[GPS_Standard] VE2ZAZ

[Dave Platt]

Bert, VE2ZAZ wrote:
> Dave,
> 
> The good thing with your setup is that it is very educational. You go
> one step at a time and try to see the effects that the changes have.

Thanks!  That was one of the big attractions to putting together
a frequency standard from your design, rather than just buying
a used rubidium standard or a Thunderbolt or whatever.  This
way, I get to understand what's going on, and learn about the
practical factors which influence the results.

> Increasing the thermal resistance and the thermal mass will improve
> stability, no doubt about this. The LM78xx series of voltage regulators
> are pretty quiet, but their temp.co. (drift vs. temperature) is not the
> best.

Yup.  It might be worthwhile generating the two critical voltages
(DAC reference and coarse-adjust-pot reference) using precision
voltage reference chips rather than 78xx parts... the current demand
is probably low enough to allow this.

> People tend to under-estimate the amount of drift that an OCXO can
> manifest when powering up after a long period of inactivity. Two
> identical units will show different behaviors. You also have to take
> that into account.

Also a good point, and it could be tricky to try to compensate for
this in a way which didn't compromise longer-term accuracy.  Having
the frequency standard "achieve a lock" more quickly might simply
give a (misleading) cosmetic appearance that All Is Well, but leave
you with a "locked" reference which was still drifting unpredictably.

There's probably no substitute for sufficient thermal inertia,
plus sufficient warm-up and stabilization time.

> Unfortunately, I am not willing at this point to spend any more time spinning
> the firmware. My list of things to do is endless (and I am only referring to my
> electronics list, not my personal life one!) 

I fully understand and sympathize!  I've been trying for years to
figure out where to send that petition I've been working on... the
one for a 30-hour day and an extra day added to the weekend... but
nobody will admit to having jurisdiction to consider it :-(
> 
> Finding the right combination is the challenge! Using Montrol's built in stats
> window will help you find a more stable combination without having to revert to
> spreadsheets or such. Of course, the tool will not reveal anything on accuracy,
> though.

I'm afraid I'm a bit of a Linux bigot, so Montrol by itself would be a bit
difficult for me to apply (I might run it under WINE but that's rather
clumsy).  Currently, I'm just concentrating on capturing the data and
then using various tools of my own fumbling-together to get a view on
what's happening (hence, PERL and gnuplot).

> Allan Deviation studies can help measure short term stability and adjust the
> system for optimum stability, but you need an atomic reference better than what
> you are trying to measure in order to make it mean something. 

I'm hoping I can persuade my friend (with the new-used rubidium
standard) to arrange to compare it against a known-good standard
at SRI or some place like that... and then loan it to me for a few
weeks so I can benchmark my GPS/FLL box against it.  No sense doing
that until I get the thermal issues dealt with and the parameters
hand-tuned as best as I can... but after that it'd be nice to be able
to fine-tune things.

The standard is already rather better than I really need it to be,
but I'll keep pushing on it for a while and see how good I can
really make it.

Thanks again!

	Dave AE6EO