[time-nuts] Z3805A cooling requirements?

Magnus Danielson magnus at rubidium.dyndns.org
Sat Dec 22 18:14:59 UTC 2012


On 12/22/2012 06:44 PM, Volker Esper wrote:
> Thank you for your statement, I have to think about it.
>
> It seems to be clear, that ADEV/MDEV aren't the convenient tools for
> characterizing such an event. My goal was to answere the question asked
> in the beginning of this thread: Does a constant ventilation of the OCXO
> affect the deviation curves and if so in what manner? More precisely: do
> I have worse deviation due to ventilation?

Well, the deviation curves was only meant to separate various 
phase-noise forms, not systematic effects, even if they may be 
illustrated to some degree using the tool. However, since noise(s) and 
systematic effects have completely different properties in how 
confidence intervals builds, so they are better measured and treated 
separately before being put back together again. For longer taus the 
systematic effects dominates, so drift, temperature, supply voltage and 
pressure kicks in.

>  > ... phase/frequency/drift plots and FT of them might be more
>  > useful.
>
> So let's have a look at the time domain. The picture shows the PPS-TI of
> the Z3805 (blue: time difference between OCXO and GPS; red: EFC).
>
> In the middle of the diagram we recognize the reaction to the
> ventilation. Before that happens, the curve has a somewhat stable "high"
> frequency noise. After starting the ventilation this noise is extremely
> varying in amplitude. IMHO that might mean, that Allan deviation will be
> affected at the short term values of tau. This seems to coincide with
> the MDEV vs time curves of my diagrams.
>
> Agree?

Agree. I wonder if the forced convection is to strong. You don't want to 
hit the limits of the oven control-range. Wonder if this is a long-term 
effect and if even small adjustment to the setup can remove the bursts 
of noise you are seeing.

Cheers,
Magnus



More information about the time-nuts mailing list