Concerning Martin Robertson's problem:
Reproducing his conditions I can see possibly where the problem arises.
Assuming his cutoff elevation of 15 degs and his lat/lon then at 1632 GMT 2
May 2002:
SV's 27 and 9 are co-planar so effectively only one sat. and at that time
they are exactly overhead Newcastle.
SV's 18 and 1 are at the same azimuth; 315 deg
So are SV's 6 and 30 - 060 deg.
See attached polar plot.
Considering only horizontal accuracy :
Apart from 27 and 9 being together (don't know why)
they are exactly overhead so do not contribute anything.
18 and 1, being at the same azimuth, are producing nearly concentric range
circles,
when projected into the horizontal, as are 6 and 30.
Concentric range circles don't help each other - no angle of cut.
Effectively, therefore, there are only two range circles contributing to
the fix
in the horizontal plane. However, they intersect at not far off 90 deg
which is usually thought to be
optimum for a geometrically good fix. And there are six visible sats so no
problem solving for time.
Maybe there's something in his software that can't resolve so many concentric
range circles.
The basic problem is the very high cutoff elevation angle. GPS cover is
predicted
on the basis of a 5 deg mask angle (see SPS Performance Standard) and if I
substitute
5 deg another sat (SV2) becomes visible at a useful azimuth (245 degs) and
fixing reverts to normal.
If it needs a 15 deg cutoff then we're looking at a basic limitation of
RTK. GPS was never designed
for RTK so you're really out on your own if it doesn't work occasionally.
Best thing is to pray for Galileo to come in quickly!
Walter.
|