ASCA Calibration Uncertainties
The in-flight calibration of the ASCA instruments is an on-going effort. This document summarizes the current status of such efforts.
Currently work is being done on the GIS response matrices
(mainly to improve the energy vs. pulse-height relationship)
Much of the information below is fairly detailed and techincal. Users may check the summary of caveats and systematics of ASCA data for a less techincal and abridged description of things to watch out for when doing ASCA data analysis.
Since the ASCA database is huge, Guest Observers who, during the course of their analysis come across additional problems or anything that may contribute to our current knowledge of the calibration are encouraged to notify the ASCA GOF by using our Feedback form.
Calibration Uncertainties (Updated 2001 January 5)
1. General issues
- 1.1 Quantitative Assesment of CCD Radiation Damage A link to a Dotani-san report.
- 1.2 XRT response and its effect on overall calibration.
The current, released XRT effective area, as embodied by the
FTOOL ASCAARF, has two empirical corrections (or fudges):
(1) Gaussians at ~1.9 keV (GIS) and ~2.2 keV (SIS). Invoked in ASCAARF by setting fudge=yes.
(2) Multiplicative factors of the order of a few percent which forced the GIS Crab spectra to be consistent with observations by previous X-ray astronomy missions. Invoked in ASCAARF by setting arffil=yes.
Note that this version of the XRT response uses modified optical constants whose real and imaginary parts do not obey the dispersion relation (unlike the above empirical factors, this is transparent to the user). The user should consult the following ASCANews #5 article for essential details:
New XRT responses have been developed using optical constants whose components do preserve the dispersion relation and also utilize more recent measurements. Details of this development can be found at
and the ASCANews #5 article
Note that these new XRT responses have not yet been released for general use because the SIS and GIS responses have to be modified in a self-consistent manner. This is an-going effort and updates will be posted on this page. Also be aware that the default in ASCAARF is fudge=yes and arffil=yes.
- 1.3 Boresight
New boresight measurements have been made by
Gendreau et al using multiple pointings of 3C 273.
Users are cautioned that these should technically be used only
when the XRT, SIS and GIS responses are all self-consistent with
the new boresights. This is because the current released calibration,
including the boresights are designed to give certain results for the
Crab (see above).
Please note, also, that the accuracy with which you determine the position of
your source (and how accurately it is centered in your extraction region) can
also affect the accuracy of the calibration. For bright sources, it is
generally possible to attain this error to be less than the boresight
uncertainty; the exceptions include 64x64 spatial resolution mode of the GIS
and a source in a crowded field.
- 1.4 Cross-Instrument and Cross-Mission Comparisons
Note that in all the following, cross-instrument and cross-mission comparisons are made using the current released XRT response (i.e. ASCAARF with fudge=yes and arffil=yes).
1.4.1 SIS and GIS The empirical "fudges" in ASCAARF were designed to bring SIS and GIS spectra and normalizations into alignment. Specifically, the individual correction factors for GIS2 and GIS3 were designed to give certain results for the Crab and these factors were averaged for the SIS (which cannot observe the Crab); the GIS normalization was forced to agree with the SIS. However there are three major problems with this: (1) the GIS cannot be used to calibrate the soft X-ray response of the SIS, (2) the SIS spectral response changes with time (see below), and (3) the flux cross-normalization factors are technically correct only for a Crab-like spectrum. Thus, in practice users may find SIS/GIS flux differences up to ~5% and even ~10%.
1.4.2 Comparison of ASCA high-energy response with XTE and BepppoSAX (MECS)
Simultaneous cross-calibration observations of 3C 273 with ASCA/BeppoSAX/RXTE were made in July 1996 and June 1998; full details can found in this report. The bottom line is that the PCA flux is more than 20% higher than ASCA, but the MECS on BeppoSAX agrees with ASCA to better than 3%. Differences in the 3-10 keV hard X-ray slope between ASCA/RXTE/BeppoSAX are not worse than 0.06.
In January 2000 there were simultaneous cross-calibration observations of 3C 273 with ASCA, BeppoSAX, and Chandra; work on these data is still in progress. The results will be reported here as soon as possible.
1.4.3 Comparison of ASCA SIS and BeppoSAX LECS in the 0.5-4 keV band
The low-energy response of the SIS is changing with time so the cross-instrument comparison in the soft X-ray band is more complicated. Results from the July 1996 BeppoSAX/ASCA observations of 3C 273 have been reported in Orr et al. 1998 (A&A 337, 685). Work is still in progress on the June 1998 and January 2000 data.
1.4.4 Comparison of ASCA and the ROSAT PSPC There have been conflicting reports on the agreement/disagreement of ASCA and PSPC data. The results of a systematic assessment will be posted here in the near future.
2. XRT, image analysis issues
- 2.1 XRT PSF: azimuthal dependence.
The XRT point response function has an azimuthal energy dependence, though the
currently available XRT response is valid for an azimuthal average.
There appears a factor of two difference at 10 keV between the lobes of the "Maltese
- 2.2 Analysis of extended sources.
Due to the energy-dependent extended lobes of XRT PSF,
data analysis of extended sources is complicated.
Details of the problems, limitations and suggested procedures can
be found in the following ASCANews articles and references therein:
1. The ASCA Mirrors (ASCANews #1)
2. XRT Ray-Tracing (ASCANews #3)
3. Extended Source Analysis (ASCANews #3)
5. PSF In The Low-Energy Region (ASCANews #5)
6. XRT Calibration Issues (ASCANews #5)
- 2.3 ASCA source positions uncertainties.
Recent calibration of the ASCA error circle shows that the satellite
pointing is stable to better than 10 arcsecs RMS and that it is possible
to measure moderate and bright sources with 5 arcsec precision.
However, the absolute sky positions measured from a given observation
have an additional large systematic offset. This offset, which is introduced
by the attitude control system (ACS) moving between pointings, results
in a total SIS source position uncertainty of 40 arcsecs (90% confidence
level) when uncorrected. For source positions measured with the GIS,
additional position dependent instrumental uncertainties are evident,
whose magnitude increases towards the edge of the FOV. A fuller
discussion, and the recipe for deriving the positions more accurately,
is available in ASCA coordinate
3. GIS issues
- 3.1 GIS RMF:
It has been found that the energy scale in the GIS RMF
version 4.0 and earlier is slightly wrong below the Xe-L edge (4.8 keV).
Using these RMFs to fit real data, means that the fitting results
give energies which are ~30 eV too low at 2 keV.
See Keith Gendreau's
ASCA Recalibration page for more details. A new version of the RMF
is expected to be released in the fall of 1999.
- 3.2 Timing, 1.56 msec pulsar detected:
The 1.5578 msec pulsation was detected with GIS from PSR 1937+21
(IAUC 7030). Most of the x-ray flux
(50-90 percent in the band 1.7-6.5 keV) was contained in the narrow
primary peak with width probably < 130 microseconds.
This indicates that the GIS relative time assignment is more precise than ~130
micoseconds at least over the observation interval (~100 ksec).
See also the
Timeconv improved page.
- 3.3 Timing:
By comparing pulse phases of the Crab pulsar, PSR B1509-58 and
PSR B1821-24 observed with the GIS, XTE and
radio observatories, these measurements agreed within
200 microseconds. This is considered to be a conservative
upper-limit of the inaccuracy of the GIS time assignment.
Note that GIS events can have time stamps precise to 61 microseconds.
Fractional stability of the time assignment is less than
3e-8 (T/10^4 sec)^-1.
Details can be found in the ASCANews articles
"Time Assignment of ASCA GIS"
(volume 4) and
"Accuracy of the GIS time assignment" (volume 5).
- 3.4 GIS gain: spatial dependence.
The determination of how the gain varies
spatially across the GIS field of view is uncertain by
about +/- 1 % for over the detector (within 25 arcmin from the center).
This uncertainty may actually represent a fundamental limit based on
the properties of the instrument. That is, further calibration may not
improve the accuracy of the spatial gain calibration. Users should be
aware of this uncertainty when analyzing their data.
See the "Status of the GIS Gain Calibration" page
for the latest information on the GIS gain issue.
- 3.5 Increase of the non X-ray background during the mission. The GIS team reports the GIS non X-ray background has increased by about 10 % over the first two years of the mission. This is presumably due to buildup of some radioactive elements. Please refer to the ASCA News article "Reproducibility of the GIS Non X-ray Background " and the GIS Night Earth Event File page for details.
4. SIS issues
- 4.1 SIS gain.
The gain calibration of the two default chips (SIS0-chip1 and SIS1-
chip3) agree to within 0.5 per cent. The other chips are not as well
calibrated, and may differ by up to 2 per cent.
Due to radiation damage, the gain is changing with time. There is
also a spatial dependence of the CTI. The
calibration of this secular effect is an on-going process.
The latest performance details of the SIS can be found in
Recent Progress of SIS Calibration
and Software (ASCANews #5). NOTE that the current most up-to-date chip-to-chip
and temporal gain corrections are incorporated in the file
The CTI values are
linearly extrapolated for more recent observations, which is now suspected to
cause a significant
mis-calibration, particularly for SIS-1 data taken during 2000.
- 4.2 SIS gain: energy scale offset.
The on-board determination of zero-level of CCDs may be incorrect by
several ADUs (i) for 4-CCD mode data and for recent 2-CCD mode data;
(ii) when CCDs are illuminated by optical light (Sun angle less the 80
deg and taken during satellite daytime or within ~30 deg Bright Earth limb);
or (iii) when the CCDs are warm (above -60C, particularly serious above -59C).
When the zero-level is determined on the ground using the current version of
FAINTDFE (which can be used both for Bright and Bright2 conversion), any
remaining inaccuracies are thought to be small. Combining on-board Bright
mode data taken during orbital night with converted Bright mode data may
offer a good compromise in statistics and systematics.
- 4.3 RDD correction
An experimental tool was released with FTOOLS v4.0 called correctrdd
to restore data suffering from the worsening RDD problem
(RDD stands for Residual-Dark-Distribution). Data suffering from
this problem has degraded energy resolution and mis-classified SIS
event grades. Consult the article in ASCANews #5
Recent Progress of SIS Calibration
and Software for details. Also, a
recipe for using correctrdd is available.
- 4.4 SIS low-energy response
It has been difficult to calibrate the low-energy response of
ASCA due to the fact that no instrument before the SIS has observed
astrophysical sources in the 0.4-2 keV band with the same
sensitivity and energy resolution. Since the SIS cannot
observe very bright sources, and non-variable sources
tend to be extended and complex, there is thus a lack of a suitable
calibration targets. The SIS low-energy response has also been
changing with time and work is in progress
to derive a proper physical model of the degradation.
Meanwhile, one can
estimate the magnitude of the effect even though software to
correct the data does not yet exist.
- 4.5 SIS "0.5 keV" feature.
It has long been known that unexplainable and often
statistically significant features are found in the SIS response
below 0.6 keV, for a variety of X-ray sources and models (thus
suggesting an error in the SIS+XRT effective areas).
While many possible explanations for the effect have been
advanced, an adequate explanation has not been found.
Users are advised either to not use data below 0.6 keV or else
be conservative about interpreting spectral features below this
- 4.6 SIS 2-pixel events and the event threshold.
It has been found that the current calibration of 2-pixel events when the event threshold is high may lead to inaccurate results.
Information submitted by-- Keith Arnaud, Charles Day, Ken Ebisawa, Eric Gotthelf, Koji Mukai, Richard Mushotzky, K. P. Singh, Nick White, and Tahir Yaqoob
Return to the ASCA GOF
This file was last modified on Wednesday, 02-Aug-2006 15:20:28 EDT
HEASARC Guest Observer Facility
If you have any questions concerning ASCA, visit our Feedback form.