Hi CJ,

Your understanding is basically correct. OpticStudio does a paraxial raytrace of the system to establish all the first order properties such as EFFL, f/# and pupil locations and sizes. They are then used by all field points, which is why you get aberrations such as field curvature and coma. These fields are ‘aberrated’ with respect to the same paraxial reference.

For surfaces other than the image surface, we use a paraxial thickness solve after the surface to come to ‘best focus’ and then do the same paraxial setup with respect to this new image surface.

The difference with respect to the paraxial reference sphere is indistinguishable from the ‘Raw’ data. If the reference sphere radius is 50 mm say, and the opd of a ray is lambda/20, then the raw data is just 50 + lambda/20. Using the reference sphere means we focus on the lambda/20, which is the ‘interesting’ part. It also means the calculations can be performed on 64-bit machines as you don’t need to use up all your precision just on the 50 mm figure.

This is entirely analogous to an interferometer with a reference sphere. Try removing the reference sphere and all you’ll get is the bulk path length of the rays: the beam itself won’t show the tiny phase profile on top of the massive constant path length. You need the reference sphere to subtract out the bulk length and leave you with only the variance.

But if you want to do it, look at the System Explorer under Advanced, at the Reference OPD setting:

You may want to experiment with the Absolute and Infinity settings. See the documentation for full details. But the exit pupil refernce is the one to use in all but very odd circumstances :-)

Hope that helps,

• Mark