Solved

FFT PSF: Sampling too low, data inaccurate!

  • 8 February 2022
  • 4 replies
  • 997 views

Dear all,

I am trying to obtain the PSF of an imaging system, which consists of an objective lens, a doublet lens, and a reflecting surface. The reflective surface and lenses for forming an image on an image plane were delineated in non-sequential mode, but the objective and doublet were delineated sequentially, so I'm working in mixed mode. 

My first attempt at evaluating the system is to use the FFTPSF, but I only get the message “Sampling too low, data inaccurate!”. After modifying the sampling and the Image Delta, I still get the same results. Thus, I think it may be an issue with the mixed-mode system that cannot calculate the wavefront. My original plan was to convert all components to sequential, then get the PSF, but they did not work properly once I converted them. 

 

I would be happy if anyone could come and discuss the issue. 

Kind regards,

Amir

icon

Best answer by Mike.Jones 13 February 2022, 05:57

View original

4 replies

Userlevel 5
Badge +3

First, do an optical path difference (OPD) plot.  If the wavefront error (WFE) is just a few waves, the 128x128 FFT might handle it OK.  If greater than a few waves, you’ll need to increase your FFT sampling.  Start with 256x256, if that doesn’t work, keep increasing until it does.  Also, try the Huygens PSF as well, and increase the sampling as well. 

Look at your ray plots near your image plane.  If they’re not forming reasonably small focal points, you’ve got something else wrong, and neither FFT nor Huygens will give you answers.  Geometric encircled energy will tell you how bad your focus is.

Userlevel 5
Badge +1

Hi all,

I just wanted to quickly chime in and agree with Mike’s comments. The FFT PSF will be good to use if your system is well-performing and not significantly limited by geometric aberrations. We also have another forum post on a related topic -- while the discussion here does cover more specifically using the ZPL to compute the FFT PSF, I think some of the discussion applies here also: FFT PSF: "Computation aborted, invalid results!" | Zemax Community

Lastly, just wanted to share some relevant Knowledgebase articles to give some background on FFT/Huygens PSF:

Thanks!

Thanks both of you! 

In FFT PSF even if I increase the sampling it takes forever to compute the PSF and finally the same error. 

I can get the Huygens PSF with low pupil sampling but have no clue why the FFT PSF doesn’t work.

Also, in Huygens PSF I can’t get a 3D shape of the PSF. 

 

Kind regards,

Amir

 

Userlevel 7
Badge +3

Hi Amir,

Generally speaking, the FFT PSF should work fine in mixed mode -- just the same as in pure sequential mode, assuming of course that aberrations are sufficiently constrained.  There is likely some issue with your model.  If you can attach a copy we might be able to help.

Regards,

Jeff

Reply