The rays are being drawn to some dummy surfaces. In your Lens Data Editor, go to the Draw properties for the non-optical surfaces and check “Skip Rays To This Surface:”

This seems like a very good answer.  There are coordinate breaks both before and after that lens, and those surfaces did not have “Skip Rays To This Surface”.  However, doing this suggestion did not solve the problem.  I was curious to note that checking “Skip Rays To This Surface” for the actual top and bottom surfaces of the lens changed the ray’s trajectory inside the lens, but did not affect the ray’s trajectory either before or after the lens --- in other words this erroneous drawing did not affect the final destination of the ray at the detector, several lenses further on.  I’m inclined to believe that this is just a glitch in the drawing and not a glitch in the optimization calculation.

@Antony.Stark 

Can you fletch the rays in your 3D Layout and show us that same screenshot?

Although we cannot exclude a glitch, I’ve never come across such behavior so far.

Take care,

David

@Antony.Stark

Could you uncheck Skips Rays To This Surface and Do Not Draw This Surface for all surfaces and give us the screenshot again? Sorry I should have said it before but seeing those rays and surfaces might help understand the behavior of those rays.

Have you tried using the Analyze..Rays & Spots..Single Ray Trace to get more details about the glitched ray?

Take care,

David

OK, you’ve solved it!   The problem was, as shown below, is that there is a dummy surface after the second lens.  That surface is a flat plane, and it has no material (and so doesn’t do anything to the optics), but there were some variables that allowed it to tilt.  There were constraints on it in the Merit Function that should have prevented it from tilting too much (it’s supposed to represent the position of a local image of the source), but these did not work properly and this “null” surface tilted enough so that part of it was actually before the lens that precedes it in the sequential lens list.  Since the surface had “Do Not Draw This Surface” checked, this error was not apparent. ZEMAX did the best it could with it, running the ray backwards up to that surface, but since the surface was a dummy surface with no effect on the optics the ray then continued downward from the lens as it should.