The European Organization for Nuclear Research, known as CERN, had a problem. In order to collect data on particles travelling close to the speed of light, it needed to time sync its large experiments such as ATLAS. Because the ring is 27 kilometers, there is no way that a signal generated in a control room can arrive in time to trigger an event that needs to happen at a precise time, down to nanoseconds or even picoseconds without some precise time synchronization.
Synchronizing many distant time sources has been a problem since communications and travel became fast enough for it to matter. The advent of railways and especially the telegraph meant that time in two different places could be compared and synchronized.
Young Albert Einstein, as a Swiss patent clerk, may have reviewed at least one if not more applications for time sync mechanisms.
A time sync mechanism named jointly for him and the mathematician Poincaré involved shooting beams of light between distant locations, and, knowing the precise distance and the speed of light, one could then accurately determine when the light left. That is, if both light source and receiver were stationary with respect to each other. If a clock is moving at a significant fraction of the speed of light, time synchronization between two clocks becomes difficult because they run at different apparent rates. It may be that thinking about this mechanism led Einstein to the theory of relativity.
In his early writings, Einstein seemed to consider time sync between clocks in relative motion to be impossible but never considered the alternative to light speed signal exchanges. Slow time synchronization involves synchronizing a clock at one location with a stationary clock and moving it to another and then synchronizing it with that clock. By transference, both clocks are synchronized to within the accuracy and drift of the clock that was moved. This phenomenon is similar to how longitude was determined once accurate shipboard clocks were invented. The clock could be sync’d with, say, Greenwich Mean Time in England, then as the ship moved, it would be compared to the time as told by the Sun. The difference in time would indicate how far East or West of the Prime Meridian the ship was.
Slow sync is also how you can sync two clocks in (special) relativistic states of motion with respect to one another because the third clock you move from one to another changes its state of motion from one to the other. Thus, it allows you to match the two clocks to the third.
But, what if you want to synchronize many different clocks that are all far apart and keep them synchronized?
One option is to use what’s called a disciplined oscillator approach. In this case, you distribute an analog wave, like a square wave or a sine wave or some more complex wave, to many different locations from a master oscillator. The oscillations are precise so that from trough to crest is always the same distance in time. You then provide what is called a Local Oscillator which is your local clock and you force it, electronically, to adjust itself periodically to match the rising voltages of the master oscillator. This utilizes an invention called a Phase Locked Loop. It also happens to be the way TVs and radios stay tuned to particular frequencies by “locking” the receiver to the frequency itself. Your time synchronization, in this case, will be as good as your oscillators.
Starting in the early 2000s, CERN began to develop this idea into an approach to time synchronization that was not exactly new, but by combining several popular time sync protocols, they aimed to achieve a complete synchronization solution. In doing so it wanted to go well beyond digital time sync protocols like Network Time Protocol; CERN was looking for precision below 1 nanosecond, down to about 100 picoseconds, that it needed to plumb the secrets of the universe.
The key to developing this mechanism, known as White Rabbit, included some well known technology, Ethernet, as well as some precise ways of measuring signals that propagate through a digital network. Unlike standard Ethernet, WR is deterministic, meaning that data packets always arrive at their destination in the same order at the same time. It also makes use of an underlying set of synchronization signals between switches and networking hardware based on the SyncE protocol. The SyncE protocol is essentially like the disciplined oscillator approach above, but it is paired with a digital time sharing system so that data can be passed through the Ethernet switches as well as timing signals. Another protocol called Precision Time Protocol keeps all the connected computers system clocks synced as well as precisely stamping the time when any piece of data arrives at any computer in the network.
All of this means that massive networks of sensors can be connected over kilometers all sharing a single timing signal that is precise enough to be able to know exactly, to within less than a billionth of a second, when a signal arrives at each element in that array of sensors.
This means that massive kilometers wide radio telescopes can be built using independent digital sensors rather than dish based approaches like the late Arecibo radio telescope in Puerto Rico.
Huge networks of, well, anything can be built this way from telecommunications systems for 5G to giant arrays of lasers that could destroy space debris.
But this made me wonder just what are the limits of time synchronization? How close can you get two clocks?
We know that atomic clocks can be built that have incredibly slow rates of drift. In fact, in the last few years the most accurate (ytterbium-)strontium lattice clocks were shown to be accurate to 3.5 ticks in 10 quintillion, meaning they would not lose or gain a second within the age of the universe.
This advancement has powerful implications towards the study of the universe. Back in 2004 I asked Kip Thorne why gravitational wave detectors measured space warping rather than time warping. He explained that we could not measure time that accurately. Now we have such precise clocks they may allow tests of general relativity that were not possible less than 20 years ago.
These clocks have already been used to try to detect violations in special and general relativity (without success) and whether fundamental constants of the universe can change over time (again, no dice), even to detect dark matter (still nothing).
In a recent experiment, the National Institute of Standards and Technology (NIST) compared their strontium lattice clock to an aluminum-ion clock and a ytterbium lattice clock located at different labs at NIST Boulder and JILA, a related institute known for their ultra-cold ion experiments.
These comparisons are so accurate that they exceed the current standard definition of the second as a unit of time! That is defined by the vibrations of the Cesium atom at 9 billion vibrations per second and was defined in 1967. In the case of NIST’s work, which are accurate to within 6 to 8 parts in a quintillion, the precision is so accurate, they must measure time accuracy in terms of the ratio of frequencies of the different atoms (ytterbium-strontium, ytterbium-aluminum, and aluminum-strontium). Because a frequency ratio has no units, it is independent of any standard and may soon become the new standard itself.
In order to achieve this precision, NIST scientists use a method developed in 1999 called an optical frequency comb.
This device uses a laser to generate extremely short pulses of light that appear as spikes at regularly spaced intervals, creating what looks like the teeth of a comb. Each pulse is of a different color or frequency, and the comb spans the entire visible spectrum. Researchers could line up one of the teeth of the laser light to the oscillations of an atom radiating in optical frequencies, lock the laser frequency to that line, and use sophisticated electronics to tally the transitions. This opened the door for creating optical atomic clocks.
No less impressive is the ability to synchronize these over the air. Over-the-air time synchronization using a two-way method eliminates effects from turbulence and vibration. These methods, part of the field of photonics, could mean synchronizing distant points without running fiber cables, and, as far as timing accuracy, it blows White Rabbit out of the water, with accuracies in the femto and even attoseconds, less time than it takes light to cross a molecule.
As in the days when a comparison of a mechanical clock time and the position of the Sun could tell you your longitude, so comparing two atomic clocks to great accuracy can tell you where you are as well. But in this case, so precisely that new GPS satellites might be able to give positions so accurate they could determine where on your body a GPS is located to within a centimeter.
They can also determine the strength of the Earth’s gravitational field just by how fast or slow they tick. Using the principle of Einstein’s general relativity, which says that clocks in a weaker gravitational field tick more quickly than those in a stronger one, the precise density of the Earth at a given location could be determined based on how fast one of these clocks ticks.
In other words, these clocks are so accurate they can measure spacetime itself, much the way that the ultra-accurate LIGO gravitational wave detector does for space alone.
But how accurate can we get?
It turns out that it all has to do with frequency. As you increase the frequency of your oscillator, in this case pulsed light from a laser, you can measure more and more accurate time. The strontium lattice clocks oscillate at 429 TeraHertz.
The smallest possible measurable time is simply related to the smallest possible measurable length and the fastest possible speed. The smallest measurable length turns out to be something called the Planck length, which is the size of the smallest possible black hole. The reason it is the smallest possible is illustrated by a simple thought experiment. If you had a laser that produced light such that the wavelength was the Planck length, it would have so much energy that it would create these tiny black holes. If you tried to make a laser with a smaller wavelength, it would have even more energy and so would make bigger black holes. Since you can’t see into a black hole, the Planck length turns out to be a limit. The frequency of our Planck wavelength laser is the inverse of the Planck time, which is the smallest resolvable time.
If we could find a substance that oscillated at this rate, we could measure time to within the accuracy of the Planck time.
It is difficult to think of anything that would oscillate this fast. Anything that did would likely be extremely unstable, but it does show that there is a lot of room to improve accuracy. To give a comparison, the tick of a Planck time clock is as accurate compared to the strontium lattice clock as the strontium lattice clock is compared to a clock that ticks once every four months!
Time synchronization is one of those unsexy areas of research that nevertheless could change the world. Much of our technology today rests on the premise that clocks cannot be precisely synchronized over large distances in real-time. Yet, when we do manage it, as with the Event Horizon telescope that gave us the first image of a Black Hole, it can truly change the world.