2
$\begingroup$

I heard that it is important to have all oscillator clocks at different places be synchronized for example in a geographically distributed system such a telecommunication network. Of course to some error margin.

They fix GPS receivers in the base stations to synchronize the oscillators. I know GPS transmits time, as in our watches.

Could someone please explain what does this time of the day has to do with synchronizing oscillators?

Thanks a lot.

Improve this question
$\endgroup$
4
  • $\begingroup$ "Synchronization" can be used to mean many things in different contexts, even with respect to "GPS synchronization." I've seen GPS used to synchronize a system's absolute time reference, the frequency of an RF receiver's reference oscillator, and in some cases even the phase of an RF receiver's reference oscillator (although this can be hard with some PLL-based frequency synthesizers). Whether you would want/need to do any of these things differs upon the application. In telecom networks, time synchronization is a big deal, so that's likely the primary use. $\endgroup$
    – Jason R
    Nov 22, 2013 at 2:46
  • $\begingroup$ @JasonR Thanks. So my question is if two people in two different parts of the world agree on the time (time of the day by GPS) and only by that information can they make their respective oscillators have same phase? Assuming the frequency is agreed upon. $\endgroup$
    – triomphe
    Nov 22, 2013 at 19:30
  • 1
    $\begingroup$ Yes, it is possible. They could have the agreement that, for example, "our oscillators will run at 10 MHz and will be at zero phase at the top of every second." Then, the phase offset between them is proportional to the errors in their individual timebases. The idea is to lock their two time references to be as close together as possible, thus enabling phase and frequency synchronization as well. $\endgroup$
    – Jason R
    Nov 22, 2013 at 19:35
  • $\begingroup$ @JasonR Thanks a lot. So I guess that answers the doubts I had. $\endgroup$
    – triomphe
    Nov 23, 2013 at 14:31

1 Answer 1

Reset to default
3
$\begingroup$

Part of the GPS signal is a 1 pulse per second clock that can be used in distributed systems for timing synchronization.

This patent has some diagrams that might help, e.g.

enter image description here

Regarding how accurate it is, I'm not sure. This paper has the following graph... which seems pretty good to me.

enter image description here

Improve this answer
$\endgroup$
4
  • $\begingroup$ Thanks. So is one pulse per second enough to synchronize clocks at ns level? $\endgroup$
    – triomphe
    Nov 21, 2013 at 20:01
  • $\begingroup$ @triomphe: See the update. They suggest that the accuracy (I assume standard deviation of the timing) is about 1.3ns, which seems pretty good to me. $\endgroup$
    – Peter K. ♦
    Nov 21, 2013 at 20:07
  • $\begingroup$ Thanks again. I have another question regarding this. Suppose two people in different parts of the world agree on the time (the time of the day of GPS) to a very high accuracy. Can they make two oscillators that have the same phase? I mean not using this other GPS signal of 1s pulses. $\endgroup$
    – triomphe
    Nov 22, 2013 at 19:32
  • $\begingroup$ 'The same' phase might be hard; they can certainly make signals that are synchronized, provided they are both keying off the same signal / time stamp of the GPS. The issue will be can they recognize the right GPS time and start their oscillators over the same duration. That might cause an off-set. $\endgroup$
    – Peter K. ♦
    Nov 22, 2013 at 21:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.