What are the parameters of the ARF correcting XSPEC tool (acisabs)?
George Chartas is the expert on this tool since he wrote it. However, one can go to the web site and the parameters are listed:
ACISABS Parameters:
par1 = Days between Chandra launch and ACIS observation
par2 = norm parameter in decay rate equation
par3 = tauinf parameter in decay rate equation
par4 = tefold parameter in decay rate equation
par5 = Number of carbon atoms in hydrocarbon
par6 = Number of hydrogen atoms in hydrocarbon
par7 = Number of oxygen atoms in hydrocarbon
par8 = Number of nitrogen atoms in hydrocarbon

Back to Presentation
Back to Agenda

ACIS Background - M. Markevitch

Since the dark moon observations are time dependent, could they be unsuitable for determining background?
The dark Moon observations (compared to the Histogram Mode data) showed that the background at the stowed position inside the SIM and at the aimpoint are similar. They have served their purpose and are now superseded by a longer exposure with ACIS stowed, which does not include the variable soft X-ray background component.

What does the spectrum of the soft background flares look like below 0.5 keV?
The same power-law model continues to lower energies (to 0.3 keV at least).

How do you know if your observation was affected by soft background flaring?
One should check a ligth curve for chip S3 in the 2.5-7 kev band (or in the 2.5-6 kev band for chip S1 if S3 is completely covered by the target emission). Even if it looks "constant", compare the rate with the nominal rate from the corresponding blank-sky background dataset. If the BI chips show a flare, but the FI chips do not, then it is a soft flare, whose spectrum can be modeled as shown.

How uniform are the flares over the face of the chips?
Nonuniform; the spatial distribution is under investigation.

VF mode can remove a large fraction of source events (~30%) - for what types of observations is VF mode appropriate?
The VF mode filtering should not be applied to bright objects at or near pileup, because it will result in loss of good events. The 5x5 pixel event islands may start to overlap even when the true pileup (when the 3x3 pixel islands overlap) is weak. Most if not all clusters are below that brightness. To be sure, one can make an image using only the VF-rejected events and see if any obvious sources are seen. Note that there is no intrinsic loss of events in the data because of the use of VF mode (provided the telemetry did not saturate).

Back to Presentation
Back to Agenda

Evolution of ACIS Charge Transfer Inefficiency - C. Grant

How could we prevent serial CTI? Where are we most likely to be affected by serial CTI?
The BI CCDs have always had serial CTI due to defects introduced in the manufacturing process, while the FI CCDs to this day do not have detectable serial CTI. Serial CTI can be increased from radiation damage to the serial register, but the particles would have to be of a high enough energy to pass through the framestore shield. The standard ACIS protection scheme (moving ACIS out of the focal plane during radiation belt passages and solar storms) will help reduce radiation damage to the detectors, however there is very little that can be done to prevent damage from very high energy particles that pass through the SIM itself. That said, there does not appear to be any appreciable increase in serial CTI since launch.

What causes the bumps and wiggles in the CTI vs time graph?
CTI is a function of the history of the charge traps encountered by each charge packet as it is transfer to the readout. If an X-ray event is preceeded (along the readout direction) by another event, it will lose less charge during transfer than if the preceeding event was removed. The preceeding event is called "sacrificial charge" and can reduce the measured CTI.
In the case of ACIS CCDs, most of the sacrificial charge is due to cosmic ray background events. The bumps and wiggles are primarily due to changing background rates.

Do results for gain evolution differ from Paul's?
The results shown here are, to the best of my knowledge, completely consistent with other similar measurements. I agree with Paul's statement that the BI CCDs show smaller changes with time than FI CCDs.

Back to Presentation
Back to Agenda

Including CTI Time Dependence in the PSU CTI Corrector - L. Townsley

Does the CTI correction account for day-to-day changes in the background?
The CTI corrector corrects ACIS events for CTI. It includes the long-term time dependence measured and parameterized as linear changes by Catherine Grant. In that sense, it accounts for long-term changes in the background. The user could attempt to account for day-to-day changes by CTI-correcting the External Calibration Source data taken close in time to the celestial data in question, then adjust the results based on residual CTI effects seen in the calibration data. Only background effects that carried over from the calibration data to the celestial data could be accomodated this way, of course. For long enough observations, the user can check the fidelity of the CTI correction by examining the spatial dependence of the gain (after CTI correction) via the instrumental Ni K-alpha line at 7.47 keV. This line is faint, though, so a long observation is required to use it as a diagnostic.

How can you use the same RMF if the gain and FWHM are changing? Do we need to change RMF as a function of time?
It takes a very large amount of calibration data gathered over many orbits to tune the CTI corrector, especially to learn about column-to-column variations (what I call the deviation map, similar to what MIT calls the trap map). So by its nature the tuning of the CTI corrector involves some averaging over time-dependent quantities. We have attempted to remove the largest of these by incorporating a linear parameterization of the CTI time dependence into the corrector. This regularizes the gain so that it is not seen to change substantially as a function of time. The FWHM is increasing as CTI increases, but so far this is not a large effect. It is certainly the case that the CTI corrector could be re-tuned as the characteristics of CTI change over the life of the mission. The trade-off is between having enough data to do a sensible tuning and combining so much data that CTI effects evolve throughout the dataset.

Back to Presentation
Back to Agenda

ACIS Response - R. Edgar

Are there CTI corrections for all the chips?
No. They exist only for the extended imaging array FI chips (I0, I1, I2, I3, and S2). We have response products (FEFs) for these chips with the CTI corrector, only for T = -120 C and only for the nominal frame time (3.2 s). CTI is a function of trap filling, and so depends on temperature, frame time, and a host of other variables.

Puzzle with S3 data at low energies?
I'm not sure what the question was, but there are gain nonlinearities at low energies on S3. We find (again at -120 C) our EO102 fits have residuals consistent with zero down to about 700 eV, but there are systematic residuals at Oxygen VIII (lyman alpha line at 653.6 eV) and Oxygen VII (Helium-like triplet at 561.0, 568.5, 574.0 eV).
We have some off-axis observations of the QSO PKS 2155-whatever with the LETG/ACIS combination. These were done with the source off axis by various amounts in the dispersion direction, thus dragging the low energy portion of the spectrum across S3 (and the other S chips). We're looking at these data to try to improve the S3 response matrix products, especially at low energies.
Gordon Garmire expressed interest in the S3 response down to even lower energies (300 eV or the carbon edge at 284 eV). We are interested in such things, but the calibration data are sparse.

Back to Presentation
Back to Agenda

ACIS Spectral Response - A. Vikhlinin

Will your method work for BI as well as FI chips?
Yes, but some modifications in the ACIS simulator are required. We need to have an access to the PHA distribution before any modifications to the CTI effects.

Will the next version of CIAO allow the read-in of multiple matrices for detector response?
No.

How do the RMFs generated with your method compare to the FEF representation which is released in CALDB 2.18 and CIAO 2.3?
They are identical for practical purposes.

Back to Presentation
Back to Agenda

Performance Tests and Verification of Current ACIS FEFs - N. Schulz

Where do the FEFs for BI chips come from? Do they account for CTI?
Yes, they account for CTI. The FEFs in the BI chips are using an advanced ACIS model for back-illuminated devices provided by the MIT ACIS IPI teams. This model is adjusted to every chip and node using the on-orbit external calibration source and LETG/ACIS-S data. So far we have verified that the model represents the redistribution function very well above 1 keV. Below 0.8 keV there are still some problems with redistribution tails. We also currently address the low energy gain and time drifts of the gain.

Back to Presentation
Back to Agenda

Future ACIS Calibration Work - D. Schwartz

Is the 5% deadtime ("dead area") due to cosmic rays on FI chips accounted for by the DTF correction or any CALDB product?
This question must be answered by the data system. It is my very strong opinion and understanding that the cosmic ray dead time is not accounted for anywhere in the CXC data system.

Glenn Allen answers:
The effect of the dead area is something that affects the flux (i.e. the ARF). The pieces involved in computing the ARF are GTIs, DTCOR, QE, and QEU.
The GTIs are only for occasions when no data is available due to a dropped frame, a flush, bad aspect, etc. The DTCOR compensates for flushes and the 41 ms readout time. The QE is from a model which does not include the cosmic-ray background. The QEU is relative to the readout, which is defined to be 1. Therefore, the dead area is not included for ACIS data. (i.e. ACIS flux numbers are systematically low by the average fraction of the dead area.)
I think it would not be too difficult to modify the various pieces of the puzzle to handle a dead area correction. I will not go into the details of such a change. What may require more work is quantifying the correction (either average values for the BI and FI CCDs or as functions of a telemetered quantity?).

Back to Presentation
Back to Agenda

Exploring the Limits of the ACIS Pile-up Model - J. Davis

Did you account for intensity variations in sources? How does this affect your model?
Yes. The pile-up model assumes that the photon arrival times are Poisson distributed. I took this into account by filtering on times where the light curves looked flat.

Does the pile-up corrector model work on higher order spectra or only on first order? What about overlapping lines?
Strictly speaking, it works only for plus and minus one orders. It can be applied to third order at long wavelengths if the first order flux at 3*lambda is small enough that it does not affect the third order.

Will this model work with the LETG?
It should, but I have not investigated it.

Will this model work for line-dominated sources?
I believe so, but this needs to be investigated in more detail.

What about 5% differences in QE between FI and BI chips?
I used Herman Marshall's QE tweaks, found here.

Back to Presentation
Back to Agenda

On the use of HETGS higher order spectra for science analysis - N. Schulz

What additional LETG/ACIS observations are needed?
LETG/ACIS data won't help here. Ideal observations would include either a very bright source where the intrinsic source spectrum is very well known or a bright continuum source that is not piled in the first orders and which is observed for a very long exposure.

Back to Presentation
Back to Agenda

Absolute Time Calibration - A. Rots

What kind of ground calibration was done for instrumental delay? Is the instrument delay fixed?
The question would probably be better answered by the instrument teams. My understanding is that there was a calibration done but not a complete end-to-end measurement with sufficient accuracy. The delay is fixed. For clarity, I should add that the following two effects are not considered included in this delay: the HRC time stamp assignment error and the time offset calculated for ACIS CC mode events.

Back to Presentation
Back to Agenda

Computation of Times of Arrival for CC Mode Events - G. Allen

Should the target RA and Dec be the true celestial coordinates, or the ones given by the aspect solution?

Should you first use the TE model to determine RA and Dec and then switch to CC mode (because of aspect error)?

These questions are very closely related, so I will answer them jointly.

Before using acis_process_events to compute the times of arrival of continuous clocking mode event data, the values of RA_TARG and DEC_TARG should be modified to be as accurate as possible (i.e. < 0.5 arcsec). Otherwise the inaccuracy of the source position will result in a corresponding inaccuracy of the times of arrival. The best values to use for RA_TARG and DEC_TARG are the reconstructed values instead of the actual source coordinates. For example, if the continuous clocking mode observation is IMMEDIATELY proceeded by a timed mode observation to determine the coordinates of the source, then use the RA and DEC coordinates of the source, as determined using the timed mode observation, should be used for RA_TARG and DEC_TARG. Use of the reconstructed position of the source instead of the actual position will help avoid inaccuracies in the times of arrival due to inaccuracies in the aspect solution. However, in some cases it is not possible to use the reconstructed position. In these cases, use the actual RA and DEC coordinates for RA_TARG and DEC_TARG. Since the aspect solution is accurate to one pixel or less most of the time, use of the actual position of the source should be fine most of the time.

Should you include pixel randomization in acis_process_events for arrival time computation?
The times of arrival are computed using the values of RA_TARG and DEC_TARG, the measured pointing direction of the telescope, and the measured position of ACIS relative to the telescope. The times are independent of the measured CHIPX coordinates of the events. Therefore, disabling the pixel randomization does not affect the times of arrival. (The times of arrival are not randomized within a pixel.)

Back to Presentation
Back to Agenda

[Back to Home Page]

Page last modified on: 24 August 2010
Comments to: axafcal (at) head.cfa.harvard.edu

Chandra Science | Chandra Home | Astronomy links | iCXC (CXC only) | Search

The Chandra X-Ray Center (CXC) is operated for NASA by the Smithsonian Astrophysical Observatory.
60 Garden Street, Cambridge, MA 02138 USA. Email: cxcweb@head.cfa.harvard.edu
Smithsonian Institution, Copyright © 1998-2004. All rights reserved.

[Back to Home Page]
Page last modified on: 24 August 2010	Comments to: axafcal (at) head.cfa.harvard.edu