Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://xmm.vilspa.esa.es/docs/documents/CAL-SRN-0109-1-0.ps.gz
Äàòà èçìåíåíèÿ: Fri Feb 8 22:35:00 2002
Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 22:10:23 2012
Êîäèðîâêà:
XMM-Newton CCF Release Note
XMM-CCF-REL-109
EPIC MOS Energy Scale
B. Altieri
February 8, 2002
1 CCF components
Name of CCF VALDATE List of Blocks
changed
CAL VERSION XSCS ag
EMOS1 CTI 0006 2000-01-01T00:00:00 CTI EXTENDED NO
EMOS2 CTI 0006 2000-01-01T00:00:00 CTI EXTENDED NO
2 Changes
An improved MOS CTI (Charge Transfer Ineôciency) correction algorithm had been developped,
that reconstructs the energy of every x-ray event to a better accuracy.
Charge transfer losses are measured, calibrated and monitored by using an internal calibration
source. This has Mn K and Mn K lines of 5.89 keV and 6.4 keV, and a uorescent Al line at
1.49 keV.
As can be seen in gure 1 (for MOS2 at the Mn line) a linear t is good approximation of the
parallel CTI degradation since launch, although solar ares (indicated with red dotted lines) tends
to create discrete jumps in the CTI. The same behaviour is observed with the Al line parallel CTI
and for MOS1.
On the other hand the serial CTI is relatively constant since launch, as the transfert area is
shielded against (soft-protons) radiation. (up-to-date plots can always be found on the internal web
at : http://xmm.vilspa.esa.es/ xmmdoc/MOS/mos cti.html )
This new algorithm required a change of structure of CCF, with more column parameters added,
and a change of the SAS cal task, which is called by the ALGOID = 1 in the CCF.
1

XMM-Newton CCF Release XMM-CCF-REL-109 Page: 2
100 200 300
0.00
0.01
0.02
0.03
0.04
CCD1
( 4.5e­05+/­3.2e­07)*day+0.00067+/­0.00013 adc/tr
100 200 300
0.00
0.01
0.02
0.03
0.04
CCD2
( 4.9e­05+/­4.1e­07)*day+­0.0013+/­0.00015 adc/tr
100 200 300
0.00
0.01
0.02
0.03
0.04
CCD3
( 4.8e­05+/­3.7e­07)*day+­0.0012+/­0.00014 adc/tr
100 200 300
0.00
0.01
0.02
0.03
0.04
CCD4
( 4.1e­05+/­3.5e­07)*day+ 0.0023+/­0.00013 adc/tr
100 200 300
0.00
0.01
0.02
0.03
0.04
CCD5
( 4.4e­05+/­3.2e­07)*day+ 0.0011+/­0.00012 adc/tr
100 200 300
0.00
0.01
0.02
0.03
0.04
CCD6
( 5.1e­05+/­3.8e­07)*day+­0.0014+/­0.00015 adc/tr
100 200 300
0.00
0.01
0.02
0.03
0.04
CCD7
( 4.5e­05+/­3.8e­07)*day+0.00020+/­0.00014 adc/tr
Parallel(Mn) MOS2
Losses
(ADC/Transfer)
Revolution number
Figure 1: MOS2 transfer losses since launch at Mn energies

XMM-Newton CCF Release XMM-CCF-REL-109 Page: 3
The CTI loss for an event in position (RAWX,RAWY) on the CCD is the sum of serial losses
and parallel losses : CT I = CT IX:RAWX + CT IY:RAWY
The previous CTI was modelled in SAS v5.2 as :
 CT IX i = ôT:rate x + (a 0Xi + b 0Xi E) for the serial losses
 CT IY i = ôT:rate y E 1=2 + (a 0Y i + b 0Y i E) for the parallel losses
for CCDi (i = 1 to 7), where(a; b) 0X;Y i are the initial CTI pre-launch values from ground-based
test, ôT = T T launch , E is the PHA energy in ADUs and rate x , rate y the serial and parallel
degradation rates.
The new proposed CTI correction is :
 CT IX i = (a 1Xi + b 1Xi E)
 CT IY i = ôT:rate yi :E i + (a 1Y i + b 1Y i E)
where (a; b) 1X;Y i (CTI at launch) are extrapolated from in- ight measurements (see gure 1)
and i is a CCD dependent power index.
6 parameters per CCD are now required to characterize the CTI: 4 for parallel losses and 2
for serial losses. The serial CTI as observed in ight is constant (although the new CCF structure
allows for a time dependance) The time-dependant parallel CTI component has now a power law
dependence of energy ( ), CCD dependent, instead of the theoritical square root dependence as
before. The power indexes varies from 0.55 to 0.7. The degradation rate has also been adjusted per
CCD.
3 Scienti c Impact of this Update
The under-correction of the old CTI correction (SAS 5.2) shifts the line position up to 30-40 eV at
Mn energies by revolution 350, as can be seen in gure 2.
The new CTI correction with MOS CTI CCF version 6, recovers much of the energy losses and
bring back the Mn and Al line position almost at the expected position.
4 Estimated Scienti c Quality
The new CTI corection algorithm gives rather good results for MOS1, but still shows some under-
correction for MOS2. The Mn line position (5899 eV) for CCD1 is about 4 eV too low for MOS1
but up to 10 eV too low for MOS2 at revolution 388 (see gure 3). This is believed to be due to the

XMM-Newton CCF Release XMM-CCF-REL-109 Page: 4
Figure 2: Mn and Al line positions with the old (left) and the new (right) SAS CTI correction at
revolution 388 for CCDs 1 to 7
Figure 3: Mn and Al line positions with the new SAS CTI correction in revolution 316 and 355 for
CCDs 1 to 7
fact that the calibration source ux is relatively large compared with most astronomical exposures,
(especially for MOS2) and thus a systematic discrepancy due to a count-rate dependency (a high
count rate tends to decrease the CTI as the charge traps are more lled on average)
The two Mn and Al lines do not provide good leverage either on energy-dependent CTI e ects
for low signal packet magnitudes and energies between 1.5 and 6keV. The serial CTI is also possibly
too crudely modelled, some CCDs showing strange evolution or even CTI improvements !

XMM-Newton CCF Release XMM-CCF-REL-109 Page: 5
However the line position accuracy with this new CTI correction in SAS 5.3, is
believed to be better than 5 eV at 1-2 keV and better than 10 eV at 6 keV up, to
revolution 350.
5 Test procedures & results
The new CTI correction has been tested with MOS CTI CCF version 6 and the SAS version 5.3
alpha that includes the new algorithm, called by the keyword ALGOID = 1 in the CCF FITS header.
The results have be shown in the previous section.
6 Expected Updates
As the MOS CTI degradation evolves in the future, tuning of the CTI parameters (e.g. degradation
rate) will be needed. There is a hint of an acceleration of the degradation rate after revolution 300
and therefore a tendency of under-correction.
The under-correction of the MOS2 CTI will have to be properly understood, corrected, and
calibrated with emission lines of celestial targets (SNR).
This CTI correction also assumes a linear degradation with time, which might not be the case
in the long term. A di erent slope might be observed when leaving the maximum of the solar are.
An apparent temporary decrease of the CTI has been observed on internal calibration measure-
ments acquired during the eclipse season when the EMAE temperature was lower than nominal at
the start of revolution after the camera switch-o during perigee. Properly speaking it is a gain
change, but that could be corrected by at the level of the CTI correction, when the e ect be properly
calibrated. However it is believed to have a ected only few science observation, since they were not
scheduled at the moment of stong temperature gradient, just after the switch-on. To give an order
of magnitude, the observed e ect is about 1 ADU (3.5 eV) per degree (ôE=ôT = 3:5)