[ntp:questions] Onboard Local Oscillator Change Improvements

Maarten Wiltink maarten at kittensandcats.net
Mon Oct 8 15:25:55 UTC 2007

<elio.gi at gmail.com> wrote in message
news:1191836018.666680.188700 at o80g2000hse.googlegroups.com...

> in my setup I'm using a pc (Stratum 1) that synchronizes its local
> oscillator from a PPS derived by a GPS (Stratum 0).
> We known that pc use normally quartz oscillators, that are poor
> quality oscilators.
> I'm intended to replace this oscillator with an OXCO oscilattor (an
> high stability oscillator).

The poor quality of PC-class quartz oscillators would be evident mostly
in an unpredictable but constant frequency error. Perhaps 'better'
crystals display less aging, but frankly I doubt it.

As a result, a simple PC is not at all bad once it has determined its
'native' drift, and the remaining unpredictability depends very largely
on temperature variations. AC cycles, doors and windows opening,
varying CPU load. A lightly loaded system in a locked closet should be
almost as good as an OCXO-equipped one. Most of the time. And almost.

> My question is:
> an oscillator replacement would create some improvement in my setup?
> Consider that my Stratum 1 pc generate a pps derived from its local
> time and this PPS is used from other pc (Stratum 2 pc).

Obviously, a real OCXO should be more stable than a flimsy crystal in
a garden-variety mostly-stable room temperature environment. I'm not
sure generating a PPS signal would be significantly better than simply
NTP over Ethernet.

> What do you think about?
> Has someone made some measures about this topic?

Replacing the oscillator has been done. What would you like to measure?
What results would change your mind either way?

Maarten Wiltink

More information about the questions mailing list