OGIP Calibration Note CAL/ROS/94-005
Region Selection based on Detector Coordinates: A Cautionary note on Extracted Spectra
Version: 1995 Jan 31
Code 668
NASA/GSFC,
Greenbelt, MD20771
This memo is intended to remind Guest Investigators (GOs) of the importance of the effects related to ``Ghost Imaging'' in extracted spectra extracted under certain criteria.
Release | Sections Changed | Brief Notes |
Date | ||
1994 Apr 13 | First public version | |
1995 Jan 31 | All | Made compatible with LaTeX2HTML software |
The fact that the ROSAT PSPC can not always properly position X-ray events
with very low energy has been well documented. This is the ``Ghost Imaging''
effect discussed at length in Nousek and Lesser 1993 (ROSAT newsletter 8).
We have had several guest investigator questions concerning spectra
extracted under certain criteria, and this memo is intended to remind
investigators of the importance of a spectral effect related to
``Ghost Imaging''.
The PSPC uses induced charge on the cathode wires to obtain photon
positions. For very low pulse-heights sometimes only one or two cathodes
register a signal above the lower level discriminator. In these cases the
position determination degenerates to a line or point (respectively).
This is known as the ``Ghost Imaging'' effect and is detailed in
Snowden, McCammon, Burows &
Mendenhall (1994, Ap. J. 424, 714) and Nousek & Lesser (1993).
Figure 1
shows a high signal-to-noise image in the detector coordinate system
in the ~ channel 12 -17 range, illustrating the checkered image
effect. Examples of ghost imaging can be seen by examination of
the detector maps in channel ranges 8-19 & 11-19 (available from the
legacy ftp area on /caldb/data/rosat/pspc/cpf/).
The cathode strips are ~ 5 arcmin (3.5 mm)
in width, this produces the size-
scale of the checkered pattern. During a typical observation, the source
is wobbled across a path several arcminutes in length over a timescale of
typically 400 seconds. If a selection is made on
detector coordinates such that spectra are preferentially extracted on or
off the wire locations, then the wrong soft count rate will be inferred and
a skewed spectrum may be extracted. As the cathode wire spacing is ~ 5
arcminutes, this becomes a pronounced problem if the extraction cell is
small (radius less than ~ 2.5 arcmin) .
Figure 2
shows a ratio of vela pulsar spectra taken in a circular cell of
radius 40 arcsec, the numerator is the on-wire spectrum, the denominator is the
off-wire spectrum. The spectrum taken at the wire intersection is very much
harder than the spectrum taken between the wires. In Figure 1,
the effect
is a factor ~ 4 difference in counts in channel 12 (the high energy
ratio is not exactly 1 because of
the slight difference in time spent at the on- and off-wire locations).
It is obvious that
these two spectra will give very different spectral fit results.
Investigators may notice a similar effect if they use a small extraction cell
in (one of the rare) unwobbled or very brief observations (such as some early
calibration observations where the exposure time was comparable to the
wobble cycle time of ~ 400 seconds), when soft enhancements and
deficits are not averaged out.
In summary, investigators should be careful when using a small extraction
cell based on detector coordinate selections, or in brief or unwobbled
observations. Spectra extracted under these criteria can have a soft count
rate (below channel 20) which is significantly enhanced or suppressed.
1 INTRODUCTION
2 DETECTOR COORDINATE SELECTION
FIGURES
The following useful links are available (in the HTML version of this document only):