[ntp:questions] Re: PPS signal that's not synchronized with second boundaries

John Pettitt jpp at cloudview.com
Thu Jun 29 01:55:10 UTC 2006


Maarten Wiltink wrote:
> "Hal Murray" <hmurray at suespammers.org> wrote in message
> news:_--dnTaJHdp4oALZnZ2dnUVZ_vKdnZ2d at megapath.net...
> 
>> Suppose I had something like a Rubidium oscillator that makes
>> a nice PPS signal, but it's not synchronized to a second boundary.
>>
>> Is there any way to take advantage of that?
> 
> If I've understood past news correctly, you can connect it as a
> PPS source and, once you have an idea of its offset, tinker/fudge
> it so it does appear synchronised to the second boundary.
> 
> The bad news is that my recollection of said past news is also that
> this didn't quite work then. I think the offset correction was
> applied at the wrong point, so it still appeared off-beat. But
> perhaps that got fixed in the meantime.
> 
> Groetjes,
> Maarten Wiltink
> 
> 


It's ok if you let ntp manage the pps but if you use the kernel pps (at least in FreeBSD) ntp
doesn't tell the kernel about the offset so it locks to the pps edge not the real boundary.  There
is a patch in an open bug for the nmea driver that would also work for the pps driver but nobody
seem interested in folding it in.

John




More information about the questions mailing list