Science can happen at the Hut!

Gathering Data on SZ Herculi - Eclipsing Binary

Variable stars are stars that change brightness. The brightness changes of these stars can range from a thousandth of a magnitude to as much as twenty magnitudes over periods of a fraction of a second to years, depending on the type of variable star. Over 30,000 variable stars are known and catalogued, and many thousands more are suspected to be variable. There are a number of reasons why variable stars change their brightness. Pulsating variables, for example, swell and shrink due to internal forces. While an eclipsing binary will dim when it is eclipsed by a faint companion; and then brightens when the occulting star moves out of the way.

 I was inspired to gather variable star data by an article in the June 2007 issue of Sky & Telescope that described the light curve of the eclipsing binary SZ Herculi.  Even though I’ve been imaging for the past 4 years I’ve never done this type of astronomy before so I really didn’t know what to expect or even if I could successfully do it from my observatory in Central Ohio .

 According to the article the middle of the eclipse was predicted to occur at 3:30 UT of 7/21 which is to say at 23:30 local of 7/20.  I arrived at Hutville (where my observatory, The Octadome is located) about 3 hours before the eclipse hoping that I would have enough time to set up and started to get ready. 

 I opened the Octadome and took the cover off the 12” Meade SCT.  The most important datum for this experiment was the time of minimum so as soon as my laptop was turned on I synchronized the computer clock with the NIST Internet Time Service.  I then connected the laptop to the Paramount ME and the USB hub that connects the ST10XME CCD camera, TCF-S Focuser and Pixys Rotator to the computer, and turned the rest of the equipment on.   Once the Paramount was homed I then slewed to one of the bright stars on the constellation of Hercules and proceeded to focus the scope. 

With the aid of The Sky 6 I looked for and quickly found SZ Herculi (is listed under the General Catalog of Variable Stars as GCVS SZ HER), slewed the scope to it and in a matter of minutes I was ready to start taking images. 

By that time the thermoelectric cooler in the camera had cooled the CCD detector to -15*C, the temperature I set the camera for summer imaging, so I started to take some test images to see what kind of exposure would give me the best results.  What I was looking for was enough exposure to clearly show the target star and at least 3 more stars that I could use as a reference for magnitude.  It is important to ensure that all these stars are kept within the linear detection range of the CCD, that is to say that none of these target stars saturate the detector.  I selected 5-second exposures unbinned with no filter as that gave me a good selection of reference stars and SZ Herculi.   

My plan was to capture 3 hours of data (1.5 hours at each side of the expected eclipse) so I could generate a reasonable light curve.  I figured that 180 data points (one image per minute) would give me the granularity I was looking for to generate a smooth light curve. 

 To my surprise the focusing and exposure time determination exercise had taken more time that I had planned for so I started my imaging sequence about an hour later than originally planned. 

When I started the sequence SZ Herculi was about 30 minutes from culmination, due to the idiosyncrasies of the german equatorial mount I had to do a flip after the mount passed the meridian.  That operation took about 10 minutes.  That is reflected in the light curve plot as a break.  Luckily the eclipse was still 8 minutes away and I was able to capture the minimum and the subsequent separation of the eclipsing stars.

 While taking the images I kept looking at SZ Herculi and how it looked relative to its nearest neighbor GSC2610:1214, after a little while the magnitude change was visually obvious.

Processing and Data Reduction

The next day I started to process the images with CCDStack.  I wanted to minimize any and all changes that could affect the determination of the magnitude of SZ Herculi so the only processing done was a dark calibration and a simple alignment of the calibrated images.

Once processing was finished I started to load the images into Maxim DL.  Maxim has an excellent Photometry tool that allows the user to generate light plots relatively easily.  The process involves selecting a set of reference stars with known magnitudes and then selecting the “New Object” star.  The program matches the stars in all the images loaded and create a comma separated value (CSV) data file that includes the Julian Date of the observation and the magnitudes of all the selected stars, it also generates a corresponding light curve from the data.

Because the images were unbinned each image was about 6 MB in size, due to memory limitations on my laptop I could load about 15 images at a time before the system started to issue warning messages.  This forced me to repeat the process above 6 times, appending the data to the one processed previously, before I was able to get all the data points needed to show the eclipse.

Reference Stars Selection

There are a few choices that allow you to determine the magnitude values of the reference stars.  One way is to visit the American Association of Variable Stars Observers (AAVSO) web site at www.aavso.org and get a chart that includes the target variable star and surrounding stars with known magnitudes.  One can then visually match the images taken with the chart and determine the magnitudes of the stars within the field of view of the image.

Another alternative is to use the Astrometry function on Maxim DL.  The process to use the Astrometry function assumes that a lot of the data needed to successfully figure out which stars were captured in the image is already available as part of the FITS file information header.  This header includes information about the type of image that was taken, whether it was binned or not, the focal length of the telescope, the time at which the image was taken, the RA and Declination at the center of the image, and other information.  This information is included in each of the FITS image files taken with Maxim.  The important requirement is that the telescope needs to be connected to The Sky and Maxim at the same time.  If the FITS image files have all the information above then one can use the Pinpoint astrometric engine that is included with Maxim to solve the plate.  Once the plate is solved astrometric information is available for the stars included in the solved image so one can use that information to find out the magnitude of a specific star (as long as the star is included and has astrometric information included in the Pinpoint catalog).   The typical catalog used by Pinpoint is the GSC 1.1 (Updated). 

 Another option to get the reference stars’ magnitude is to use the information window on The Sky 6.  When a star is selected in The Sky 6 the information window provides (among many other things) the magnitude of the star.  I used The Sky as my information source. 

 The three referenced stars I selected were: GSC2610:1116 (9.72 mag), GSC2610:821 (10.9 mag), GSC2610:1214 (11.8 mag) 

Results

The results of the data processing are summarized in the light curve shown.  It clearly shows the magnitude of SZ Herculi diminishing as the eclipsing star passes in front of it and the subsequent increase in magnitude as the occultation ends.

 Minimum was at 3:34:06 UT , just over 4 minutes after the predicted time.

  It is interesting to note that the reference stars magnitude measurements differ slightly from the one listed on The Sky.

 As shown above the difference in magnitude is not a bias so the only explanation I can give is that my measurements were not through any scientific filters so there is no normalization of data between the stars.  

 Isaac Cruz

The Octadome

8/1/07