[time-nuts] Racal-Dana 1992 switches

Bruce Griffiths bruce.griffiths at xtra.co.nz
Fri Mar 28 07:15:26 UTC 2008


Charles S. Osborne wrote:
> Now the real question is... is there a clever way to make the Racal 1992
> readout the difference in µHz between two GPS disciplined oscillators? My
> only other counter, an HP5384A, is offscale at 10.000 000.000 MHz . I'm
> referencing the counter with one Lucent RFTG-m-XO and clocking a Lucent
> RFTGm-II-XO. Things are working well enough to be beyond my counter's
> ability to see any jitter. The Racal says nanosecond time interval counter,
> so I bet there's a way to subtract and increase the resolution similar to an
> HP53131?
>
> tnx,
> Charles
> K4CSO
> Duluth, GA
>   
Charles

One way to achieve microHertz resolution at 10MHz is to use a dual mixer 
time difference system or a variant thereof.

The output from each GPDSDO is mixed with the same low noise offset source.
This produces a pair of beat frequencies (one from each mixer) outputs.

The zero crossing from one beat frequency output is used connected start 
a time interval counter whilst the zero crossing of the other beat 
frequency output is used to stop the time interval counter.

The trick is to amplify the low slew rate beat frequency signal in such 
a way that the time interval counter sees 2 high slew rate low jitter 
signals.
Just connecting the low pass filtered beat frequency outputs to the 
counter inputs isnt effective as the measured timing jitter will 
inevitably be much greater than the actual zero crossing jitter in the 
beat frequency signal itself.
A resolution approaching 1E-14/sec is possible with this technique.

Another approach is to use a series of cascaded frequency difference 
multipliers, this is the approach used by Quartzlock in their precision 
frequency comparators.
This approach can also have a resolution approaching 1E-14/sec.

If you are willing to wait a little longer to see microHertz frequency 
differences then an analog phase comparator can be used.
A resolution approaching 1E-11/sec is possible with this technique if a 
high resolution ADC (eg  a high res DVM) is connected to the phase 
detector output.

Bruce




More information about the Time-nuts_lists.febo.com mailing list