SAS watchout page

Can I use SASv6.1 with any processed event lists?

Unfortunately, sometimes you cannot. Changes in the content and format of the calibration files prevent in some cases event lists, generated with old versions of the SAS from being analyzed with SAS v6.1 Be aware that several old EPIC PPS event lists are present in the XMM-Newton Science Archive. This will be fixed with the reprocessing of all XMM-Newton data, which is expected to be completed early 2005. The SAS version used to generate the PPS products for a given observation is one of the quantities displays in the XSA: refer to the XSA Users' Guided Tour.

Here you are some examples of these inconsistencies and, whenever pertinent, an assessment of their scientific impact:

• calculations of effective areas, corresponding to spectra extracted from EPIC event lists produced with SAS v5.3.0 and earlier fail, with errors like:

  arfgen: warning (ExpiredCcfConstituent), The CCF constituent /xsadata/ccf//XMM_MISCDATA_0013.CCF has been accessed, but it is on the list of expired CCF constituents. This is the associated message:
  XMM_MISCDATA_0013.CCF : Cannot be used with SAS versions from 5.3.0 on-wards: it contains the wrong OPTICS_X/Y coordinates for PN
  Please consult the CCF release notes at the XMM-Newton SOC.

  or

  ** especget::arfgen: error (ImproperCcfConstituent), The present QuantumEf CCF `/xsadata/ccf//EMOS2_QUANTUMEF_0007.CCF does not contain pattern fraction data - please upgrade to a more recent version

• A reworking of the optical-axis positions and boresight CCF, coincident with the release of SAS 6.1 means that there may be an error of a few arc seconds when converting from sky to detector coordinates when spectra/light curves are extracted from event lists generated with earlier SAS version. This will only affect point sources very close to a chip gap where the calculation of the effective area may not correctly account for the presence of the gap.
• The CTI correction for EPIC-pn has changed between SAS versions, and the redistribution matrix generated by rmfgen using SAS 6.1, and the latest CCF is not compatible with all event files. In practice this will only affect Small Window and Large Window mode observations processed with a SAS version earlier than 5.4.1. Full Frame and Extended Full Frame observations are more or less unaffected. The problem may be minimized by producing pn spectra using the expression (PATTERN<=4 && FLAG==0) (which is the default used by especget).

  Spectral analysis using the latest SAS with current CCF will yield differences, due to the CTI, for the Small/Large Window modes below 1 keV, which will be evident as shifts in line energies of a few percent or mild changes (of the order of a few tenths in terms of spectral index) in continuum shape.

In all cases above, the safest workaround is to reprocess the ODF with the latest SAS. A list of the public SAS version is available here.