Epsilon Aurigae CCD BVRI Photometry

The next eclipse of this 27.1 year system will begin in the summer of 2009 and last nearly 2 years. There are still many unanswered questions about this system and it will surely be studied in detail during the next eclipse. Most astronomers do not get excited about the observing until the eclipse stars, however, there is still much to be learned out-of-eclipse as the epsilon Aurigae star system is anything but quiet out-of-eclipse.

At the Hopkins Phoenix Observatory we have been doing extensive UBV photon counting photometry since 2003. We also have some single channel solid state JH band infrared data.

It seems most people doing photometry now days use a CCD camera. Observing epsilon Aurigae posses some special problems for CCD photometry. First, the system is very bright (3rd visual magnitude). Second, the normal comparison star, lambda Aurigae, is too far away (5 degrees) to be included in the same image as epsilon. While there are possible comparison stars closer, they are 3 to 4 magnitudes fainter and posse a problem for the dynamic range of the CCD. Single channel photometry seems to work best, but figuring a way around the problems and use a CCD camera was a challenge.

Aperture Mask

For the first case, it has been suggested that a telescope can be used with a CCD camera by stopping down the aperture. This will certainly work and allows sufficient exposure times to get around the scintillation effects. Bob Buchheim wrote an article for the SAS Vol. 5 Number 3 Newsletter titled "Getting Ready for Epsilon Aurigae." He indicated the problem with CCD photometry and the brightness of epsilon Aurigae. By using shorter exposures (under 1 second) and other problem arises, that of atmospheric scintillation. Usually at least a 10 second exposure is needed to average that out. This would saturate a CCD camera connected to most telescopes. To get around that Bob suggested making an aperture mask.

Wide Angle Photometry

For the second case the aperture mask will not help. At the Hopkins Phoenix Observatory we have experimented with using a DSI Pro CCD camera coupled to a telephoto camera lens. We have had success using the following procedure. Note that conditions were close ideal with clear nights and measurements close to the meridian. The B and V filter measurement were compared with the B and V data from our photon counting system for the same evening. In some cases the agreements were better than 0.01 magnitudes. Other case showed difference of 0.02 to 0.05 magnitudes. It seems if a good defocusing technique can be worked out the accuracy as well as the precision should be very good.

Note: Any of the DSI Pro (monochrome cameras) series cameras will work.

The color CCD cameras will not work. Other CCD cameras may work if they can be mated to a camera lens and obtain focus with a means to use filters. The DSI Pro uses a stock filter slide that allows very close mounting of the camera lens. A filter wheel was tried, but focus could not be archived, not even close. The camera lens system allows a smaller aperture/longer exposure time and wider field of view solving both the above problems. Figure 1 shows a 15 second exposure using just an IR blocking filter with a telephoto lens.

Note: DO NOT use the IR blocking filter when doing photometry. This was a test for field of view.

Figure 1
FOV Test with IR Blocking Filter and Telephoto Lens

Figure 2 shows a DSI Pro with filter slide, Mogg adapter and 25-200 mm telephoto camera lens. The Mogg adapter has a 1/4"-20 tapped hold for mounting. Various types of mounts could be used. What was used here were two scrape pieces of 1/2" aluminum. While some adjustment in azimuth can be made, elevation adjustments are not possible with this arrangement. Because of the wide field of view, precise alignment is not needed.

Figure 2
DSI Pro/Mogg Adapter/Telephoto Camera Lens

The Mogg adapter can be purchased for around $60 US$ for various camera lens and CCD cameras. See http://www.webcaddy.com.au/astro/adapter.asp

Ideally a filter wheel could be used, but as noted earlier it was found that using a filter wheel adds too much distance and focus cannot be achieved. To achieve focus 1/8" to 1/4" would need to be milled off the Mogg adapter. Since this is already about as narrow as practical, that does not seem like a possible approach.

The DSI Pro series comes with a filter slide for 4 filters. While certainly not ideal, it does work. To make change filter positions easier a small extended handle was added to one end of the filter slide. See Figure 3.

Figure 3
Filter Slide Handle

When installing the photometric filters in the filter slide, be patient. Getting the threads started can be tricky, but DO NOT force anything. Once you have the thread started it should go smoothly. Be sure to tighten all filters finger tight. The end filters are particularly difficult to thread and get tight, but can it can be done.

There is a rubber gasket that normally fits over the filters. It may be that the photometric filters are slightly larger than the LRGB filters, because the gasket could not be installed. The slide will work fine without it. Just minimize any stray light during the observing. Figures 4 shows the installed BVRI photometric filters and discarded rubber gasket.

Figure 4
Filter Slide with BVRI Filters

To keep the filters safe. It is recommended that they filter slide and filters be removed from the camera and kept in a plastic bag when not in use. Otherwise you risk dust, finger prints and possible physical damage to the expensive filters.

Figure 5
Telephoto Assembly Mounted on Telescope for Tracking

While the telephoto lens will work, it was decided to try a smaller 50mm camera lens. The smaller lens turned out to work well and was less cumbersome than the telephoto lens. It had a slightly smaller field of view, but sufficient to get both epsilon and lambda Aurigae in the image.


Figure 6
50 mm Assembly Mounted on Telescope for Tracking


Figure 7
50mm Assembly With Filter Slide and BVRI Filters

Figure 8
50mm F/2.0 Field of View

The Experiment

For this experiment AutoStar Envisage was used to capture the images and AIP4WIN for the image processing.


While quality BVRI CCD photometry can be done using this set up there are some pitfalls.

Peak Pixel Value

With the short focal length sharp focusing can produce very high maximum pixel values for a star. With a 16 bit system the peak maximum is limited to 65,535 ADU counts. That is well beyond the linearity of the CCD, however. Tests using multiple images of epsilon Aurigae with increasing exposure times produced the graph in Figure 9.

Figure 9
Linearity Test Maximum Star Pixel ADU Counts versus Exposure Time

The peak or maximum pixel value for the star MUST be in the linear region. Values under 55,000 ADU counts are in the linear region. This value can be controlled by reducing the exposure time. Typical exposure times for this set up are 2.0 seconds for the B and I filters and 1.0 seconds for the V and R filters. These times require the image to be defocused as noted below.

Under Sampling

Because of the very sharp focusing of the short focal length, star images can be severely under sampled. While this produces a very nice visual image, it will reduce the photometry quality drastically. A way around this is to defocus the image. There is a problem is with how to tell when it is defocused enough. One way to do this is to adjust the focus for a maximum Histogram reading (as long as it is below 65,535). This will be the sharpest focus. Note the value and then readjust the focus until the Histogram reads half that value or a bit under. This will spread the star's image over more pixels. Focusing the 50 mm camera lens is extremely touchy, but with a bit of practice one can do this. The defocused image will not look good, but if done properly will produce good photometric data. This needs to be done with each filter. While the filters may be parafocal (or approximately) with the short focal length minor difference will be amplified so the procedure must be done for each filter. When the focus is set, take some sample images at different exposure times and check the peak pixel values for epsilon Aurigae. They should be under 40,000. Use the exposure time that produces peak/maximum pixel values slightly under 40,000 ADU counts. Figure 10 shows a sample B image that has been defocused. Not the Histogram count is 31,965. Figure 11 shows the same image focused and with a Histogram count of 61,991. The defocusing must be done for each filter each night. Be sure to subtract dark frames for the same exposure for all the images.

Figure 10
B Filter Image Defocused (Histogram =31,965 ADU Counts)

Figure 11
B Filter Image Focused (Histogram = 61,991 ADU Counts)


With one and two second exposure times atmospheric scintillation becomes a big problem. To get around this multiple images are taken and stacked. For the 2 second exposures 30 images should be stacked. For the 1.0 second exposures, 60 images should be stacked.

Dark Frames and Flat Fielding

Be sure to take dark frames for the exposure times (1.0 and 2.0 seconds) and subtract them. Flat fielding was not done, but may improve the quality of the data.

Image Processing

For Image Processing, AIP4WIN was used. AutoStar Image Processing could also be used with a slightly different procedure. Remember with AutoStar Image Processing the aperture and Annulus settings are diameter where as AIP4WIN are radii.

Figure 12 shows the AIP4WIN image when first opened. While you can work with it as it, it is sometimes easier to adjust the "-B" value to make it darker as seen in Figure 13.

Figure 12

AIP4WIN Initial Image

When extracting the star - sky ADU counts from the images, use the AIP4WIN Analysis option to see how the star's image is spread out. To start the measurements select Photometry from the Measure pulldown menu and the Single Image. See Figure 13.

Figure 13
AIP4WIN Star Data

Allow an aperture setting that is at least 1 pixel radii beyond the floor of the star's reading. AIP4WIN uses a default aperture of 6.0, but a large aperture should be used for these defocused images. An aperture setting of 7.0 pixel radii is suggested as a starting point. Depending on how defocused the image is it may be necessary to increase that to 7.5 or even 8.0. The default 9.0 and 15.0 for the Annulus can stay fixed. Check the star's profile using the "Photometric Analysis" profile tool. See Figure 14. This can be selected by selecting the "Analysis" checkbox in the "Single Image Photometry" window." See Figure 15.

Figure 14
Photometric Analysis Profile With Default Settings

Figure 15
Single Image Photometry Window

Data Reduction

The Star - Sky value should be noted (48,656.6 ADU Counts in Figure 14) for epsilon and lambda Aurigae along with the UT of the exposure. Note that the PV max value is less than around 40,000 ADU Counts (11,721.92 in Figure 14).

Figure 16 shows the Profile and ADU counts for lambda Aurigae (Star - Sky = 48,656.6 ADU Counts with a Maximum pixel value of 11,721.92 ADU Counts, well in the linear region of the detector)

Figure 16
AIP4WIN Comparison Star Data

Figure 17 shows the Image's FITS Header where the UT of the exposure can be seen (03:24:23 as seen in Figure 17). This is selected from the Edit pulldown menu and then FITS HEADER.

Figure 17
AIP4WIN FITS Header Data Reduction

All the data should be logged in. Figure 18 shows a suggested form for logging the data/ The Max and FWHM values are for reference and not used in the data reduction.

Figure 18
Sample Data Form

Once the data has been entered in the Data Form it can then be entered into a data reduction program. While it is possible to enter the data directly without the form, having the form is good for comparing conditions and data as well as a backup.

Various programs can be used for the data reducing. At the Hopkins Phoenix Observatory we use FileMaker Pro Database program. We developed the database program which allows entry of data, does the data reduction, archives the data and allows lists and export for plotting. Figure 19 shows the initial data entry screen.

Figure 19
Data Reduction Program Initial Data

Figure 20 shows the data entry layout. This is where the program and comparison star ADU Counts are entered. The counts per second and raw magnitudes calculated.

Figure 20
Data Reduction Program Star Data Entry

Figure 21 shows the reduced data layout. Individual bands are calculated. The Extraterrestrial magnitudes for both the program and comparison star are calculated. A differential magnitude is then determined and the normalized program star magnitude calculated. This is done three times for each filter. The average program star's magnitude is then calculated along with the standard deviation as an indication of data spread.

Figure 21
Data Reduction Program Reduced Data

Figure 22 shows a summary of the calculated Air Masses, Hour Angles, average comparison star counts per second and raw magnitudes in each filter, a list of the transformation coefficients zero points and comparison star's published magnitudes, and night extinction for each band as determined by the comparison star.

Figure 22
Data Reduction Program Constants Summary

Figure 23 shows a list of observational data.

Figure 23
Data Reduction Program Star Data List

Typical data compared to the photon counting data take the same evening of 18/19 January 2008 is seen in Table 1.


Table 1
Typical Data



While the Hopkins Phoenix Observatory will be continuing UBV observations using a single channel PMT based photon counter, the experiment with the camera lens and DSI Pro indicated quality BVRI CCD photometry can be done.

Created 24 January 2008
Modified 11 February 2008