next up previous contents
Next: The Astrovid 2000 Video Up: manual Previous: The Santa Barbara Instruments


The OptoMechanics Model 10C Spectrograph


(Rev. August 09, 2011)

General Information

Certain laboratory exercises will require you to use the Spectrograph and CCD together instead of simply doing straight CCD imaging. The CCD setup and operation are identical to that described in the first part of this manual. This section is designed to take you through the setup and operation of the spectrograph so that you can perform your laboratory without having to rely heavily on help from a Teaching Assistant. Like the CCD, the spectrograph is a delicate instrument and should be treated with care. The optical elements, in particular, are very expensive (as is the CCD). As with all delicate equipment, NEVER force any moving part beyond reasonable and expected resistance. Never move the telescope by pushing on the spectrograph or CCD, and never lean or support yourself by holding onto this equipment. Never, NEVER, touch any optical element. Oils from your skin will permanently embed into glass surfaces and optical coatings. It is preferable to leave small amounts of dust on optical surfaces rather than risk scratching or marring the surfaces with attempts at cleaning. If you are uncertain about how to operate any aspect of the spectrograph, please consult the T.A. for help. THINK BEFORE DOING.

Just as in the CCD section, this section will assume you have a general understanding of spectroscopy. You will find information on spectroscopy in the references listed in an appendix at the end of this manual if you need more information not provided here.

Use of the Spectrograph

Instrument Design

Figure 1 gives a view of the spectrograph and mounting at the tail of the 26" Clark refractor. An external view of the grating assembly including the dial can be seen in Figure 2. Figure 3 shows a layout of the internal optical design of the instrument. A sample spectral image of a star and comparison sources is in Figure 4.

The major elements of the instrument are:

Instrument Rotation Ring:
This allows rotation of the slit angle on the sky. At present, there are detents every 15 degrees of rotation, but random angles are also possible. Do not rotate slit position angle without TA assistance. The default position angle is 90 degrees.

Focal Reducer:
The first optical element of the spectrograph is a transfer lens. This lens serves to speed up the (rather slow) f/14.9 refractor telescope beam to f/10.

Slit Assembly:
At present, there are two slits available with spectrograph, a 50$\mu$m width slit which is equivalent to $\sim\!1.5$ arcseconds (calculated with f/10 and 26" aperture), and a 100 mm width slit which is equivalent to $\sim\!3.1$ arcseconds. You can tell which slit is in the beam by the color of the round knob (see Figure 1): black is the 50$\mu$m and silver is the 100$\mu$m. The ``slit'' is actually a set of three slits end-to-end. Your sky source will be imaged onto the central slit, which is 1.5 mm = 46.8 arcseconds in length. The two side slits are fed light from the comparison source via fiber optic cables for simultaneous recording of astronomical and comparison source spectra.

Figure 1: The Spectrograph and Mounting on the 26-inch Clark Refractor.

Comparison Source:
At present there are two available comparison lamps, mercury and neon/argon. In general, you will want to use both sources, although certain spectral regions are devoid of lines from one or the other. If this is the case for your particular wavelength set-up you can extend lamp life by using only the source needed for your work. The lamps have been intensity balanced to achieve approximately equal line intensities; however, these lamps do have a limited lifetime which results in gradual dimming. If you should notice peculiar line intensity ratios between the mercury and neon for the same exposure time, notify the TA. In this spirit, please DO NOT LEAVE THE LAMPS ON for more than the time required to make your comparison spectrum (typically $\sim\!1-5$ seconds). An atlas of the Hg-Ne/Ar source lines is given in Appendix A.

Guide Eyepiece:
At present, acquiring objects and guiding the telescope is done manually through the guiding eyepiece assembly. The image plane of the telescope (when in focus) is on the slit ``jaws''; an aluminized mirror surface into which the slit holes are cut. Thus, through the guide optics you observe that part of your source which does not fall through the slit. Your goal in guiding the instrument is to ensure that as much of your source flux falls through the slit as possible, and in general at the same place along the slit.

The collimator is a 225 mm focal length mirror which converts the converging telescope beam into a collimated beam.

Four gratings are available with this spectrograph with rulings of 240, 400, 600 and 1200 grooves per mm (at present, we do not have the 400 line grating). All are blazed (yield maximum reflection) near 5000Å. When used in first order, these gratings yield spectral resolutions of 295, 180, 120, and 60 Å/mm. Note that the ST8-CCD has pixels of size 9$\mu$m (at high resolution; when binned, the pixel size changes accordingly), a limitation which must be included in any calculation of the true resolution.

Figure 2: This image includes the grating knob, the counter displaying degrees of tilt angle (here, 28 degrees), as well as the lock to hold the grating at the specified angle.

You may NOT change gratings yourself. In general the grating needed for your experiment will be in place before you arrive at the Observatory. If this is not the case, please ask the TA for any changes. In order to remove the grating:

  1. Set the tile angle to 35 degrees.
  2. Unscrew (but don't remove) the 3 recessed 4-40 sockethead screws that fasten the grating assembly to the housing.
  3. Carefully withdraw assembly from housing. NOTE THAT GRATING IS UNCOVERED AND FACING CAMERA.
  4. Immediately and carefully install cover on grating cell, to protect grating during subsequent handling and storage.
  5. Turn over assembly and remove the two 8-32 sockethead screws holding the grating cell to the assembly.
  6. Install new grating cell and remove its cover just before replacing the assembly back into housing (STORE COVER IN SAFE PLACE).

The spectral region delivered from the grating to the detector is determined by the grating tilt, controlled by the lockable knob with counter (Figure 2). The counter displays degrees of tilt angle (always approach final tilt setting from lower values for reproducible results). Zero tilt angle is when the grating acts as a normal mirror. Tables of central wavelength delivered as a function of grating tilt are given in Appendix B. Note that the knob is quite loose until locked, so that if you do not lock the knob, you run the high risk of your spectral region shifting with time and telescope angle.

In principle, these gratings can be used in other than first order; in practice, both the refracting optics of the telescope as well as the CCD detector have limited spectral range (centered on the yellow part of the spectrum) which limit access to other orders. For this reason, the spectrograph does not presently have capability for adding order blocking filters. However, it is good practice when working with a spectrograph to consider the possibility of spectral contamination by higher orders.

The camera for this spectrograph is a 135 mm focal length Nikon lens for a 35 mm camera. It is focused in the usual way a camera lens is focused, by rotation of the knurled outer ring.

The standard detector for the spectrograph is the ST8 CCD, although other detectors (e.g., a 35-mm film camera) may also be used. Do not attempt to remove the CCD without TA assistance. With spectrographs on large telescopes, the first operation of an observer is to ensure that the detector is rotated so that the spectrum is aligned straight along CCD rows. However, as this is a delicate operation with the 10C, only the T.A. is allowed to perform this operation, which will be accomplished before you arrive. However, you should check the detector alignment by checking the placement of the same line on the pairs of comparison source on either side of your image.



Figure 3: The Optical Design of the Spectrograph.

Spectrograph Set Up

Generally, when you arrive at McCormick Observatory, the Spectrograph mounting will already be mounted to the tail end of the 26'' refractor, with a specific grating and slit appropriate to your experiment, and the ST-8 CCD connected to the camera optics of the Spectrograph. The equipment is generally protected with the silver cover, which you may need to remove. In addition to the CCD setup as described in section 1.1 of this manual, there is only 1 additional step required for preparing the spectrograph.

The spectrograph has two comparison lamps in a black housing connected to the main (blue) portion of the spectrograph. These lamps need to be plugged in to the powerstrip on the PC cart with the extension cord provided. You do not need to turn on the comparison lamps until they are needed, so for now simply plug in the lamps but leave them off.

Spectrograph Operation

The ST-8 CCD camera is usually used as the instrument detector for the 10C Spectrograph. Familiarize yourself thoroughly with operating this device before proceeding.

Before you start observing, you need to initialize the ST-8 using the MaxIm DL imaging software . Operation of the CCD is no different when using the spectrograph than if you were doing straight CCD imaging. However, since with spectroscopy you will often be working closer to the background levels of the CCD, it is important to make sure that the CCD temperature is cold (at least -10 C) and stable throughout your observations.

Using the spectrograph is slightly more complicated than straight CCD imaging, and you should follow the steps below. In some cases, you may need to iterate between steps.

Slit and Grating

Verify that you have the appropriate slit and grating for your experiment (see above). Check the position angle (PA) of the slit on the sky and see if it matches your needs. In general, PA = 90 degrees (East-West) is recommended. If the position angle needs to be rotated, contact the TA.

Camera Rotation

It makes sense, whenever possible, to align the dispersion axis of the spectral images along the rows of the CCD (i.e., along the long dimension). You obtain about 50% more spectral coverage on one image if you place the dispersion along the long axis of the CCD than along the short. If the CCD is not oriented in this way, discuss the matter with the TA. The image in Figure 4 was made with the dispersion in the less favorable CCD orientation.

Take a short exposure with a comparison source on, and verify - by comparing the position of Hg-Ne/Ar lines on each side of the image - that the CCD detector is aligned with the spectrograph (to within a pixel or two on either side). If the two comparison spectra are not well aligned (parallel), notify the T.A. Check also that the camera is tightly held onto the spectrograph; if the camera is loose, you may get unwanted rotations introduced into your images as the telescope moves around.

Spectral Range

Based on the goals of your experiment, you should have a clear idea of the spectral region you require. Based on the grating dispersion and the fact that the ST-8 has 1530 pixels of 9$\mu$m size along the dispersion axis, you can calculate your approximate accessible spectral range. You will want to vary the central wavelength so that that available range contains your desired astrophysical spectral features. For example, if you are observing planetary nebulae, which have the prominent [O III] emission line doublet at $\lambda =
(4959, 5007)$, you will want to have the central wavelength set to $\sim\!5000$Å. Appendix B lists the appropriate angles and wavelengths for a given grating. The spectrum in Figure 4 is with the low resolution, 240 line mm$^{-1}$ grating and centered near 5000Å.

Tilting the grating determines the central wavelength of the resulting spectrum. To set the grating tilt angle, you simply unlock and then turn the knob on the side of the spectrograph housing. It has an ``odometer''-like counter next to the knob which tells you the tilt angle in degrees. To unlock the knob, slide the black lock underneath the knob to the left. Then, twist the knob so the meter reads the angle required. When it is set, slide the lock back to the right. When setting the angle by twisting the knob, you will notice there is a significant amount of backlash. Therefore, you should always approach the desired value from a lower value of the angle. In this way, you are most likely to match your setup should you need to return to it in the future.

Because the tilt knob turns so freely, please take care to slowly move the grating.

Figure 4: Example spectrum of a V = 5 M star taken with the 240 line mm$^{-1}$ grating centered at about 5000Å. Exposure time was 45 seconds for the star and 1 second for the calibration lamps; with 2x2 binning, the maximum flux on the red end was 12,500 ADU. However, due to the effects of chromatic aberration from the 26-inch lens and the focal reducer, the image of the star fans out as it goes out of focus in the blue. As a consequence of this and the transmission properties of the system, the spectrum quickly drops in level to 0 ADU.

Focusing the Spectrograph

There are two focus operations involved in the use of any spectrograph: (1) You must first focus the exit beam from the grating onto the detector, and (2) you must focus the entrance beam from the telescope into the spectrograph onto the image plane of the slit. Astronomical spectrographs require one additional focus operation, which is that of the guiding optics on the image plane of the reflecting slit surface. Before taking a spectrum of your program objects, you will need to carry out these three focus operations. Remember that your spectrograph is most efficient when both the telescope and camera are in best focus, as this concentrates light from your source onto the smallest number of CCD pixels (increasing signal-to-noise). In addition, the best spectral resolution with any particular grating/slit setup is achieved when the camera is in best focus.

First, you must focus the spectrum on the CCD using the calibration lamps. This is done by turning the camera lens which is connected to the CCD head just as if you were focusing a normal 35 mm camera. You will also need to take short 1-3 second exposures. You can evaluate the focus by using MaxIm DL to extract vertical plots of the calibration spectra. First, turn on one of the calibration lamps to use as the spectrum for focusing (we recommend the Ne lamp because of the numerous amount of available lines). Open the `Focus' tab in the `CCD Control' window and take a short exposure of the entire CCD (make sure `Continuous' is not checked). Place the cursor as a thin box traversing a close line doublet on the calibration source spectrum to select a subframe. Check the `Continuous' option and click `Start Focus' to run the feed at 1-3 second exposures. With the `Line Profile' tool, place a vertical line cut across the doublet (which should appear as a double humped curve in the plot).

Begin with the camera lens turned to one limit and slowly adjust the camera lens focus by twisting the barrel of the lens. Look for the point in the graph at which the lines in the doublet become the most distinct from one another. This should coincide with the point at which the peak flux in the lines is greatest. At first, you may require large turns to get the focus ``in the ballpark"; but when closing in on the best focus, small turns of the lens yield noticeable results. You may need to iterate back and forth to find the very best position. The best focus is achieved when spectral lines are narrowest, when the maximum count value is the highest, and when pairs of lines appear most clearly separated.

It is important to remember that lenses suffer from chromatic aberration. Because the system you are using has three lenses (telescope, focal reducer and camera) it may not be possible to have your entire spectrum in focus at the same time. Figures 4 & 5 shows what happens when chromatic aberration affects a spectrum. Based on your experimental goals, you will have to decide where you want your best focus. Note that simply focusing on the center of the CCD chip means that the majority of the spectrum will not be in best focus. If you are interesting in achieving the best possible focus along the entire dispersion, it is often best to focus at about 1/4 of the way from either end.

Next focus the guide eyepiece on the slit. First put an eyepiece in the black tube projecting from the side of the spectrograph if there is not one already there. Depending on the nature of your source (bright versus dim, compact versus extended), you may find that you need a higher or lower magnification eyepiece for optimal guiding.

Turn on one of the comparison lamps using the toggle switches on the lamp housing. Twist the eyepiece in and out of the tube until the slit (the black lines visible against the comparison lamps) is in sharpest focus. Do not be confused by the comparison lamps; when the slit is in focus in the eyepiece, the comparison lamps will be out of focus.

Finally, you will need to focus the telescope image on the slit. To do this, first point the 26'' at a bright star or planet. If you are pointed correctly, you will see your object in the spectrograph eyepiece. Note that you will see the silhouettes of the comparison source fiber optic assemblies against the (brighter) night sky. With your bright object in the center of the field of the spectrograph eyepiece, crank the telescope focus until the image is in focus. The most accurate way to focus your object is to go through the focus and then go back slowly until you have the image as small as possible. Because the entrance beam to the slit is fairly slow (about f/10 after the focal reducer), it may be hard to tell when you have achieved best focus. In this case, you should take a series of CCD images with a bright star centered on the slit. Take exposures sufficiently long to integrate over the seeing (i.e., more than a few seconds), but short enough that your stellar spectrum is not saturated. At first, you may want to test the focus once every three or four complete cranks of the telescope focus, until you can zero in on the approximate region of best focus. Then do another focus series, with steps of 2 cranks in between, and so on. Depending on the seeing, the smallest focus difference you may be able to discriminate may be 1-2 cranks. Best focus is achieved when the spectrum is narrowest or has the highest flux in the central pixels.

Observing with the Spectrograph

Once you have set the focus of all three components of the spectrograph system, you are ready to begin observing your program objects. First point the telescope at the object you wish to observe. Once it is aligned in the finder, it should be visible through the guide eyepiece on the spectrograph. You will then need to adjust finely the position of the object so that it is bisected by the slit. In some cases (e.g., to observe two nearby objects simultaneously, or to align along the major axis of a galaxy or nebula) you may need or desire the slit to be in another orientation on the sky. Ask the T.A. to rotate the spectrograph position angle for you.

If there are no other considerations, we recommend using the spectrograph in a position angle of 90 degrees (i.e., an East-West slit). The slow telescope motion is much easier to control in the declination than in the right ascension direction, so it easier to place objects on the slit if the slit is at PA = 90 degrees. To place an object on the slit in this orientation, first roughly center the object in right ascension above or below the slit. Then lower or raise the object onto the slit with the declination slow motion.

When the object is aligned with the slit you are nearly ready to observe. Remember that you must make sure the object remains on the slit in the same position throughout the entire exposure to maximize efficiency. It is a good idea, therefore, to guide the exposure throughout the integration. Your goal in guiding is to ensure that as much object flux falls through the slit as possible, and in the same place along the slit. Thus, you will be guiding on the edges of objects - those parts that do not fall through the slit. Here, again, we have found using PA=90 degrees to be an advantage. The 26-inch seems to track fairly well in right ascension over the course of several minutes, and is even more steady in declination. Thus, it makes sense to keep the narrow dimension of the slit in the north-south direction, where there will be the smallest telescope movement (though refraction will come into play for long exposures at high zenith angle). At PA=90 degrees, any mistracking of the telescope in hour angle will be along the slit and, while not optimally concentrating flux in the fewest number of pixels, at least no flux is lost in this situation.

Take exposures using your favorite software package such as MaxIm DL. You will need to compare the spectrum of your object to the Hg-Ne/Ar comparison lamps for wavelength calibration purposes. Therefore you need to turn on the comparison lamps briefly during your exposure. Be sure not to leave them on too long, or they will saturate and you will not be able to accurately wavelength calibrate your spectrum. You may also risk ruining your object spectrum from charge leakage by supersaturated pixels. You will also shorten the lifetime of the comparison lamps. We have found that with 2x2 on-chip binning of the CCD, comparison lamp exposures of about 1 second are fine with the lowest resolution grating (240 lines mm$^{-1}$).

Keep in mind that the chip saturates at 65,535 ADU, and both your object spectrum and calibration lines must be below this to be useful. As a guide, with 2x2 on-chip binning and the 240 lines mm$^{-1}$ grating, magnitude 4-5 stars will saturate the chip in about a minute.

During your observations take care that the instrument will not run into anything (pier, ladder, fellow students) and be mindful of all cables.

When you are finished observing, follow the usual CCD shutdown procedures. The only additional step necessary for shutting down the spectrograph is unplugging the comparison lamps, rolling up the cord, and putting the cover on the spectrograph.

Reduction of Spectrographic Data


Once you have finished observing and are ready to analyze the data, you can log into one the Department of Astronomy's Astro 3130 Windows XP workstations (Room 233). See the the Computing Handbook ($\sim$hbp4c/comp-uting/handbook/Computing.pdf) for more information. You will be able to use the MaxIm DL and/or MIRA software package (whichever you prefer) to reduce and analyze your CCD data. The astro3130 account should already be set up to run these programs. Contact your TA if this is not the case.

First it is important to make a directory for the raw images (e.g., ``raw") that were taken at the telescope and a separate directory for all your processed images (e.g., ``proc"). This will prevent you from overwriting your raw data, which should be kept as is. These subdirectories can be made in your group directory on the astro3130 network drive (Orion). You should start by making copies of the raw images and putting them in the ``proc" directory.


In order to dark subtract an image, have your image frame open along with the corresponding dark frame of equal integration time. Open the Pixel Math command via the Process menu. Set the following options:

Hit OK and save the new image under a file that indicates it has been dark subtracted (ie. ``Image_sub.fits'').

With the Line Profile tool, take the mean over many columns by using a vertical box selection method in order to increase your signal-to-noise and to overcome the differences in focus with wavelength. After choosing the Vertical Box and Mean option in the Line Profile window, draw a selection box down the length of the spectrum that has a width approximately equal to the size of the out-of-focus flared portion.

Click the Export button to output the resulting plot to a .csv file to be opened with Microsoft Excel or some other spreadsheet program. Using the comparison lamp spectra as a guide (Appendix A), place a wavelength scale on each spectrum.


Have two images open in MIRA, the combined dark image and the object image. Open each image separately (i.e., not as an Image Set). Make the object image window the `active' window. This is done by clicking on that image.

Under Process$\backslash$Math, select `Image Arithmetic'. Under Image, choose the frame that contains the spectrum you want to analyze, choose `Subtract' as the Operation, and choose your dark frame as your Operand Image. Click Process and then save the new object image using `Save as'; select a new name for the image so as not to overwrite the old image (ie. ``Image_sub.fits'').

Repeat this same process for all of your observations.

Ne-Hg/Ar Comparison Sources

Wavelength source: National Institute of Standards and Technology (NIST).

For reference, H-Alpha and H-Beta occur at 6563 and 4861 Angstroms, respectively.

Figure 5: A composite of the wavelength calibration spectra. The left spectrum corresponds to the Neon calibration lamp, while the right shows the Mercury/Argon calibration spectrum (Argon lines are the broad, dark lines above 7000 Angstroms). The middle strip contains the combination both lamps. These spectra were taken using the low resolution 240g/mm grating.

Central Wavelength vs. Grating Tilt

  240 g/mm Grating 400 g/mm Grating
Wavelength (Å) Tilt ($^{\circ}$) Dispersion (Å/mm) Tilt ($^{\circ}$) Dispersion (Å/mm)
3500 2.60 290.2 4.35 176.0
4000 3.00 290.9 5.00 176.6
4500 3.35 291.6 5.60 177.2
5000 3.70 292.2 6.20 177.8
5500 4.10 292.8 6.85 178.3
6000 4.45 293.5 7.45 178.8
6500 4.85 294.1 8.10 179.4
7000 5.20 294.7 8.70 179.9
7500 5.60 295.3 9.35 180.3
8000 5.95 295.9 9.95 180.8
8500 6.35 296.4 10.60 181.2
9000 6.70 297.0 11.25 181.6
9500 7.10 297.5 11.85 182.0
10000 7.45 298.1 12.50 182.4
  600 g/mm Grating 1200 g/mm Grating
Wavelength (Å) Tilt ($^{\circ}$) Dispersion (Å/mm) Tilt ($^{\circ}$) Dispersion (Å/mm)
3500 6.50 118.7 13.15 60.9
4000 7.45 119.2 15.05 61.2
4500 8.40 119.7 17.00 61.4
5000 9.35 120.2 18.95 61.6
5500 10.30 120.7 20.90 61.7
6000 11.25 121.1 22.95 61.7
6500 12.20 121.5 24.95 61.7
7000 13.15 121.8 27.05 61.5
7500 14.10 122.1 29.15 61.3
8000 15.05 122.4 31.30 61.0
8500 16.00 122.7 33.50 60.6
9000 17.00 122.9 35.75 60.1
9500 17.95 123.1 38.10 59.5
10000 18.95 123.2 40.50 58.7

References on CCD Imaging and Spectroscopy

The preceding sections of the manual all assume a general knowledge of CCD Imaging and Spectroscopy. Prior to performing a CCD or Spectrograph laboratory exercise you should have received instruction on these instruments in class. If you wish to read about these subjects in more detail, however, the following list of references may be useful:

File Transfer to UNIX Workstations

After your observations have been completed, you will want to analyze your new data. You will most likely be using the UNIX based astronomical data reduction software IRAF (Image Reduction and Analysis Facility) or IDL (Interactive Data Language). Both of these packages will read FITS (Flexible Image Transport System) images, so it is imperative that when you save your images on the PC, to save them as FITS files and not TIFF or PC compressed/uncompressed files.

To use IRAF or IDL to analyze your images, you first need to transfer the files to the Astronomy Department's UNIX workstation cluster. To do this requires the following steps:

  1. Wheel the PC cart to the ``museum area'' of the observatory.

  2. Connect the co-ax ethernet cable to the port on the back of the PC.

  3. Boot up the computer.

  4. If the computer is not already at the DOS prompt, exit Windows.

  5. At the Dos prompt, type: ftp

  6. Login to either your account, or the class account (e.g. astr3130) on astsun.

  7. cd to the directory on astsun where you would like the images to be stored.

  8. type: lcd \ccdops to get to the CCDOPS directory on the PC.

  9. type: binary to ensure your images are transfered in the proper format.

  10. type: put filename replacing filename with the name of the image you have stored on the PC's disk.

  11. repeat the item 7 until you have transferred all of your images.

  12. type: bye to exit ftp.

With several classes using the PC for CCD operation, the disk will eventually fill up. You will therefore need to eventually delete some of your images from the PC disk after you have transferred them to the UNIX machines. However, DO NOT delete your images from the PC until you have tested them on the UNIX machines to make sure they were transferred properly.

Reduction of Spectrographic Data with IRAF

Viewing and Extracting Spectra

Once your spectral image is transferred to the Unix environment (Appendix D: File Transfer to UNIX Workstations), you can begin sophisticated analysis. If you are already familiar with the IRAF environment you know a number of ways to display and manipulate images. The following section supposes a familiarity with IRAF, and is intended primarily as a suggested approach to begin evaluating data and testing images from the spectrograph.

The spectra imaged by the spectrograph may not be perfectly aligned along columns, and the transmission optics of the spectrograph introduce some geometric distortions. Thus a more sophisticated approach is generally needed to view a spectrum than simply plotting columns or rows, and better sky subtraction is possible when these geometric problems are accounted for. The IRAF apall package is designed to do this preliminary geometric calculation, spectral extraction, and sky subtraction. The following is a cookbook to get started with apall, but you are strongly urged to read the help manual for apall to branch out from the technique outlined below.


Once in the IRAF environment, you must load the spectrographic IRAF processing packages; enter the following that are needed:

cl$>$ onedspec$<$cr$>$,
on$>$ twodspec$<$cr$>$,


tw$>$ apextract$<$cr$>$.

If spectra lie along CCD rows, enter:

ap$>$ dispaxis=1$<$cr$>$.

If along columns:

ap$>$ dispaxis=2$<$cr$>$.

Next, set the processing parameters for apall. Apall is a program that converts your two-dimensional image into a one-dimensional, extracted spectrum that is also sky subtracted. It does this by coadding pixels optimally across the dispersion direction (along the slit) of your spectrum at each pixel position along the dispersion, and then subtracting evaluations of the adjacent sky along the dispersion.

If there are doubts that the previous user may have left the apall parameters set in disarray, enter:

ap$>$ unlearn apall$<$cr$>$

Then, enter:

ap$>$ epar apall $<$cr$>$

and, with the $<$down$>$ or $<$up$>$ arrows, examine the parameters and reset the ones mentioned below. Any that are not mentioned may be left as is. To change a parameter, move to it with the $<$up$>$ or $<$down$>$ arrow, type in the new value, then move on. You do not need to type a $<$cr$>$ with each new entered value. The editting process is terminated with a control-$<$d$>$.

Parameters to set or modify:

First group:
format: multispec
interactive: yes
find: yes
recenter: yes
edit: yes
trace: yes
fittrace: yes
extract: yes
review: yes
extras: no

Default aperture parameters:
lower: -2.5
upper: 2.5

Default background parameters:
b_funct: chebyshev
b_order: 1
b_sample: -8:-4, 4:8
b_nave: 1
b_niter: 3

Automatic finding and ordering parameters:
nfind: 1
minsep: 25

Tracing parameters:
t_nsum: 25
t_step: 25
t_nlost: 3
t_niter: 1

Extraction parameters:
background: fit
skybox: 1
weights: variance
pfit: fit2d
clean: yes
readnoise: 15
gain: *

*To be determined.


You must convert your FITS format frames to IRAF format. By way of example, a FITS frame called ccd117.fits would be converted as follows:

cl$>$ rfits ccd117.fits$<$cr$>$

then answer

IRAF filename: ccd117.imh$<$cr$>$

to the ``IRAF filename:" output file query. Unlike with most IRAF tasks, the extensions .fits and .imh should be typed in full.

The use of apall described here is in interactive mode. This is recommended for getting your feet wet with the process. There are many layers of complexity you can add to this simple cookbook reduction with increased knowledge of apall. For example, cosmic rays in longer exposures of faint objects may occasionally require deviations from the process described. In this case, however, it is generally best to take multiple exposures (at least three) and attempt cosmic ray cleaning by stacking the images with the IRAF combine task, after setting the combine option to ``median" and the reject option to ``avsigclip", for example.

Two processes are here described: one for bright objects where a continuum spectrum is obvious across the entire spectrum, and one for faint objects where the continuum is faint or nonexistent.

For Bright Objects:

After converting to IRAF format, start the reduction of a frame with the example name ccd117.imh as follows:

ap$>$ apall ccd117.imh

Answer the first three questions asked with the default response (yes), which can be entered simply with a $<$cr$>$.

The first graph shown in your gterm window is a section cutting across the spectrum along the slit. The finding algorithm will indicate what it thinks is the location of your object spectrum. If you had centered the object in the slit, then the spectrum should be in the center of the image. If the automatic selection looks right, typing a $<$q$>$ will accept it and move to the next step of processing. Occasionally the selection will not be correct, as in the case of a very bright cosmic ray contamination, or in the case of a brighter source somewhere else in the slit, or in the case that the comparison source was selected. If you wish to alter the selection of object spectrum, strike a $<$d$>$ to ``delete" the automatic selection, place the cursor on the correct feature, and then strike $<$n$>$ to select a ``new" choice; then hit $<$q$>$ to move to the next step. Respond with $<$cr$>$ acceptance of default ``yes" answers until the next graph comes up. If you wish to abort the procedure at any time, begin entering ``no" responses to queries until you exit the program.

The second graph shows the location on the chip where the spectrum was traced at each point along the dispersion. It corresponds to the centroid of the cross-section of the spectrum along the slit at closely spaced intervals along the dispersion. Not infrequently, a hot pixel or cosmic ray will introduce unacceptable deviations. When this happens, the program may automatically delete points in the interactive fitting procedure, or you can delete the points by hand by placing the cursor on the point and typing $<$d$>$. Refit the spectrum by typing $<$f$>$. You may want to improve the fit to the trace by changing the order of the fitting function (to, say, third order), by typing:

:order 3

or by changing the fitting function to a Legendre polynomial or a spline3,

:function spline3

Always refit the spectrum with a $<$f$>$ after any such alterations. You may finish the fit by typing $<$q$>$. Apall will then ask if this fit information should be added to the database, to which a $<$cr$>$ response will give the default (yes) response. Next, you will be asked to review the spectrum (yes), and to show the spectrum (yes). If all is well, a graph showing your spectrum will appear. However, if the spectrum was too faint, the program may bomb here with a variety of error messages, and you may have to use the other method for extraction described below. The graph displayed is the raw spectrum. If the object is faint, or the exposure is long enough to accumulate cosmic rays that dominate the graph scaling, the ordinate may be set so that your object spectrum is unusably compressed. You may redo the display by placing the horizontal cursor just above your spectrum and typing a ``$>$" symbol, and then below your spectrum and typing a ``$<$" symbol. Alternatively, you may quit the plot with a $<$q$>$ and redisplay using the splot task. When you quit, a $<$cr$>$ response to the query about whether to write the plot to the disk. The program will ask you the name of the file into which to write out the extracted spectrum. It will pick a name it likes, but this name may be too long for your liking (the IRAF default is designed to accommodate multiple spectral extractions from one image), and you may wish to type a new name. Whatever you type will be appended with the suffix ``.ms.imh". For example, if your image name is ``ccd117.imh", you may type in that you want the extracted, one-dimensional spectrum to be named ``ccd117", and the raw, 1-D spectrum will ultimately be put into an IRAF format file called ``".

You may then examine your spectrum with the IRAF program splot, by typing

ap$>$ splot $<$cr$>$

The splot package allows a very broad variety of manipulations that the user should peruse typing

ap$>$ help splot $<$cr$>$

For example, you may set the x and y windowing of the plot with ``:/xw" and ``:/yw" commands. A very useful feature is boxcar smoothing excessive noise in your spectrum by typing $<$s$>$ and entering the smoothing width in pixels (3 or 5 are good values) followed by a $<$cr$>$.

For Faint Objects:

The procedure for faint objects is almost the same as given above for bright ones, with the exception of two parameter changes. An object that has a very faint continuum can no longer be extracted properly, since the centroid of the spectrum may no longer be traceable. In this case, you must have another exposure of a bright source in the same slit position as a reference. You need to turn off the automatic "fittrace" routine and tell the program to refer to the other image. You can do this with the invocation of the apall routine as follows:

ap$>$ apall ccd117 refer=ccd116 -fittrace $<$cr$>$

where in this example the reference spectrum is contained in the image ccd116.imh, which has already been reduced into beforehand. The minus sign before the ``fittrace" turns off that option. Note: because of the effects of atmospheric dispersion, there may be differential shifts of wavelengths along the slit (see Fillipenko, 1982, PASP, 94, 715 for a description of the effects of atmospheric dispersion on spectrophotometry and how properly to minimize this problem with slit position), especially in the blue. In order to minimize the differences between reference spectrum curvature and target spectrum curvature, it is best to take both spectra at comparable sky position (airmass and hour angle) and the same slit orientation and positioning.

Extracting the Comparison Spectrum

In order to do the wavelength calibration, you need also to extract your comparison spectrum. Hopefully your chip was well aligned along columns or rows, so that there is not significant rotation which will cause shifts in rows/columns between the comparison and target spectra. If this is the case, you will need to extract BOTH comparison sources, do individual wavelength calibrations for each, and then interpolate the wavelength shifts as a function of distance between them.

Extracting the comparison source spectra is done similarly to the target object spectra above, with a few exceptions. First, you need to store the extracted spectrum in a place other than the apall default name, by explicitly specifying a name. Second, because the comparison sources are quite extended, you want to ensure that the ``background subtraction" region picked is outside the area on the image occupied by the comparison spectrum. In the following example of an apall call, the b_sample keyword is specified to account for: (1) the image is 2x2 binned, (2) we are dealing with the right hand comparison spectrum in Figure 4, and (3) the background sampling region is made further away from the center of the comparison source to be in a ``blank" part of the image:

apall bs8520b2 b_sample="-35:-25,40:50" output=bs8520b2.comp

The output image name will be

When the first graph is displayed by apall, ensure that you are on the correct place for the comparison spectrum you want (compare the imtool position for example). If not, move the extraction to the correct place.

The graph showing the trace will show a lot of scatter, but this is a function of the few well exposed places of spectrum (i.e., where the lines are) it has to work with. Therefore, it is recommended that a low-order fit, say linear, be used for the trace.

Wavelength Calibration

For wavelength calibration of your spectrum, use the IRAF identify command. Before doing so, you need access to a data file containing a list of line identifications for the comparison sources you used (Hg, Ne, or Hg+Ne). In Appendix A, an atlas of these sources, as well as a line list for IRAF is given. You should ask your TA for the location of this linelist file so you do not need to type it in.

Invoke the identify command on your image, specifying the linelist file (in the example here, which has both Ne and Hg, the linelist is given the name idhehg.dat) specifically:

identify bs8520b2.comp coordlist=idhehg.dat

You will then be presented with a graph of the extracted spectrum, as in Figure 6a. Pick a line that you think you recognize, place the cursor over it, and type an ``m". The program will ask you for the wavelength of the line and it's name, which you should take from the linelist. In the example, the feature at pixel 363 is the 4358.35 Åline of Hg, and you would reply to the query

4358.35 Hg

Do this again for at least one, but hopefully two other widely placed lines that you recognize. For example, the feature at pixel 154 is

5460.753 Hg(I)

and the feature at pixel 80 is

5852.4878 NeI(6).

At this point, the program will attempt to fit all of the other lines that it can. Type an ``l" to match other features with lines in your linelist. As soon as you do, you will be presented with a graph where the features are now shown in wavelength space, as in Figure 6b. Note that the little dashes show the locations of expected features in your linelist; they should match the features in your spectrum.

Typing a ``?" at any time will result in the presentation of a list of options for reworking the wavelength calibration. For example, use the ``+" and ``-" keys to step through the line identifications. If you screwed up your identifications, you can type an ``i" to initialize. Typing an ``f" will give rise to a graph as in Figure 6c, which shows how well the current fit of the dispersion works for your linelist. In general, the RMS residual should be less than 1 and you should have no residual in the pixel to wavelength mapping more than a few pixels. You seen in FIgure 6c that the fit is worse in the blue. This is because there are fewer blue lines in this comparison spectrum to constrain the fit. You also see that a first order spline3 function was used to to the fit. If you would like the fit to remove certain points, either remove the points you don't like with the cursor and the ``d" key, or type

:niter 10

to have the program iteratively throw out n-$\sigma$ outliers, where the n is given by the low_rej and high_rej options. Make sure to type an ``f" after each change in parameters. Use the h, i , j, k, and l keys to look at various representations of your fit. For example, ``h" gives the actual fit, instead of the residuals (Figure 6d), ``i" gives you the fitting errors in terms of Doppler velocity shifts (Figure 6e), whereas ``l" shows the non-linear component of the fit (which in the example of Figure 6e we can see there is some curvature in the dispersion solution).

Type a ``q" to leave the fitting window. Type another ``q" to leave the program and create a database file with your solution. In the case here, the database file is


and it contains the list of matched lines and the fitting parameters selected. This database will be needed to correct your target spectrum.

Figure 6: Doing the wavelength calibration in IRAF.

next up previous contents
Next: The Astrovid 2000 Video Up: manual Previous: The Santa Barbara Instruments

[an error occurred while processing this directive]