Collecting  Photons

1) An astrograph to go.

The idea was to have a very compact wide field imaging setup to bring along for the occasional outing to dark sites. The imaging bit is a compact telephoto lens from Canon: the 135 mm f 2.0 lens. This is a very well regarded lens in the "normal" photography community. It is light and compact and fits beautifully over the SBIG STL 11000 CCD. In the past I used it often over the modified DSRL, but I was a bit unsure about its performance on the much larger sensor of the 11000. The following of the story proved that the lens is indeed a good performer even on this large sensor.

The lens is mounted over the STL body by a small adapter manufactured by Astronomik. It is made of anodized alluminum and chromed brass. The chromed ring is a bit weird, because places some VERY shiny material nearby the light cone. Indeed, under a light polluted sky, this can produce internal reflections, and even allow some light in from the outside. I painted the ring flat black, but this operation require great care, because if the paint flakes, can place some VERY large dust bits on the filters! The adapter was not easy to find: I've got it from a very friendly store in Germany: www.astro-shop.com

The lens is mounted on a double Losmandy dovetail bar by means of a Losmandy camera adapter. There is no guide telescope since I hoped that I could use the internal chip for guiding. Indeed it worked perfectly, and it was always easy to find good guide stars even while imaging with a 6 nm Hα filter. Indeed, given the small image scale, the guiding routine can get a bit confused, and, at first, I could not have CCDsoft to work properly. Eventually I found that the following procedure worked reliably.

1) Acquire one image on the internal guide chip. 1-2 sec exposure should suffice.

2) Crop a smallish rectangle around the most isolated star that is reasonably bright.

3) Calibrate the guiding with long movements since the image scale is so small. 20 sec should work.

4) Select a guide star and start guiding.

I also use the CCDsoft plugin AutoDither by Paul Kanevsky and works just fine, provided that a delay of at least 50 sec precedes each sub.

2) Selecting the f number.

In spite of the high quality of the lens, I never expected it to work properly at maximum aperture, especially with a large sensor such as the KAF 11000. The preliminary imaging effort went into identifying a proper working point. The image on the left shows a magnified detail of the upper right corner of the image obtained at full aperture. Rolling over the image with the mouse will show the same detail at f 2.8. The star images are obviously better. Since I desired to remain as fast as possible, I did not evaluate closer diaphragms since I did not want to be tempted!

A detail of the four corners and centre at f 2.0 and f 2.8 are visible at those links (magnified 2 times). Here you can see that the star images at the four corners appear slightly different, suggesting a less than perfect collimation of the lens. CCDinspector computed a FWHM at the image center of 1.47 (f 2.0) and 1.03 (f2.8); at the corners the FWHM increased to 2.17 and 1.78.

3) The focusing device

During the early trials, and even before, when using this lens with a DSRL, it became clear that I needed a way of controlling the focuser ring in a precise and reproducible way. Here I used two old rings from a cheap Skywatcher guide telescope that fasten very nicely around the focusing ring (upper ring) and at the base of the lens. A precision micrometric screw is fastened to the fixed ring and pushes against the focusing brace. A steel spring ensure minimum backlash. The screw provides a very convenient and quantitative way of controlling the focus.

Two different views of the device can be seen here and here.

4) Focusing the four corners

When experimenting with the focusing process there are two aspects to consider: 1) the star shapes along the field as altered by field curvature and other extra-axial aberrations and 2) the residual chromatic aberration that shifts the best-focus position of the RGB images. We will look at these two aspect separately.

The plot on the left shows the average FWHM in four fields at the frame corner and at the center in function of the lens focus Images measured with Iris). The graphs refers to the green filters, and a document including also the data acquired with the red and blue filters can be found here. The continuous line is the average of the readings. Since best focus is reached at different positions for the central and peripheral fields, it is clear that there is some field curvature. Furthermore, the different behaviour of the corners suggests that the lens is not properly collimated. On the bright side, it is pretty clear that nearby the focal plane the data points move considerably closer.

5) Focusing the different colours.

The data relative to the five field positions are averaged to better appreciate the effects of wavelength on the focal plane position. The size of the FWHM computed by IRIS for the three wavelengths is shown by the three plots. It is clear that as the lens moves toward focus, the three plots approach considerably. A best focus position is found between positions 120-140 of the micrometric screw.

6) Focusing the different colours: CCD inspector.

The same images have been analysed by CCD inspector. The algorithm used are likely different and infact the absolute FWHM values that are returned by this program are different, however the same results hold. In summary:

1) the lens can produce stars with a FWHM averaged on the entire field equal or smaller than 2 pixels.

2) The green channel is brought to focus a little bit closer to the lens than the red a blue and red channels.

3) The best strategy is to reach focus from the intrafocal position finding the optimum focus for the green channel. The optimal focus for the blue and red channel is then only 10/20 screw units away. Although it would be preferable to acquire the green and red-blue channels separately a good intermediate position can be found that returns stars reasonably well focused.

7) Estimating the field curvature.

The same set of images was analysed by CCDinspector to estimate field curvature. Not surprisingly, these data show that the best focus interval correspond closely to the region of best flat field.Also these data identify the same focus interval .

8) Flat-fielding the little beast.

It is difficult to take proper flats with this lens. At first I tried my usual on field approach by using the screen of the laptop as illuminator. This procedure was extremely unsatisfactory: the calibrated images had a very strong gradient, with the edges of the field far overcompensated. This was due to the anisotropic light emission of the LCD screen, leading to fewer oblique rays and a consequent under-illumination of the edges during the flat acquisition. The situation much improved with the flat box.

A fake colour look up table shows well the peculiar aspect of the flat field image. The central part of the image is pretty flat (see graph below) but it is also rather elongated. Reasons for this are totally unknown. However illumination is reasonably uniform along the field, while the small asymmetry might be related to the collimation error also demonstrated by the focal series analysis