Hello John,

I don't know if you have seen it but we have a learning path on Illumination Systems: https://www.zemax.com/blogs/free-tutorials/illumination-systems-fundamentals. It is in the knowledgebase article section.

In this path, there is an article called How to optimize non-sequential optical systems.

Let us know if that helps and do not hesitate if you have any further questions.

Sandrine

Hi John, 

My experience has been that using a large area detector for a specific target, in this case, a specific angle of 90 degrees, may not work well. We end up with many empty pixels that don't provide any function. In my previous experience, I sometimes used small tiles and tiles of detector rectangles in the shape of a pseudo-hemisphere because I needed the resulting data in a specific format. I think something similar is needed here. 

My advice would be close to your own suggestion, use a different detector but try to shape it so that the range is very small. For example, although the cylindrical object set to a detector may not work here, you can make a very slim and long rectangular object and set it far away from your source. Let's say the height of the rectangle is equal to the range from 89 to 91 degrees. We can set this rectangular detector far from the lens, and optimize for maximum flux (or intensity). I think the symmetry in your system may be beneficial that this can work. We can set multiple detectors for optimization as well, so don't be shy about using, say six long-but-thin detector rectangles and use them around your lens (partially), and optimize for all to be maximum flux. 

-Kats

I think Kat has a great idea. (As usual)

If the system has axial symmetry, then you shouldn't need to cover the entire circumferance. It should be sufficient to use one detector on axis and one on the side. The merit function targets zero power to the on axis detector to direct power to the side. The merit function targets full power to the side detector with small SpotY (or X depending on orientation) and a Ycentroid at the center. Hopefully, if the initial design puts some power on the side detector, the gradient will tend toward what we want.