Up to top level
AO15   AO16   AO17   AO18   AO19   Backgrounds   Calibration   Conference   Data   Docs   EPICMOS   EPICpn   Feedback   Gallery   Misc   OM   Pending   PhD_Theses   Publications   RGS   RadMonitor   SAS_Hardware   SAS_WS   SASv16.0   SASv16.0_Installation   SASv16.1   SASv16.1_Installation   SASv17.0   SASv17.0_Installation   SASv18.0   SASv18.0_Installation   SciSim   Simulators_other   Suggestions   Trash   Visibility   XMM-bouncing   XMM-news   XRPS   XSA   esas   incoming  

Logged in as guest

Viewing Calibration/8441
Full headers

From: lewis@astro.psu.edu
Subject: Intercalibration of PN, MOS and RGS
Compose reply
Download message
Move To:
7 replies: 1 2 3 4 5 6 7
7 followups: 1 2 3 4 5 6 7

Private message: yes  no

Notes:

Notification:


Date: Mon, 21 Apr 2003 16:29:01 GMT
From: lewis@astro.psu.edu
To: xmmhelp@xmm.vilspa.esa.es
CC: lewis@astro.psu.edu
Subject: Intercalibration of PN, MOS and RGS
Full_Name: Karen Lewis
Submission from: (NULL) (128.118.147.115)


Hello,
   I am analyzing an observation of 3C111, a radio loud AGN, which was taken 
in March of 2001. The data sets have quite good photon statistics and I am 
having three major problems. (These issues are best illustrated with plots, 
which I will have to send to you via email.) But, I can briefly describe
the problems I am having.

   The first is that when fitting an absorbed powerlaw to the data, which 
generally describes the data well, there is a very strange, double-humped
absoprtion in the PN data near 2 keV. I would have assumed that this was 
simply due to a calibration error in the ARF near the edge, but I have
spoken with other observers with high quality data sets and they have not 
had this problem, so it seems to be unique to my data set. In the RGS data
there are suggestions of similar features, but they are not quite the
same, so I don't believe that these features are real. (Furthermore, they 
don't show up in MOS data.) 

   I also notice that while the RGS data and PM/MOS
data agree fairly well above 0.5 keV, the agreement below 0.5 keV is 
terrible; I am not sure which, if any, of the data sets are reliable below 
0.5 keV.

   The third issue is that the results of the fits to the PN and the MOS
data are inconsistent. At the 95% confidence level, the column density range 
inferred from the PN and MOS data are 0.62-0.68 and 0.79-0.83 respectively.
The ranges in photon index are somewhat more consistent 1.62-1.67 in 
PN and 1.68-1.73 in MOS. 

    I realize that there will always be calibration problems between the 
various data sets, but I have not heard/read of other people having the same 
problems, so I wanted to contact you about them. 

     I am sure you will need more information than this to help me with my 
problems, but I thought I'd wait for you to tell me what you need. Thank
you very much for any help you can give me.

Karen Lewis


Reply 1

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: lewis@astro.psu.edu
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Date: Mon Apr 21 17:14:38 2003
Dear Karen,

 as a general suggestion, one should not use EPIC MOS deta below 0.5 keV,
and pn data below 0.35 keV.

 There are some known differences in the cross-calibration between the
EPIC cameras, which may become apparent with high-quality data. Have you
compared the results you get with the latest version of the EPIC and
RGS calibration documents? If not, here you are a couple of useful
references:

http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0018-2-1.pdf
http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0030-2-0.ps.gz

 A recent XMM-NEWS bullettin:

http://xmm.vilspa.esa.es/external/xmm_news/news_list/xmm_news_031.shtml

contained some important updates on the EPIC calibration.

 Last, but not least, please check carefully that your observation
is not affected by pile-up. This may cause a substantial (and camera-
dependent) spectral distorsion.

 If all the above is still not enough to clarify the calibration
issues you have encountered, please let me know and we will perform
an "ad-hoc" investigation.

 Regards, Matteo

P.S.: be reminded to specify in your reply which SAS version are you
using for the processing of your data, and when the Calibration Index
File you are using was generated.


Followup 1

Compose reply
Download message
Date: Fri, 25 Apr 2003 18:22:34 -0400 (EDT)
From: Karen Lewis <lewis@astro.psu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Hello,

  Thank you for pointing me to the new information on the calibration
status. Some of the information was very useful, but this information does
not solve all my problems.

  I re-reduced the data recently. I am using SAS 5.4.1. I generated the
CIF file on Feb 17th, 2003 using the calibration files which were released
in late January.

  I have checked for pile-up in several ways. The count rates in my data
sets is 11.4 counts/s for PN (Large Window mode) and 3.9 counts/s for MOS
(Small Window mode) which is below the limit suggested in the handbook.
When I run epatplot on the data, I actually find that I have slightly
fewer doubles and more singles that expected. Finally, I tried fitting
spectra in which I progressively removed more of PSF core and I find no
significant change in the column density or spectral slope. It doesn't
seem like my data are piled-up, but if you know of additional tests I
could run, I would be very grateful to hear about them.

  I did find the source of the problem that caused huge discrepancies
between the RGS data and PN/MOS data below 0.5 keV, so that is no longer a
problem.

  The problems I am having are best illustrated with figures. Rather than
send you large PS files, I thought it would be easier for you if I made
the pictures available on a website.

http://www.astro.psu.edu/users/lewis/XMM/xmm_problems.html

I've included my comments on the figures in the captions and summed up the
two problems I'm having at the end. If you prefer to get a bunch of PS or
PDF files, please let me know and I will send them. I apologize for the
very large amount of information I am throwing at you, but these problems
seem pretty significant and are affecting my ability to analyze the data.

  Thank you very much!

Karen Lewis




Reply 2

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: lewis@astro.psu.edu
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Date: Tue Apr 29 07:19:07 2003
Dear Karen,

1. the fact that the residuals at the shortest wavelenghts in the RGS
(< 7.5 A) have the same shape as the pn ones is purely coincidential.
Nonetheless, it is true that RGS spectra should be analyzed only
for E<1.65 keV. Scattering from the grating is not yet perfectly
calibrated in this regime. An empirical correction is being worked out
and should be included in a new release of the corresponding CCF,
scheduled to be ready by the end of May. Please refer to:

http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0030-2-0.ps.gz

for more information.

2. the residuals around 2 keV are most likely due to the MOS gold
edge not perfectly accounted for. If MOS and pn are fit together,
this appears as an excess in the MOS residuals. Please, have a
look at the curves in Section 1.7 of the EPIC calibration status
document:

http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0018-2-1.pdf

3. from your Fig.5, it seems unlikely that the RGS can significantly
constraint the N_H in this source, and be decisive in deciding which
measurement is right between MOS and pn. Anyhow, column density
determination with the pn are currently believed to be more accurate

 I hope this may help. Please, do not hesitate to ask any further
if something remains not clear.

 Regards, Matteo


Followup 2

Compose reply
Download message
Date: Tue, 29 Apr 2003 15:03:35 -0400 (EDT)
From: Karen Lewis <lewis@astro.psu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Hi,
   I am not sure that I explained my problems very clearly, because
I am very puzzled by some of your comments.

> 1. the fact that the residuals at the shortest wavelenghts in the RGS
> (< 7.5 A) have the same shape as the pn ones is purely coincidential.

I don't see that the residuals in the pn and RGS have the shape below 1.65
keV (7.5 A). If anything, the mos and RGS residuals look very similar in
the region. Where the RGS and pn look strikingly similar is in the
residual at 2 keV.

> 2. the residuals around 2 keV are most likely due to the MOS gold
> edge not perfectly accounted for. If MOS and pn are fit together,
> this appears as an excess in the MOS residuals. Please, have a
> look at the curves in Section 1.7 of the EPIC calibration status
> document:

But, if I fit the pn data, not with the model from the mos data, but it's
own best fit model, I cannot fit this feature. (In figure 2, the fit
I show is for the pn data.) The problem isn't even with fitting the data
sets simultaneously. I can't even fit the pn data alone, unless I remove
the region from 1.5-3.1 keV.

> 3. from your Fig.5, it seems unlikely that the RGS can significantly
> constraint the N_H in this source, and be decisive in deciding which
> measurement is right between MOS and pn. Anyhow, column density
> determination with the pn are currently believed to be more accurate

Figure 5 is misleading in the sense that each data set was fit with its
best fit model. The NH in the model for the PN is much lower than for the
RGS and mos data. If I tried to fit all the data sets with the mos model,
it would be more clear that the RGS data are well-fit by the model, while
the pn data are not. The RGS does not constrain the column density well

RGS 1, order 1: 0.75-0.91 (95% confidence)
RGS 1, order 2: 0.65-0.91
RGS 2, order 1: 0.71-0.86
RGS 2, order 2: 0.66-0.91

So, although the range is still quite large, the discrepancy between the
mos and pn columns is so large that the RGS data do favor the mos column.
The RGS data are only marginally consistent with the pn column, and only
in order 2 which has a very low count rate and doesn't extend below 0.8
keV.






Reply 3

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: lewis@astro.psu.edu
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Date: Wed Apr 30 12:42:35 2003
Dear Karen,

1. 

> Where the RGS and pn look strikingly similar is in the residual
> at 2 keV.

 This is exactely what I mean: this agreement is purely coincidential.
Anyhow, RGS data should not be currently analyzed at energies >
1.65 keV

2. concerning the residuals you see in the pn spectrum around 2 keV
and the mismatch between the N_H determined with the MOS and the
pn, I believe that your findings are due to a combination of old
processing and pile-up. I have reprocessed the pn data, starting from
the ODF available in the XSA. Even if I do not excise the PSF core
from the spectrum extraction region, I get residuals in the 2 keV
region within +\-5%, therefore significantly lower than you observe,
as well as no significant soft excess (I fit the data in the 0.5-15
keV energy range). However, "epatplot" suggests that in this case
the source could be substantially affected by pile-up. By removing
the PSF core (~5 arcminutes), the pile-up seems to disappear, and
the N_H determined by the pn is in better agreement with that
determined with the MOS:

N_H = 0.75 +\- 0.01
\Gamma = 1.624 +\- 0.013

although now it is the spetral shape is admittedly much harder than
in the MOS. This may be still due to unproperly removed pile-up.
Please, refer to:

http://xmm.vilspa.esa.es/external/xmm_news/news_list/xmm_news_031.shtml

for the latest news on piled-up sources.

 I would invite you to reproduce the steps above, and confirm me that
you get the same results. If you provide me with an anonymous ftp
site, I can upload a couple of plots describing my results.

 Regards, Matteo



Followup 3

Compose reply
Download message
Date: Wed, 30 Apr 2003 10:50:08 -0400 (EDT)
From: Karen Lewis <lewis@astro.psu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Hi Matteo,
   I will do as you suggest. Just to make sure I repeat your steps, did
you remove the flaring intervals of the data using the provided GTI tables
or did you create your own? Also, I'm assuming you used pattern 0-4 in the
PN data.

  If you would please upload the plots you mentioned

ftp.astro.psu.edu

and place the files in  /pub/incoming/lewis/

I will be able to verify that I get the correct results. Once again, thank
you for all your help. I am sorry I have been so confused.

Karen


* * * * * * * * * * * * * * * * * * * * *
* Karen T. Lewis                        *
* Department of Astronomy               *
* 525 Davey Lab                         *
* University Park, PA 16802             *
* (814) 863-0182                        *
* http://www.astro.psu.edu/users/lewis/ *
* * * * * * * * * * * * * * * * * * * * *

On Wed, 30 Apr 2003, Matteo Guainazzi wrote:

> Dear Karen,
>
> 1.
>
> > Where the RGS and pn look strikingly similar is in the residual
> > at 2 keV.
>
>  This is exactely what I mean: this agreement is purely coincidential.
> Anyhow, RGS data should not be currently analyzed at energies >
> 1.65 keV
>
> 2. concerning the residuals you see in the pn spectrum around 2 keV
> and the mismatch between the N_H determined with the MOS and the
> pn, I believe that your findings are due to a combination of old
> processing and pile-up. I have reprocessed the pn data, starting from
> the ODF available in the XSA. Even if I do not excise the PSF core
> from the spectrum extraction region, I get residuals in the 2 keV
> region within +\-5%, therefore significantly lower than you observe,
> as well as no significant soft excess (I fit the data in the 0.5-15
> keV energy range). However, "epatplot" suggests that in this case
> the source could be substantially affected by pile-up. By removing
> the PSF core (~5 arcminutes), the pile-up seems to disappear, and
> the N_H determined by the pn is in better agreement with that
> determined with the MOS:
>
> N_H = 0.75 +\- 0.01
> \Gamma = 1.624 +\- 0.013
>
> although now it is the spetral shape is admittedly much harder than
> in the MOS. This may be still due to unproperly removed pile-up.
> Please, refer to:
>
> http://xmm.vilspa.esa.es/external/xmm_news/news_list/xmm_news_031.shtml
>
> for the latest news on piled-up sources.
>
>  I would invite you to reproduce the steps above, and confirm me that
> you get the same results. If you provide me with an anonymous ftp
> site, I can upload a couple of plots describing my results.
>
>  Regards, Matteo
>
>
>
>





Reply 4

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: lewis@astro.psu.edu
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Date: Wed Apr 30 15:41:59 2003
Dear Karen,

 the answer to your questions is a double "no": I have not removed any
background flares (just to make things quickly, but it is of course
advisable to do so in the real life), and I have not used pn single
and double events. Using only "PATTERN=0" events (singles) decreases
the pile-up.

 I have put a couple of plots in the anonymous ftp area you indicated.
I hope they are self-explaining. If not, please do not hesitate to ask.

 Regards, Matteo


Followup 4

Compose reply
Download message
Date: Tue, 20 May 2003 14:57:27 -0400 (EDT)
From: Karen Lewis <lewis@astro.psu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Hi,

  I am sorry that it has taken me so long to get back to you. I was
running into problems and noticed some curious things, so I wanted
to run tests on some other data sets. After clearing up some problems, I
do in fact reproduce your results by excising the core and using the
single pixel events only.

I do have a question and comment though.

1) I agree that there is pile-up in my data. But I am wondering if you
know why epatplot showed that there were MORE single events and fewer
double events than expected from theory. Isn't this the opposite of what
one would see in piled-up data?

2) The feature at 2 keV is not significantly weaker when only the single
pixel events are used. In both the plots I sent you and in your plot
without the PSF core excision, the ratio of the feature to the model is as
low as 90%. When the core is excised, the feature is still there, but it
is buried in the large noise. I would never have noticed it in a core
excised spectrum and it no longer affects the fit, but when you plot it
against a data set which has the core, you can see that the feature is
still there and not substantially weaker.

   I spoke with other people in my department who have data with pretty
good photon statistics which are piled-up, often more severely than in my
case, and none of them see a similar feature in their data. So while this
feature no longer affects the analysis of my data (because I need to
excise the PSF core and the S/N decreases greatly), I think there may
still be something strange going on which can't be explained by pile-up.


Thank you for all your help!

Karen Lewis



Reply 5

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: lewis@astro.psu.edu
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Date: Fri May 23 07:48:19 2003
Dear Karen,

1. indeed, pile-up normally yields a pattern migration, in the sense
that lower pattern (single events) "migrate" towards higher pattern
(multiple events). Could you please put the POSTSCRIPT file of the
"epatplot" output on an ftp site? I would be interested to have a
look.

2. my point concerning the 2 keV feature was that the reprocessing
of the data with the latest SAS makes it smaller, not that it
decreases when the PSF core is excised. I do not have here any
longer the plots I prepared for you, but I believe you that
the main effect between the spectra with and without the PSF core
excised is an increase of the statistical noise, which may
"bury" systematics features. Unfortunately, the accuracy of the
pn response matrix in the 2 keV regime is of the order of 10%. Please
compare Fig.1.16 in
http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0018-2-1.pdf. This
is a fact of life. Such a figure represents, however, a "typical"
systematic uncertainty.

 Regards, Matteo


Followup 5

Compose reply
Download message
Date: Fri, 23 May 2003 11:54:54 -0400 (EDT)
From: Karen Lewis <lewis@astro.psu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Hi Matteo,
  The epatplot output can be found in

ftp ftp.astro.psu.edu
cd /pub/lewis

Thank you once again for all of your help in tracking down my problems.

Karen



Reply 6

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: lewis@astro.psu.edu
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Date: Tue May 27 07:28:06 2003
Dear Karen,

 unfortunatley, from the analysis of the "epatplot" output, it
seems that your pn data are affected by X-ray loading, which
produces a distorsion of the offset map, and, consequently of
the gain reconstruction (it is X-ray, and not optical loading as
the brightest optical source at the 3C111 position has a
V-magnitude of 18 only).

 X-ray loading introduces spectral hardening. Unfortunately,
X-ray loading can be measured, but there is currently no way to
correct for it. It probably explains the systematic harder pn
spectrum, with repect to the MOS ones.

 I apologize for the bad news. Regards,

 Matteo


Followup 6

Compose reply
Download message
Date: Tue, 27 May 2003 10:48:19 -0400 (EDT)
From: Karen Lewis <lewis@astro.psu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Hi,

  I am a little confused by what you mean by X-ray loading. I can't
imagine that the source (even in the flaring intervals) was actually
bright enough to saturate the detector and fill the wells completely. Is
X-ray loading related to the telemetry? The telemetry was saturated during
the flares, because the count rate in the source decreased during the
flares. But these intervals have been removed, so the remaining data
should be OK.
  I have been unable to find any information about X-ray loading in the
XMM websites. If you know of any sites, could you please let me know so I
can read about this problem some more. I hate to throw away the entire PN
data set until I at least understand the problem. Thanks.

Karen


On Tue, 27 May 2003, Matteo Guainazzi wrote:

> Dear Karen,
>
>  unfortunatley, from the analysis of the "epatplot" output, it
> seems that your pn data are affected by X-ray loading, which
> produces a distorsion of the offset map, and, consequently of
> the gain reconstruction (it is X-ray, and not optical loading as
> the brightest optical source at the 3C111 position has a
> V-magnitude of 18 only).
>
>  X-ray loading introduces spectral hardening. Unfortunately,
> X-ray loading can be measured, but there is currently no way to
> correct for it. It probably explains the systematic harder pn
> spectrum, with repect to the MOS ones.
>
>  I apologize for the bad news. Regards,
>
>  Matteo
>
>
>



Reply 7

Resend
From: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
To: lewis@astro.psu.edu
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Date: Fri May 30 09:13:56 2003
Dear Karen,

 perhaps my previous answer did not give you enough details to fully understand
the nature of the problem, which affects your pn data. I am trying to amend
here.

 X-ray loading occurs whenever a too large number of X-ray photons hit a pn
pixel
during the offset map calculation. The offset maps are calculated on board to
estimate and remove the electronic noise associated with dark current and to
correct
for optical loading. In other words, the offset maps allow properly to evaluate
the
zero energy level. The offset per pixel is determined from a sample of frames
from
which those with the three highest PHA values have been excluded. Nominally,
this
exclusion will avoid X-ray contamination of the offset map. X-ray loading
occurs
when the X-ray count rate is such that more than three X-ray photons are
included
in the sample, thus leading to an additional and incorrect offset shift for
that
pixel. The energy reconstruction of all subsequent events associated with that
pixel
will thus be wrong.

 One of the effects related to an incorrect subtraction of the offset is
"pattern migration" from double to single events. In other words: some double
events are not recognized as such. If this effect dominates (there are
competing
effects such as the offset to lower energies, and the generally simultaneous
presence of pile-up), the pattern distributions in the "epatplot" output (which
are
pattern fractional distributions normalized to one) will show an excess of
single and
a corresponding deficit of double events. This is exactly what your have
observed.

 The reason for X-ray loading to occur is the same as pile-up: the choice of an
inappropriate instrument mode, i.e. a mode whose readout time is too large to
cope with the source count rate. Unfortunately, there is no recipe yet how to
correct for X-ray loading, although studies are ongoing. In your specific case,
the effect of optical loading is unfortunately not negligible, and is probably
the reason for the systematically harder pn spectrum that you get.

 One possibility is to account for the energy shift is via the function "gain"
in XSPEC. Preliminary analysis of your observation carried on the pn instrument
team
suggests that the shift to lower energies of the central pixel events can be
accounted for by a "gain" fit with parameteres: slope=1.0; intercept=0.17 keV
(i.e. -34 ADU). Nonetheless more analysis would be necessary to consolidate
this
result. On the other hand, I am not sure you would like to use this strategy,
and I would not recommend it myself.

 Please, do not hesitate to contact me uf somnething is still unclear.

 Regards, Matteo


 Regards, Matteo



Followup 7

Compose reply
Download message
Date: Fri, 30 May 2003 06:18:36 -0400 (EDT)
From: Karen Lewis <lewis@astro.psu.edu>
To: Matteo Guainazzi <xmmhelp@xmm.vilspa.esa.es>
Subject: Re: Intercalibration of PN, MOS and RGS (PR#8441)
Hi,
  Thank you for your detailed reply. I understand much better what has
happened. I think I will just make do with the MOS data and our XTE
observation. The MOS data are pretty decent, because the source is bright
and we used the small window mode. I am sorry to have taken so much of
your time.

Karen


* * * * * * * * * * * * * * * * * * * * *
* Karen T. Lewis                        *
* Department of Astronomy               *
* 525 Davey Lab                         *
* University Park, PA 16802             *
* (814) 863-0182                        *
* http://www.astro.psu.edu/users/lewis/ *
* * * * * * * * * * * * * * * * * * * * *

On Fri, 30 May 2003, Matteo Guainazzi wrote:

> Dear Karen,
>
>  perhaps my previous answer did not give you enough details to fully
understand
> the nature of the problem, which affects your pn data. I am trying to amend
> here.
>
>  X-ray loading occurs whenever a too large number of X-ray photons hit a pn
> pixel
> during the offset map calculation. The offset maps are calculated on board
to
> estimate and remove the electronic noise associated with dark current and
to
> correct
> for optical loading. In other words, the offset maps allow properly to
evaluate
> the
> zero energy level. The offset per pixel is determined from a sample of
frames
> from
> which those with the three highest PHA values have been excluded.
Nominally,
> this
> exclusion will avoid X-ray contamination of the offset map. X-ray loading
> occurs
> when the X-ray count rate is such that more than three X-ray photons are
> included
> in the sample, thus leading to an additional and incorrect offset shift for
> that
> pixel. The energy reconstruction of all subsequent events associated with
that
> pixel
> will thus be wrong.
>
>  One of the effects related to an incorrect subtraction of the offset is
> "pattern migration" from double to single events. In other words: some
double
> events are not recognized as such. If this effect dominates (there are
> competing
> effects such as the offset to lower energies, and the generally
simultaneous
> presence of pile-up), the pattern distributions in the "epatplot" output
(which
> are
> pattern fractional distributions normalized to one) will show an excess of
> single and
> a corresponding deficit of double events. This is exactly what your have
> observed.
>
>  The reason for X-ray loading to occur is the same as pile-up: the choice
of an
> inappropriate instrument mode, i.e. a mode whose readout time is too large
to
> cope with the source count rate. Unfortunately, there is no recipe yet how
to
> correct for X-ray loading, although studies are ongoing. In your specific
case,
> the effect of optical loading is unfortunately not negligible, and is
probably
> the reason for the systematically harder pn spectrum that you get.
>
>  One possibility is to account for the energy shift is via the function
"gain"
> in XSPEC. Preliminary analysis of your observation carried on the pn
instrument
> team
> suggests that the shift to lower energies of the central pixel events can
be
> accounted for by a "gain" fit with parameteres: slope=1.0; intercept=0.17
keV
> (i.e. -34 ADU). Nonetheless more analysis would be necessary to consolidate
> this
> result. On the other hand, I am not sure you would like to use this
strategy,
> and I would not recommend it myself.
>
>  Please, do not hesitate to contact me uf somnething is still unclear.
>
>  Regards, Matteo
>
>
>  Regards, Matteo
>
>
>
>


Up to top level
AO15   AO16   AO17   AO18   AO19   Backgrounds   Calibration   Conference   Data   Docs   EPICMOS   EPICpn   Feedback   Gallery   Misc   OM   Pending   PhD_Theses   Publications   RGS   RadMonitor   SAS_Hardware   SAS_WS   SASv16.0   SASv16.0_Installation   SASv16.1   SASv16.1_Installation   SASv17.0   SASv17.0_Installation   SASv18.0   SASv18.0_Installation   SciSim   Simulators_other   Suggestions   Trash   Visibility   XMM-bouncing   XMM-news   XRPS   XSA   esas   incoming  

Logged in as guest


Please make your (short) question the subject of your request!


Web interface using JitterBug ... back to the XMM home page