Post on 29-Dec-2015
NASSP Masters 5003F - Computational Astronomy - 2009
Lecture 19
• EPIC background
• Event lists and selection
• The RGA
• Calibration quantities
• Exposure calculations
NASSP Masters 5003F - Computational Astronomy - 2009
Background• Background from instrumental noise
– Worse at low energies & higher chip temperatures.
• X-ray background– Cosmic– Fluorescence
• Si of course, but also Al and Cu from support structure.
• Particle background– Hard, penetrating – “cosmic rays”.
• Fairly constant in time;• Fairly isotropic.
– Soft protons (~100 eV).• Flaring time behaviour.• Funnelled by the mirrors.• These weren’t suspected before launch! A major headache,
because too strong a flare can damage the CCDs.
NASSP Masters 5003F - Computational Astronomy - 2009
Background examplesNote the background in the masked areas.
Mostly from flares.
Cu fluorescence. Instrumental noise at low energy.
(Masking here is done via software.) (Masking here is done via software.)
MOSpn
pn pn
dec
RA
NASSP Masters 5003F - Computational Astronomy - 2009
Background – what to do with it
• Significance of background depends on what you want to do.– Spectra: obviously one needs to know the
spectrum of the background as well as possible.
– Images, in particular source detection and flux measurement: spatial properties of the background are important.
• Cosmic ray, x-ray and flares all have different spatial behaviour – so working out the proportions is important.
– Time series:• Soft proton flares dominate the problem.
NASSP Masters 5003F - Computational Astronomy - 2009
Other mainly spatial problems with EPICs:
Optical loading from a bright visible-lightsource (filters minimize this)
Single-reflection arcs from far-fieldsources
NASSP Masters 5003F - Computational Astronomy - 2009
Event lists• In high-energy astronomy, we deal not
with voltages or brightnesses (essentially floating-point quantities) but with lists of events – 1 event per photon.
• Each event comes with the following data:– Its pixel position on the CCD.– If necessary, the number of the CCD.– Its frame number.– Its energy. (XMM: the column is called PI.)
• Maybe also: a quality flag, event pattern, etc.
• In XMM output the events are stored in a table in a FITS file.
NASSP Masters 5003F - Computational Astronomy - 2009
Event selection
• The aim is to separate ‘interesting’ events from ‘boring’ events – eg divide the events into those which probably come from a source and those which don’t.
All events
Good Bad
r
E
t
Define a selectionvolume
•Limits in defining volume shapes.•Problems integrating over overlapping volumes.•FITS format for storing selections: Data SubSpace (DSS)
NASSP Masters 5003F - Computational Astronomy - 2009
Diagnostic plots:It’s helpful to plot 2 of the eventcoordinates – here energy vs time.
PN
Cu fluorescenceline
Al fluorescenceline
Time
Pho
ton
ener
gy
‘So
ft p
roto
n’
bu
rsts
NASSP Masters 5003F - Computational Astronomy - 2009
Diagnostic plots:MOS 1
Al fluorescenceline
‘Gatti’ events
NASSP Masters 5003F - Computational Astronomy - 2009
V
Gatti process – a kind of dithering.
Histogramof events withvoltage V.
ADC levels are analog- thus not evenly spaced.
Distorted digitizedhistogram.
+
V
t ADC
-
Undistortedhistogram.
V
t
=
NASSP Masters 5003F - Computational Astronomy - 2009
The Reflection Grating Spectrometer (RGA)
• Each MOS has one.
• They divert about ½ the x-rays.
• Diffraction grating array of 9 CCDs.
• Pixel position in the dispersion direction is a function of x-ray energy.– But not a linear function (I think there is a
cosine term in it).
• Energy resolution is much sharper than via amount of charge the photons generate.
• Spectral orders overlap –– but the 2nd order has even finer resolution.
NASSP Masters 5003F - Computational Astronomy - 2009
RGA –plot showing the event pixel locations:
NASSP Masters 5003F - Computational Astronomy - 2009
The ‘banana plot’
NASSP Masters 5003F - Computational Astronomy - 2009
An example RGS spectrum:
Spectral resolution:about 2 eV
NASSP Masters 5003F - Computational Astronomy - 2009
An example EPIC spectrum:
Spectral resolution:about 100 eV
NASSP Masters 5003F - Computational Astronomy - 2009
Charge redistribution• Photons of a single, narrow energy give rise to
broadened charge redistribution spectrum.– Partly because of Poisson (quantum) statistical variation;– Partly because of smearing out during the transfer of charges
from row to row during readout.• The relation between true spectrum S and measured
spectrum S':
• R is called the redistribution matrix (RM).• As the chips degrade with age (due mostly to particle
impacts), the RM changes and has to be recalibrated.• The philosophy with x-ray spectra is not to subtract
background or deconvolve RM, but to begin with a model, and add background and RM-convolve this before comparing it with the measured spectrum.– See the program XSPEC.
ESEEREdES ,
NASSP Masters 5003F - Computational Astronomy - 2009
MOS RM cross-section at 800 eVEnergy of the x-rays
NASSP Masters 5003F - Computational Astronomy - 2009
Evolution of the energy dispersion
Black: pnRed and Green: the MOS chips
MOS temperatures werelowered here.
1.5
keV
6.0
keV