Re: [Ntp] Timescales, leapseconds and smearing

[Kurt Roeckx <[e-mail]>]

On Tue, Dec 08, 2020 at 08:44:29AM -0800, Steve Allen wrote:
> On Tue 2020-12-08T10:31:04+0100 Miroslav Lichvar hath writ:
> > On Mon, Dec 07, 2020 at 11:27:35PM +0100, Kurt Roeckx wrote:
> > > The proposed draft has support for the first 4, and offsets
> > > between some of them. Note that it calls the NTP timescale
> > > UTC, which it's not.
> >
> > How do you suggest it should be called?
> 
> More generally, I think that names of the time scales in the document
> all need an introductory explanation of what they mean.
> 
> Look at monthly issues of BIPM Circular T and it becomes evident that
> there is no "UTC" which is available in real time.  Sources of time
> which can be used by an NTP server there are only
> UTC(pick a laboratory)
> 
> Note in the issues of Circular T that the Bulgarian Institute of
> Metrology (BIM) has had UTC-UTC(BIM) values around 12 microseconds
> during the entire year 2020.  Nevertheless, UTC(BIM) is the legal
> time in Bulgaria, and devices in Bulgaria might be constrained by
> regulations that require they use UTC(BIM) rather than another
> UTC which is in better agreement with the rest of the world.
> 
> The draft does not mention the source or authority for the values of
> UTC, UT1, TAI which are provided by NTP.  The draft almost certainly
> cannot specify particular sources of time values, and in the case of
> TAI there can be no source because TAI is not defined until next
> month when Circular T is published for this month.

Whatever value is used in NTP, it's never the true value, it's
always an estimate. The TAI value we would use is an estimate of
TAI with error bounds. If BIM is off by 12 us, and BIM is used as
a source for NTP, they can either compensate for that 12 us, or add
the 12 us in the error calculation.

If the 12 us difference is important, I suggest they use an
extention to distribute that 12 us.


Kurt