Документ взят из кэша поисковой машины. Адрес оригинального документа : http://xmm.vilspa.esa.es/docs/documents/CAL-SRN-0183-1-0.ps.gz Дата изменения: Mon Nov 15 12:35:00 2004 Дата индексирования: Mon Oct 1 22:29:02 2012 Кодировка: Поисковые слова: arp 220

XMM-Newton CCF Release Note
XMM-CCF-REL-183
EPIC-pn spectral response
R. D. Saxton
11 November 2004
1 CCF components
Name of CCF VALDATE Blocks changed XSCS ag
EPN QUANTUMEF 0015.CCF 2000-01-01 FRACTION CHANNEL, FRACTION ENERGY
QE TOTAL, EBINS FRACTION NO
EPN REDIST 0009.CCF 2000-01-01 NOISE PARAMS, PARTEVENT PARAMS
THRESH LOSS NO
2 Changes
Recent calibration work has shown that the Epic-pn redistribution function is in practice
narrower than that represented in the current matrices. In this release the redistribution,
especially at low energies, has been sharpened and is now in better agreement with the
results from ground calibration.
The number of energies has been increased from 1319 to 2067 to provide a better descrip-
tion of the redistribution function around sharp features.
A technical change has been introduced in this release with some hard-coded redistribution
parameters being moved from the code into the EPN REDIST CCF le. This will allow
the redistribution matrices to be modied between releases of the SAS software by CCF
upgrades. In detail the EPN REDIST CCF now contains an extension NOISE PARAMS
which contains a 6 element vector for each observing mode, and independently for single
and double-pixel events, that can be used to calculate the width and height of the main
redistribution peak. It also contains the extension PARTEVENT PARAMS that holds the
arrays TAU IN, S IN and L IN as a function of photon energy. A placeholder extension
THRESH LOSS has been added which may eventually be used to hold the thresholding
parameters for each detector mode. This is currently not used in the code and is set to zero
in this release. The normalisation of the shelf is contained in the attribute FLATSHLF in
the header of the primary array and the Silicon escape fraction in the keyword SIESCFRN.
The constant, C, used in the redistribution calculation is held in the keyword C IN within
1

Figure 1: Residuals from a simultaneous t to an observation by Epic-pn, MOS and RGS of the blazar
MKN 421. The dierence between the MOS and pn spectra has reduced from a peak of 30% at 0.5 keV
with the current pn redistribution function to 15% with the new one.
the header of the PARTEVENT PARAMS extension.
In the quantum eфciency CCF no calibration change has been introduced. The pattern
fractions in energy space and total quantum eфciency have been extended to cover 2067
bins and the fractions in channel space have been recalculated to be consistent with the
new redistribution function.
3 Scientic Impact of this Update
While the discrepancy between Epic-pn and MOS spectra continues to be a source of
intense calibration activity this update demonstrably improves the agreement of the three
cameras below 1 keV. Figure 1 shows the change observed on a revolution 807 observation
of the blazar MKN 421.
The redistribution has been dened at ner energy intervals to enable spectral tting
software to more accurately calculate line centre positions. For instance around neutral
Iron K (6.4 keV) the redistribution function is now stored at 15 eV intervals whereas
it was previously tabulated only every 50 eV. This means that the best t line energy
may be found with xspec to the nearest 15 eV whereas previously the line energy was
quantised to the nearest 50 eV.
2

Now that the important parameters of the redistribution function are contained within
the CCF it will be easier to disseminate improvements in the calibration to the user base.
4 Estimated Scientic Quality
This update improves the agreement between the MOS and PN cameras at 0.5 to 1 keV
to about 15%. However, this is a complicated subject with observations made at dierent
times appearing to show varying levels of agreement between the cameras. The reader is
urged to consult XMM-SOC-CAL-TN-0052 for a more in-depth analysis.
5 Expected Updates
6 Test procedures
Test 1: These changes have been tested within the response matrix generation task
rmfgen by producing matrices for single, double and single plus double pixel event spectra
for all the observing modes and comparing them against canned matrices produced by
separate software at MPE.
7 Test results
Test 1: Figure 2 shows that the redistribution function for a single-pixel event spectra
taken from a PrimeFullWindow mode observation is consistent with the canned matrix.
The results for a double-pixel small window mode spectrum and a single plus double-pixel
Timing mode observation are shown in Figures 3 and 4. In all cases the SAS generated
matrix is nearly identical to the canned matrix.
8 Compatibility issues
The change introduced in the CAL and CALPNALGO code is not backwards compatible
and so, while the new CCF les EPN REDIST 0009 and EPN QUANTUMEF 0015 can
be used with earlier versions of the SAS, the new SAS, Version 6.1, can only be used with
these or later CCF versions.
3

Figure 2: A comparison of the redistribution functions for a full frame, single-pixel-only event spectrum
of a 0.5 keV emission line. The SAS-generated and canned matrices are indistinguishable.
Figure 3: A comparison of the redistribution functions for a small window mode, double-pixel-only event
spectrum of a 0.7 keV emission line. The dierences between the SAS-generated and canned matrices
are minimal.
4

Figure 4: A comparison of the redistribution functions for a timing mode, single-plus-double-pixel event
spectrum of a 0.6 keV emission line. The SAS-generated and canned matrices are indistinguishable.
5