A rather obvious systematic variation of the azimuth residual as a function of elevation was seen in the data of UT20050307, and prompted this analysis of the year's data so far.

Recall: the current pointing model is based on SCUBA data of 15 December 2004

The 281 SCUBA pointing data taken so far this year (until SCUBA was removed from service for repair on 02 Feb) are plotted below:

Allowing for slight nightly variations in collimation, and for a slight trend between the elevation residuals and elevation (central panel), the data seem essentially unafflicted by systematics, and confirm the validity of the model (for SCUBA).

This year's pointing data for RxA, RxB and RxW_C are shown left-to-right below:

Each of them shows a strong systematic in the behaviour of azimuth residuals as a function of elevation (upper centre panel), with a bilinear function seeming necessary to fit the RxB data. As a result, the rms scatter in the azimuth residuals is somewhat greater for the heterodyne receivers (2.0" or 2.1") than that of SCUBA data in the same period (1.7"). For just January - the period that coincides more precisely with the SCUBA dataset - the RxA data are much fewer (24) and more ambiguous, but the (91) RxB data show the aforementioned systematic even more strikingly.

Formal fitting of the SCUBA data using TPOINT brings the rms residuals in elevation down from 2.6" to 2.4" (a 'tweak') - but do nothing to improve the azimuth performance (1.7"). TPOINT fits to the data for RxA, RxB, RxW yield the following:

                   --- RxA --      --- RxB ---       --- RxW --
       coeff       change  +-      change  +-        change  +-
 
  1     IA         +6.0   3.0      +3.8   2.2        -1.3   4.4
  2     IE        -63.4*  0.6      +8.2*  0.4       -35.3*  0.9
  3     NPAE      +15.9*  3.8     +13.1*  2.9        +5.1   5.1
  4     CA        -40.8*  4.6     -32.2*  3.4        -4.4   6.4
  5     AN         +1.3*  0.2      +0.7*  0.2        +2.1*  0.3
  6     AW         -0.2   0.2      +0.6   0.2        +0.1   0.3
  7     TF         -5.5*  0.9      -2.7*  0.6        -8.2*  1.3

        * - 4-sigma significance

Despite the qualitative similarity in the daz-vs-el systematic for the three receivers there is no obvious single adjustment set one could make to the pointing model that would rectify the systematic errors. IA, CA and NPAE are well-known for their interconnectedness: it is possible to adjust them together by quite large amounts and not influence the quality of the model. The only parameters with a degree of independence that show significant and significantly similar changes in all 3 fits are AN (mean of 3 suggested adjustments = 1.1+0.3) and TF (-4.3+1.4). TF is (ostensibly) the flexure of the secondary support structure and changing it would not only affect only the elevation residuals, but, as for AN, would also affect the SCUBA data. The apparent consistency to the suggested changes to these terms must be coincidental.

However, a difference between the heterodyne data and SCUBA data does exist, and it varies with elevation. The obvious difference between SCUBA and the heterodyne receivers is that the latter are in the cabin and the former not - so one is left wondering if there are cabin flexure terms unaccounted for ? and if so, if they are, in some sense, 'new'.