[stebalo]

According to the BBC, this clock is accuracy to 1 second over the (currently estimated at 15 billion years) life of the universe

Re:accuracy

[stebalo]

In physics this kind of estimation is what we call an "order of magnitude" estimation and if you are within a factor of 10 ( 10* or /10) then you consider your answer close enough.

so a factor of 2 is nothing to sweat :)

Re:Millimeter GPS?

[cpl almost]

I believe upgrading the clocks would improve the accuracy. The sats all "know" their relative positions, and update constantly, to a maximum resolution based on the clocks, and it would make sense that they include a ground location(s) defined as stationary. I think the most significant limit to the system is the clocks.

Millimeter GPS?

[Kadin2048]

So, if you had one of these clocks and put it in a GPS satellite, wouldn't it mean that you could get the GPS system that much more accurate? As I understand it, at the heart of GPS sats are atomic clocks; would more accurate clocks = more accurate positioning?