[time-nuts] Measuring ADEV using TBolt-Tic tester
ws at Yahoo
warrensjmail-one at yahoo.com
Fri Oct 7 16:03:50 EDT 2011
It would be great to have a direct Tbolt driver on TimeLab.
Right now it is so much trouble and time to use it, it takes away of the
great real time capabilities of TimeLab.
According to Tom his really good Cs has an Flicker noise floor of almost 10
days using 4 ns rms for phase noise.
A really good Tbolt set up will hold around 2 ns of phase noise,
so I think that means the flicker noise floor of a Tbolt will surpass Tom's
Best Cs after 5 to 20 days.
>From what I've seen it takes a few hours to a day to do better than the more
I have a plot showing a Tbolt GPSDO doing better than 1e-13 (which seems
pretty typical of most Cs) over a day using a compensated LPRO.
John Miles jmiles at pop.net
> Very Important note, The above is NOT available directly from
> ADEV plots (at least not yet - Mark is a revision coming?),
> What one needs to do is to log the Tbolt's Freq and Phase data at the 1sec
> rate and then use that data with an external ADEV program such as Ulrich's
> Plotter or John's TimeLab.
> Note for TimeLab users. LH adds some extra comment data lines in it's log
> file that need to be manually removed before using the data file with some
> versions of TimeLab.
I had a 'native' Thunderbolt driver in the old TI.EXE program that I could
probably bring across to TimeLab without much trouble. It didn't work with
the COM port directly, but with a Heather TCP/IP server. That means that
it takes some extra work to set up since you have to run two programs, but
it also means that you can run LH and TimeLab on the same Thunderbolt at the
same time, with no need to monkey around with logfiles.
I'll have a look at it to see if I can bring that code back from cold
> . . .
> If you want even more accuracy, watch the PHASE change over a few days.
> This > can check your best Primary Cs standard's frequency.
Eventually a GPS clock will reach something like a "flicker floor" where the
ADEV trace flattens out and stops decreasing over time. I don't really know
where the flicker floor is on a Thunderbolt-class clock versus where it will
be on a local cesium standard. The sigma-tau statistics for the more common
HP 5061-era clocks were never really specified for periods more than a day,
while at the same time, there don't seem to have been many long-term tests
run on Thunderbolts and other GPS clocks to see where their performance
flattens out. It would be good to see some more long-term trials.
Said, do you have data for the long-term (multi-week) floors of the various
Jackson Labs GPSDOs?
Intuitively, I don't believe a GPSDO can outperform an HP 5071A-era clock
over periods greater than a few hours. But it may be reasonable to
benchmark 5061A-class standards with a good GPSDO setup. We really need
some more data from trials lasting multiple weeks.
More information about the time-nuts