[time-nuts] low-g OCXO GPSDO
magnus at rubidium.dyndns.org
Thu Jun 5 14:51:50 EDT 2008
From: "Matt Ettus" <boyscout at gmail.com>
Subject: Re: [time-nuts] low-g OCXO GPSDO
Date: Thu, 5 Jun 2008 10:25:43 -0700
Message-ID: <ad9f78230806051025i22149aeci191f71bb7f7f602 at mail.gmail.com>
> > A "normal" OCXO would drift significantly when being turned around in any
> > direction.
> I've actually been wondering about what physical mechanism that causes
> this. I could understand how it could cause a phase shift, but I
> can't envision the cause of frequency shift. Does anyone know?
Anything that deforms the actual crystal will slightly change the acoustical
waveequation and thus the frequency. Crystals also deforms due to
electrostatics, which have been used by NIST for finegrained phase modulation.
The interested can make the exercise to model the crystal in detail by solving
the Schröder equations and the various stress mechanism and then how that
changes the acoustical waveequations. It's that "simple". :)
Actually, today I have been looking for articles relating to Kalman filter
estimation of exponential frequency drift. The traditional phase, frequency
and frequency drift does not match data very well. Curve-matching shows much
better aggrement with exponential frequency. However, normal curve-matching
does not match well with real time estimation. Most articles seems to use the
linear drift. Ah well.
More information about the time-nuts