Problem:

This exposure of a bright star shows the signature of internal reflections as doughnuts around, and centered on, the bright star.

Part of the light hitting the CCD surface is reflected back toward the dewar window and the filter, then reflected back toward the CCD surface where it is perceived as out of focus (thus the doughnuts). There seem to be multiple internal reflections in this example here, as there as a number of different-sized doughnuts around the stellar image. In general, the farther away from the CCD a reflecting surface is, the larger of an image it will produce. Depending on this particular setup, the smaller doughnut might be a reflection off the dewar window and the bigger ones from the filter.

The fact that the centers of the doughnuts are centered on the star indicates that neither of the reflecting surfaces is tilted, and that the star is centered in the image. See below for links to bright stellar images and internal reflections in which either the reflective surfaces in the optical path are tilted, or in which the star is off the center of the CCD and thus causes the doughnuts not to be centered on the star.

Although this might at first glance appear as a collimation problem, the fact that the "hole" in the doughnuts is exactly in the center indicates that the collimation is not a problem.

The diffraction spikes in this image (which are at a slight angle with respect to the y-direction on the chip along which the saturated star is "bleeding" electrons) are caused by (the Fourier Transform of) the support struts which hold the secondary mirror in place. The very same support struts also cause the dark lines in the doughnuts which are almost perfectly aligned with the diffraction spikes.

Fix: