This paper describes an iterative process of surface improvements made to the central 1.7m zone of a 2.5 metre
hyperbolic reflector constructed of cast and machined aluminium. Throughout all stages of the process, the mirror
surface was measured using a laser tracker, with initial maps taken by scanning the tracker target over the surface at low
spatial resolution. While the overall RMS of the full surface was in excess of 200 microns, the central area of interest
was in excess of 60 microns. The final goal of the program was to achieve 40 microns or better in this central area.
Surface maps showed a major low area on the mirror surface covering many tens of square centimeters, plus several
smaller high spots. The high spots were removed progressively by sanding with an orbital sander. Frequent pauses were made to take repeat surface measurements with the tracker.
A total of 22 grinding iterations were interspersed with tracker measurements at differing spatial resolutions, allowing
the RMS surface error to be reduced from 63 to 35 microns best measurement. Sanding periods lasted from 15 seconds
to 4 minutes at each sanding spot, while tracker measurements took approximately 15-20 minutes to acquire from 600
data points (low spatial frequency) to 6800 points at high resolution.
We present details of the surface improvement program with emphasis on the assurance of metrology integrity. We
discuss data fitting to the desired hyperbolic shape, sampling strategies, identification of sanding zones, and tracker
performance outside of operating environment specifications.