[time-nuts] 15 Seconds error...??

[mike cook]

Le 20/01/2012 09:37, Attila Kinali a écrit :
> On Thu, 19 Jan 2012 10:02:56 +0100
> mike cook<michael.cook at sfr.fr>  wrote:
>
>> Hmmmm (from a time serf).  Some labs provide UTC(k) but do not generate
>> a TA(k) , maybe as there not enough clocks available.  Some UTC(k)
>> providers have only one clock,.
> This somehow doesn't make sense. Given you have a clock that generates
> UTC(k). How can you not also generate TAI(k) at the same time?
> (modulo leap seconds of course)
I was just restating data from the BIPM annual report. It would not be 
possible to generate TAI as it a cooperative effort, but a good stab at 
TA could be made.
>
> > From a technical point of view, if I had to run a clock that generates
> UTC(k), i would let it run as TAI(k) and apply an offset to derive
> UTC(k) from it. And i would have thought that this is the way how the
> UTC(k) references are generated. Or am i missing something important?
For the  labs cited in the report; the implication is that, the sites 
are using GPS time transfer to discipline a local Cs reference with UTC 
. As TAI is not available to them, they could reconstruct a local TA(k) 
from the thus disciplined UTC(k) I guess, but it is not indicated in the 
report.
>
> 			Attila Kinali
>