Status of the PSPC Spectral Calibration--Guenther Hasinger, Almudena Prieto, Steve Snowden
While work is continuing at both MPE and GSFC on the spectral calibration of the PSPC, no changes have been made yet in the detector response matrix. The recommendations to Guest Observers (GO's) set out in ROSAT Status Report #64 (WG #20) for spectral analysis of PSPC data still remain in effect. Specifically, we suggest that observers use the old matrix (DRM_06, pspcb(c)_92mar11.rmf for the convolved matrix) for observations through AO-1 and the new matrix (DRM_36, pspcb(c)_93jan12.rmf for the convolved matrix) for observations in AO-2, AO-3, and AO-4. A convenient break point between usage of the matricies is the date of the gain change: 11 October 1991. The difference between fits with the old and the new matrices can be interpreted as estimates for the systematic errors. The matricies can be found on the legacy.gsfc.nasa.gov anonymous ftp account under /caldb/data/rosat/pspc/cpf/.
Temporal Variation of the PSPC Spectral ResponseIt is becoming more apparent that the spectral response of the PSPC is changing in a non-linear manner with time. Qualitatively, as noted in ROSAT Status Report #64, this is seen as a shifting of the mean of a pulse-height distribution to lower PI channels. This effect is most clearly seen in spectra with limited dynamic range (and which thus resemble, to a small extent, a monoenergetic distribution). The strongly absorbed spectrum of the LMC SNR N132D provides a good example. Its distribution is peaked near PI channel 80 and its width is not much greater than the intrinsic spectral response of the PSPC.
Since ROSAT Status Report #64 was distributed, several other monitors of the variation have been analyzed. The scattered solar X-ray contamination has been extracted from a number of pointed observations and the variation of the O K alpha emission line mean channel studied. Over the time period where useful data exist (up to spring of 1993), the data show the same qualitative variation as in N132D. The mean channel has dropped by ~1.4 percent per year. The fall-off of the 1/4 keV band peak in background data and the spectrum of 3C273 have been studied to determine the magnitude of any variation at lower energies. Figure 1 shows the yearly shift in channels at the limited number of effective energies that can be analyzed. The data are listed in Table 1.
Table 1: Channel shift per year Source Effective energy Shift Chan/Year Uncertainty Al K alpha 1.49 0. N132D 0.90 -0.79 0.07 O K alpha 0.53 -0.86 0.20 O K alpha 0.53 -0.68 0.13 C edge, Bkd spec 0.30 -0.09 0.10 C edge, 3C273 0.30 -0.61 0.14 Meaty (WD) 0.18 -0.18 0.03We stress that the spectral variation is non-linear. The gain-correction procedure that is used to convert raw pulse hight to PI channel continues to normalize the Al K alpha calibration photons to a constant mean channel (PI channel ~151). Modifying the detector response matrix by a linear scaling is therefore inappropriate, even if it does produce improved fits.
Additional Ground Calibration DataAdditional calibration data were collected by MPE personnel at Neuried during the month of February. The data were collected at various gains and over a more extended range of energies. Previously, there were no useful calibration data that were collected at the present operational gain of the flight instrument. In addition, the high-energy end of the PSPC response was not well sampled. These additional data will allow a better understanding of both the low-energy response and the gain saturation effect. With a better knowledge of the PSPC response to monoenergetic input, we hope to use flight data to empirically fit the temporal variation observed on orbit.
Please use the Feedback link if you have questions on ROSAT.