instrumental features and problems
Vignetting on detectors 1 and 4
Adjacent echelle orders are isolated in the spectrograph by usage of an additional slit after the pre-dispersing prism (so-called intermediate slit). For settings below 1450 nm, the slit had to be set up in a way that the outer parts of detectors 1 and 4 are not fully exposed. See Sect. 15 of the User Manual for a detailed description of the usable wavelength ranges for each setting. Since October 2007, updated values for the wavelength range that is not affected by adjacent orders are written into the header keywords INS WLEN MIN and INS WLEN MAX.
Fixed-pattern noise and detector non-linearity
The CRIRES detectors have significant gain variations from pixel to pixel which are strongest in dispersion direction on detectors 1 and 4 (and in cross-dispersion direction on detectors 2 and 3). This results in an odd-even column effect in the extracted spectra for these detectors. The gain variations can be corrected with flat fields but in several cases a small pattern remains present after flat-fielding. Since 7 June 2007, daily flat fields are measured with DIT = 2 sec (instead of 1 sec). This significantly improved correction for fixed-pattern noise.
The detectors also become non-linear for exposure levels above about 4000 ADUs. This effect is measured by a series of dedicated flats with increasing exposure level (so-called detlin flats). A quadratic function is fitted for each pixel and the coefficients can be used in the reduction cascade for correcting raw frames.
Static calibration files (including coefficient files for correction of non-linearity) can be dowloaded here.
2010, July to September: tilt between intermediate and entrance slit
CRIRES underwent a technical intervention between 8 and 17 July 2010 to solve the problems with the entrance slit. After this intervention it has become apparent that the problem could not be solved entirely. Whereas the slit width is adjustable to different sizes again, a tilt between the intermediate and entrance slit was detected. This tilt caused a strong vignetting of the 1st and 4th detector and flux losses along the slit. Nevertheless, useful observations can be obtained for the 2nd and 3rd detector using a small nod throw of about 5 arcsecs. The slits were finally aligned again early September 2010. All data obtained after 5 September are of nominal quality.
2010, May to July: slit width varied along length of slit
After the May 2010 intervention, the width of the entrance slit varied along the slit length; the width was smaller on the bottom of the science detector and larger on the top. This resulted in a change of the spectral resolution of about 20% from bottom to top. There was also a steep gradient in flux, visible especially in sky background and in flat fields. All observations taken after 28 May 2010 until 7 July are affected.
2009, January: detector position shifted
After the January 2009 intervention, the position of the detectors has shifted in Y direction. This affects all measurements taken in February 2009 and later. Calibrations taken before the intervention cannot be used for observations after the intervention. This is especially important for the correction of detector non-linearity. New non-linearity coefficients have been determined in March 2009; the files can be downloaded here.
2008, April to June: reduced slit length
After the instrument upgrade early April 2008, the field-of-view along the slit (and therefore the effective slit length) was reduced to 12 arcsec. On the detector, only about 200 pixels in Y direction were illuminated. The full slit length was restored for observations after 11 June 2008.
2007, March to May: additional vignetting during early Service Mode observations
Accurate positioning of the pre-disperser
and the intermediate slit turned out to be difficult during the first
weeks of CRIRES Service Mode operations in P79 (March to May 2007). Although measures have been taken
to stabilise these elements, some observations were affected by additional
vignetting and/or contamination from adjacent orders on detectors 1 and
4. All observations have been checked for these effects and have been
repeated where appropriate.