Pointing model update

Pointing data taken during first shift on 981213 are shown below :

The strong trend of daz with elevation (reflecting an error of 6" in the telescope model parameter #3 (the non-orthogonality of the major axes) is subsequently removed by running the data through FIT9.

The new model has expected scatters in (daz,del) of (1.2",1.4") and was installed during the evening of HST 981214, although not in time for use that night. Subsequent use on 981215 is shown below :

The improvement - observed rms scatters are now (1.7", 1.0") - is most satisfactory, and the residual linear trend of daz with azimuth would seem to lend itself to some solution.

It first suggests perhaps that the conversion of inclinometry data into a track model may actually suffer from baseline removal - the old q5-q1 problem.

Linear baseline removal (RLB) is applied to all channels of the inclinometry on the assumption that the behaviour of the antenna in quadrant 1 is the same as in q5. However, the record of the baseline slopes removed from each channel show some consistency from one run to the next, indicating that the observed change q5-q1 is not random. For instance, the q5-q1 differences for LY and RY (the components of yaw ), amount to approximately +1.5" and -1.0" respectively. It is possible then that the F1 term in the model is being unnecessarily corrected by -2.5" at large azimuth values. The best fitting straight line to the daz -vs- az data above shows a change of about -6" across the full azimuth range 0-450, but at least in the correct sense.

Note however, that F1 is not the only term affecting daz, and even the maximum corrections to F1 suggested here (2.5") must be multiplied by cos(el) to contribute to daz.

I shall experiment with removing the RLB process from the track model creation if further data shows this trend to persist.