[time-nuts] Frequency Acuracy at short gate times
bruce.griffiths at xtra.co.nz
Thu Sep 17 10:22:06 UTC 2009
Martyn Smith wrote:
> Can anyone advise how I can measure the frequency accuracy of my 10
> MHz standard at gate times of 1 ms and 100 microseconds.
> I have the Wenzle program that calulates Allan Variance from phase
> noise, but don't know how reliable that is.
> The 1 second Allan Variance is about 1.2 x 10E-12 at 1 second and 3 x
> 10E-11 at 1 ms according to the Wenzel program .
> Any ideas to actually make a measurement.
> I have a HP3048A and SR620 timer counters.
You will need to mix the signal with an offset source to produce a beat
note of 1kHz or so.
If you then timestamp the beat note zero crossings (use the SR620 with
the beat frequency connected to START) and 10Mhz connected to STOP.
The resulatant phase measures can then be unwrapped for calculating ADEV.
Off course this requires an accurate and stable offset source.
Alternatively use a dual mixer scheme with both the source under test
and a stable 10mHz reference as inputs producing to 1kHz beat notes the
zero crossings of which are time stamped.
Since the offset source is common to both channels most of its phase
noise and instability cancels out.
In principle one could use a sound card based COSTAS receiver to measure
the phase differences between the 2 beat signals.
More information about the time-nuts