Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://xmm.vilspa.esa.es/external/xmm_sw_cal/icwg/presentations/RXTE_PCA_JAHoda.pdf Äàòà èçìåíåíèÿ: Mon Jun 19 13:03:55 2006 Äàòà èíäåêñèðîâàíèÿ: Sat Dec 22 14:59:24 2007 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: storm

PCA Calibration Status
· "Calibration of the Rossi X-ray Timing Explorer Proportional Counter Array" Jahoda, Markwardt, Radeva,
Rots, Stark, Strohmayer, Swank, and Zhang, 2006, ApJS, 163, 401
­ Energy: residuals < 1% (E< 10 keV); < 2% (below 20 keV) ­ Background: (excess variance)1/2 < 2%/channel on time scales > 1600 sec
· ~ 1% (2-10 keV); ~ 0.7% (10-20 keV)

­ Timing: time tags accurate to better than 4µs ­ PSF: 1' positions for bright transients


Outline
· Characteristics of PCA relevant for calibration · Philosophy, components of energy response, which seems of highest interest for crosscalibration · Emphasis on limitations, future prospects for PCA or lessons for community


PCA characteristics
· Advantages of PCA
­ No Mirror
· No energy dependent arf
collimator Mylar windows

­ No Imaging
· No pointing dependent rmf variation
propane xenon

· Disadvantages
­ 1500 cm2 per detector

· Even short observations are systematics dominated

Tagged Am241 source



Energy Response Philosophy
· Use .pha files as observed
­ Incorporate any time dependence in channelenergy or rmf parameters

· Physical models with reliably determined parameters · Make calibration parameters accessible, but
­ Energy to channel conversion delivered via CALDB, Ftools distribution, or GOF ­ Rmf parameters delivered via pcarmf.par ­ Result is extra complexity in use


Discontinuous events
· · · · · · · · Dec 1995 - launch Mar 1996 - intermittent breakdown, PCU 3,4 Mar 1996 - HV lowered, duty cycle of PCU 3,4 reduced April 1996 - HV lowered - SC disables HV during breakdown April 1999 - intermittent breakdown, PCU 1 Apr 1999 - HV lowered, PCU 1 duty cycle reduced Mar 2000 - pinhole in anti-coincidence volume, PCU 0 2006 - several PCU 0 breakdowns, duty cycle reduced

Response matrix generator keeps most parameters ( I.e. amount of xenon in first layer) constant across discontinuities. Energy-channel relationship changes discontinuously


Time dependent gain
Am spectra are continuously available; Gain drifts is slow (~1% over 7 years) but significant
241


Time dependent efficiency
Xenon leaks continuously from signal volume into anticoincidnce volume; peak efficiency is reduced by ~1%/year
Model assumes linear buildup of Xe in anticoincidence layer; ignores loss of Xe in main layer


Quantum Efficiency
· Parallel slab
­ ~0.0035 gm cm-2 mylar with ~4x10-5 gm cm-2 Al ­ 0.026 gm cm-2 Propane + 10-4 gm cm-2 Xe + (5 x 10-8 gm cm-2 Xe/day) T ­ 2nd mylar window (identical to first) ­ ~0.0068 gm cm-2 Xe - layer 1 ­ ~0.0006 gm cm-2 Xe - dead layer ­ ~0.0055 gm cm-2 Xe - layers 2 and 3 ­ Plane parallel slab is a simplification, particularly for PCU 0 after loss of propane layer

· Photo-electric absorption only. Compton scattering becoming important (but ignored) at 60 kV


Energy scale
· Proportional counter signal is proportional to amount of ionization
­ NOT proportional to energy · Ep = 22 E/wf(E) · wf = (w-22)f + 22 · f ~ 0.4 is best fit for PCA, dominated by L-edge ·
eV per electron 22.5 22 0 23

10 20 30 Photon Energy (keV) From Dias et al

40

f ~ 1 gives measured E (170 eV) at K-edge



Xe absorption edge in layer 1


In Flight Energy Scale input


In Flight Energy Scale input

For each epoch: ch = (A0 + A1T + A2T2) + (B0 + B1T + B2T2) Ep + C0 Ep2 time dependence for quadratic term, cubic term don't improve results


Redistribution matrix
· Resolution (FWHM)
­ ch = (aE + b)1/2B
· · · · a B = 0.121, b = 0.44 is energy to ch slope E/E ~ 17% @ 6 keV E/E ~ 8% @ 22 keV

· Escape peaks
· ­ no L escape above K edge Partial charge collection ­ Inoue parameterization
­ estimated (poorly) from fits to Crab


Putting the pieces together
Pick f, evaluate energy to Channel relationship For many Crab observations, fit quantum efficiency and redistribution parameters Average q.e. and redistribution parameters Fit Crab or other sources, develop figure of merit Choose best solution Ftools v5.3, v6.0, vNext 2003 Crab data PCU 2, layers separately and combined. Largest feature in ratio plot is at 33 keV


33 keV feature could be

- Xe K line from uninstrumented xenon l1-3 eq widths are 650, 450 450 eV - Escape peak from Sn (shielding) fluoresence photon (no indication of photopeak)

Does not appear to be re - eq width much bigger - ad-hoc 2 channel shift K-edge removes, but no energy scale this far.

lated to energy scale than expected jump in energy scale above plausible model shifts


Substantiates claim that systematic residuals are very small below 20 keV


Absolute areas
· · · · · For many Crab monitoring observations Dead time (~6%) corrections consistent with 2.1 for all detectors/times Area scaled to yield constant flux Zombeck Crab parameterization (A = 10 phot s-1 cm-2 keV-1 @ 1 keV; = 2.05; NH = 3 x 1021 cm-2) gives 2-10 keV flux of 2.34 x 10-8 ergs s-1 cm-2. However:


Absolute area (2)
Zombeck is among the highest of the available normalizations PCA flux scale is likely to be ~10% high wrt to other instruments


Limitations
· (almost) no ground data at mission HV · No ground data available at current gain, efficiency · No energy scale monitor below the L-edge; determing the L-edge "channel" is model dependent. · Too little ground data in vicinity of Xe-L edges
­ Model has three discontinuous features (gain, quantum efficiency, partial charge collection)

· Inadequate appreciation of state of knowledge about the flux of the Crab


Missing tools
· Standard advice
­ Make background model ­ Extract pha data (source and background) ­ Make response matrix

· This proceedure
­ Over estimates errors associated with background (artificially reduces 2) ­ Does not provide systematic errors (2 increases with total counts) ­ Users often approximate by adding constant (1%) systematic

· PCA team currently trying to develop a more nearly correct proceedure