[time-nuts] Re: Some MADEV Results

[Magnus Danielson]

Hi,

Yes, there is convergence issues the way that classical "three corner 
hat" works. This is because how noise into the squaring behaves, where 
cross-correlation and Grosslambert processing is better. See the works 
of Prof François Vernotte. Also, if there is a tiny systematic like hum 
going into squaring, it makes things more unsolveable and does not 
converge nicely.

Cheers,
Magnus

On 2024-08-24 16:24, Bob Camp via time-nuts wrote:
> Hi
>
> Cool !!!! Thanks for sharing.
>
> I think I would grab the best units from the first and third groups and run them against the units in the second group.
>
> The “three corner hat” / ensemble method does have its limits. The closer you can get to comparing similar devices, the better the results typically will be.
>
> Bob
>
>> On Aug 24, 2024, at 9:02 AM, Jim Schatzman via time-nuts <time-nuts at lists.febo.com> wrote:
>>
>> Time Nuts-
>>
>> I attached a few plots of ensemble MADEV results for three sets of oscillators, measured in my lab with the ensemble method, posted here for general information.
>>
>> a) FTS1050A and FTS1150A oscillators. The two FTS1150A units are the best and worst of this bunch. Note that FTS1150A1 has a BVA oscillator.
>>
>> b) HSO14 and OSA8600 oscillators.  It is evident that my testing is pushing the limits of what is measurable with the ensemble method for the HSO14 units. If only I had a few more...
>>
>> c) Cesium beam oscillators plus one 5065A rubidium. 5061B4U is the 5061B4 unit tested with the cesium beam turned off. 5061B6 is nominally working but the MADEV curves suggests that it is not locking well. I am still trying to determine why.
>>
>> Jim
>> <FTS1050-1150.png><HSO14-OSA.png><Cesiums3.png>_______________________________________________
>> time-nuts mailing list -- time-nuts at lists.febo.com
>> To unsubscribe send an email to time-nuts-leave at lists.febo.com
> _______________________________________________
> time-nuts mailing list -- time-nuts at lists.febo.com
> To unsubscribe send an email to time-nuts-leave at lists.febo.com