EBOUNDS Extension in SIS Response Matrices
1996 Jul 30
It has come to our attention that v1.0 of sisrmg has been released with 'linear smoothing of the EBOUNDS extension' as the default (both in the original FTOOLS v3.5 release and in the ASCA-specific update).
The EBOUNDS extension in a response matrix contains the approximate channel to energy conversion; it is used by XSPEC for plotting only (and not for fitting). This conversion, for ASCA SIS, is not completely linear, even in the PI space: there is a departure from linearity above 10 keV due to incomplete charge collection, and some nonlinearity around the Si edge.
The linear smoothing option smoothes all these non-linearities out. However, since E>10 keV is included in the smoothing, the resultant scale is roughly 3.6 eV per channel compared with the correct average below 10 keV of 3.65 eV per channel. This results in a 6.4 keV Fe line apparently plotted at 6.3 keV instead; a 6.4 keV line model will reproduce a 6.4 keV line in the data, and both will be plotted at 6.3 keV. Even though the fit is unaffected, the appearance of the plot can be confusing, and may lead the user to incorrectly initialize the line energy (which may cause XSPEC not to converge properly).
We therefore recommend 'no smoothing' as the default when running sisrmg. One way of achieving this is to change one line in the parameter file (sisrmg.par in the bin.host directory of FTOOLS) from:
gains,i,h,-1,-1,12,"Amount of Ebounds smoothing"to
gains,i,h,0,-1,12,"Amount of Ebounds smoothing"
If you have any questions concerning ASCA, visit our Feedback form.