[LEAPSECS] preprint about timekeeping for neutrino experiment

[Warner Losh]

On Sep 30, 2011, at 6:45 PM, Tom Van Baak wrote:


>>> And the clocks are not locked to a receiver, they are free but the offset

>>> is continuously monitored through those CV measurement.

>> 

>> Would you not lock the GPS (GNSS) receivers to the CS-clocks being compared?

>> 

>> Björn

> 

> It's pretty common with high-performance timing to NOT lock the

> local clock to GPS. There are several reasons for this. One is that

> you get the both the best stability of the undisturbed local clock

> plus a continuous log of the difference between the local clock

> and the external timing source (GPS). Locking an oscillator to

> GPS introduces its own set of noise. Locking implies real-time

> processing of the GPS signal (you miss a lot of performance

> that way).

> 

> The other reason is that they are at least using common view, if

> not other forms of post-processing, to greatly improve accuracy.

> There's always a trade-off between good GPS timing in real time

> or superb GPS timing if you wait days or weeks and crunch all

> the numbers.


It is a lot easier to post-process the data if there's no third party changing the frequency behind the scenes to keep things on time and on frequency.

Warner