[time-nuts] Interesting (meaningless?) measurement.

[Brian Alsop]

The standard deviation using the 5335 internal oscillator is 0.39 nS.

Regards
Brian

On 7/23/2013 11:32, Poul-Henning Kamp wrote:
> In message <51EE68CE.6090604 at nc.rr.com>, Brian Alsop writes:
>
>> Good, bad or meaningless?
>
> On grounds of principle I will say "probably meaningless" until I
> see you reproduce the same number with the counter running on the
> internal OCXO.
>
> That said, I've done the same thing, and found numbers of the
> same magnitude.
>
> For additional fun:  Play with the phase-angle setting on the
> generator :-)
>
I acquired an HP-8657B refurbished by a well regarded individual. 
Beautiful printout of unit's performance versus the test matrix.

One thing I was interested in is how well it "locks" to an external 
source.  The previous generator I had produced 5 -10 ns random jumps 
observable on a 'scope.  I've seen none of that with the 8657B.

The next question: So how good is the lock?

The experiment was to use my HP5335A to measure time difference between 
the 8656B and the Rb source locking it.  The HP5335A was locked to the 
same Rb source.  This might be pushing the 5335A's ability to reliably 
measure time differences in the 100 ps range.

The experiment produced a standard deviation of 0.27 ns at 10 MHz for a 
couple hour run.  Each data point was a 100 sample average.  Obviously 
the results have components from both the 5335A and 8656B.  Perhaps 
another measurement could be made with the internal high stability 
oscillator in the 5335A.

There is no statistically meaningful drift in mean value with time.

Good, bad or meaningless?
Brian


-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2012.0.2242 / Virus Database: 3204/6013 - Release Date: 07/23/13