Date: Wed, 25 Jun 2003 13:36:22 GMT
Subject: MOS/PN flux discrepancy
Full_Name: Daniel Evans
Submission from: (NULL) (22.214.171.124)
I have been analysing a source that I know to be piled-up. My investigations
showed that the pile-up was essentially confined to a circle of radius 20
arcseconds centred on the point source.
Therefore I extracted spectra using an annulus with an inner radius of 20
arcseconds and outer radius 50 arcseconds for the MOS1, MOS2 and pn cameras. I
used FLAG=0 and PATTERN=0 filtering to minimise the extent of any further
I believe that each of the source and background spectra were extracted OK,
although I did get a warning message backscale::arfgen: warning (zeroSumDetmap),
The filtered detector map has a zero normalisation - returning zero area. Try
using a flat detector map and/or more bins.
I then proceeded to generate rmf and arf files using the syntax:
rmfgen format=var rmfset=mos1_corespectrum_annulus_50_20_pattern0_detpsf.rmf
modelee=yes detxbins=100 detybins=100
and a similar syntax for the pn camera.
After grouping the spectra, I viewed them in XSPEC. The spectrum of my source is
quite complex, so I decided to limit my analysis to the 4-8 keV band, as I know
that there is just one component that vastly dominates the spectrum here.
The problem I find is that the flux and normalisation of the MOS1 and MOS2
cameras is 30-35% higher than that of the pn.
I note that extracting MOS and pn spectra using a circular extraction region
with a radius of 100 arcseconds does not produce any discrepancy in the flux.
So there's obviously a problem somewhere! Any help would be greatly
Thanks in advance,