[time-nuts] Testing 2 Cesiums & GPS

[Brooke Clarke]

Hi John:

The subject of the FTS patent number that's stamped on my 4060/S24 is 
the micro controller program that finds the correct peak.  See:
http://www.pacificsites.com/~brooke/FTS4060.shtml#C
It sweeps the control voltage and looks for a peak that has two adjacent 
valleys that are about the same magnitude.  If you look at a wrong 
adjacent peak you'll see that the valley nearest the correct peak is 
deeper than the valley away from the correct peak.  This morning I'll 
reread the manual to see if I can come up with a problem that would 
allow the lock LED to be on and yet leave the OCXO free running.

The other thing I've learned about Cesium standards that I had not 
realized is that you can turn them on and they are on frequency.  In the 
case of my s/n 1227 It's been off for over 2 years, yet turned on in the 
E-13 area.  This may relate to the difference in magnetic field where it 
was calibrated (many states away) and my local magnetic field.  My hope 
would be that if I calibrate it and leave it alone it would come right 
back to where it was when powered on.

I'm shopping for a function generator that has a BNC output connector 
and very fine frequency setting capability and hopefully is small (not a 
big rack unit like the HP 3325) to see how well the Zeeman C field 
adjustment works.  Also for other uses it would be nice to have the 
ability to input an audio frequency and be able to shift it up in 
frequency by a controllable amount.  The Berkeley Neucelonics 625 looks 
nice in this regard.

Can you add to your web pages diagrams of the setup you are using and 
instrument settings?  I agree that it's not a trivial thing to make a 
measurement good to 14 digits.  A friend and metrology expert pointed 
out to me that time (or frequency) is the thing that can be measured 
more precisely than any other physical quantity.  For example the very 
best voltmeter (HP 3458) is a 9 digit instrument.  So now most other 
physical quantities are measured by somehow getting the thing to be 
measured converted into a time or frequency.

Have Fun,

Brooke Clarke, N6GCE

-- 
w/Java http://www.PRC68.com
w/o Java http://www.pacificsites.com/~brooke/PRC68COM.shtml
http://www.precisionclock.com



John Ackermann N8UR wrote:

> Brooke Clarke wrote:
>
>> Something is very wrong with s/n 1013.  But the green lock light is one.
>> When I turn off the modulation and open the loop, sweeping the 
>> control voltage produces a definite peak.  If the true peak was 
>> outside the control voltage range how much frequency error would 
>> there be?
>>
> Hi Brooke --
>
> When Corby Dawson walked me through firing up my 5061A, he had me 
> first do a rough check with a frequency counter to look for 5.000 000 
> MHz because if the C field was far enough off you could get a lock on 
> a secondary peak (there are seven total, with the two nearest the real 
> peak being the strongest) which would result in a frequency error of 
> about 0.25Hz .  It looks like your error is much less than that, but 
> it might not hurt to double check and make sure you're on the right peak.
>
> By the way -- I've been doing some runs of the 5061A versus Z3801A and 
> was quite pleased with the offset I apparently had -- after two days, 
> it was statistically flat (a few parts in 10e14).  But then last night 
> I redid the rack mount which I'm sure jarred the unit a bit, adjusted 
> the loop gain as after several days of run time it had drifted up from 
> the nominal 40 to 50, and switched to the long time constant.  I also 
> adjusted the trigger point on the counter.  Now, a 12 hour run is 
> showing about -1x10e-12.  Not sure which of the changes caused that, 
> but it's certainly clear to me now that this is a very touchy business!
>
> I'm now dumping all the plots I generate (using the stable-stats.pl 
> tool I mentioned a week or so ago) into 
> http://www.febo.com/time-freq/plots/ if you're interested.  Not a lot 
> there now, but I hope to soon be doing some simultaneous 
> intercomparisons of the Cs, Rb, and GPSDOs that'll be plotted there.
>
> John
>
> _______________________________________________
> time-nuts mailing list
> time-nuts at febo.com
> https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
>
>