[time-nuts] DDS, OCXO and ADEV

[Tom Van Baak]

Hi Vlad,

Here are some comments. Please read carefully.

Look at your data and notice that every reading is the same 3.505082408E-7, except one is 3.505082409E-7. That's not going to give you very good plots at all. Also you're not going to get much information out an ADEV tool with just 13 points.

It's not clear from your post what those readings are. They do not look like phase difference. They do not look like frequency, or frequency offset, or fractional frequency error. Instead I will guess they are period, since 1 / 3.505082408E-7 is 2852999.968 and that is close to the frequency you mentioned: 2853000 Hz.

My ADEV tools do not accept period or frequency data as input. This is by design. They expect a series of phase measurements. That is, a periodic sampling of the phase difference between the REF and the DUT.

There is no need to convert a number like 3.505082408E-7 to a number like 0.0000003505082409000000. Software knows how to read floating point numbers.

You can convert your period measurements (seconds) to frequency (Hz) by inverting them. Some tools will then take this frequency data as input.

You can convert period measurements (seconds) to fractional frequency error by inverting, subtracting 2853000 Hz, and dividing by 2853000 Hz. Some tools will take fractional frequency error as input.

Or, you can convert your period measurements to phase difference by subtracting 1/2853000 and accumulating (running sum) the phase error for each sample. There's a tool for you that does this, p2t, and the arguments are the nominal phase and the sample rate ( www.leapsecond.com/tools/ ).

In other words, you can get correct ADEV statistics from your gpib.dat file using:

    sed < gpib.dat "s/...//" | p2t 1/2853000 10 | adev1 10

I won't get into zero dead time measurements yet. One step at a time.

I'm not sure if my tools are the best for you. Command line tools work best when you understand what's going on and can keep track of units and such. Instead consider using John's TimeLab ( www.ke5fx.com/timelab/readme.htm ) which is a GUI with an extensive help system.

Your DDS vs OCXO plot looks suspect. You must remove that huge outlier point before you do any statistical analysis (like ADEV). The y-axis of that plot doesn't look right: all the grid lines are zero. Also there are no units.

The "Allan Variance Plot Y" you included is odd. I would not expect a y-axis scale to go from -31 to -45. But this is probably a side-effect of your feeding period readings into an ADEV tool instead of a phase difference time series.

If you want send me the full raw data file (more than 13 points), I'll make a set of plots that you can compare against.

/tvb

----- Original Message ----- 
From: "d0ct0r" <time at patoka.org>
To: "Discussion of precise time and frequency measurement" <time-nuts at febo.com>
Sent: Saturday, May 02, 2015 6:53 PM
Subject: [time-nuts] DDS, OCXO and ADEV


> 
> Hello,
> 
> 
> I would like to create some charts for ADEV for following setup:
> 
> HP5386A counter connected to External REF. (OCXO). The input of counter 
> connected to my DDS VFO. The frequency on VFO is 2853000 Hz.
> 
> Here is what I got from counter via GPIB:
> 
> S +3.505082408E-7
> S +3.505082408E-7
> S +3.505082408E-7
> S +3.505082408E-7
> S +3.505082408E-7
> S +3.505082408E-7
> S +3.505082408E-7
> S +3.505082409E-7
> S +3.505082408E-7
> S +3.505082409E-7
> S +3.505082408E-7
> S +3.505082408E-7
> S +3.505082408E-7
> 
> I converted it to following form:
> 
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082408000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 0.0000003505082409000000
> 
> Then I tried to use ADEV by tvb (long life to Tom!). See attached file 
> 1b.bmp
> 
> And I tried to create the charts for the collected data. See attached 
> file 1a.bmp.
> 
> But it seems I am doing something wrong. Any advises will be greatly 
> appreciated. Thanks !
> 
> 
> -- 
> WBW,
> 
> V.P.


--------------------------------------------------------------------------------


> _______________________________________________
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
> and follow the instructions there.