Coordinating The World

March 20, 2014 4:07 AM

Chief Scientist Demetrios Matsakis gives us a tour of the U.S. Naval Observatory's Time Services and explains where time comes from.
posted by gman (33 comments total)

Fascinating, especially the idea of getting too accruate and the problems it creates.
posted by Brandon Blatcher at 4:23 AM on March 20, 2014

The man with one atomic clock knows what time it is.

The man with 100 atomic clocks is never sure.
posted by pjern at 5:22 AM on March 20, 2014 [8 favorites]

Philomena Cunk explains where time comes from on Charlie Brooker's show.
posted by crocomancer at 5:35 AM on March 20, 2014 [7 favorites]

Not only does altitude in a gravity well affect time in general relativity, but my understanding is that because the density of the Earth immediately beneath you affects the force of gravity infinitesimally, just walking past the edge of a caldera^ will alter the rate time flows at for you relativistically. Presumably, whether you're at sea or on land would also matter?

I can't articulate why, but somehow a future in which you could have your own watch or clock sensitive enough to perceive these differences as you move around seems cooler than jetpacks and flying cars.
posted by XMLicious at 5:52 AM on March 20, 2014 [1 favorite]

I was really shocked that the US Navy was going to explain to us how time comes to exist.

I was less shocked by the actual content of the video, but still, really cool! thanks.
posted by So You're Saying These Are Pants? at 5:57 AM on March 20, 2014 [1 favorite]

I wonder if at some point geologists will be able to detect subterranean magma flows and other volcanic activity because the clocks change. Or if oceanographers will be able to map the flows of water currents at different temperatures, or of more and less saline water, in realtime because they have different densities.

(Not that there aren't already lots of sophisticated ways that they do those things, it would just be cool to participate in such a project through giving them data via a phone app or something, if phones end up with atomic fountain clocks in them.)
posted by XMLicious at 6:10 AM on March 20, 2014

That bit where 'this is how your iPhone gets its time' is not how I understand it to work. You'd hope for a bit more... er... precision. Plus, relativity issues with timekeeping are a lot more interesting and involved than the gravity gradient across a clock (although those are plenty fun).

So, good video but could do better.

Incidentally, you can have your own atomic frequency standard for a few dollars, these days, as Ebay has loads of ex-equipment rubidium modules. They don't tell you the time, just provide a 10 MHz pulse, but can be GPS-disciplined and built into other stuff for as much local accuracy as you probably need. I have an abandoned project for turning a $10 LED bedside alarm clock into a rubidium device, Just Because. The major danger is turning into an amateur time nut
posted by Devonian at 6:17 AM on March 20, 2014 [2 favorites]

Sly, possibly inadvertent joke at 5:50 -- "It's something like being a doctor who may know how to keep somebody alive..."
posted by ricochet biscuit at 6:25 AM on March 20, 2014 [3 favorites]

What a compelling video! What was that phone number again?

I hope they paid him partially in bookends for this interview.
posted by oceanjesse at 6:44 AM on March 20, 2014 [1 favorite]

The man with 100 atomic clocks is never sure.

The man with 100 atomic clocks builds a timescale and is optimally sure, statistically speaking.

Also, fun fact about this excellent place... The conspicuous circular shape of the USNO grounds was part of the original vibration isolation specification for the master pendulum clock that kept US time before atomic standards (or so I was told).
posted by fatllama at 7:05 AM on March 20, 2014 [1 favorite]

I was really shocked that the US Navy was going to explain to us how time comes to exist.

There's probably a good reason for that. The Navy needed clock time the most at one point, or they could have lost all ships in one sudden crash into rocks at night or in fog. They knew how to navigate latitudes and sail around the globe using stars, but couldn't know where exactly they were longitudinally without distance measurement. Only clocks could measure the distance they traveled on the ocean. Developing those clocks was a type of arms race that even involved exotic self-lubricating wood.
posted by Brian B. at 7:06 AM on March 20, 2014 [1 favorite]

XMLicious said

I wonder if at some point geologists will be able to detect subterranean magma flows and other volcanic activity because the clocks change. Or if oceanographers will be able to map the flows of water currents at different temperatures, or of more and less saline water, in realtime because they have different densities.

Well, it doesn't use clocks, but Gravity Gradiometery is an established discipline, which uses the differential of gravity in many ways.

My particularly cool toy from this is the Full Tensor Gradient system developed by Bell Aerospace to support the Trident 2 system of submarines. They have maps of the sea in 3 directional gravity that they then use with inertial and dead reckoning systems to figure out where a submarine is on the planet, without emitting any signals. I'm delighted to find that those systems are small and portable enough to put on airplanes. I want one to play with.
posted by MikeWarot at 8:15 AM on March 20, 2014 [3 favorites]

He is educated stupid.
posted by ocschwar at 8:15 AM on March 20, 2014 [5 favorites]

What a compelling video! What was that phone number again?

202 762-1401

Of course, if you call it from a cell phone or VOIP number it won't be especially accurate because of the delay that digitization introduces. But it's good enough to set a wall clock from.
posted by Kadin2048 at 8:38 AM on March 20, 2014

Call WWV at [phone] and tune your piano while you're at it.
posted by mykescipark at 8:55 AM on March 20, 2014 [2 favorites]

The man with one atomic clock knows what time it is.

The man with 100 atomic clocks is never sure.

I wonder how they deal with synchronization. At the scale he works on, there are noticeable delays in transmitting measurements between devices, so I'm curious how they decide on which is the one measurement to sync to, and how they account for those necessary corrections.
posted by Blazecock Pileon at 9:16 AM on March 20, 2014

He is educated stupid.

That's a pretty inflammatory statement with no followup for such a benign topic. Care to elaborate?
posted by Dr. Twist at 9:18 AM on March 20, 2014

Dr. Twist: 1 Educated Are Most Dumb.
posted by Skorgu at 9:26 AM on March 20, 2014 [1 favorite]

Loved this. I think I've mentioned before that my father used to work for the Smithsonian Astrophysical Observatory, pre-Internet and he would take a super-precise clock around the world setting the time at various satellite stations.
posted by jessamyn at 9:35 AM on March 20, 2014 [3 favorites]

Devonian: "I have an abandoned project for turning a $10 LED bedside alarm clock into a rubidium device, Just Because. The major danger is turning into an amateur time nut"

You know what they say... "When the going goes weird, the weird go pro!"
posted by symbioid at 9:44 AM on March 20, 2014

That's a pretty inflammatory statement with no followup for such a benign topic. Care to elaborate?

A reference to Time Cube, a web site that is the love child of an unfortunate with a theory of physics and profound schizophrenia.
posted by ocschwar at 9:46 AM on March 20, 2014 [1 favorite]

I don't like making fun of Gene Ray any more, but whatever's going on in his head, he seems to function in society just fine and able to manage his own affairs well enough, and he can hold perfectly normal conversations when the subject isn't the time cube.
posted by ROU_Xenophobe at 10:38 AM on March 20, 2014 [1 favorite]

That bit where 'this is how your iPhone gets its time' is not how I understand it to work. You'd hope for a bit more... er... precision.

I didn't see anything terribly wrong with the explanation. It's a really high-level popsci explanation, but it's not really wrong.

My understanding is:

1. The USNO Master Clock signal is communicated to the Air Force Space Command Master Control Station, which uses it to synchronize the satellites in orbit. The satellites obviously have their own atomic clocks, but need to be kept periodically in sync with an external reference. (Whether this is because each satellite has a slight drift that needs to be corrected, or because of relativistic differences between their orbits, I'm not sure and I've never seen discussed. I'm also unclear how they actually transmit the time signal from the USNO to the MCS, which must be challenging.)
2. Cellular systems (at least those using TDMA, maybe less so with other schemes) have to be very closely synchronized to each other in order to do call handoff between cells without dropping the connection. The easiest way to do this is by putting a GPS receiver on the tower and getting the time signal from it. It could probably be done other ways (over the phone lines, NTP, WWVB, etc., but GPS is cheap, robust and easy)
3. The time signal is received by various handsets as part of the communication with the tower, and used to set the phone's internal clock.

One interesting effect of this is that some handsets -- until recently, most Android handsets -- took the "GPS time" provided by the cell tower and displayed it to the user directly. Since GPS time doesn't incorporate leap seconds added to civil time / UTC since 1980, it's noticeably off from UTC. The iPhone adds leap seconds back in, though.

Obligatory NdGT discussion about the Android vs. iPhone time difference.
posted by Kadin2048 at 11:59 AM on March 20, 2014 [1 favorite]

I could be off on this, but I was just reading about it in Physics Today, so my understanding of why you need the ensemble of clocks at USNO and you need to periodically update individual atomic clocks (like GPS satellites) is that the second is defined as a certain number of oscillations of a hyperfine transition of an isolated Cesium atom, and to do so you put it in a resonant cavity and measure the frequency. But that cavity has other cesium atoms, which can perturb each other, or there could be small electric fields, or other things that cause that line to shift minutely in a way that's mostly particular to that specific clock. So their drifts are apparently somewhat stable (if I remember correctly), but of course to measure the drift you need a non-drifting reference, which has to be an ensemble of clocks. And if you only have one clock, you need to remove that drift every so often to put you back in sync.

Ok I think that was vaguely related. I thought it was interesting.
posted by kiltedtaco at 12:45 PM on March 20, 2014

The US Navy knows the most about time, given how they were the ones who travelled through it.
posted by Apocryphon at 1:13 PM on March 20, 2014 [1 favorite]

As I understand it, the GPS system uses a separate clock run out of an AFB in Colorado. Monitor stations measure the orbits and apply corrections to its time base, which are sent to each satellite and eventually received by your GPS receiver to use in its calculations.

GPS time to UTC involves a separate correction for leap seconds, and is also steered to UTC time using a linear correction which changes every decade or so. The GPS satellites broadcast these two constants which your receiver use to give you UTC.

GPS satellites apply several corrections based on relativity, and your GPS receiver also performs a periodic correction to each satellite signal based on its orbit's eccentricity.

And there's just a crapload more.
posted by RobotVoodooPower at 1:23 PM on March 20, 2014

Ah, this is good - it's finally got me to research my assumptions about network time and handsets. It turns out that in GSM, it's complicated, and the fact it's never worked for me or my friends in the UK with whom I've discussed it is due to a combination of patchy network and handset support. Which is why (when I could be bothered) I used NTP on my old Android phones.

My shiny new LTE handset does do proper RTC network time synchronisation (as it bloody well should). I just went out in the rain to check. Yay, waterproof handsets. No yay, British weather.
posted by Devonian at 1:47 PM on March 20, 2014

this was pretty good!
posted by rebent at 1:52 PM on March 20, 2014 [1 favorite]

Gravity Gradiometery

I recall first hearing about this during the Apollo program. Flight controllers were puzzled why their clocks were not accurately synchronized with the predicted orbit of the CM around the Moon. They attributed it to unexpected "masscons" (concentrations of mass) inside the Moon, causing gravity gradients.
posted by charlie don't surf at 3:58 PM on March 20, 2014

I wonder how they deal with synchronization.

In general, accurate time transfer is all about accurate measurement of the one-way signal delay. Often, the best way to do this is measure the round-trip delay and assume you can divide by two, which is equivalent to assuming the paths A->B and B->A have symmetric delays. If it is transfer down the hall, this means a cable delay, and the assumption is good. If it is transfer between continents, this means atmospheric delays and all sorts of transceiver delays (the assumption is good in practice down to the 5 ns level from the US to the EU, or about five feet of light travel). If it is transfer on the internet, it means assuming the network paths and queues in each direction are the same (only a good assumption at the several ms level).

To use a system like GPS, you need a different solution to determine the one-way delay, because you can only listen to the satellites (well, more correctly, they don't listen to you). You only get A->B; there is no B->A measurement. So, your GPS receiver has to know everything about each satellite's orbit parameters... fortunately, this is part of the coded message the satellites transmit, along with their unique ID number. More modern receivers can get a head start by looking up the predicted orbit parameters online or in an internally stored table (independently knowing the approximate current time, which is also part of the GPS message, is the key here).
posted by fatllama at 6:32 PM on March 20, 2014 [2 favorites]

I don't like making fun of Gene Ray any more, but whatever's going on in his head, he seems to function in society just fine and able to manage his own affairs well enough, and he can hold perfectly normal conversations when the subject isn't the time cube.

FACT: Every sentence and diagram on http://www.timecube.com:80 is correct and true. Any incorrect information received is an error on our part. I think it just takes a great deal of close study.

I have a hypothesis on the 4 simultaneous 24-hour days - I think the 4 is arbitrary or symbolic, it's 4 quadrants of the Earth, meaning the days are roughly morning, afternoon, evening, night which occur simultaneously across the globe. Our notion of any one time UTC or whatever is a construct and an imposition, one Gene Ray rails against ... I think this is something of a theme, he hates when signifiers overpower the signified. I should reread my Baudrillard...

Word has no inherent value, as it was invented as a
counterfeit and fictitious value to represent
natural values in commerce. Unfortunately,
human values have declined to fictitious
word values.

I doubt there's any novel philosophy there but it's not word salad or content-free.
posted by save alive nothing that breatheth at 9:34 PM on March 20, 2014

So, that's the origin of the "Master Clock" voice.
posted by JHarris at 3:20 AM on March 21, 2014

Fred Covington, is his name. The name of the man whose voice reads the Observatory's Time Of Day clock. I've admired the reading before. Seems he was an actor, who had roles in Roots and In The Heat Of The Night. He died in 1993, but his voice keeps telling us the time.
posted by JHarris at 3:29 AM on March 21, 2014