The Giant Magellan Telescope (GMT) primary mirror consists of seven 8.4 m light-weight honeycomb mirrors that are being manufactured at the Richard F. Caris Mirror Lab (RFCML), University of Arizona. In order to manufacture the largest and most aspheric astronomical mirrors various high precision fabrication technologies have been developed, researched and implemented at the RFCML. The unique 8.4 m (in mirror diameter) capacity fabrication facilities are fully equipped with large optical generator (LOG), large polishing machine (LPM), stressed lap, rigid conformal lap (RC lap) and their process simulation/optimization intelligence called MATRIX. While the core capability and key manufacturing technologies have been well demonstrated by completing the first GMT off-axis segment, there have been significant hardware and software level improvements in order to improve and enhance the GMT primary mirror manufacturing efficiency. The new and improved manufacturing technology plays a key role to realize GMT, the next generation extremely large telescope enabling new science and discoveries, with high fabrication efficiency and confidence.

Precision optics can be accurately fabricated by computer controlled optical surfacing (CCOS) that uses well characterized polishing tools driven by numerically controlled machines. The CCOS process is optimized to vary the dwell time of the tool on the workpiece according to the desired removal and the calibrated tool influence function (TIF), which is the shape of the wear function by the tool. This study investigates four major topics to improve current CCOS processes, and provides new solutions and approaches for the next generation CCOS processes.The first topic is to develop a tool for highly aspheric optics fabrication. Both the TIF stability and surface finish rely on the tool maintaining intimate contact with the workpiece. Rigid tools smooth the surface, but do not maintain intimate contacts for aspheric surfaces. Flexible tools conform to the surface, but lack smoothing. A rigid conformal (RC) lap using a visco-elastic non-Newtonian medium was developed. It conforms to the aspheric shape, yet maintains stability to provide natural smoothing.The second topic is a smoothing model for the RC lap. The smoothing naturally removes mid-to-high frequency errors while a large tool runs over the workpiece to remove low frequency errors efficiently. The CCOS process convergence rate can be significantly improved by predicting the smoothing effects. A parametric smoothing model was introduced and verified.The third topic is establishing a TIF model to represent measured TIFs. While the linear Preston's model works for most cases, non-linear removal behavior as the tool overhangs the workpiece edge introduces a difficulty in modeling. A parametric model for the edge TIFs was introduced and demonstrated. Various TIFs based on the model are provided as a library.The last topic is an enhanced process optimization technique. A non-sequential optimization technique using multiple TIFs was developed. Operating a CCOS with a small and well characterized TIF achieves excellent performance, but takes a long time. Sequential polishing runs using large and small tools can reduce this polishing time. The non-sequential approach performs multiple dwell time optimizations for the entire CCOS runs simultaneously. The actual runs will be sequential, but the optimization is comprehensive.