[time-nuts] Features of a Precision Clock?

[Magnus Danielson]

From: "Poul-Henning Kamp" <phk at phk.freebsd.dk>
Subject: Re: [time-nuts] Features of a Precision Clock?
Date: Sun, 08 Oct 2006 21:12:16 +0000
Message-ID: <46535.1160341936 at critter.freebsd.dk>

> In message <000301c6eb1d$5cdd8c40$0900a8c0 at cyrus>, "Bill Hawkins" writes:
> 
> >Since control is not precise, the subject is probably off
> >topic for this list.
> 
> It all depends on your integration time.
> 
> I was told from Stockholm that they hadn't lost track of a single
> cycle in 50 some years, so that's:
> 
> 	20msec / 50 y = 12.675e-12

You rarely need that kind of stability for tau = 50Y.

> As far as I know, no other frequency reference can do that.
> 
> Obviously the numbers are highly suspect, because he didn't
> know what they measured against...

Not counting the times we have been WITHOUT cycles at all I assume? We had a
large outage in the 80thies. I remember it clearly as I just came out of the
subway when it all shut down.

I don't think they run high in frequency to come back cycle-wise, do you?

Outage isn't common, even if my office have experienced it a little too often,
and certain other areas (Kista) also have experienced it my home have been
spared most of the times. The Kista incidents have been the result of lack of
redundancy, which is a bit annoying since it is such an IT-industry
concentrated place.

Cheers,
Magnus