J. Eur. Opt. Society-Rapid Publ. 21, 37( 2025) 73
Fig. 11.( a) Roughness profiles before and after SAG finishing,( b) roughness evolution across different finishing runs, and( c) surface condition of the channel surface before and after SAG finishing.
that P-V height is 16.2 lm and 17.8 lm for experiments and simulation, respectively, which seems to be in good agreement. Small variation is removal footprints, as seen in Figure 7b, and removal rate can be attributed to the eccentric wobbly motion and run out of the wheel caused by low rigidity of the wheel assembly and the jig.
Next, the measured surface of the workpiece was mapped against the design and the error map was obtained, shown in Figure 8. Feed moderation was applied to calculate the feed rate scheduling based on the error map and removal characteristics of the tool obtained from the influence function. The 9 lm NBD SAG pad was used on the wheel for form correction, due to high removal rate capability of the NBD pads.
Results of form correction are summarized in Table 2. Correction time, targeted correction and achieved P-V error for different corrective runs are given in the table. The target correction is the amount of correction targeted for a particular corrective run, which can be input in the CAM software ZephyrCAM. For example, the P-V error before a particular run is 50 lm and the target correction chosen is 25 %, then the expected P-V error after the corrective polishing will be( 50 � 0.25 50) = 37.5 lm. The complete 100 % correction in single run is usually not considered due to uncertainty of the grinding and polishing processes. To analyze the P-V error, a clear aperture of 97 mm was chosen and the remaining 4 mm length( 2 mm from both the edges) of the channel surface was cropped down. The reason behind leaving the 2 mm length near the edges was that the wheel tool was susceptible to abrasive pad tearing and rubber slippage when moving near the sharp edge of the channel. Therefore, the raster path was generated considering 2 mm under-hang from both sides near the edges. The P-V form error in the clear aperture was 42.9 lm. After four correction runs, the form error came down to 7.84 lm P-V from the initial error of 42.9 lm P-V in the clear aperture of 97 mm. Form error convergence over different runs can be clearly observed from Figure 9. The form error was a bit high near the edges of the channel. 2-D profiles taken along the center of the channel are plotted in Figure 10. The form error was found to be within 2 lm P-V along the center line.
For further surface finishing of the channel, different types of SAG pads were used sequentially. The roughness profiles before and after finishing are shown in Figure 11a. Surface roughness Ra across various finishing runs is plotted in Figure 11b. To start with, 9 lm NBDpadcouldtake down the roughness to 15 – 20 nm from the initial roughness
~ 1. 12 lm. Next, 3 lm RBD abrasive pad was used on the
finishing wheel. This could achieve a highly shiny surface with Ra ~ 3 nm. Enhanced surface micrographs before and after finishing are shown in Figure 11c, where the evolution from a severely pitted surface to a polished surface could be observed.