The ability to efficiently calculate root-mean-square (RMS) spot size and wavefront error in an optical system is of critical importance for fast optimization. This calculation involves tracing rays from various points in the field-of-view of the system at various wavelengths representing the spectral distribution of the source. This article describes the use of Gaussian Quadrature sampling for determining the optimal set of wavelengths to use in optimizing an optical system for performance over a broad input spectrum.
Zemax LLC thanks Dr. Brian Bauman of Lawrence Livermore National Laboratory for his guidance in the development of this feature.
Authored By Sanjay Gangadhara
Introduction
In 1988 Dr. Greg Forbes introduced the invaluable Gaussian Quadrature (GQ) algorithm1 that used a small number of rays in the entrance pupil to obtain an accurate value for the RMS spot size through knowledge of the aberrations within the system. In 2010, a more general algorithm was developed based on Forbes's work.2 Using GQ principles, this algorithm works with the Sellmeier dispersion formula to characterize a broad spectrum with a limited number of wavelengths. This allows for efficient optimization of systems that use broad polychromatic sources.
OpticStudio has used GQ sampling of the pupil for optimization of systems since its initial development. In this article, we will show how it is used to accurately model broad wavelength spectrums.
Using GQ wavelength sampling in OpticStudio
At the 2010 International Optical Design Conference (IODC), Dr. Bauman provided a ZPL macro that could be used to define the wavelengths and weights associated with his algorithm. The macro is attached to this article. The GQ algorithm developed by Drs. Bauman and Xiao has since been incorporated directly into OpticStudio.
To choose wavelengths using GQ sampling, the user needs to specify the wavelength range of interest and the number of wavelengths desired:
As always, wavelength values are in microns. The number of wavelengths - or Steps - that can be chosen can be any even number from 2 to 12. Once the “Gaussian Quadrature” button is selected, OpticStudio will calculate the appropriate wavelengths and weights to characterize the source over the input wavelength range. For example, using four wavelengths to characterize a source over the range of 0.4 to 0.7 microns results in:
The input values for the minimum and maximum wavelength are identical to those provided by the user, within round-off error. Note that OpticStudio will always reset the wavelength range in the GQ dialog box based on minimum and maximum wavelengths defined in the system (this is true regardless of how those wavelengths were defined). Also note that OpticStudio uses a suitable “middle” wavelength as the Primary wavelength, but this can be changed by the user at any time.
Evaluating the RMS spot size of an apochromat
In this section an example is provided to illustrate the benefit of GQ sampling for wavelengths.
To start, open the sample file "Apochromat3.zmx" provided in the {Zemax}\Samples\Sequential\Objectives folder. Then add a small off-axis field point to the system:
Save this file to a new name, e.g. “Apochromat3mod.zmx”.
The performance of this system will be evaluated over the wavelength range of 0.4-0.7 microns. We’ll start by using four wavelengths uniformly spaced and weighted to define this range. Open the Wavelength Data dialog box and enter:
A plot of the RMS spot size vs. field can be generated under Analyze...RMS...RMS vs. Field.
For the settings shown below, hit the “Save” button (for later use). The results show:
How sensitive are these results to our wavelength sampling? We can increase the number of wavelengths systematically and then re-evaluate the system to find out. This process has been automated using a ZPL macro. The macro is provided as an attachment.
The basic structure of the macro is as follows:
- Load the lens file
- Change the number of wavelengths and the wavelength values (weights always = 1)
- Retrieve values for the RMS spot size vs. field using the GETTEXTFILE keyword
The macro is used to investigate the results for RMS spot size vs. field for 4, 6, 8, 10, and 12 wavelengths. In all cases, the spacing between wavelengths is constant making the sampling uniform. The macro is then used to compare the results to those generated using four wavelengths (and weights) given by GQ sampling. The results are then plotted using the PLOT keyword:
The data shown above represent the polychromatic results for the 6 cases considered (blue curve = 4 wavelengths; green = 6 wavelengths; red = 8 wavelengths; gold = 10 wavelengths; purple = 12 wavelength; black = GQ with 4 wavelengths). Note how the results generated by uniform wavelength sampling converge towards those generated by GQ sampling. For this system, the results indicate that we would need three times as many uniform wavelengths to achieve the same level of accuracy as provided by GQ sampling.
Points to consider
As indicated by Drs. Bauman and Xiao in their 2010 paper,2 the GQ algorithm used here for wavelength sampling works well for glasses whose dispersion is given by the Sellmeier formula (e.g. glasses from the Schott and Ohara catalogs) and for glasses whose principle absorption occurs near 0.1 microns (the algorithm uses this absorption peak as a reference). Thus, this algorithm may not be appropriate for systems containing dissimilar materials with varying dispersion characteristics or for systems containing materials with significant absorption at longer wavelengths. In addition, this algorithm may be less effective for systems operated at wavelengths longer than about 1.5 microns, since the effect of the IR absorption line will become significant under such conditions.
The effects of wavelength sampling will generally be significant in systems dominated by chromatic effects. One way to characterize this is to look at Analyze...Aberrations...Chromatic Focal Shift.
Re-open “Apochromat3.zmx” and change the minimum and maximum wavelength to 0.4 and 0.7 microns, respectively. The Chromatic Focal Shift in this system is given by:
When the variation in back focal length with wavelength has a complex structure, as shown above, it will be difficult to characterize such structure with a few uniformly-sampled wavelengths. Under such conditions, GQ sampling will be a much more efficient method of characterizing the wavelength dependence of the system.
References
[1] G.W. Forbes, “Optical system assessment for design: numerical ray tracing in the Gaussian pupil”, J. Opt. Soc. Am. A, Vol. 5, No. 11, p1943 (1988).
[2] B.J. Bauman and H. Xiao, “Gaussian quadrature for optical design with non-circular pupils and fields, and broad wavelengths”, Proc. SPIE, Vol. 7652, p76522S-1 (2010).