Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stsci.edu/ftp/instrument_news/FOS/pub/fos_isr101.ps Äàòà èçìåíåíèÿ: Wed Oct 26 22:45:39 1994 Äàòà èíäåêñèðîâàíèÿ: Sun Dec 23 02:12:09 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: solar system

where m is the magnitude of the object, and is the valuefrom the graph.
References
Baity,William,A.,ed.; ``Faint Object Spectrograph Science Verification Report''; STScI; May
1992.
Kinney,A.L.; ``Faint Object Spectrograph Instrument Handbook''; STScI; April 1992.
Rosa, M.; ``Correcting for FOS background''; ST­ECF Newsletter; August 1993.
This research has made use of the Simbad database, operated at CDS, Strasbourg, France.
Straylight V 15
=

Observations of HD144858
An example of the excess flux due to straylight can be seen in figure 1a,b where the observation of
16 CYG B are compared to a reference spectrum of a G2V star. The extra flux detected by the
diode array is readily apparent. Figure 1c shows the difference spectrum (i.e. the straylight).
Dividing this spectrum by the inverse sensitivity function, the actual number of counts/s detected
by the diode array is obtained. Figure 1d shows that the straylight is constant across the diode
array. There are approximately 1046 counts/s/A of straylight detected by the instrument for a G2V
star with V=0.
Predicting the Amount of Straylight
Spectra from the BPGS spectra library were used as the data to construct the graphs of the amount
of straylight as a function of spectral type. The spectra were all normalized to the same V magni­
tude by convolving them with a Johnson V passband. Each spectrum was then multiplied by the
detector's quantum efficiency and the aperture throughput for the 1.0 arcsecond round B aperture,
(B­3), then summed along the entire wavelength range to get the total light detected by the undis­
persed FOS. Finally, the plot was scaled to actual observations (e.g. of 16 Cyg B) and renormal­
ized for a V magnitude of 15 (Figs 3­10). Since observations of unreddened stars are not very
common (so direct comparison between our curve and the data would be meaningless), we have
also constructed curves for reddened spectra, using the same technique (the spectra were reddened
before the normalization to the same V magnitude). Fig1 shows the various curves and the
observed values of straylight from some observations. The agreement is quite satisfactory, espe­
cially since this method is more a predictive tool than a way to actually remove the straylight after
the observation.
It is clear from figure 3 that blue stars produce a larger amount of straylight than late type stars.
This is not surprising, given the sensitivity of the diodes to the UV and the fact that in our hypoth­
esis they are detecting essentially a fraction the whole spectrum. However, the `contamination' by
straylight is a problem only for objects with negligible UV flux: for hot stars, the amount of stray­
light represents only a very small percentage of the total flux detected, and although one may want
to subtract it, no obvious `distortion' of the contimuum is expected to be seen in these spectra.
Amount of Straylight in FOS Observations
To compute the estimate amount of straylight, one simply has to read the value of straylight
expected for the spectral type of the object and scale it to the appropriate magnitude using the
equation:
Straylight 10 0.4 m 15
--
( )
-- Straylight V 15
=
â
=

Instrument Science Report
CAL/FOS­101
8 June 1994
NOTE: THIS ELECTRONIC DOCUMENT DOES NOT INCORPORATE FIGURES
FOUND IN THE PAPER VERSION OF THE DOCUMENT!!
Removal of Straylight in FOS Observations
Ellyne Kinney and Roberto Gilmozzi
Introduction
The FOS suffers from a straylight problem that is especially apparent in observations of late­type
stars taken in the 1150­2100A region. Flux calibrated observations of these objects show an artifi­
cial increase in flux in the shorter wavelengths. The problem is easily seen when comparing
observations with observations from other instruments such as the HRS. Since the UV flux from
late type stars is very weak, the presence of straylight from the optical part of their spectrum often
completely swamps the spectral features of interest, making it very difficult to detect them over
such an abnormal background. In order to provide a way to estimate the amount of straylight to be
expected (so that exposure times can be adjusted to yield the appropriate signal to noise for the
spectral features one wants to study), we have developed a very simple `model' for this effect and
used it to predict the amount of straylight in actual observations.
The main idea behind this method of straylight removal is that the light is not due to grating scat­
ter. This assumption is supported by two observational evidences: that the straylight is constant (at
least in first approximation) across the entire diode array when viewed as raw counts/s/A (Fig 1),
and that its amount does not depend on the grating used (Fig 2), but only on the magnitude and
spectral type of the observed object. The conclusion we draw from these two points is that each
diode is behaving like a photometer (with a quantum efficiency curve which is the same for all
diodes, i.e. the undispersed one) which `sees' the same amount of light as all the others. In this
interpretation, the pixel to pixel variations that are indeed present (but ignored in our approxima­
tion) would not be wavelength dependent, but rather variations in illumination (due perhaps to
unknown geometrical effects).
Starting from this simple assumption we have been able to construct a curve of the response of
each diode to the light from stars of equal magnitude and of the various spectral types. We have
also produced curves for various values of reddening, and used actual observations to determine
the curves' zero points. (Figs 3­10)
A detailed analysis of the various backgrounds affecting FOS spectra has been made by M. Rosa.
His treatment of the straylight problem is very similar to the one described here, and we refer to
his paper for a more comprehensive description of the background removal from FOS spectra.