ROSAT ROSAT PSPC

Status of the PSPC Spatial/Temporal Gain Calibration

Analysis of the in-flight Al Ka data

ROSAT PSPC

Status of the PSPC Spatial/Temporal Gain Calibration

Analysis of the in-flight Al K-alpha data

and

Code 666/668,
NASA/GSFC,
Greenbelt, MD  20771

P. Predehl and A. Prieto

MPE,
Garching

Version: 1995 March 9

SUMMARY

The SASS PSPC data processing software corrects for time and spatial variations in the detector gain as part of the conversion from detected pulse height to pulse invariant channels. Analysis of the in-flight Al Ka calibration data has shown that there is an error in the spatial gain correction currently applied in SASS, and also that there are uncalibrated effects which need to be considered, requiring some additional correction based on detector coordinates and arrival time (DETX, DETY, and TIME) of each event. The Al Ka data allow us to construct a gain map which can be used to correct the PI data. Whether or not the required correction is pulse-height dependent is not known, and will require considerably more work to determine.

1  Introduction

The calibration corrections performed on PSPC events in the SASS processing are described in detail in ``Calibration Corrections to individual PSPC events'' by Hasinger and Snowden (1990). These include corrections for electronic variations (detailed characteristics of the anode amplifier chains), electric-field variations due to the (slight) variable spacing of the anode wires, and long-term temporal gain variation. The corrections are performed using event arrival position, time, and pulse height. The PHA to PI corrections are performed in SASS within the CT module, the spatial gain correction is applied within the DCORG subroutine, using the two ground calibration files gnampl_new.dat and gain_kor3_b.dat (or the equivalent PSPC C detector files) which contain energy and (electronically corrected) X-coordinate elements of the spatial gain correction (respectively).

The PSPC in-flight aluminum Ka (1487 eV) calibration-source data are becoming available for analysis as the Rev2 processing proceeds. In-flight calibrations have been taken every ~ week throughout the mission lifetime. The results of Prescott fits to the Al Ka profiles have always been used internally by SASS (through the near-real-time analysis) in the temporal gain correction process, but were not originally written out to event files for subsequent analysis.

2  Spatial and Temporal Variations in PHA and PI channels determined by analysis of the Al Ka data

We constructed detector images of the Al Ka data from the available data of two epochs, the start of AO-1 and the end of the mission (18 ROSAT days between days 263 and 349, ~ 0.78×106 events, and 22 ROSAT days between days 1387 and 1586, ~ 1.38×106 events). While the detector is not uniformly covered by the events, the central region is well sampled. We expect the PHA data to show variations across the detector face, depending on the event arrival position relative to the anode wires, but the PI channels were designed to be corrected to be uniform across the detector face with a mean channel at ~ 151. Figures 1a,b show the actual distribution of the mean PHA channel across the detector for the two epochs. The data were fit with the full model for the monochromatic response of the PSPC (Jahoda & McCammon 1988). Individual spectra with 1000-3000 events were created on a 2 arc minute grid with the sample region of varying size. (In order to get relatively uniform statistics from a very nonuniform coverage sample of data, data were included for individual spectra which lay within a ring of variable radius from the grid point. This radius varied from ~ 3 arcmin for high event-density regions to ~ 10 arcmin for low density regions. This will obviously smooth out any fine-scale structure.) In order to compensate for the variation in absolute gain state for the two epochs for display (they were collected at high and low gain, mean channels 147.95+/-0.02 and 105.26+/-0.01, respectively for epoch 1 and epoch 2), the contour spacing is in units of 1% of the average value and the shading indicates the same range, from -4% to +4%, for each plot. The range within the inner circle of the window support structure was 145.6 to 150.3 (-1.6% to +1.6%) for the high-gain data, 101.4 to 106.2 (-4.4% to +0.9%) for low-gain data.

Figure

Figure 1: Mean PHA channel for Al Ka calibration data at the a) start (on the left) and b) end (on the right) of the mission as a function of position in the detector. The contours are in steps of 1% while the shading level changes only every second contour. The small closed contour near the center in a) is at -1%, in b) it is at -2%.

In epoch 2 data, there is a marked minimum evident near the detector center, where the target source was placed during most of the pointed observations. This feature may be an enhanced aging effect of the area of the detector having the greatest bombardment of photons, perhaps caused by the polymerization of the counter gas and the coating of the anodes thus reducing the electric field (and therefore gain) in a localized region.

The spatial gain correction in the PHA to PI conversion software was instituted to eliminate the general structure in the PHA mean channel distribution. In no way was it designed to correct for the unexpected gain decay at the center of the detector. Figures 2a,b show the mean-channel distributions after the PHA to PI channel conversion. The average mean channels for epoch 1 and epoch 2 are now 151.77+/-0.03 and 151.15+/-0.02, respectively, indicating the rough consistency of the general conversion. However, of concern is the greater apparent variation of the general structure when compared to the PHA data. The contours and shadings of Figure 2 are the same as for Figure 1. The total PI range within the inner circle of the window support structure is 146.60-155.17 (-3.4% to +2.2% for epoch 1 and 145.10-155.54 (-4.0% to +2.9% for epoch 2. Now, large temporal variations in PHA channel are expected as the PSPC high voltage was changed during the mission. The PHA to PI temporal correction applied in SASS is based on an interpolation between Prescott fits to the nearest Al Ka observations before and after the target observation, so we should see no residual time variation in the PI data. Figure 2, however, illustrates that on-axis, this is not the case. The important point to note here is that while the regions of higher gain are the same, the region of minimum gain has shifted to the on-axis position where the effective gain for on-axis data has dropped from 148.88 at the start of AO-1 to 144.28 at the end of the mission. This is a -3.2% gain shift over ~ 4 years. The linearity with time of this shift will be determined as the mission data are reprocessed and the calibration data become available.

Figure

Figure 2: Mean PI channel for Al Ka calibration data at the a) start (on the left) and b) end (on the right) of the mission as a function of position in the detector. The contours are in steps of 1% while the shading level changes only every second contour. The small closed contour slightly to left of the center in a) is at -2%, the smallest closed contour at the center in b) is at -3%.

For another way of looking at the data, Figure 3 shows a profile of PI channel versus position on the detector face (taking a cut across the line of highest signal-to-noise).

Figure

Figure 3: The mean PI channel for cuts across the detector. The open circles are data from the low gain state, the crosses are from the high gain state.

3  PCPICOR: A software tool to fix the PI data

As described above, analysis of the Al Ka data show that there are errors in the PHA to PI conversion applied in SASS, and also, that there are uncalibrated effects which we need to consider. We have used the Al Ka data to create calibration maps, similar to those shown in Figure 2, for several epochs covering the PSPC lifetime. These calibration maps can be used to calculate a correction for existing PI data, and we are developing an ftool, PCPICOR, which will apply this correction to the event data. PCPICOR interpolates between the nearest two calibration maps, before and after the target observation, to calculate a correction for each photon, based upon DETX, DETY and TIME. Application of the correction tool will hopefully avoid a need for reprocessing of the data with any corrected version of SASS. Figure 4 shows this algorithm applied to a set of Al Ka data used in creating the calibration maps. Comparison of the corrected and uncorrected data indicates that this tool provides a significant improvement to the PI data. We also tested the tool on an astronomical calibration source, N132D. Figures 5a and b show the ratio of PI spectra of N132D before and after the later data were corrected using PCPICOR. In this Figure, the numerator is rp160084 (from May 1991), and the denominator is rp142011 (from September 1994).

Figure

Figure 4: Mean PI channel after correction for Al Ka calibration data at the end of the mission as a function of position in the detector. The contours are in steps of 1% while the shading level changes only every second contour. Note that the major contour now running through the figure is at 0% indicating that the mean is dithering around the nominal value. In Figures 1, 2, and 4 the uncertainties are ~ 0.3%.

Figure

Figure 5: Comparison between two N132D PI spectra taken during two widely separated periods in the mission. The top panel shows the ratio of these spectra with no correction, the bottom panel shows the ratio after the PI channel correction has been applied to the rp142011 data. The channels (x axis) are linearly binned down from 256 to 32 with a factor 8. This places the peak of the N132D pulse-height distribution between channels 11 and 12.

We are continuing to work on providing more calibration maps at different epochs, to improve the temporal calibration. Whether or not the required correction is pulse-height dependent is not known, and will require considerably more work to determine.

4  Conclusions

a) The spatial gain correction currently carried out in SASS is not appropriate.

b) The existing PI data require a spatial re-correction based on DETX, DETY and TIME.

c) The Al Ka data allow us to create calibration maps with which we can correct the PI data.

The correction tool, PCPICOR will shortly be available from NASA/GSFC as part of the FTOOLS analysis package. Initially, it will only have data to allow a linear interpolation between the distributions at the start and end of the mission. With reprocessing, intermediate points will be added.

REFERENCES

G. Hasinger and S Snowden, 1990 TN-ROS-ME-ZA00/027

K. Jahoda and D. McCammon, 1988. Nuclear Instr. and Methods, A272, 800

USEFUL LINKS TO OTHER HTML PAGES

The following useful links are available (in the HTML version of this document only):

File translated from TEX by TTH, version 2.25.
On 22 Aug 2001, 22:56.