Is it common to reverse the whole system and use the field data editor as a image plane?

Hi Lukas,

You could try reversing the system design, but in that case your field points would become ideal point sources propagating to the image plane (which you would then treat as your actual source plane, so you would try to optimize the optical path to best mimic your actual source distribution when all fields are combined together).

Or, you could consider a forward-path design and use a set of apertures to isolate your individual target image locations.  This could be done by applying an aperture in front of the image plane and using the Multi-Configuration Editor (MCE) to define a set of configurations, each one decentering the aperture to a specific target location.  This way you can locally isolate each of your target image plane locations.  In the Merit Function Editor (MFE) you can then use the CENY operand for each configuration to calculate (or optimize) the set of local y-centroids.

Regards,

Jeff

Hi Jeff,

is it possible you describe the forward-path method a little further? 

Best regards, Lukas

Hi Lukas,

I can try, but first it would be helpful if you can supply a copy of your model or a more complete description of exactly what you are trying to accomplish.

Thanks,

Jeff 

Application is a vehicle headlight in the near infrared range consisting of a laser source and 4 lenses. I already had headlights with LED matrices where the pixels were represented by the fields. Using the CENY operands, one can then create a distortion and define the point of incidence of the beams on the image plane at 25 meters. Typical of vehicle headlights, many light beams strike near the optical axis and fewer farther away.
If the LED matrix is replaced by a laser light source, I have only one field and can no longer position beams on the image plane.
I have the exact specifications of the impact points of the beam bundles from a simulation, which gives me the best possible light distribution for sensors in the near-infrared range. In the end I want to realize a certain light distribution. Here, a two-dimensional view is sufficient for the time being.
 

Here you can see the different fields positioned using the CENY operand. The system is so small that you can't see it. The distance is 25 m from the last lens to the image plane. In this way I want to create a system with one field. 

Best Regards, Lukas

Hi Lukas,

I still don’t fully understand what your proposed optical path looks like.  All I see are some rays originating from a small region and fanning out to a distant illumination plane.

In any event, to use one field to generate an array of beams, it would seem that you need some sort of a splitting function.  This could be a lens array that takes a collimated laser beam and creates an array of focused spots in the back focal plane; those focused spots can then be collimated and directed into different angles with a subsequent larger-diameter single lens.  Or, it could be a diffraction grating that splits a collimated beam into multiple angles.  You may have a different way, but if the entire set of beams in the array are simultaneously present, then to calculate the centroids of the individual spots/beams at the illumination plane, you can use the Multi-Configuration Editor (MCE) to create a set of apertures, so that each configuration isolates one beam.

Here is an example in which I generate a 5 x 5 array of focused spots with a lens array, then collimate these spots into beams propagating in 25 different directions.  To isolate an individual beam at the illumination (image) plane, I simply place a small aperture, roughly centered, on the beam of interest using the APDY operand in the MCE.  The aperture is big enough to pass the entire individual beam, but it blocks all of the other beams.  The CENY merit function operand for the corresponding configuration then calculates the y-centroid for that particular beam.

Here’s a zoom view of the front-end optics:

Regards,

Jeff

Thank you Jeff so far!

I was looking for a way to optimize the system in such a way that the target points of the beams on the image plane are treated as targets in the merit function. So that the system optimizes itself in this way. When I use apertures, the light is absorbed at the point, but the system is not optimized so that the light collects at the points of interest.

Using a lens array is a good idea to create multiple beams first. Inserting apertures via the MC editor is also a good idea, but optimizing the system with respect to the focus points on the image plane doesn't quite work out as desired yet.

Do you have any thoughts on this?

Kind regards, Lukas

Hi Lukas,

I don’t know what you mean when you say that optimization doesn’t quite work out as desired.  It should be okay.  The spots are isolated by configuration, so you can add whatever operands you like to the merit function to adjust the individual spot properties in some particular fashion for your application.  The variables in your optical path (i.e., the ones that you select for optimization) should be chosen so that when they change, the merit function operands change in a sensible way.  For example, if you want independent control over spot positions, then your optical system should be constructed so that it provides a sufficient number of independent variables.  If a variable is strongly coupled to multiple spot locations, then it’s not a good candidate for setting the positions of the spots independently from one another.  But I leave that detail up to you…

Regards,

Jeff

So I have also expressed myself incomprehensibly or thought much too complicated myself. The solution of the problem was the GENF operand. In combination with the OPVA operand I can define a radius in which a certain part of the brightness must be contained. The same goal was previously pursued with the CENY operands, but it was a workaround due to the number of fields and a simplified assumption in which the rotational symmetry of the system was exploited. That's why my example looked asymmetric. With a single field, these "simplifications" do not have to be made and other operands can be used for the same goal, which from the outside make the actual goal clearer.

Thanks anyway for the really nice help Jeff!