In summary, our results are consistent with the hypothesis that the corrector was fabricated and assembled quite well. The 2m difference between the measured and predicted focal lengths is a slight focal change caused by random manufacturing variations of the parameters of the optics within the specified tolerances. Image quality is excellent, as far as we are able to tell, with intrinsic image quality in perfect seeing of .5" or better in the center of the field.

Table 3 gives the final OFAD to use for the PFCCD in the default location with the correct window and filter thicknesses. It is straightforward to extrapolate these results to determine the focal length and estimate image quality for other configurations by using the offsets from the B focal lengths in Table 4 and the -7.67 mm/mm relationship between e.f.l. and BFE. These changes should not have a significant effect on d3 and d5.

The image plane of the PFCCD in its proper configuration is achromatic even though there is a 3.7mm shift in focal length from U to I. This is a manifestation of chromatic difference in distortion, not change in focal plane. No focus change with color is required if the filters are all the standard thickness and composition. In Table 2 the PF Camera has a 4.4mm focal length shift over the same range, of which .5mm is a color shift in the focal plane caused by the filter and the window being thinner than the design calls for. With filters of differing thicknesses and composition the focal length shifts will be different from those shown here. The proper values are easy to compute if filter thicknesses and indices of refraction are known.

It is clear from Table 4 that the image plane of the PFCCD should be moved to a better location. Installing a spacer 2.1mm thick will keep the BFE to under 1mm with any filter in current use and not cause significant image degradation, though the focal length will obviously change as filter thickness is varied. This solution to the back focus problem is reliable and easy to implement. It will significantly improve image quality and will provide a new fixed focal plane location which can be used to attempt a more precise determination of the system's focal length than we have been able to obtain so far.

Moving the detector farther from the corrector in this way will change the telescope's focal length to a value very near to the optimum for filters 7.5mm thick. There will be a very slight focal shift with color. This shift will have the opposite sign of the shift in the photographic camera because the PFCCD has too much refractive material in the beam rather than too little, as is the case from the PF camera.



The author would like to thank members of the CTIO staff, especially Daniel Maturana, John Filhaber, Gabriel Pérez, Nick Suntzeff and Alistair Walker for their invaluable help in accumulating and presenting the data given here.