In some cases, MAKEE will either exit with an error or crash with a core file. These failures may be due to a failure to find a good echelle trace, or the program finds too many cosmic rays, or it cannot find the object profile, etc. If the message 'makee: All done.' does not appear on the last line of the log file, then MAKEE has failed to finish. In some cases, MAKEE will finish but with bad results. These could include non-physical artifacts in the spectra, or a spectrum without a calibrated the wavelength scale. These problems usually occur when the program is dealing with unexpected data. A discussion of the ideal input files for MAKEE is here . Listed below are examples of problem input data.

• Faint or poor trace source.

  The ideal 'trace' source is a hot, bright star such as a white dwarf standard (Feige 34, BD +28, etc). A fainter source (possibly the primary object itself) may yield a poor or failed echelle order trace solution, especially red objects which will show very faint blue orders. Also if the object has a lot of absorption, especially broad deep lines, MAKEE will have problems determining the trace. A flat field with the pinhole decker (D5) will usually yield a adequate trace (see 'Checking the Trace' ).

• Wide profile, short slit.

  If the object profile FWHM is wide relative to the slit length, the object extraction boundaries sometimes fall on the slope of the profile. Since the echelle orders are curved the extraction window will include fractions of pixels. An approximate correction is made for partial pixels, but this approximation fails when the profile is highly sloped at the extraction boundary. This occurs when the profile fills the slit window. Here is an example of a wide profile: (wide profile) (PS). This generally causes a jagged pattern in the resulting spectrum: (wide profile spectrum) (PS). The period of this pattern is related to the number of pixels in the dispersion direction for one whole pixel change in the position of the object.

• Slit too long.

  If the slit is too long, the echelle orders will overlap. MAKEE is designed to tolerate some overlap in the blue orders as long as there are enough non-overlapping echelle orders on the same CCD (MAKEE can extrapolate the boundaries of the slit window). However, is more than half the echelle orders overlap the reduction may fail. (In some cases, this will occur intentionally, if the observer does not need the data from the blue CCD, and prefers a long slit for the redder CCDs.)

• Object peak off center in the slit window.

  This will cause similar problems as the "wide profile, short slit" since the object extraction boundary will fall on the slope of the profile. Note that since the object position changes with wavelength due to differential atmospheric refraction, an object which is centered at 7000 Angstroms may be shifted in the slit by 2 or 3 arcseconds at 3500 Angstroms. Example: (shifted profile) (PS). This produces a very jagged, irregular spectrum: (shifted profile spectrum) .

• Complex profile, multiple objects.

  A complex profile can also create extraction problems if MAKEE has trouble determining the object center and boundaries. (MAKEE does not assume a gaussian profile for modelling and cosmic ray rejection, so a non-gaussian profile (by itself) will not necessarily produce extraction problems.) If there are distinct, separated peaks MAKEE will likely choose bad object boundaries and may use the area containing one of the peaks as background. You CAN extract multiple objects in the slit by running MAKEE with the 'uop=' option (see 'Controlling the Object Boundaries" ). In fact, this feature was added (in part) to allow extraction of spectra of gravitationally lensed quasar components separated by only 2 or 3 arcseconds.

• Object has resolved features (non-point source).

  If a source has spectral features (such as emission lines) which have a different profile than the continuum (such as in nearby AGN) MAKEE will have difficulty in rejecting cosmic rays. MAKEE assumes that the profile does not change rapidly along the dispersion direction. In this case, the spectrum may look very bad near the resolved feature. The observer may still get a good extraction by choosing a different extraction mode on the MAKEE command line (such 'mode=3' which will mask CRs in the sky only and do a straight sum of the object). See the command syntax: makee_syntax .

• Misalignment of object image and flat image.

  If the cross disperser angle (XDANGL) for the object image and flat field is not the same within about 0.02 units, the spectral extraction may fail (either no data or bad data). The boundaries of the slit for each order on the CCD image is determined by the flat field image. If these two images are misaligned, the background may be not be computed correctly and the relative flux will be wrong. You can see this effect in the 'profiles-###.ps' plot ( more info ). It is possible that the cross disperser will be misaligned even if the 'XDANGL' values are the same (although this should be very rare).

• Setup outside range of calibrated arc lamps.

  The MAKEE automated wavelength calibration is done by comparing the extracted arclamp spectra to a pre-existing database of calibrated arclamp spectra. The program correlates the arclamp spectra with the existing database and then uses the wavelength scale of the database spectrum to identify arclines. As long as reasonably close to an existing database entry the wavelength calibration should succeed. See existing arclamp calibrations for the RED cross disperser (as of June 2008).

• Object too faint.

  If the object is very faint, MAKEE may have problems finding the object center on the slit and the extraction boundaries. MAKEE will extrapolate the source centroid from adjacent echelle orders, but if there are several faint orders this may fail. (If there were clouds during the observations, and the object centroid cannot be found by MAKEE, the extraction will fail).

• Object too bright.

  If the object is very bright, the optimal extraction may fail. In this case, the profile shape is not well enough defined for the high S/N and the cosmic ray rejection may fail (too many rejections). (You can try a 'mode=3' MAKEE reduction to suppress optimal extraction).

• Bad Input Files

  If the ECHANGL and XDANGL are not closely matched in the input data files, MAKEE may fail, see: tolerances . Also, if the decker, slit size, or order blocking filters do not match this can also cause problems. If the quartz or Thorium-Argon lamps are turned off, or are not working, this will obviously cause problems. When trouble-shooting a failed reduction, check the input file parameters.

  ---

  Back to "MAKEE User Guide" main menu...

  (June 2008)