UNSWIRF Observing Procedure

Observing with UNSWIRF differs little from the normal mode of observing with IRIS, but does require a little more setting up. Here we outline the basic setup procedure, and point out some traps for the unwary.


  1. If IRIS is already running, you may still wish to restart it, particularly as a new data directory needs to be created each day. Type exit, and answer y to Exit from Observer also? (y/n).
  2. On an NCD Xterm, connect to aat40a, and login as OBSERVER.
  3. At the 40a> prompt, enter system irisct, followed by xon node, where node is the name of the xterm you are working at (If IRIS is being run on the 4th floor rather than on the telescope, then you will also need to issue the commands ccdu_vac and iris_vac to specify the CCD and motor controllers to be used).
  4. To enable communication between UNSWIRF and the VAX, enter DITS_NETSTART (This is in addition to the dits_netstart command that needs to be issued from the Sun workstation when establishing communications with the UNSWIRF PC).
  5. To have the Unix server automatically create both FITS and IRAF versions of the raw files, login to aatssf, and issue the (seemingly redundant) command xhost aatssf. When the first image is taken, a window will pop up showing the status of file creation (make sure both "FITS" and "IRAF" are selected).
  6. To get the IRIS control system started, enter rvicl ccdn , where n is the number of the CCD controller given on the whiteboard on the telescope console. This will bring up an IRIS control window labelled IRISCT. Wait a couple of minutes until you see the Iris: prompt in this window. If attempts to initialise IRIS are met with MAJOR SYSTEMS FAILURE, check that the telescope control computer is running (Right Ascension display is updating), or else the XMEM may need to be restarted. Error messages that have scrolled past in the IRISCT window can be recalled with the Prev Screen key.
  7. At the ICL> prompt (not the Iris: prompt in the other window), enter LOAD DISK$USER:[SLL.IRIS]DISPn (where n is the number of the CCD controller given on the whiteboard on the telescope console). This will give a more useful XMEM display, including a greyscale lookup table with which any saturated pixels will appear red. It may take some minutes before all the actions are completed.
  8. To define the extra software needed to control the FP in tandem with IRIS and telescope offsetting, enter LOAD DISK$DATA:[OBSRED.UNSWIRF]UNSW.ICL at the next ICL> prompt.
  9. In the IRISCT window, check the following:
  10. If necessary, set Idle Mth to IRIS_IDLE1 by typing IME 1 in the IRIS window, and set the idle time with IT 1.5.
  11. Generally, use Observing Method 1 (ME 1) and Idle Method 1 (IME 1) for tests, and Observing Method 4 (ME 4) and Idle Method 2 (IME 2) for actual observing. Setting the idle time to the same as that used for a dark exposure is useful for looking for lines in real time, but remember to set it back to 1.5 before observing, in order to cut down on overheads.
  12. Commands can be sent manually to the FP by entering in the ICL window the line obeyw aatssy##UNSWIRF SEND command, where command could be beep (beep the UNSWIRF terminal), "z 500", "slide out", or any of the other UNSWIRF commands.
  13. It is a good idea to stow the FP under the slide cover during the day (to minimise dust falling on it), which is accomplished with the slide out command. To return the FP to the observing position, do a slide in, followed by slide 150 (to better centre the etalon in the IRIS field).


Although individual exposures and FP settings can be commanded manually, the repetitive nature of infrared observing makes it much more efficient to set up a file specifying the parameters for each observation (object name, FP spacing, telescope offset, etc.). This file can then be executed by ICL, and the whole sequence of exposures repeated if necessary.

These run files can be created in any directory on the VAX, provided it has World readership permission. We suggest using the directory DISK$DATA:[OBSRED.UNSWIRF], which can also be accessed from the Sun network as /vaxdata/obsred/unswirf/. Use your favourite editor on either the VAX or Sun to create a file like the example shown below:

! command              name       z  TI  CY PE RP dR dD
aatssy##UNSWIRF M83nuc_2.17_z720 720 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z720 720 120  1  8  1  0 300
aatssy##UNSWIRF M83nuc_2.17_z770 770 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z770 770 120  1  8  1  0 300
aatssy##UNSWIRF M83nuc_2.17_z780 780 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z780 780 120  1  8  1  0 300
aatssy##UNSWIRF M83nuc_2.17_z790 790 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z790 790 120  1  8  1  0 300
aatssy##UNSWIRF M83nuc_2.17_z800 800 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z800 800 120  1  8  1  0 300
aatssy##UNSWIRF M83nuc_2.17_z810 810 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z810 810 120  1  8  1  0 300
aatssy##UNSWIRF M83nuc_2.17_z820 820 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z820 820 120  1  8  1  0 300
aatssy##UNSWIRF M83nuc_2.17_z830 830 120  1  8  1  0   0
aatssy##UNSWIRF M83nuc_2.17_z830 830 120  1  8  1  0 300

The first line of this file is a comment line, indicated by the ! mark. The first entry on each line is a command to the UNSWIRF controller via the SparcStation aatssy. The next entry is the object name you want inserted in the image header. The first number is the etalon z spacing you desire for the line of interest at the object's redshift (you can get this by using Galactic sources such as OMC-1 or M17 as references, or the program /home/aatssy/unswirf/fpz). The second number is the total integration time (120 seconds), consisting of 1 cycle (CY=1), multiple readouts separated by 8 seconds (PE=8), and no repeats (RP=1). Note that owing to a bug in the ICL, requests for repeats in any line of this file other than the first will be ignored. The last 2 numbers are offsets in RA and Dec in seconds of arc (of polar axis rotation in the case of RA) from the telescope position prior to the run file being executed. After the run file has been executed, the telescope will return to this original position (unless the run file was aborted for some reason, in which case you will need to ask the night assistant to return the telescope to the original coordinates, after which you type tidy (twice), followed by base).

Although moving the telescope back and forth from object to sky position (here 300" E) will incur greater overheads, we strongly recommend observing this way, rather than doing all the object z settings, followed by all the sky z settings. Repeat sequences should use (say) an offset of 300 arcsec W, so that stars in the sky frames will not be repeated. Do not deliberately "jitter" exposure sequences like this on the object if accurate intensities and velocities are required, otherwise the fits generated with the assistance of arc-line width maps will be invalid. Most of the bad pixels and even the "holes" caused by stars in the sky frames can be removed in the reduction, continuum-subtraction, and cube-fitting, or if necessary, using imedit in IRAF.

The Perl script cube.pl can be used to generate these run files semi-automatically. You will need to edit the following parameters:

When you are happy with the new parameters, save the file and execute it with something like ./cube.pl > /vaxdata/obsred/unswirf/m83.dat to generate the run file.

Before executing the file, check that the required narrow-band blocking filter is in place, select Method 4 or Method 1, and if necessary, set Period to some value greater than zero, but less than the total integration time.

Having edited the file to your satisfaction, you can execute it from the ICL> prompt with a command of the form unsw m83.dat. Chances are, this will be greeted with an error like Cannot open file; don't panic! Just press the up-arrow key, and re-issue the command. If this doesn't work, try deleting any Emacs backups of your file, as these can confuse the VAX-SUN disk naming system. Next, check that the file has world read and execute privilege: SET PROT=(W:RWED) DISK$DATA:[OBSRED.UNSWIRF]M83.DAT on the VAX, or chmod a+x /vaxdata/obsred/unswirf/m83.dat; chmod a+r /vaxdata/obsred/unswirf/m83.dat on the SUN. You may also want to try issuing the ICL command tidy twice. Give the unsw m83.dat command again, and keep trying: it should be accepted eventually!

The run file should now execute the desired sequence of observations, but you will still need to keep a record of the exposure sequence. Should the sequence crash for any reason, remember to issue the tidy command twice in ICL before continuing.


  1. Focussing

    With one of the 1% bandpass filters in place, ask the night assistant to slew to a star of 4th magnitude or brighter. When roughly centered on the array, switch to the smaller window (WINDOW IRIS25) and allow the night assistant to run through the standard FOCOFF procedure. In some cases, it may be necessary to override the auto-scaling of the IDLE display, with (say) DISPLAY SCALE,IDLE_ER,20000,30000. When focussing is completed (the focus does not change with the filter used), return to the normal display with DISPLAY AUTO,IDLE_ER and WINDOW IRIS128.

  2. Standard Stars

    For flux calibration, we recommend selecting one of the spectroscopic standard stars in Appendix 1.2 of the IRIS manual, at an airmass similar to that of your program object. A single integration of 10 seconds (with 10 NDRs) on a 4th magnitude star should suffice, but you should obtain two images with the star displaced 15 arcsec either side of the field centre. It is also suggested that you obtain images of the standard with all of the FP settings used on your program object to assist in scaling and continuum subtraction.

  3. Flatfields

    The variations in sensitivity of IRIS + UNSWIRF are such that you MUST obtain flatfields at each FP setting and with all the blocking filters you used for program objects. Because of the way the etalon may drift out of parallelism during the day, it is highly recommended that you obtain all necessary flatfields as soon as the night ends. WARNING: UNSWIRF observing is not for those who like to go to bed before the Sun rises! Be kind to your night assistant, and offer to close the mirror covers and switch on the building lights for him when you have finished.

    For flatfields, use the floodlamp on maximum brightness pointed at the white patch, with all the mirror covers open. Integration times of 20 seconds, with 2 second periods will give you of order 10000 counts above the background. You may want 3 flatfield images at each setting to allow medianing, but one will probably suffice. A global "dark" frame (lamp off) for each filter should suffice. Flatfield sequences can be commanded by a run file; make sure there are no RA or Dec offsets at the end of each line, since the telescope tracking will be turned off. You will still need to command filter changes yourself from the IRIS window.

  4. Arc Line Scans

    Since most sources will be too weak for the cube-fitting routine to treat line width as a free parameter, the best way to constrain it is by assuming the line to be unresolved, and using high S/N observations of an arc line to construct a map of line width over the etalon area. Use cube.pl to take a series of 16 images spanning the arc line centre, at intervals of 2 Z units. Use docube.cl and unswspec.cl to examine the resulting cube, to make sure the peak can be well determined over most of the image area. The following table gives suggestions for appropriate arc lines and exposures to match most lines of astrophysical interest:

    Line/filterArc lineExposureApproximate
    (µm)(µm)(Ti/Cy/Pe/Rp)z peak
    1.64 Ar 1.6436 40/1/4/1 50
    1.65 Ar 1.6519 10/1/1/1 -48
    2.12 Kr 2.1165 40/1/4/1 -262, +813
    2.16 Ar 2.1534 60/1/6/1 -280
    2.21 Kr 2.1902 10/1/1/1 -320
    2.25 Kr 2.2485 60/1/6/1 200
    2.34 Kr 2.3340 40/1/4/1 92

  5. Observing Strategies

    For astronomical objects, total integration times of 180 seconds per etalon setting are sufficient to guarantee sky-limited backgrounds in all filters, while still enabling a sequence of etalon settings on object and sky to be completed before the sky background changes appreciably. The whole sequence can then be repeated as often as required until the desired signal-to-noise is achieved. Note that the velocity differences between objects, even in our own Galaxy, can be significant (as much as 30 Z units), and unless accounted for using /home/aatssy/unswirf/fpz, can cause you to miss the line peak altogether.


  1. No Arc Lines Visible: Is the telescope power switched on? (Main Power key to left of observing console). Has the arc lamp fallen from the top of the slide? Has the CS100 gone out of lock? (See below).
  2. XMEM / IRIS Crash: If the XMEM should crash, i.e., the XMEM monitor display does not update (try swapping between Idle Methods) and/or the IRIS control window has crashed, then go over and press the black button marked COLD START on the XMEM control rack. After a minute or so, the words AAO XMEM V4.3 should come up on the XMEM display monitor. In the meantime, you will need to EXIT from the ICL window, and answer Y to Exit from OBSERVER also? Reissue the rvicl ccdn command, and then run through the entire startup procedure again.
  3. UNSWIRF Communication problems: While the ICL script is running, keep an eye on the (x,y,z) values returned by the UNSWIRF PC; if you get a warning of a possible communications problem, you may need to abort the sequence (use Ctrl-c twice). Possible causes include the PC needing to be rebooted, or (more likely) the /home disk of the Sun controlling UNSWIRF is 100% full (ask the night assistant to fix this).
  4. CS100 problems: The CS100 controller may occasionally go out of lock (the arc line image will go from roughly uniform illumination to an a semi-circular arc, the green Close Loop light on the front of the CS100 will go off, and one of the parameters x_overload, y_overload, z_overload in status will be non-zero). If this happens, you will need to slew the telescope back to the zenith (ask the night assistant to MARK the present telescope position first), go into the Cass cage, flick the switch below the Close Loop lamp down, cycle the CS100 on/off switch, depress briefly the Reset switch, then push up the Close Loop switch. The green light should come back on. If the etalon is still out of lock, check the seating of the cables to the etalon, as well as to the back of the PC.

Utilities and Analysis Software

A suggested software list and directory structure is as follows:

  1. ICL scripts in /vaxdata/obsred/unswirf/:
  2. Scripts in /opt/obsred/unswirf/ on aatssf:
    • cube.pl - Perl script used for generating UNSWIRF run files of astronomical sources and sky positions, or for arc line scans and/or flatfield sequences (for which no telescope offsets are required).
    • dokeys - a shell script that outputs the value of a specified keyword for a sequence of images.
    • login.cl - the usual IRAF startup script, specifying the directories in which the .imh and .pix files are to be stored (be sure to check that the environment variables home and imdir are appropriate for the machine you are working on).
    • loginuser.cl - a supplementary file that defines a number of useful UNSWIRF-related cl scripts described below.
    • sdf2iraf - a shell script that converts a sequence of .sdf files directly to their IRAF .imh and .pix equivalents. You will need to set the environment variable IRAFDIR to point to your IRAF home directory (e.g., setenv IRAFDIR /data/ssf/1/obsred/iraf/ccd_1/991121).
  3. IRAF reduction cl scripts in /opt/obsred/unswirf/iraf/scripts :
    • docube.cl - takes a series of arc-line scan images, subtracts a single dark frame from each, and stacks them into a cube, ready for processing in IRAF with unswslope.cl.
    • doscube.cl - same as docube.cl , but allows one to use a matching sequence of sky frames, instead of a single sky frame.
    • unswcube.cl - procedure for fully processing datacubes prior to fitting, i.e., sky-subtraction, flatfielding, rotation, cleaning of bad pixels, register images, scale and subtract continuum.
    • unswflat.cl - procedure to form a series of dark-subtracted dome flats automatically from a sequence of lamp-on / lamp-off images (if only one lamp-off image is available, then use unswflat2.cl ).
    • unswlin.cl - procedure to "pre-process" all raw UNSWIRF frames. Converts from ADU to e- by multiplying by 9.5, gets the etalon Z value from the comment line (or header), and places it in the keyword Z_ETALON.
    • unswphot.cl - for photometry of standard star images. Extracts the names, identities, and exposure times from a sequence of images, subtracts images of the same etalon Z (but opposite stellar offsets), flatfields the result, allows the user to mark the positions of the positive and negative stellar images, then carries out the photometry and calculates the sensitivities as a function of etalon Z.
    • unswproc.cl - more basic version of unswcube.cl. Sky-subtracts, flatfields, rotates, and cleans, but does not stack images into a cube.
    • unswspec.cl - allows the user to plot a spectrum (intensity as a function of etalon Z, or cube plane number), averaged over some region in x and y.
    • unswblank.cl - executes MCBA's Lorentzian fitting routine, then allows the user to interactively blank out pixels in both the intensity and velocity maps which fail to both exceed an intensity threshold and yield a sensible velocity.
    • unswfit.e - Michael Ashley's IRAF task for fitting Lorentzian instrumental profiles to each pixel of the datacube. The width can be a free parameter, fixed as a constant, or set by the width found from scans of an (unresolved) arc line of similar wavelength. The file unswfit.par should also be present in this directory.
    • unswslope.cl - procedure for fitting Lorentzian profiles to arc-line scans (using unswfit.e), followed by fitting of a plane to the "velocity" map to assess the etalon parallelism.
    • unswvel.cl - corrects the fitted velocity map of an astronomical object for any residual curvature, as mapped by the "velocity" image of a matching arc-line scan. Converts from etalon Z to km s-1, but the absolute velocity scale is arbitrary; an offset of 100 km s-1 is added to avoid velocity values near zero.
    • unswdisp.cl - task for sequentially displaying each plane of a cube at 5 second intervals ("movie mode").
    • xset.cl - a procedure that informs IRAF of the number of rows in your Xterm window. Useful to prevent help screens scrolling off the top of the page. Execute at startup.

If any of these programs are missing, you can always find the latest versions in /home/aatssy/unswirf/ (note that you need to be logged in to aatssy in order to see this directory).

Return to the main UNSWIRF page.

Last modified: November 21, 1999.
Stuart Ryder