[time-nuts] Thunderbolt precise positions needed

[Tom Van Baak]

> With a good (geodetic/survey) quality antenna I get my location to
> within a few inches.  With a conical timing antenna,  I get around
> 8 inches of error.  With a cheap patch antenna,  around a foot.

HI Mark,

A couple more thoughts.

I still question the value of all this for a cheap receiver such
as a Thunderbolt.

Before you get too far measuring your sub-meter position, or
putting code into LH, it would be helpful to know what effect,
if any, this has on the actual performance of the GPSDO.

The plot I'm looking for is how the time/frequency output degrades
as a function of error in fixed position. I'm not convinced an error
of a couple of cm, or even a couple of meters is going to make a
real difference in the stability of the 1 PPS or the 10 MHz output.

The question is how far off does the surveyed position have to
be before the effects of position error rise above the sum of
all the other normal sources of bias and noise in the GPSDO?

True, these issues are important for high-end, dual-band GPS
receivers, with choke-ring antennas, clear views of the skies,
and post-processing, etc. But I question where the line is for
your average GPS timing receiver, such as a TBolt.

Have you or anyone measured the short- or long-term output
accuracy (1 PPS) and stability (10 MHz) as a function of the
position error (say, from 10 cm 10 m)? For a TBolt, or M12
or any other time-nut class of GPSDO?

This might also shed light on why the TBolt defaults to (only) a 
2,000 second survey. And why the TSIP LLA initial position is
single- not double-precision.

/tvb