If (uaz,uel) are the means of the observed offsets as recorded in the last two columns of the TPOINT .pnt file then a TPOINT analysis of the data using the following command sequence

• fix
• use IE CA
• fit

will produce values of CA and IE that are related to the default values, CA_default, IE_default, as stored in the collimation files above, and to uaz and uel as follows:

• CA = CA_default + uaz
• IE = IE_default - uel

(See this example data analysis.)

These new values of CA,IE may be entered as the defaults for future improved performance, i.e. for residuals (uaz,uel) that should be more preciesly (0,0).

However, there is a sign convention that must be obeyed, as was revealed in a previous document. Simply put, if you point with JCMT/primary-FE and observe pointing offsets in azimuth and elevation of (uaz,uel) -- and you wish to rectify the pointing model so that future pointings (with any receiver) will appear at offsets of (0,0) -- then you should run the data through TPOINT, FIX all 7 parameters, USE IE CA, note the changes to these two collimation values, and change

*         CA   to    old_CA + d(CA)
*         IE   to    old_IE - d(IE)

Consider first the acquisition of a large pointing dataset obtained with receiver Rx in circumstances where no change to the model is expected. Analysis by TPOINT should suggest no changes to the values for CA, IE, meaning, in reality, no changes to the differences in collimations between Rx and the primary-FE, meaning in effect no changes to the collimations for Rx, since those for the primary-FE are maintained at (0,0). If non-zero offsets are seen you are faced with determining, separately (how ?) whether the problem lies with the model or the receiver (collimations).

Suppose instead that there is a need to redetermine the model using Rx under circumstances that ought to preserve the relative collimations. Suppose that

• the old receiver collimations were (X_Rx,Y_Rx)
• when processed with TPOINT the new data show non-zero changes in (CA,IE) of (uaz_data,uel_data), as well as to the other model parameters.

What to do with the collimations ? and the others parameters ?

• Changes to the other parameters reflect structural misalignments which should be common to all receivers and may be incorporated into the TPOINT pointing model irrespective of the RX used.
• The collimations for the primary-FE may be left as (0,0).
• Those for Rx must remain as (X_Rx,Y_Rx) . . .
• . . so the new values of CA,IE suggested by the TPOINT analysis may then be used as if they had been determined by use of primary-FE data; i.e. change CA to (CA + uaz_data), IE to (IE - uel_data).

If data are taken with Rx when both the model and the relative collimations may have changed then any resulting model may work (i.e. generate no further observational offsets) only for Rx itself. The relative collimations will need establishing separately and the system reverted eventually to one based on the primary-FE.