This issue is due to the special definition of Paraxial surfaces in OpticStudio.

For a general lens with thickness, when the marginal rays propagate in the lens, it will have longer distance propagation with ray angles compare to center ray. Because of the thinner thickness at the lens edge, the optical path of marginal rays are no difference compare to center ray, as shown in the figure below:

However, for Paraxial Lens, there is no definition of the thickness. The method in OpticStudio is to directly add phase change to all rays so that the rays can be propagated under the influence of a certain surface/lens power. Such method cannot consider the optical path with thinner lens thickness of marginal rays mentioned above.

Therefore, in order to ensure that all rays have the same optical path when they finally arrive at the image plane, the optical path of corresponding marginal rays will be compensated accordingly in subsequent propagation. This is why, in the propagation described above, a 16 mm path length propagation will have a return value less than 16 mm. In general, we do not recommend using optical path calculation PLEN and OPTH operands for systems with pure Paraxial surfaces.

Please feel free to contact us at Support@zemax.com if you need more detailed instructions on phase change and definitions about Paraxial surfaces in OpticStudio!

This is a good point. There’s only so far you can push an idealized, infinitely thin surface in terms of modeling a real, thick lens. Check out the description of the Mode parameter for the Paraxial lens to see the various options for computing the OPL added by the paraxial surface.

BTW, this is precisely why the NSC paraxial lens cannot compute the added path length. We cannot assume that all rays incident on one side come from the same conjugate point. The NSC paraxial surface has no equivalent of the Mode flag.