50th birthday of the favorite radio station of clocks
July 5, 2013 5:25 PM   Subscribe

Today is the 50th birthday of WWVB, a low frequency broadcast time service and frequency standard station of the U.S. National Institute of Standards and Technology that can be received in much of the Northern hemisphere. Along with high frequency sister stations WWV and WWVH, WWVB continues to evaluate and deploy new technologies to provide easily received and decoded time and frequency signals, which unlike GPS signals, will penetrate buildings and propagate past many terrain feature shadows, like hills and canyons.
posted by paulsc (37 comments total) 31 users marked this as a favorite
 
This is the reference radio source of what are colloquially called "atomic clocks", i.e., the self-setting consumer clocks. So-called because it's synced to the NIST's actual atomic clock which is the national time standard.

But speaking of WWV, I vividly recall finding that station on my first little portable shortwave receiver when I was a teenager. The idea of knowing the true time to the second was a big deal to me. Maybe it's an OCD kind of thing. But it bothered me that back then everyone set their clocks to other somewhat inaccurate clocks. Plus, I'd just leave the station on, listen to the little ticks and then the minute announcement. It made me happy.

I'm glad to be reminded of this because up until this moment I'd thought the genesis of my time fetish was from when I worked as a disc-jockey just after high school — it was AM, with network news on the hour, and so generally I was aware of the exact time to the second for my entire shift every night. And I became accustomed to that. But now I realize that it really started with shortwave station.

When the "atomic clocks" first appeared around 1999, I was ecstatic. I bought a nice little analog watch, that didn't in any way appear to be anything other than a (elegant, to my eye) regular analog watch, which I loved. I broke it a few years later. Since then I've had a couple of other "atomic" watches that I don't quite like as well, but I simply can't tolerate an inaccurate watch any more. Or any other inaccurate clocks around the house. I think all timepieces should be self-setting to the NIST standard. The tech is cheap. It's the twenty-first century.

Don't get me started on NTP/SNTP and networked computers that aren't using it.

Yeah, I'm kind of obsessive on this topic.
posted by Ivan Fyodorovich at 5:50 PM on July 5, 2013 [17 favorites]


Interesting article, although not a lot of new info for those of us who are already familiar with NIST's time stations. Back in the early-mid 90s, I'd make a long distance call to NIST's modem bank every week or so to set my computer's clock. I think the accuracy on that was claimed to be around 250ms. Pretty good for bailing wire and string.

That said, I'm feeling an irresistible urge to nitpick. Your phone doesn't display time "beamed down from space", unless you've got an odd duck of a phone. The displayed time is either free running, set occasionally from the cell network (2G GSM phones often work this way), or slaved to the cell network so the whole damn thing has a chance of working (CDMA-type phones).

Granted, much of the time those cell sites use GPS as a reference clock, but they almost all also have relatively accurate oscillators that can maintain accurate time to within a couple of milliseconds for a week or two should there be a GPS problem. Sometimes they use network clocking, but that's getting more and more rare, I assume because the telco's networks are often no longer synchronous end-to-end and therefore aren't so useful.
posted by wierdo at 5:51 PM on July 5, 2013 [1 favorite]


There's a nice soundbite history of WWVB at

http://soundcloud.com/shortwavemusic/sets/at-the-tone-a-little-history/
posted by the Real Dan at 5:54 PM on July 5, 2013 [1 favorite]


At the third stroke it will be ten -- four -- precisely.
posted by localroger at 5:57 PM on July 5, 2013 [1 favorite]


Most boring station ever.
posted by 2N2222 at 5:58 PM on July 5, 2013


I worked at a place some years ago where the computers were all set up to set their clocks to a central server, which was about 7 minutes slow. This annoyed me, so I downloaded an NIST tool to set my computer's clock correctly. genius, right?

No. I would turn my computer on in the morning, it would misset the time. Then, a bit later, the NIST tool would set it correctly. Then the institutional server would set it back. So, instead of being a regular 7 minutes off, I never knew what time it was. I eventually uninstalled the tool and just used my watch.

In other news, I suppose these stations figure in conspiracies somewhere....
posted by GenjiandProust at 6:06 PM on July 5, 2013 [1 favorite]


I'd just leave the station on, listen to the little ticks and then the minute announcement. It made me happy.

I remember once when I was home sick from school I left it on for several hours. It was fascinating and comforting for some strange reason. I even wrote to the station and got a magazine sized booklet that explained about the atomic clock, how it worked, etc.
posted by cropshy at 6:10 PM on July 5, 2013 [6 favorites]


The first time I heard of this was a few years back, when a friend took the guts from a couple of WWVB clocks, wired them to a custom circuit with a bundle of LEDs, and placed them high in trees at opposite ends of the North Shore mountains facing Vancouver. Once an hour they would wake up and do a little light show together. Thanks to WWVB they stayed in synch forever, or until the batteries wore out, which was maybe a month or two. From downtown it was kinda surreal to see these two pinpricks of light, 10 km apart, flashing in unison.
posted by jeffj at 6:26 PM on July 5, 2013 [11 favorites]


Most boring station ever.
You've never lived in a radio market with THREE right-wing talk stations. #3 is as repetitive at WWV, but without the factual content.
posted by oneswellfoop at 6:32 PM on July 5, 2013 [13 favorites]


It's worth noting that "atomic" clock means "uses atoms", rather than "is nuclear powered"... it always seems to be misunderstood.
posted by underflow at 6:34 PM on July 5, 2013 [1 favorite]


I'm a huge fan of NTP, the sane way to set clocks on the Internet. But the stratum 1 servers have to get their time from somewhere. For the longest time that place was WWVB. By the late 90s most stratum 1s had already switched to GPS but WWVB is still an important alternate source using an independent technology stack.
posted by Nelson at 6:40 PM on July 5, 2013


I'm one of those kids that listened in as well. It was on a beat up shortwave radio my grandpa gave me. I found it completely by accident and it was this private little secret to me. A far away station tracking "coordinated universal time"! What the hell was that?!?

I soon figured the rest out, but there's something reassuring and soothing about these old transmissons. Ever cross the Atlantic on United where the pilot lets you listen in to his radio traffic? Once you're in the middle of the ocean, late at night, all you hear is Gander or Shannon checking in over VHF, thousands of miles away, letting you know how isolated you are yet still being watched. WWV always felt like that.
posted by JoeZydeco at 6:54 PM on July 5, 2013 [5 favorites]


Don't get me started on NTP/SNTP and networked computers that aren't using it.


Ivan, I would LOVE to get you started on any fun technical knowledge or anecdotal tales!
posted by theartandsound at 6:55 PM on July 5, 2013


I once had the privilege of being copied on a fascinating interoffice helpdesk exchange about running time servers for experimental support of a physics department, and it seemed like an awesome sysadmin challenge. Some intrepid worker spent hours reading docs on predictable consistency of delay between devices, identifying cracks in DIMM sockets as a source of trouble, the difficulty of using external comparison systems that with their own flaws confuse where the flaws lie, flutter between switches...

One of the fun recommendations was to not run two time servers, because clients would change out of fickleness in their preference code and cause ripples in consistency. It was also a choice to figure out if consistent slowness over a reliable short span was adequate. In the end, it was stated that maintaining perfect clock synchronization to hundredths of a second is tough.

It would be interesting to hear from researchers that deal with measurement in very narrow time frames.
posted by zangpo at 6:56 PM on July 5, 2013 [3 favorites]


Aren't the GPS satellites synced to the NIST clock, too? I mean, I know that "synced" is confusing in this context given that the atomic clocks aboard the GPS satellites drift because of general relativity. But their standard is the NIST standard, right?

Also, though, the WWVB signal provides only millisecond-level precision while GPS can get into nanosecond-level precision. So, if your computer isn't on the internet, it's better to use a GPS receiver for your LAN NTP than WWVB. Unless you have trouble with receiving the GPS signal. But I'd thought that more common was dedicated hardware that makes dial-up modem connections to NIST.

"It's worth noting that "atomic" clock means 'uses atoms', rather than 'is nuclear powered'... it always seems to be misunderstood."

Yeah, now with the consumer "atomic" clocks, there's two levels of misunderstanding. I really, really dislike that nomenclature. I don't know who is responsible for that coinage — Oregon Scientific, maybe. I'll avoid it in conversation and say something like, "my watch is automatically set to the national time standard via radio" or something.
posted by Ivan Fyodorovich at 6:56 PM on July 5, 2013 [1 favorite]


"Ivan, I would LOVE to get you started on any fun technical knowledge or anecdotal tales!"

I fear that you're being sarcastic. But, actually, this is just an obsession and pet-peeve of mine and I don't have any funny or interesting stories about it.

I've been using NTP for a long time, though. Not as long as the folk who did UNIX and Internet admin prior to the early 90s, but I was using SLIP (which came before PPP) and Trumpet's Winsock stack for Internet access on Windows 3.1 almost the week it became available, sometime in the winter of 1994-95. This is relevant because I found pretty much all of the TCP/IP apps available at the time, incidentally made a web page cataloging and linking them (mine and TUCOWS were the only two sites like this for Windows internet apps for a while), and, interestingly, one of the very first Windows winsock apps that was available was an SNTP client. So I've been using SNTP/NTP on my own PCs since 1994. Although starting with Win2K MS implemented w32time, it was sucky. Anyway, it's like it's taken twenty years for NTP on network connected computers to become the norm, outside of networks where they had reason to especially care about it.
posted by Ivan Fyodorovich at 7:08 PM on July 5, 2013 [3 favorites]


Oh I was being totally sincere. I love hearing about interesting esoteric knowledge or stuff not typically known. Thanks for the tales!
posted by theartandsound at 7:16 PM on July 5, 2013 [1 favorite]


Holy shit, I also was using Trumpet Winsock the instant SLIP was available on my ISP. What was your site? I probably used it.

I worked at a place some years ago where the computers were all set up to set their clocks to a central server, which was about 7 minutes slow. This annoyed me, so I downloaded an NIST tool to set my computer's clock correctly. genius, right?

No. I would turn my computer on in the morning, it would misset the time. Then, a bit later, the NIST tool would set it correctly. Then the institutional server would set it back. So, instead of being a regular 7 minutes off, I never knew what time it was. I eventually uninstalled the tool and just used my watch.


Netware uses exact time for [something...]. It has a robust time sync feature. But if the master time server is wrong, all hell breaks loose. I was at a site where they accidentally set the time back a year. Everyone's password didn't work because it expired in -335 days, which the client didn't like at all. So they called the helpdesk, got their passwords reset, and off they went. Until the time was fixed, and their passwords were now 336 days expired and thus their accounts were security locked. And god help someone if they created a file at the wrong time.
posted by gjc at 7:41 PM on July 5, 2013 [2 favorites]


"Holy shit, I also was using Trumpet Winsock the instant SLIP was available on my ISP. What was your site? I probably used it."

I'd actually forgotten. I just had to google to find something that referenced it — I don't think I have the pages archived anywhere. So long ago.

Anyway, it was Keith's "Best Of" List of Winsock Apps. I forgot about Stroud's winsock app list. I actually think it was mine and Stroud's sites that were first, then TUCOWS came a bit later. Googling reveals a lot of pages that still mention my page and/or link to it. Talk about link rot. It hasn't existed since 1996.
posted by Ivan Fyodorovich at 8:08 PM on July 5, 2013 [1 favorite]


Last summer I fixed up my mom's old timey tube radio with AM and shortwave bands. WWVB helped me calibrate it, plus it was about the only English-language radio that wasn't super-religious or anti-government survivalist.
posted by dirigibleman at 8:16 PM on July 5, 2013


The unfortunate thing about WWVB is that it doesn't tell you the time. There are two important omissions: the total number of leap seconds (so that you can compute the approximate TAI), but more importantly the century is omitted, so that you cannot make a receiver that works for more than 100 years. (however, an explicit leap day flag is included so you can find out whether there's a february 29, xx00). Surprisingly to me, this was not even rectified when the "phase modulated time code" was added sometime in the last few years (well past y2k)
posted by jepler at 8:22 PM on July 5, 2013


WWVB is to Earth as the cosmic background radiation is to the Milky Way.
posted by localroger at 8:31 PM on July 5, 2013


My first experience with a WWVB application in a commercial setting happened in 1975, when I was working for a General Electric Broadcasting television station in Nashville, that operated on analog Channel 2, and in competition with neighboring Channel 2 station assignments in St. Louis and Atlanta. In our mutual fringe areas, like the Land Between the Lakes area of southwest Kentucky, and the Lookout Mountain area of southeast Tennessee/northeast Alabama, viewers often had great difficulty discriminating between our stations, where our primary signals were of approximately equal strengths, but carrying different networks and programming. For years, many viewers in such areas attempted to solve this by building large multi-element high gain directional antennas, which could be rotated to provide 9 to 13 db of antenna gain to the station desired to be watched at a particular time; but this solution was only partly successful, due to terrain and meteorological factors, which could create temporary and unpredictable multipath conditions, that overcame antenna gains.

In the mid-'70s, commercial equipment in the form of high precision rubidium frequency standards became available, allowing cooperating stations to introduce precision carrier offset systems (4 page .pdf file), whereby an intentional 20Hz frequency comb was created in the receivers of viewers, which intentionally overlaid bands of desired signal picture with bands of undesired signal picture, in a pattern which lightened desired picture, and darkened undesired picture automatically. That gave an automatic 3 to 6 db "boost" to whichever signal was at least momentarily stronger at the reciever, and made viewable pictures out of otherwise interlaced noise. For tens of thousands of viewers in fringe areas operating nothing better than rabbit ear antennas, it made viewable television much of the time, and for those with the big yagi stack antennas, it meant a lot less antenna rotation and fiddling, trying to get rid of multi-path signals.

But the system required each cooperating station to install an expensive rubidium frequency standard as its primary transmitter carrier exciter, capable of long term control of their carrier to an accuracy of ± 1 Hz, and to then offset their assigned picture carrier center frequency by plus or minus 20 Hz, to create the mixing signal conditions in receivers. Thus, if our station went "up" 20 Hz, St. Louis and Atlanta agreed to go "down" 20 Hz, and all was better in the fringe areas.

Our rubidium standard came in the form of a 1/2 rack height grey box that cost $50,000, and had Nixie tube displays of primary frequency, offset frequency (to 3 decimal places of fractional precision), and transmitter output frequency (measured as a feedback control input on its internal phase locked loop). It also required and used a WWVB signal input as a secondary frequency check, to allow it to sanity check itself, and initiate a graceful shutdown and failover to our transmitter's normal internal carrier generation circuitry, whenever it lost precision due to loss of power, or environmental factors like temperature drift outside nominal conditions.

I was fascinated with the thing when we first got it, as I'd never seen anything so precise, over such a long period of time. It took a few weeks after we started it, but it actually got more stable with time, as its manufacturer predicted it would. And eventually, it became just another set of measurements to log on the hour, every hour, when you were transmitter operator...

But anyway, here's to WWVB, which has measured out the time for its own existence, and most of mine, one 60,000th of a second at an instant, for now, 50 years!

(raise a beer, and in passing shed a technical tear, for all that was once wondrous, and is now taken for granted)
posted by paulsc at 11:55 PM on July 5, 2013 [17 favorites]


Most boring awesome station ever.
posted by mykescipark at 1:21 AM on July 6, 2013 [1 favorite]


I remember sitting at the shortwave around 7 Eastern on some Dec. 31 to hear the leap second ...
posted by gubo at 3:33 AM on July 6, 2013 [2 favorites]


I don't see it mentioned anywhere, but the male voice of the person reciting the time on WWV belongs to the late radio talk show host Lee Rodgers.

Yes it's true we are surrounded now with gadgetry that automagically adjusts to the proper time, but once in awhile it's just strangely soothing to tune in and listen for a few minutes.
posted by Bindyree at 4:36 AM on July 6, 2013 [2 favorites]


As an aside, WWV/B/H are some of the few stations with the initial W call letter located west of the Mississippi. Normally stations west of the Mississippi start with a K. According to this link on Wikipedia, WWV was originally registered in Maryland, which explains its W call sign.
posted by tservo at 5:04 AM on July 6, 2013 [1 favorite]


I hope that WWVB got a nice card from its European cousins MSF and DCF77.

Rubidium frequency standards are now stupidly cheap (~$100). Yes, you could have your own atomic clock controlling your NTP server.

jeffj, your friend's synched light show works the same way air navigation lights do. They're all controlled by the PPS signal from GPS. In theory, all the light flash at once. In practice, there's always one, somewhere, somehow, that doesn't.
posted by scruss at 5:14 AM on July 6, 2013 [1 favorite]


now can they please use one of these technologies to synchronize all the turn signals in cars?
posted by jepler at 5:40 AM on July 6, 2013 [1 favorite]


A little screed for the technophiles...

I once heard an important, but subtle bit of wisdom. "Metrology is often not about standards but about scaling." I think this means that, for example, while it is all well and good to define the second as exactly 9,192,631,770 counts of a microwave oscillator locked to a cesium atom, we really should worry about the nuts and bolts detail involved in translating such a definition into real a time or frequency measurement, stable over decades, disseminated everywhere, while introducing as little uncertainty as possible. It's a little like the scaling problem in defining the micrometer but having to measure kilometers, or the opposite, all the while having secondary-standards (rulers) with jagged edges.

In timing metrology, a timescale is a sort of algorithm, implemented in both hardware and software, that tries to derive both its short term stability and long-term accuracy from the best-performing of the clocks in its ensemble while at the same time derive utility from but not allowing the worst of its clocks to add error. Simultaneously, it must be absolutely robust against component failure, loss of electrical power, and acts of god. It's a tricky business.

WWV(B) derives its timescale not from the or even an atomic clock, but a particular ensemble of hydrogen masers (optimized for short term stability) and cesium beam systems (better long term accuracy). This ensemble is located near the transmitter station and is disciplined to another ensemble at NIST, an hour's drive south. This ensemble is occasionally compared with (but technically, not steered by, I think) the cold atomic cesium fountain apparatus to generate the timescale UTC(NIST), the distinction about which I'll rant about in a bit.

GPS time, which should not be conflated with either UTC or UTC(NIST), is solely the job of a very fine technical institution, USNO. Both USNO and NIST strive to realize the international Universal Coordinated Time (UTC), but at any given moment the two timescales may be offset. UTC(NIST) strives to be absolutely independent of GPS time, which may become an important fact if GPS is ever jammed or experiences a fault. UTC itself is a funny beast: it is an international timescale that is only realized after the fact. Metrology labs have two jobs associated with the atomic time part of UTC (so, I'm ignoring the exciting leap second bits), one of which you can think of as a feed-forward operation and the other is sort of a feedback operation.

First, each lab tries to predict UTC by (perhaps) feeding steering corrections into their timescale. Doing so requires some understanding of the latent drift and sensitivities in one's own timescale as well as some understanding about latent drift and sensitivities in the weighted average of the rest of the world (which, due to it being a huge average, is hoped to be zero). Each month (or so), each lab transmits their measured offsets between the local realization of UTC and a timescale shared by geostationary satellite to the BIPM. Then, BIPM produces something like a feed-back operation: they publish a table of computed corrections to UTC(x) for the previous month. So, if you really really cared, you would log a list of events occurring at, say, UTC(NIST) time epochs, and then a month later adjust those time tags each by some number of picoseconds. Life is rough.

Anyway, back to WWVB. It ought to be stressed that the red/white towers visible in photographs of the antenna (e.g.) are not, in fact, the driven antenna itself. The driven antenna (we're talking ~50 kW) is suspended between four of those towers and is really working together with the four horizontally stretched feed lines that form a sort of "top hat" with respect to the earth. Forgetting about the fact that a quarter-wave dipole at 60kHz would be something like 1 km tall, I think (by which I mean: I've been told that) the shape of the top-hat is a better "mode match" into the ionosphere, which greatly aids in the total range of the antenna array. I've also been told that the soil and ground-water properties around Ft. Collins are better than average for this sort of ionosphere-earth waveguide transmission mode. The technology was derived wholesale from submarine communication operations invented by the Navy (again, I think).

And, of course, there are two of these antennas (so, eight red/white supporting towers). The two are driven with a particular phase/amplitude match that augments the signal in the east-west direction, where the extent of the country is largest. The electric field in the vicinity of the antennas is immense, something like 50 or 100 V/m. The working transmitter vacuum tubes, so elegantly displayed behind a glass cabinet, look like they are on loan from the Doctor Who prop department.
posted by fatllama at 6:58 AM on July 6, 2013 [12 favorites]


Wow, fantastic comment, fatllama. That's the sort of thing you were hoping for, eh, theartandsound?
posted by Ivan Fyodorovich at 7:21 AM on July 6, 2013


We had a shortwave radio in the 60's.
I remember tuning in to WWV a lot, but I'm not sure I knew about WWVB.
And as to Most boring station ever., I prefer it to WWVA
posted by MtDewd at 7:38 AM on July 6, 2013


"... The working transmitter vacuum tubes, so elegantly displayed behind a glass cabinet, look like they are on loan from the Doctor Who prop department."
posted by fatllama at 9:58 AM on July 6

High power RF final stage electron tubes, used in transmitters which are not entirely solid state, don't even look like vacuum tubes, what with their massive copper and aluminum forced air cooling heat sinks and ceramic envelopes. But they're nothing to the immense water/steam cooled tubes built by Powell Crosley in 1933, when he increased power at his WLW Chicago AM station, from 50 KW, to 500 KW.
"On March 22, 1922,[1] Crosley and his Crosley Broadcasting Corporation began broadcasting with the new callsign WLW and 50 watts of power. Crosley was a fanatic about the new broadcasting technology, and continually increased his station's capability. The power went up to 500 watts in September 1922, 1000 watts in May 1924, and in January 1925 WLW was the first broadcasting station at the 5000 watt level. On October 4, 1928, the station increased its power to 50 kilowatts.[1] Again it was the first station at this power level, which still is the maximum power currently allowed for any AM station in the United States.

At 50 kilowatts, WLW was heard easily over a wide area, from New York to Florida. But Crosley still wasn't satisfied. In 1933 he obtained a construction permit from the Federal Radio Commission for a 500 kilowatt superstation, and he spent some $500,000 (at least $17 million in 2010 dollars using a CPI conversion factor of 0.13[9]) building the transmitter and antenna.

It was the first large amplifier used in the United States for public domestic radio broadcasting and was in operation between 1934 and 1939. It was an experimental amplifier and was driven by the radio station's regular 50 kW transmitter. It operated in class C with high-level plate modulation. The amplifier required a dedicated 33 kV electrical substation and a large pond complete with fountains for cooling. It operated with a power input of about 750 kW (plus another 400 kW of audio for the modulator) and its output was 500 kW."
From 1933 to 1939, WLW (still called "The Big One" and prominently marked near the center of the dial on old Crosley AM consumer receivers) was regularly heard during night hours in Chicago, in far away parts of Australia, usually by skywave bounce, but reportedly, occasionally by ground wave!
posted by paulsc at 7:58 AM on July 6, 2013 [3 favorites]


Upon a bus-ride reflection, I should correct my comment above: I'm almost positive that the radio station does not include any masers in its local timescale, only cesium beams and maybe rubidium cells. Masers are ~5x more costly than cesium beams and the radio station just doesn't require a maser's short-term stability because such stability can't be easily disseminated over the long- or short-wave protocols.

Internet time transfer (e.g. NTP) is another fun subject, and given that I'm currently learning a lot about it (at NIST) I'd welcome technical questions or concerns about it via private memail. I'm not sufficiently knowledgable about it yet to say anything interesting publicly, never mind that doing so would probably annoy some excellent colleagues who built the thing.

However, as a fun trick for this crowd, I'll mention that NTP is just one of the internet time protocols supported by NIST. NTP is sophisticated enough that one really needs client software to use it. However, simpler protocols, referred to as TIME and DAYTIME are more easily accessed manually by humans. Just open up a terminal and type: "telnet time.nist.gov 13". The details of the returned string are documented here. All told, NIST replies to billions of requests per week for NTP and the other protocols.
posted by fatllama at 8:08 AM on July 6, 2013 [1 favorite]


Aren't the GPS satellites synced to the NIST clock, too? I mean, I know that "synced" is confusing in this context given that the atomic clocks aboard the GPS satellites drift because of general relativity. But their standard is the NIST standard, right?

No, it's the GPS standard. GPS Time Zero was set at 0000UTC 06-Jan-1980 and has counted seconds since then. GPS Time does not have leap seconds, so currently, GPS is currently 16 seconds ahead for UTC. As I typed this, it is week 1747 and second 572203 on the GPS clock, and the date is Cycle 1 Week 0723 Day 6.

GPS sats are synced to the GPS master clocks at the US Naval Observatory. When they were started, they were synced to UTC-NIST, but it's fundamentally a distinct timescale. The GPS sats have their own master clocks on board -- the earlier sats were cesium clocks, the Block II-R and later had rubidium and cesium clocks onboard. Some also have laser retroreflectors to allow direct measurement of the distance to the sat as a check. They are set to the GPS master clocks before launch, and then free run. They're monitored by comparing to other sats and the master clocks on the ground.

The LORAN clock was time zero at 0000UTC 01-Jan-1958, and continues from there. It's currently 25 seconds ahead of UTC.

TAI, the canonical timescale without leap seconds, was defined earlier, when UTC itself was defined. TAI is always ahead of GPS by 19 seconds, and of LORAN by 28 seconds, and is currently ahead of UTC by 35 seconds

Conversion between GPS/LORAN/TAI is simple math. Conversion of those to UTC is also simple math, except there's no direct way to derive what to add/subtract, because it's based on observation.

UTC is a continuation of GMT, also based on observation of the earth's rotation. The reason for the difficulty is the definition of one day is one rotation of the earth, not a given number of seconds. So, a TAI based clock will drift against the day over time. Since GMT was defined by the earth's rotation, so it didn't drift, it just effectively made each second a *tiny* bit longer or shorter to match. UTC is designed with the conflicting goals of maintaining a consistent second and a consistent day. This cant be done directly, so UTC fudges the day by, every so often, making a given day one second longer (or shorter, but I don't think that's ever happened.)

So, UTC is a discontinuous timescale, but it keeps Noon and Midnight where we want them. Thus, it's the civil clock.
posted by eriko at 8:08 AM on July 6, 2013 [4 favorites]


Thanks!
posted by Ivan Fyodorovich at 8:24 AM on July 6, 2013


My understanding of GPS time is that it is a precision versus accuracy issue. Since GPS location depends on receiving time signals and then doing the math on the time offsets from the different signals, GPS time needs to be very precise. Every second needs to be an exact second, the exact same length of time as every other second it has sent out. Kind of a really precise metronome. But it doesn't really care what day it is, so GPS doesn't need to include messy things like leap seconds and whatnot. As eriko says, it's not meant to be a time source. It's just convenient and good enough for most purposes.
posted by gjc at 8:37 AM on July 6, 2013


« Older Audio recordings of 1964 interviews with Civil...   |   RIP: Randy Udall Newer »


This thread has been archived and is closed to new comments