The Accurate Method for Removing Misalignment Errors from Testing Rectangular Optics

Zhu Li, Xiaokun Wang, Li SongProceedings of the 2nd International Conference on Computer Science and Electronics Engineering (ICCSEE 2013) 2013 · DOI

Abstract

The limitations of current interferogram reduction software using Zernike polynomials in analyzing rectangular interferograms are illuminated. Zernike polynomials are no longer orthogonal over a rectangular region and lack explicit physical meanings in this context. A reasonable mathematical model for removing translation errors is established based on least squares fitting. The experiment is carried out with a 200mm x 75mm aspherical mirror by null-compensation. Misalignment errors can be eliminated from the original phase distribution exactly, allowing for an accurate surface map and providing a guarantee for super-precision fabrication.

Introduction

Modern optical systems, particularly Off-axis Three-Mirror Anastigmat (TMA) telescopes, increasingly utilize rectangular aspherical mirrors. These geometries are preferred for their ability to reduce system weight and eliminate central obscuration. However, the non-circular nature of these components presents significant challenges during interferometric testing. Standard reduction software typically employs Zernike polynomials to model and subtract misalignment errors—such as piston, tilt, and defocus—from the measured surface map. While Zernike polynomials are orthogonal over a circular unit disk, they lose this property over rectangular domains. This lack of orthogonality leads to mathematical coupling between modes, where "tilt" adjustments may erroneously influence "defocus" calculations, resulting in inaccurate surface characterization.

Methodology

To address the "orthogonality trap," a dedicated mathematical model for rectangular domains was developed. The positioning error $\epsilon(x,y)$ is defined by a four-term polynomial: $$\epsilon(x,y) = ax + by + c(x^2 + y^2) + d$$ where $a$ and $b$ represent tilts in the $x$ and $y$ directions, $c$ represents defocus (power), and $d$ is the piston offset.

Instead of projecting the data onto Zernike modes, the coefficients are determined using a Least Squares Fitting algorithm. This approach minimizes the sum of the squares of the residuals between the model and the discrete data points collected from the rectangular interferogram. By treating the rectangular data directly, the method ensures that alignment errors are decoupled, allowing for the precise extraction of the true surface figure.

Experimental Validation

The proposed method was validated using a 200mm × 75mm rectangular aspherical mirror. The test was conducted using a Zygo digital interferometer in a null-compensation configuration. In this setup, a compensator lens is placed between the interferometer and the asphere to produce a wavefront that matches the ideal aspheric surface, simplifying the detection of deviations.

Experimental setup for null-compensation testing of a rectangular aspheric mirror.

The raw phase data obtained from the interferometer was processed using both the standard MetroPro software and the custom least-squares model implemented in Matlab.

Results and Discussion

Comparative analysis revealed significant discrepancies between the standard Zernike-based approach and the rectangular model. For instance, the "Power" (defocus) coefficient calculated by the standard software was 0.038, whereas the proposed method yielded 0.071—a difference of nearly 87%.

Phase map of the rectangular mirror showing surface errors and alignment metrics.

The surface accuracy metrics also showed variations:

  • Standard Software: PV = $0.359\lambda$, RMS = $0.041\lambda$
  • Proposed Method: PV = $0.355\lambda$, RMS = $0.038\lambda$

While the numerical differences in RMS and PV appear small, they are critical in the context of super-precision fabrication. The coupling effects in standard software can mask or exaggerate specific surface features, potentially leading to incorrect corrective polishing steps.

Conclusion

This research demonstrates that standard Zernike-based interferogram reduction is insufficient for high-precision rectangular optics. By employing a direct least-squares fitting model tailored to the rectangular geometry, misalignment errors can be accurately decoupled and removed. This method provides the necessary accuracy for the fabrication of advanced off-axis optical systems and can be generalized to other non-circular geometries, such as hexagonal segments in large-scale telescopes.

Summary updated

Abstract

The limitations of current interferogram reduction software with Zernike polynomials in analyzing rectangular interferogram are illuminated. The Zernike polynomials are no longer orthogonal over an rectangular region and have no explicit physical meanings. The reasonable mathematical model for removing translation errors is established based on least squares fitting. The experiment is carried on with a 200mm75mm aspherical mirror by null-compensation, the misalignment errors can be eliminated from the original phase distribution exactly, thus the accurate surface map is gained, so it provides a guarantee for the superprecision fabrication.
BibTeX
@inproceedings{li2013accurate,
  author = {Zhu Li and Xiaokun Wang and Li Song},
  title = {{The Accurate Method for Removing Misalignment Errors from Testing Rectangular Optics}},
  booktitle = {Proceedings of the 2nd International Conference on Computer Science and Electronics Engineering (ICCSEE 2013)},
  year = {2013},
  doi = {10.2991/iccsee.2013.767},
}

Loading related content...

Figure