Diversions: How the US Atomic clock works.

I can't stand puzzles. A jigsaw puzzle that others are working on over a Christmas vacation, I might put in a token piece. Little wooded puzzles (like the one seen here), I refuse to touch. I recently decided that I do enjoy puzzles: I enjoy designing systems. Systems (like computer systems, a piece of software, a web application, etc.) have specific problems to solve, constraints (usually technological, often psychological, sometimes political as well). They are some of the more complex puzzles and I quite enjoy them.

One morning, I found myself wondering how the US Atomic clock works this morning. More specifically, the synchronization part.

I have a clock in my home that receives the signal from Denver / Boulder, CO and keeps its clock correct. I had never really thought about -how- that works, until this morning. How does it work?

I treated it as a problem that I might be designing a solution for. I started with some of the constraints and usages of the "system:"

  • The needed receiver electronics must be small
  • They must be cheap (cost of $20 / unit)
  • It doesn't require any connection (to power, internet, or other)
  • It requires the user to input their timezone (I couldn't remember if this was the case - did this just "figure out" your timezone?, I assumed that I had entered it at some point)

I started with my long-held belief that this was some sort of shortwave / HAM radio message. Shortwave can travel a long way, as I learned early in my life, but the delay could be a problem for synchronizing time. I did some mental calculations based on radio waves traveling at the speed of sound (which, for some reason, my brain has permanently lodged in my brain from physics class: 340 m/s) and figured that depending on a variety of factors, there might 10-12 seconds of delay in a best case scenario (don't get ahead of me on this one...). I've also experienced the slowdown of radio waves - talking on a phone via radio has horribly slow delays (it was the 80s, I was in South America, technology was different then).

So then, how might it work?

I thought of a few possible solutions:

  1. You measure decay of the signal to obtain adjustment / correction.
    This would require a signal of known strength (easy), an accurate measurement (hard - remember the "cheap" requirement), and the atmospheric interference to be negligible (I don't know, but I doubt this is true)
  2. Send multiple messages
    If each radio frequency behaves differently (to atmospheric or other interference that would slow it down), then from the delay on 2 signals coming along different frequencies, you might be able to accurate obtain decay information and therefore calculate the distance & correct for it.
    This again requires accurate measurement which is likely not cheap.

I was stumped. But it was a fun exercise.

Google, of course, had the answer (though not in Wikipedia, nor as easily found as I would have imagined - all about the Atomic clock, Cesium-133, etc. but little about the signal itself).

Radio waves travel not at the speed of sound, but as all electromagnetic radiation (microwaves, television waves, waves at wi-fi frequencies, and light), at the speed of light. It's a bit higher than the 340 m/s (300 Mm/s). So much for science & calculations done in the shower. BTW, the signals are broadcast at 60 kHz VLF.

Such a speed means that you simply send a radio signal with the current time and receive it - no delay and therefore, no calculation required.

The phone delay I remembered, I'll chalk up to it being old technology and the delay was likely in the intermediary electronic processing or the electronic signal rather than the "radio" portion.

While searching for the answer, I also stumbled across a variety of interesting items. The most interesting might be the adventures of a family who, armed with very precise atomic clocks, went up a mountain to compare the speed of time at a higher elevation (Einstein's special theory of relativity suggests that they should - 22 ns over a weekend): http://www.leapsecond.com/great2005/

My point in relating this little story? Well, if I were Aesop, I might say that

But really, I just enjoy designing systems and I thought you might too.

Monday, June 30, 2008, 12:00 AM

tagged: atomicclock, design, ecuador, puzzles, radio, science, system

series: Diversions (2 other posts in this series)