[time-nuts] Datum FTS 4065B Plot
magnus at rubidium.dyndns.org
Wed May 14 20:07:27 EDT 2008
From: "Tom Van Baak" <tvb at LeapSecond.com>
Subject: Re: [time-nuts] Datum FTS 4065B Plot
Date: Wed, 14 May 2008 14:49:46 -0700
Message-ID: <007b01c8b60c$6ffe03c0$0800a8c0 at pc52>
> > If you look at the plot I attached as a .pdf you can see that the current slope
> > looks like a nice straight line covering almost 4 days so temperature is not a
> > factor. The only factor I can think of is magnetic field. I guess the next
> > step is to rotate the box 90 degrees and see what happens unless there's some
> > other suggestions.
> Since you are plotting phase (not frequency) the nice straight
> line you see simply means you have a frequency offset. This
> is normal and expected with any standard compared to GPS.
> What may be temperature effects are the little wiggles above and
> below the line. You can almost see they form a diurnal pattern.
> If you subtract the frequency offset before you plot, you'll get
> phase residuals, which will reveal patterns in greater detail.
> There are Excel functions that will help you fit lines to the data
> and remove trends, etc. Email me your XLS file offline and I'll
> show you.
A little software tip would be to try out the fityk software. I use it for
various purposes, and can fit curves for a number of functions while also
allowing for some simpler form of dataprocessing.
More information about the time-nuts