After roll changes, rolling force fluctuated heavily; uneven wear of the housing window liners was suspected of causing roll crossing, but no in-situ method existed for the large plane.
Solution: MC030-MAT300/48/0.5 (±0.5 arcsec error) with a 200 mm bridge mirror surveyed a 13×7 grid over the 2.4 m × 1.1 m window; software reconstructed least-squares flatness.
Result: the operator-side window showed 0.31 mm flatness with diagonal twist. After shimming at four points per the error map, flatness converged to 0.08 mm and rolling-force fluctuation fell 40%.




