Hello! We have a question about how we frequently get unexpected results in RMS spot radius when we optimize using variable aspherics. The question is best posed using an example:
We can optimize a two-doublet system with (light through a doublet, to a 45-deg. fold mirror, through another doublet, to the image) for one wavelength and achieve an RMS spot radius on the order of 11 microns. This is done using the optimization wizard set to image quality: SPOT and a merit function which outlines all of our important constraints, and this is only for STANDARD surfaces. We expected that, when choosing Find Best Asphere or setting aspheres to variable (V), we could optimize and improve performance; this assumption was incorrect, since repeatedly (it could be every single time, actually), our performance gets worse by multiples of 6, 10, or larger, in terms of RMS spot radius.
What is going on here? We were under the impression that by allowing OpticsStudio the option of adding an aspheric component (or not), the performance surely shouldn't get worse!
Thanks a lot. Hope this is clear enough. Happy to provide clarifications on our question. Fundamentally, we have some learning to do about the nature of optimization in general.
+2Hello Elliot. You are correct that your chosen performance metric should not get worse, so there must be some other factor at play here.
My initial guess would be that the spot size calculation in your merit function is undersampled with respect to the order of asphere you are using. You can modify the Rings setting in the Optimization Wizard to address this. When using the Gaussian Quadrature pupil integration algorithm n rings will allow the effects of aspheres (or aberrations) of order r^(2n - 1) to be computed accurately.
If the number of rings is set to 3 then you can use aspheres up to 5th order (2n-1). If you're using an asphere with contributions from 6th order, or higher, terms then you'll need to increase the number of rings and rebuild the merit function before optimising.
If this is not the cause of the problem then if you could share your affected file it would greatly help to find the source of the issue.
My initial guess would be that the spot size calculation in your merit function is undersampled with respect to the order of asphere you are using. You can modify the Rings setting in the Optimization Wizard to address this. When using the Gaussian Quadrature pupil integration algorithm n rings will allow the effects of aspheres (or aberrations) of order r^(2n - 1) to be computed accurately.
If the number of rings is set to 3 then you can use aspheres up to 5th order (2n-1). If you're using an asphere with contributions from 6th order, or higher, terms then you'll need to increase the number of rings and rebuild the merit function before optimising.
If this is not the cause of the problem then if you could share your affected file it would greatly help to find the source of the issue.
Hi Chris,
Your suggestion worked perfectly -- our optimization improved drastically. Thank you very much for the insight.
Your suggestion worked perfectly -- our optimization improved drastically. Thank you very much for the insight.
+2Hi Elliot, I'm glad I could help.
Reply
Enter your E-mail address. We'll send you an e-mail with instructions to reset your password.
Need more help?
To Chinese users:
Do not provide any information or data that is restricted by applicable law, including by the People’s Republic of China’s Cybersecurity and Data Security Laws ( e.g., Important Data, National Core Data, etc.).
不要提供任何受适用法律，包括中华人民共和国的网络安全和数据安全法限制的信息或数据（如重要数据、国家核心数据等）。