The performance of JCMT, particularly as regards pointing accuracy when using
receivers with small beam-size, and as regards phase stability when used in
interferometry experiments with the CSO, should not be limited by the antenna
track. At the highest frequencies observed at JCMT (850GHz, 350
m)
beams are about 7
FWHM, so ensuring gain variations are kept below,
say, 10% requires global pointing stability of better than 1.3
rms in
the vector sum of both coordinates, i.e. better than 1.0
rms in
each coordinate.
There are a number of factors which affect the pointing - the levelness of
the track, the accuracy of the encoders, mechanical and thermal stability
of the telescope, the stability of the atmosphere, and so on. As a crude
approximation, given:
e
= the total pointing error,
e
= the mechanical error,
e
= the servo error (including the encoders),
e
= the thermal error,
e
= the error due to the track,
e
= the seeing,
then we might assume:
e
= e
+ e
+
e
+ e
+
e
Assuming on the best nights the seeing contribution is negligible,
then to achieve the current performance of around 1.6
implies
that all of the remaining terms are around 0.8
(assuming
they contribute equally). These values seem reasonable; MTPPN004
shows that the expected contribution from the track is it is now
is between 0.5
and 1
. A more detailed analysis of
the error budget calculation is given in Appendix A.
To get to 1
rms, we either need all of the above terms to be
0.5
, or if we think that some terms may be less tractable,
we need to make correspondingly larger improvements in the others.
We therefore conclude that the rms error contribution to the pointing from the az track should be of order 0.3".
Although this criterion could be applied over the full 360
\
range, we propose to take advantage of the fact that local pointing
corrections (i.e. fivepoints) can be made. These adjustments correct
for all contributions to the pointing error, including the local
track contribution. We therefore propose that the rms criterion of
0.3
rms be applied over any 30
arc of the sky, with
respect to a local zero-point.
Were the pointing errors due to the track randomly distributed, we could
simply characterise the performance requirement via the rms value.
We know however that this is not the case - in particular the effects
of the track segments are different when one or more wheels are
in the vicinity of a track joint.
To allow for this, we made the additional requirement that the peak-peak
pointing error due to the track should nowhere exceed 2
.