Pupil Image Sharpness and aperture size

During CD16, we serendipitously noticed that the pupil images are sharper when the aperture is opened  up. 

When we take pupil images, we typically must use a short wavelength filter while crossing a slit with a decker dash, cutting down on light. Slit5 x 14pix decker dash is typical. 

Here is a what we got with the ZnSe pupil lens, filter=N04679, slit5 x 14 pix decker during cool down 16. The second image is a 

version that's blown up and rotated by 22.5 deg, so that the two parallel wires line up with columns and the single target wire lines up along rows. The FWHM of the pupil target wires was 15-20pix. 

Then we saw marked improvement in the sharpness of the pupil target when we used the middle of the central decker arc crossed with slit 5. 

The FWHM of the two parallel target wires became 7.8pix. The FWHM of the single target wire perpendicular to the wire pair was still around FWHM ~25 pix.

The explanation is that when the aperture is widened, we use more of the pupil lens diameter, which collapses the depth of focus. (If I got that wrong, please correct me). 

I'm not sure if we expect a focus difference for the differently oriented wires due to the aperture asymmetry, but that will be clarified experimentally in our next cold cycle, when we a newly acquired 10% ND filter will allow us to take pupil images with the full circular aperture.