DLL (User-Defined Surface): Sequential Immersion objective lens model

  • 15 September 2020
  • 6 replies
  • 251 views

Userlevel 5
Badge +2

Dear OpticStudio users,


In the context of my research at the MRC - Laboratory of Molecular Biology, I am tolerencing a custom light-sheet microscope, which uses commercial immersion objective lenses.


When tolerancing an optical system, which uses non-immersion objective lenses, I would often assume that this element can be replaced with a paraxial lens in a first approximation. However, when it comes to immersion objective lenses, this assumption breaks Abbe's sine condition [1]. Consequently, it is usually not possible to satisfy the specifications, namely focal length, back-aperture diameter, and numerical aperture, of an immersion objective lens with a single paraxial lens.


A direct solution to this problem involves patent search, and cumbersome reverse engineering to try and get your hands on the actual design of the commercial immersion objective lens. However, this is time-consuming, and requires a different kind of expertise.


To address this issue, Hwang and Lee published an immersion objective lens model [1] in 2008. This model uses a single spherical surface with a modified refraction law to reconcile Abbe's sine condition.


I have created a DLL, and a sample file (attached to this thread), that reproduce the model of Hwang and Lee (see Fig. 1).



Figure 1 (A) Reproduction of Fig. 3 in [1] using the DLL in OS. n' is the immersion medium refractive index, and f is the objective lens focal length. (B) Zoom of (A) at the sample plane showing how the different fields focus to a single point. ENPP: entrance pupil position.


Unfortunnately, I haven't been able to reproduce Hwang and Lee's results yet, but I think it is related to my implementation of the real immersion objective lens that is used for comparison (Fig. 6 in US patent US7199938 I believe). I am willing to discuss this issue with whoever is interested, and I will update this thread accordingly. In this regard, one might want to know that there could be a mistake in US patent US7199938 in page 16 around line 50. The r6 radius of the imaging lens should be positive I think (as seen in Fig. 17 of the patent).


Please also note that Eq. 5 of the paper is an approximation when the angle phi is zero. The complete formula involves solving a quadratic equation. Consequently, it also changes the form of Eq. 6, but luckily the norm of the correction vector corresponds to the path t in OpticStudio standard DLLs.


[1] S.-U. Hwang and Y.-G. Lee, Opt. Express 16, 21170 (2008).


Installation: If you load the archive attached in OS, it should also place a copy of the DLL in your document folder.


Parameters: the parameters of the DLL are the immersion medium refarctive index, and the focal length of the objective lens.


Let me know if you find this useful or if you want to try to reproduce the paper's results.


Kind regards,


David


6 replies

Userlevel 5
Badge +2

@David.Nguyen Hi David! I think your post would be a good one in the Code Exchange. Let me know if you would be happy for me to move it.

Userlevel 5
Badge +2

Hi @Sandrine Auriol,

 

Happy new year, I hope all is well for you :)

 

Feel free to move this post wherever you think its best. In the end, I did not really use it because the results I obtained differed quite a bit from Hwang and Lee’s paper. My hope is that someone might be interested into this and give me a second opinion on the paper.

 

Take care,

 

David

Badge +2

David,

I couldn’t get a solution like this to work effectively for me either.  That’s why I solved it using the ZOS-API and non-sequential ray tracing.  I presented this work at a Zemax Users’ Group in 2017 (predecessor to the Envision conference).

There are some “rules” one can infer from a perfect microscope objective:

  1. It must behave consistently with paraxial optics
  2. Almost all are designed to be telecentric

I’ve read Hwang’s work.  I found that Rimas Juskaitis has written several articles and book chapters that are useful in characterizing and understanding the behavior of microscope objectives.

When I looked at all the “rules”, the only solution that made sense to me was to:

  1. Trace rays to the Exit Pupil of the objective
  2. Detect these rays (position and direction cosine)
  3. Compute in a separate program the ideal emission
  4. Create a source file to trace rays forward from the Principal Sphere to the sample

I’ve built this model for (x2) different applications and it works really well and matches what we observe in the hardware.

Best,

Brian

 

 

Userlevel 5
Badge +2

Hi Brian,

 

Thank you for the excellent post. Do you know if your presentation was recorded and whether it is possible to access it online?

I’ll definitely study and try your methodology and perhaps report back here.

Take care,

 

David

Badge +2

David,

That’s a question for the Zemax staff.  I authorized its redistribution, but I don’t know where they archived the presentations.

Best,

Brian

Userlevel 5
Badge +2

Hi David and Brian

I contacted our Marketing team but it seems that we unfortunately didn’t keep the presentations from back then. We have presentations of newer ones only (and recordings when the event was online). 

Brian, would you be able to post your presentation here? Sorry about that.

Reply