Dealing with EPIC Out-of-Time (OoT) events

Introduction In EPIC imaging modes, photons are not only registered during the actual integration interval, but also during the readout of a CCD. These so called Out-of-Time (OoT) events are assigned incorrect RAWY values, leading to a wrong energy correction. OoT events broaden the spectral features, and create a strip of events with wrongly reconstructed position. The fraction of OoT events scales with the (mode-dependent) ratio of integration and readout time, and is highest for pn Full Frame (6.3%) and Extended Full Frame (2.3%) modes (the user is referred to the XMM-Newton Users Handbook (EPIC chapter, section EPIC out-of-time events) for more details). Examples of the effects of OoT events in pn images and spectra are shown in the User Guide to the XMM-Newton Science Analysis System (EPIC chapter, Section EPIC pn Out-of-time events). It is important to stress that for most targets a correction of OoT events in the spectrum is not necessary. In any case, a correction is only necessary if OoT events overlap the source being investigated. Expected Outcome This thread will allow the user to create an image or spectrum cleaned from out-of-time events. SAS Tasks to be Used epchain Prerequisites Set up your SAS environment (following the SAS startup thread) Useful Links XMM-Newton Users' Handbook, Section 3.3.10 User Guide to the XMM-Newton Science Analysis System, Section 4.10 Caveats

---

Procedure

The OoT events correction is a "two-step" process:

1. generation of an OoT event list (together with a standard observation even list)
2. removing of OoT events from an image or spectrum

These steps are described below. In the following example, it will be assumed that a user intends to remove OoT events from a pn exposure, taken in Full Frame Mode.

Generation of an OoT event list

1. run epchain to generate a OoT event list (PiiiiiijjkkPNblllOOEVLInmmm.FIT)
   
   epchain runbackground=N keepintermediate=raw withoutoftime=Y
   
   
2. run epchain to generate an observation even list (PiiiiiijjkkPNblllPIEVLInmmm.FIT)
   
   epchain runatthkgen=N runepframes=N runbadpixfind=N runbadpix=N
   
   

In order to save time one does not have to re-create the attitude file, re-run epframes, badpixfind and badpix either. epchain will create an output event file with the name PiiiiiijjkkPNblllPIEVLInmmm.FIT (the standard name of a calibrated pn imaging mode event list).

Removing OoT events from an image

1. generate an image from the OoT event list
   
   evselect table=PiiiiiijjkkPNblllOOEVLInmmm.FIT imagebinning=binSize \
     imageset=PN_OoT_image.fits withimageset=yes xcolumn=X ycolumn=Y \
     ximagebinsize=80 yimagebinsize=80
   
   
2. generate an image from the observation event list
   
   evselect table=PiiiiiijjkkPNblllPIEVLInmmm.FIT imagebinning=binSize \
     imageset=PN_observation_image.fits withimageset=yes xcolumn=X ycolumn=Y \
     ximagebinsize=80 yimagebinsize=80
   
   
3. rescale the OoT image to the expected fraction of OoT events (6.3% for pn in Full Frame Mode; the user is referred to the EPIC chapter, section EPIC out-of-time events of the XMM-Newton Users Handbook for more details). The FTOOLS farith shall be used for this purpose
   
   farith PN_OoT_image.fits 0.063 PN_OoT_image_rescaled.fits MUL
   
   
4. subtract the rescaled OoT image from the observation image
   
   farith PN_observation_image.fits PN_OoT_image_rescaled.fits \
     PN_observation_clean_image.fits SUB
   
   

The image PN_observation_clean_image.fits is cleared from OoT events.

Removing OoT events from a spectrum

1. extract a source spectrum (PN_source_spectrum.fits) from the observation image:
   
   evselect table=PiiiiiijjkkPNblllPIEVLInmmm.FIT withspectrumset=yes \
     spectrumset=PN_source_spectrum.fits energycolumn=PI spectralbinsize=5 \
     withspecranges=yes specchannelmin=0 specchannelmax=20479 \
     expression='(FLAG==0) && (PATTERN<=4) && ((X,Y) IN circle(..,..,..))'
   
   
2. extract a spectrum from the OoT events
   
   evselect table=PiiiiiijjkkPNblllOOEVLInmmm.FIT withspectrumset=yes \
     spectrumset=PN_oot_spectrum.fits energycolumn=PI spectralbinsize=5 \
     withspecranges=yes specchannelmin=0 specchannelmax=20479 \
     expression='(FLAG==0) && (PATTERN<=4) && ((X,Y) IN circle(..,..,..))'
   
   
3. change the name of the COUNTS column in the OoT event spectrum to CTS_OOT
   
   fparkey value=CTS_OOT fitsfile=PN_oot_spectrum.fits+1 keyword=TTYPE2
   
   
4. copy the CTS_OOT column of the OoT event spectrum into the source spectrum
   
   faddcol infile=PN_source_spectrum.fits+1 colfile=PN_oot_spectrum.fits+1 \
     colname=CTS_OOT
   
   
5. multiply the values in the column CTS_OOT by 0.063 (6.3% for pn in Full Frame Mode; the user is referred to the EPIC chapter, section EPIC out-of-time events of the XMM-Newton Users Handbook for more details). As the PN_source_spectrum.fits file will be modified in the next two steps, the user may wish to create a backup first
   
   fcalc clobber=yes infile=PN_source_spectrum.fits+1 outfile=PN_source_spectrum.fits \
     clname=CTS_OOT expr=CTS_OOT*0.063
   
   
6. subtract the rescaled values of the CTS_OOT from the COUNTS column of the source spectrum
   
   fcalc clobber=yes infile=PN_source_spectrum.fits+1 outfile=PN_source_spectrum.fits \
     clname=COUNTS expr=COUNTS-CTS_OOT
   

Last Updated: 16 April 2010

---

Caveats

---