[time-nuts] TBolt and Z3801A performance data

[John Miles]

> Hi John,
> 
> On Sat, 2008-08-09 at 17:48 -0400, John Ackermann N8UR wrote:
> 
> > The longer tau readings are interesting mainly to see how the GPS loop
> > kicks in; tau below about 1000 seconds seem related to OCXO performance
> > on the TBolts, while the Z3801A seems to have a longer time constant,
> > and somewhat more gradual attack.  (By the way -- the TBolts were
> > running with the loop parameters they were set to when I got them).
> > 
> > John
> 
> Would the fact that the bump is that "big" indicate that a more
> "aggressive" loop parameter setup would improve those two individual
> units performance?

It would be an interesting thing to try.  The default loop bandwidth is 0.01 Hz, but since the peak of the hump is right there too, I'm not sure if GPS or the OCXO is responsible for it.  The OCXO could be relatively stable at a 100-second scale but dragged around inappropriately by the GPS receiver, or the OCXO could be relatively unstable and corrected too slowly by the GPS receiver.

It looks like HP uses a much lower loop bandwidth.  Even so, it's interesting that we see no hint of convergence between the two GPS-locked HP oscillators at longer timescales.  

It's unfortunate that this kind of test takes so long to run, because it's the sort of thing you need to repeat several times to make sure you're getting at the truth of the matter.  

-- john, KE5FX