The baffle of the telescope has the role of stopping unwanted light, from reaching the detector. A technique that can be used to see if the baffle is doing its job is to obtain images of the telescope through a pinhole. We have done that by placing the pinhole at the position of one of the WFI filters. This means that we are imaging the telescope from just below the upper triplet (see the WFI manual for details). The figure below shows a gif with the image through the pinhole and the sky-baffle with and without inner rings. The image was obtained pointing to the dome flat field screen. Notice that all this light will go into the psf; in a perfect system only the light in the telescope pupil should be there.

We can see three main components to the light reaching each pixel: (1) the light that corresponds to the appropriate sky area, represented by the illuminated pupil above; (2) scattered light by the optics, seen as a diffuse patch of light in the above image; and (3), internal reflections, seen as the many concentric rings in the above picture.

What the above image does not show is a gradient in the pupil illumination, The figure below shows an enlargement of the pupil area. The image to the left was obtained with the pinhole and pointing to the sky at twilight. The image to the right was obtained pointing to the dome flat screen. The dynamic range of the dome image is much larger than the sky one; the upper right quadrant of the sky image looks slightly darker due to the fact that the images have not been flat fielded. We can see in the dome image that the upper half is a factor of two less bright than the lower part, probably due to the fact that we are illuminating the flat field screen using a single lamp.

The upper part of the pupil corresponds to the zone further away from the lamp. This feature is not visible in the pinhole image obtained using the sky flats where the illumination is uniform. This immediately suggests a way to improve the quality of the dome flats by adding extra sources of light to more uniformly illuminate the flat field screen.

Despite the obvious improvement for the image taken with the baffle with rings, there is one disadvantage, and that is the fact that in the present configuration they introduce vignetting. This can be seen in the rations of the flatfields taken with and without rings. The figure below shows the ratio of a flat taken without baffle rings over a flat taken with baffle rings, with the Rc filter: brighter areas are at a higher level. That is, the flat with baffle rings is relatively fainter near the edges, a clear signature of vignetting. Keep in mind that the Rc is a large round filter which introduces by itself minimal filter vignetting. The amount of vignetting can be as high as 15% near the corners.

We repeated the above test with the smaller square V filter. The figure below show the ration of the flat without rings over the flat with rings for this filter. Here we can see, in addition to the vignetting introduced by the baffle rings a clear component introduced by the filter holder. This component of vignetting is stronger in the flats taken without baffle rings presumably because the beam already vignetted by the sky baffle being narrower is less affected by the filter holder.

Perhaps more important than the vignetting itself is the stability of the flatfield. When we have vignetting introduced by intermediary structures which have not being designed for mechanical stiffness, the degree of vignetting can change when observing at different positions. To test this effect we have obtained twilight flats at the zenith and at a zenith angle of 60 degrees, for the same two filters.

The first figure shows the ratio of a flat obtained 60degrees to the east to that obtained at the zenith, with the large circular Rc filter. There is an east west variation on the order of 2% peak-to valley probably due to sky gradient at high Z. There is also a circular structure probably associated with a slight change of vignetting by the baffle rings: the area within the circle is probably not vignetted; the area outside is; a slight movement of the M! Baffle tube is probably responsible for the effect.

The same ratio for the square V filter, but this time taken 60 degrees west of the meridian is displayed below. It shows the same circular structure, with a similar east-west gradient (reversed) of 2%. There is almost no hint of the filter holder showing that this structure is rather stable. But not so the structure associated with the baffle rings.

CONCLUSIONS: The baffle rings introduce unwanted vignetting which can vary up to 15% peak to valley. Although this amount of variation is unacceptable, we can see that properly fitting the sky baffle with rings considerably reduces the contribution of unwanted straighlight. It is recommended that a new raytraicing be performed and that the diameter of the baffle rings be reduced appropriately so that they do not vignette when pointing the telescope to any position on the sky.