Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://xmm.vilspa.esa.es/docs/documents/CAL-TN-0001-1-0.ps.gz Äàòà èçìåíåíèÿ: Tue Jul 8 12:38:21 2008 Äàòà èíäåêñèðîâàíèÿ: Mon Oct 1 21:14:48 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: zodiacal light

XMM
PHS Tools ­ EPIC Optical Loading
XMM­PS­TN­40 Issue draft
Written by: D. Lumb
November 16, 2000
i

european space agency
agence spatiale europÈenne
XMM Science Operations Team
Document No.: XMM­PS­TN­40
Issue/Rev.: draft
Date: November 16, 2000
Page: ii
1 Introduction
This note briefly describes issues related to one of the SSD­developed PHS tools, used to determine
if the correct optical blocking filter has been chosen for a particular exposure.
The EPIC CCDs are highly sensitive to visible light, therefore if an astronomical target has a
high optical­to­X­ray flux ratio, there is a possibility that the X­ray signal becomes contaminated
by optical photons. The resulting analysis of data would be impeded in three ways:
1. Shot noise on the optically generated photoelectrons will degrade overall system noise,
and hence the energy resolution. Spectral fitting will be inaccurate because the cali­
bration files will assume narrower spectral line features
2. The energy scale will be incorrectly registered, because a nominally zero signal pixel
will have a finite offset. For each optically generated photo­electron which is regis­
tered, the energy scale shifts about 3.6 eV. This is comparable with the accuracy with
which we expect brightest emission line features could be centroided.
3. Excess signal and noise fluctuations can affect the detection efficiency as well, by
disguising single pixel X­ray events as events split between pixels
To prevent this, the EPIC cameras include aluminised optical blocking filters, and also an
internal ``Offset Table'' which is calculated before each exposure to subtract the dc level of light
or other systematic zero shifts.
If these measures work perfectly, the above problems are minimised. However the use of a
thick blocking filter capable of minimising light contamination for all scenarios would necessarily
limit the softest X­ray response. EPIC has therefore decided to equip each camera with 4 separate
filter selections, named THICK, MEDIUM, THIN and OPEN. It follows that it is necessary for
the GO to select the filter which maximises the scientific return, by choosing the minimal optical
blocking that is required for the target of interest. To aid in the evaluation of the selection, a
ESTEC SSD­provided PHS tool is being produced to check the amount of light that contaminates
a measurement. It is TBD if the tool will be supplied to GOs for the AO­1, certainly the summary
of results will be provided as part of the Proposer's Guide, together with a set of guidelines for
use.
2 The Model
According to Zombeck's Handbook of astrophysical data, the rough measure of flux from a A0
star of zero visible magnitude is about 10 3 photons per second per sq cm and per Angstrom. For
sanity check purposes, then one could assume a telescope area 1800cm 2 , CCD efficiency 50% and
a waveband of about 4000 š A. This leads to a total flux from the star of 4 10 9 photons/sec in an
OPEN position, approximately focussed as the PSF.
The effective area of the telescopes for optical light will depend on the geometrical area, mirror
reflectivity and (for the MOS cameras) blocking due to the RGA. Visible waveband reflectivity
of gold is high, but a wavelength dependent distribution with a maximum value in the range 95
­ 97% was not available. Given the large uncertainties with other terms in our model, we simply
took the measured visible/UV geometric area measured at CSL as representative of the in­flight
performance (1800 cm 2 for the PN camera, with a 45% throughput due to RGA blocking).
Figure 1 shows the assumed filter wavelength­dependent transmission coefficients, based on
data from early engineering samples of filters. This data will require updating for the flight filters
when available.
The key parameter for performance is actually the amount of light per pixel per CCD readout
frame. There are some preliminary data showing that due to diffraction the optical light Point
Spread Function may have a Half Energy Width of 30 arcsec, rather than a Ÿ20 arcsec response
at X­ray wavelengths. However to be conservative, we assume that the fraction of light that is
detected at a pixel in the peak of the PSF distribution will be 2% and 15% for the MOS and PN
pixel sizes respectively.

european space agency
agence spatiale europÈenne
XMM Science Operations Team
Document No.: XMM­PS­TN­40
Issue/Rev.: draft
Date: November 16, 2000
Page: iii
Because the EPIC CCDs may operate in a wide range of readout modes, each presenting a
different frame time collection, the basic model proceeds to calculate the optical loading per unit
time, so that the exposure specific details may be abstracted.
The most problematic bright sources of optical loading will be point­like objects with visible
magnitudes brighter than 10, therefore for most purposes we can estimate the loading based on
simple stellar spectra. To perform this calculation we take simple black­body formulae as
f(–) = 6:73 \Theta 10 10
T e
10 \Gamma0:4m b P (–; T e ) (1)
photons cm \Gamma2 s \Gamma1 š A \Gamma1 where P is the Planck function
P = 1:26 \Theta 10 \Gamma8
– 4 (cm)T 3
e
1
e hc
–kT \Gamma 1
(2)
where T e is the effective temperature of the star, and m b is the bolometric magnitude of the
star, found from the bolometric colour correction apropriate to the stellar class.
A look­up table will be produced from this model. It will contain for each CCD camera and
filter combination a calculated value for each broad specral type of the optical photons detected
per second per pixel, for a visual magnitude of 0.
The observer would have to supply only the spectral type and visible magnitude, and the PHS
tools will calculate the detected photons per pixel per CCD readout frame, based on the mode
parameters of the readout and the magnitude conversion from the look­up table.
If the observer cannot provide a spectral type, we might assume a M­type (worst case value,
and is representative of many galaxies that might be in any case be the target of some of those
observations with unknown stellar type)

european space agency
agence spatiale europÈenne
XMM Science Operations Team
Document No.: XMM­PS­TN­40
Issue/Rev.: draft
Date: November 16, 2000
Page: iv
Figures 2a and 2b shows the assumed CCD detection efficiencies for visible light. This is
extrapolated from known astronomical CCD performance, but accounting for the fact that the
EPIC CCDs have no anti­reflection coatings.
Note with the very deep PN camera depletion depths, the long wavelength detection efficiency
is very high, leading to high sensitivity to late­type stars.

european space agency
agence spatiale europÈenne
XMM Science Operations Team
Document No.: XMM­PS­TN­40
Issue/Rev.: draft
Date: November 16, 2000
Page: v
3 Results
Stellar Te BC MOS PN
Type (K ) (=m b ­mv ) THICK MED THIN OPEN THICK MED THIN OPEN
M8 2660 ­4.2 14.4 8.09e4 4.7e6 5.74e8 233 1.32e6 7.6e7 9.33e9
M0 3920 ­1.2 2.0 1.18e4 6.85e5 8.11e7 32.8 1.93e5 1.11e7 1.32e9
K7 4160 ­0.9 1.6 9.65e3 5.61e5 6.57e7 26.6 1.58e5 9.09e6 1.07e9
K0 5240 ­0.19 0.98 6.0e3 3.48e5 3.93e7 15.9 9.78e4 5.64e6 6.38e8
G8 5490 ­0.15 0.95 5.87e3 3.41e5 3.81e7 15.4 9.59e4 5.53e6 6.19e8
G0 5920 ­0.06 0.88 5.46e3 3.17e5 3.5e7 14.2 8.91e4 5.14e6 5.69e8
F8 6200 ­0.05 0.85 5.40e3 3.14e5 3.44e7 13.9 8.83e4 5.09e6 5.58e8
F0 7240 ­0.08 0.82 5.30e3 3.07e5 3.29e7 13.3 8.62e4 4.97e6 5.34e8
A7 8190 ­0.12 0.77 5.06e3 2.93e5 3.09e7 12.5 8.24e4 4.75e6 5.03e8
A0 10800 ­0.40 0.72 4.83e3 2.79e5 2.86e7 11.6 7.85e4 4.52e6 4.66e8
B9 12400 ­0.66 0.73 5.00e3 2.88e5 2.94e7 11.9 8.10e4 4.67e6 4.78e8
B0 30000 ­3.17 0.11 7.82e3 4.47e5 4.46e7 18.0 1.26e5 7.24e6 7.24e8
O9 31900 ­3.34 0.11 7.85e3 3.73e5 4.47e7 18.1 1.27e5 7.28e6 7.26e8
Pixel loading in photons per pixel per second

european space agency
agence spatiale europÈenne
XMM Science Operations Team
Document No.: XMM­PS­TN­40
Issue/Rev.: draft
Date: November 16, 2000
Page: vi
4 Operational Implications
The first interesting parameter we can derive from the above table is the maximum brightness
allowed per point source if the calibration accuracy is not to be affected (i.e. assume the detected
signal must be Ÿ1 electron per CCD frame time). The table is therefore recast below assuming
the default operating modes of MOS = 2.6 secs FULL FRAME image mode, and PN = 0.07 secs
FULL FRAME image mode. As a rough guide a source 1 ­ 2 magnitudes brighter could be handled
with one of the EPIC window modes in each case. The units are in v magnitude.
Stellar MOS PN
Type THICK MED THIN OPEN THICK MED THIN OPEN
M 3.9 13.3 17.7 22.9 3.0 12.4 16.8 22.0
K 1.55 11 15.4 20.8 0.67 10.1 14.5 19.7
G 1.0 10.5 14.9 20.3 0.08 9.6 14.0 19.2
F 0.86 10.4 14.8 20.2 ­0.03 9.5 13.9 19.1
A 0.75 10.3 14.7 20.1 ­0.14 9.4 13.8 19.0
B 1.14 10.8 15.2 20.6 0.25 9.9 14.3 19.5
O 1.14 10.8 15.2 20.6 0.25 9.9 14.3 19.5
An additional factor has also to be considered, particularly for the OPEN position. There
is the existence of the zodiacal background which varies with ecliptic co­ordinate, and also the
contribution from the galactic plane. The latter can be partially the diffuse Galactic UV light, but
also the combination of confused point sources.
As a rule of thumb, in the Galactic plane, the number of unresolved stars per square arcsec
is about m v 21.5 equivalent. Thus in the PN camera the equivalent magnitude is 18.5 per pixel,
decreasing to 21 at the galactic pole. In the MOS camera the equivalent numbers are 23.8 and
21.3. This stellar density may add to the overall noise on the pixels.
The zodiacal light appears as a truly diffuse source with G­type spectrum, varying from a
magnitude of 23.4 to 21.6 per square arcsec, so the equivalent ranges per pixel are: 23.2 to 21.4
(MOS) and 20.4 to 18.6 (PN). At favourable galactic and ecliptic co­ordinates the EPIC cameras
are able to utilise the OPEN position without degradation, but in general the unresolved light is
comparable with the point source brightness limit.
If the observer's target is an extended source the magnitude must be entered as equivalent per
point source. The conversion is probably not required to great accuracy, because one can expect
that even the brighter nearby galaxies are classed as magnitudes 8 ­ 12 on spatial scales of several
arcminutes, which correspends to an equivalent magnitude per pixel of about 10 fainter.
Final considerations are more subtle practical ones. In the MOS camera the offset calcula­
tion is performed as averages along columns and rows. Therefore it may be possible that while
the correct filter choice is made for the target of interest, brighter optical sources elsewhere in
the same set of columns and/or rows may perturb the offset calculation with a negative energy
delta applied at the interesting target, leading to incorrect energy assignment and event selec­
tion criteria again. (See for example technical note XMM­PS­TN­17 on the XMM SOC webpage
http://astro.estec.esa.nl/XMM/tech/socdoc top.html).
More generally the preceeding analysis assumes no stray light in the form of earth and moon
limb, or Solar system objects at close to the viewing constraints, nor stray light from pinholes in
the instrument and spacecraft structure, poorly baffled filterwheel paths etc. etc.. As with the
conservative margins adopted in the above calculations, we can only await actual measurements
in­orbit before confirming these performance data.