Previous Article Next Article Table of Contents


Fringing Properties of the Loral 3K CCD on the...(1Dec95) R-C Spectrograph (from CTIO, NOAO Newsletter No. 44, December 1995) The Loral 3K x 1K CCD now offered in the Blue Air Schmidt camera for use on the Blanco 4-m telescope R-C, Echelle, and Argus spectrographs has, as expected, proved to be popular with users due to both its large format and very high quantum efficiency--especially at UV wavelengths. (See NOAO Newsletter No. 42 page 29 for a general description of the properties of this detector.) It has been our hope that this CCD would serve for all spectroscopic programs on the R-C spectrograph, and for all low-resolution applications on the Echelle spectrograph, thus decreasing the number of detector/camera combinations that we must support on these spectrographs. However, worries about the ability to remove the significant fringing of the Loral 3K CCD at wavelengths longer than 7500 A have kept us from withdrawing the Tek 1K/Folded Schmidt camera combination as an alternative for spectroscopy with the R-C and Echelle spectrographs in the red. We have now demonstrated that this fringing can be simply and successfully removed from R-C spectrograph data. Accordingly, the Tek 1K/Folded Schmidt camera combination will no longer be an option with the R-C spectrograph. Test were carried out with the R-C spectrograph during an engineering night in August. These were aimed at understanding more thoroughly the fringing properties of the Loral 3K CCD and determining how well the fringing could be removed through flat fielding. There were some pleasant surprises. Figure 1 shows normalized dome flat spectra obtained at four different tilts of grating KPGLF. (A low-dispersion grating was selected for these tests since past experience suggests that "de-fringing" is most difficult at the lowest dispersions.) Note that, because of the fast beam of the Blue Air Schmidt camera, the fringing amplitude depends not only on the wavelength of the light, but also on how far off axis that wavelength is imaged. The fringing observed at 9000 A thus varies from 12% peak-to-peak when that wavelength is centered on the CCD to less than 5% when lying near the edge of the detector. Hence, for some projects, a judicious choice of the center wavelength may result in a substantial decrease in the fringing at the wavelengths of greatest interest. For example, if one were interested in obtaining high signal-to-noise with this grating in the wavelength region of the Ca II triplet lines (~8500-8700 A), a tilt centered at around ~7500 A would give a factor of 2 less fringing at 8600 A than would a tilt centered on the latter wavelength. (We have informally dubbed this effect as the "Crotts Effect" in honor of the first observer who called it to our attention, Arlin Crotts.) Another interesting discovery made during this engineering run was that, contrary to our expectations, the fringe amplitude is not a strong function of the slit width. This is illustrated in Figure 2 where normalized dome flats for slit widths of 224 um (1.5"), 448 um (3"), and 1000 um (6.7") are compared. A close comparison of the 224 um and 1000 um slit width flats show that the fringe pattern and amplitude is the same to high precision. Hence, those of you who like to take both narrow and wide slit observations to obtain the highest spectrophotometric precision may not need to take two different sets of flat fields (at least for "de-fringing" purposes). Although it may be possible to minimize the fringing of the Loral 3K chip for some projects, the ability to remove the fringes present in the raw data depends on the quality of the flat-fielding that one can achieve. Clearly if there is significant flexure in the spectrograph or camera, the fringing pattern will be slightly shifted with respect to the pattern present in the flat field images. Because of this, our initial guess was that "de-fringing" would be most effective if the flat field exposures were taken at the same telescope position as the observations. Hence, we took quartz lamp exposures using the comparison lamp projector system of the spectrograph, and attempted to use these to flat-field observations taken at the same telescope position. To our disappointment, these tests revealed that the fringe pattern of the quartz lamp exposures was significantly different than the fringing observed in exposures of the night sky taken at the same position. This is undoubtedly due to the fact that the comparison lamp projection system does not perfectly match the f/7.8 beam of the telescope. [Figures not included] We, therefore, were forced to flat field our observations using dome flat exposures of the white spot obtained during the afternoon preceding the observations. To our pleasant surprise, the dome flats proved to be remarkably effective at removing the fringing. A typical example is illustrated in Figure 3. The lower plot in this diagram shows the extracted spectrum of the nucleus of the Seyfert 2 galaxy IC 5063 with no flat field correction applied. The fringing in this spectrum, which covers the approximate wavelength range 7000-9300 A (grating tilt = 5140; cf. Figure 1), is clearly visible, with the amplitude peaking just blueward of the Ca II triplet. The upper plot shows the same spectrum after flat fielding with the dome flat. Note that the fringing has essentially disappeared. The signal-to-noise level achieved in the "de-fringed" spectrum near the Ca II triplet lines is approximately 100:1. The fact that the dome flats are so successful in removing the fringing is a tribute to CTIO mechanical engineer Gabriel Perez and his group, who worked very hard to minimize the flexure in the Blue Air Schmidt dewar which houses the Loral 3K CCD. The conclusion, therefore, is that low-dispersion spectra taken with the Loral 3K CCD can be "de-fringed" to acceptable precision using standard dome flats. In the interest of decreasing the number of spectrograph/camera/CCD combinations that must be supported, we have therefore decided to withdraw the Tek 1K/Folded Schmidt camera as an option with the R-C Spectrograph. We will, however, continue to offer the Tek 1K/Folded Schmidt combination on the Echelle spectrograph until tests can be made comparing performance of this option with the Loral 3K/Blue Air Schmidt and the Tek 2K/Long Cameras combinations. Mark Phillips (mphillips@noao.edu), Steve Heathcote (sheathcote@noao.edu)
Previous Article Next Article Table of Contents