One of the issues we are currently investigating in the BI response is the behavior around the the Si-K edge at 1.84 keV. Here the relation between pulse height and energy has a discontinuity. (Figure 2.1)
In general this does not affect the application of the released response PHA spectra, but could in some cases cause a discontinuity at the Si-K edge in PI spectra. Current effects give residuals of the order of 2 sigma at most. In general the effect is subtle for most applications. Here are a few examples:
A comparison of spectra from the extended cluster Abell 1068 in pha-(Figure 2.2) and pi-(Figure 2.3) space. In pha space the residuals are essentially flat, while in pi-space there is a little kink at the position of the Si K edge.
Figure 2.4 shows a fit of a region of the supernova remnant Cas A. Even with a bright source we observe some extended residuals near 1.8 keV. In an attempt to fix the problem, we actually worsened the effect. In Figure 2.5 we show (as an exaggeration and for illustration purposes only) how badly pi spectra can fail in this region because in this case the fitting engine chose a negative normalization for the Si-K line.
A similar issue can arise near the position of the Ir edges. Again the effect - to our current knowledge - appears in TE mode only in pi-spectra and sometimes in cc-mode spectra and represents a slight mismatch of the actual position of the Ir-edge in the response (rmf) and the ancillary response file (arf). Figure 2.6 shows several pi spectra over various extended ring regions around the center of the pulsar synchrotron nebula in 3C58 (courtesy of Pat Slane, CfA). We clearly see a spurious 1-2 sigma "absorption trough" in the residuals at about 2.05 keV.
One of the problems still outstanding is the gain of the BI CCDs below 1 keV. We currently have calibration of the scale between 1 and 10 keV to better than roughly 0.5%. However, the current scale is degrading rapidly as we go below 1 keV and our previous estimates ranged from 10% at O VIII to almost 20% at the C K edge.
We have performed grating (LETG) measurements of the bright quasar PKS 2155-304 at large off-axis angles in order to put the spectral band between 15 and 45 A (275 to 820 eV) across S3.
Figure 2.7 shows the raw and fluxed spectrum at various off-axis angles. The data show nicely the O K edge at 23.3 A and the C K edge at 43.6 A. The feature at 34 A is caused by a QE difference between node 1 and 2 on S3.
Figure 2.8 shows the ratio of the LETG energy to the energy assigned to the mean pulse height of the CCD for the 8' offaxis observation. The data are binned into 1 A intervals. It can clearly be seen that these is systematic drift between 4% at oxygen 8-10 % between 35 and 40 A.
Figure 2.9 shows this in a closeup. This analysis tells us two things:
- The deviation of the current low energy gain appears to be better than 10% down
to the C edge and is thus a bit better than we previously advertised.
- We will now be able to fix the gain to a better than the 1-2% level (depending on energy) and we will do so in the next release.