How to use optimization function to collimate light in NSC mode?

  • 24 June 2020
  • 3 replies
  • 1281 views

  • Monochrome
  • 3 replies

Hi all,


I would like to test a variety of lenses in a 2-lens Keplarian beam reductor. What is the best way to optimize this design to get the BEST collimation with the pair of lenses? I have used the optimization wizard with a variable Z-axis distance. The criteria settings are attached. Occasionally, the optimization function does not change the Z distance of the second lens even after I put it in an incorrect Z-position. Can someone please advise?


Thanks!




 


3 replies

Unfortunately I am not getting the desired outcome. After I run my optimziation, the beam is clearly not collimated as can be seen in the ray diagram.


When I reset my merit function and run the optimization, sometimes I do not even get a changed value for my Z-distance variable - even after I put a random distance in my lens data editor. 


I have moved to a sequential mode for this analysis because I appear to be stumped. But, now I am unable to generate a merit function due to the following error: 'Error in target 5. TIR at surface 4!'. I am trying to run the optimization with the default settings. I have attached my settings below, as well the lens data editor and my 3D layout.


I don't know what to do now, both non-sequential and sequential modes do not appear to be working as easily as they should be. Can you please advise?




Userlevel 4
Badge +1

Hi Martin,


Thank you for contacting us with your inquiry!


To optimize for collimation, one chould use the default merit function with Criterion set for RMS Angular Radius. I looked at the screenshot of your optimizaton wizard. It appeared proper. The NSC optimization usually takes longer than Sequential optimization, because in Sequential mode, with Gaussian Quadrature or Rectangular Array, OpticStudio usully only needs to trace couple hundreds of ray, while in NSC, we are looking at million rays or more which could also split and scatter. This is the main reason why the NSC optimization takes much longer. Usually we recommend using Othogonal Descent for NSC optimization, but you could also try switching between DLS and OD to see which algorithm works better for your system. 


Based on the layout plot, if your system contains only these two lenses, I would recommend maybe starting the design and optimization in Sequential mode first. Sequential mode also provides a default merit function criterion for Angular spot size. This way you could take advantage of the fast optimization in Sequential mode. When done, you can then convert this system into NSC mode, and run further optimization in NSC mode if you are interested in using a specific source object in NSC mode. This will speed up the optmization process. 


Let me know if this idea works for you, or if you have any other questions.


Best regards,


Hui

Userlevel 4
Badge +1

Hi Martin,


Would it be possible for you to share the file? I can see the LDE from your screenshot but it's missing the System Aperture definition. I tried to build it but didn't get identical LDE as the one you showed. If you don't want to share your file in the forum, you can either email it to support@zemax.com, or create a case in MyZemax and attach the file there. 



As for the TIR error message in the Sequential merit function editor, you can swtich the Pupil Integration from Gaussian Quadrature to Rectangular array and check Delete Vignetted, that should get rid of that error message. 


I can see in your layout, most of the rays are collimated with some outliers. I think this partly is due to the fact you only allow one varable, which is the separation between the two lenses. With just this one variable, it's as expected the output beam won't be perfectly collimated. If you can introduce more degrees of freedom, for example radius of curvature on the lenses, you'll probably achieve better collimation.


Hui

Reply