9.9.1 CCD Orientation

The angles $\alpha$ and $\beta$ were calculated from the extracted elements of ${\bf A}$ as

$$\tan\alpha = -\frac{a_{21}}{a_{11}}$$ (9.30)

and

$$\tan\beta = +\frac{a_{12}}{a_{22}}.$$ (9.31)

Figure  shows the values of $\alpha$ and $\beta$ as function of a chronological image serial number for the 68 images. Vertical dashed lines demarcates the different nights. The observations come in sets of 4 images, one in each filter. These sets are labeled with the field name, with ``PG'' being PG1633+099, and ``SA'' being SA110-503.

The observation history is, that on night 1 there was no encoder on the rotator. After images with serial numbers 1 to 8 had been observed, the rotator was turned, and therefore it is expected that images with serial number 9 to 12 have a different orientation than the above. On night 2, an encoder was available.

It is seen, that for all the PG images, $\alpha$ and $\beta$ agree well. $\alpha = \beta$ is expected, since it means that the CCD is right-angled. However, for all the SA image, there is a systematic difference between $\alpha$ and $\beta$ of about $0.1^\circ$. This is probably due to the fact, that the 18 GSC stars in the SA field comes from two different Schmidt plates, with 9 stars from plate 0192, and 9 stars from plate 050T. For the PG field, all 12 stars come from one plate, 006K. This can be seen from the PLATE_ID column in the image.tab tables. The hardcopy documentation for finder actually warns against using data from more than one plate at the time, so it seems justified to exclude the SA data.

The PG data only is plotted in Figure . The horizontal dotted-dashed line shows the mean of these data of $+0.29^\circ$. The data points scatter around this value within $\pm 0.3^\circ$ (min-max). The step size of the encoder is $0.2^\circ$, which almost explains the scatter.