Req ; 4.1.2; FM-ILT3 Spectrometer Spatial Calibration

Transcription

1 PACS Test Analysis Report FM-ILT Page 1 Req ; 4.1.2; FM-ILT3 Spectrometer Spatial Calibration 4.1.1; 4.1.2; A. History Version Date Author(s) Change description A. Contursi First issue A. Contursi Added spec FOV in sky coordinates Nov 07 A. Contursi Added real background cross checks Added description of how the spatial calibration file have been derived Nov 07 A. Contursi Added PV phase table Dec 07 A. Contursi Added Flat field extraction and attempt to correct for cross talk from module 11 in the RED ; 4.1.2; B. Summary This document reports on the PACS spectrometer spatial calibration results, based on data taken in FMILT 3. First the test description and the analysis up to the reconstruction of the PSF in each module are given. Then, we describe the main results obtained, namely: PSFs characteristics in each module for each chopper position Comparison with theoretical PSF Detector fingerprint in RED and BLUE for each chopper position and for each spectral pixel Distortion: dependence on chopper and grating movements Cross talk Raster maps Ghosts In the second part this report describes the details about how calibration files are generated, and some useful numbers such as the estimated BB flux and background level when PACS looked at the outside world during FM ; 4.1.2; C. Data Reference Sheet See tables 1, 2 and ; 4.1.2; D. Test Description

2 PACS Test Analysis Report FM-ILT Page 2 Table 1: 27 x 27 raster file names, detector settings, on Board reduction, raster step, dwell time per raster position Archive filename Capacitor Bias Zero Bias R ro per ramp OB algo Ratser step in x/y Dwell time (pf) (mv) (mv) (mm)/ pixel sec Blue/Red (Blue/Red) (Blue/Red) FILT x27 chop0 01.tm 100/ / / LSF 0.427/ FILT x27 chop pl 01.tm 100/ / / LSF 0.427/ FILT x27 chop ml 01.tm 100/ / / LSF 0.427/

3 PACS Test Analysis Report FM-ILT Page 3 Table 2: 9 x 9 raster file names, detector settings, On Board reduction, raster step and dwell time in each raster position Archive filename Capacitor Bias Zero Bias R ro per ramp OB algo Ratser step in x Dwell time (pf) (mv) (mv) (mm)/ pixel sec Blue/Red (Blue/Red) (Blue/Red) (mm)/ pixel FILT x9 chop0 grat tm 100/ / / LSF 0.427/ FILT x9 chop0 grat tm 100/ / / LSF 0.427/ FILT x9 chop0 grat t 100/ / / LSF 0.427/

4 PACS Test Analysis Report FM-ILT Page 4 We execute three XY stage rasters, with a hole of 1.5 mm and an external Black Body at temperature equal to 1000 o C. The raster size has been conceived such that it slightly oversizes the spatial FOV size. All rasters start from module 24, proceeding from 24 to 20, from 15 to 19 etc. The step size in both the X and Y stage directions are mm which corresponds to 1/4 th of a spectral pixel, assuming that 1.7 mm is equal to one spectrometer pixel. Note that the X stage coordinate corresponds to -Y detector coordinates (assuming the origin of the detector coordinates in module 0) and the Y stage coordinate corresponds to the X detector coordinates, (see figure 1 this document and figure 12 of PACS-ME-DS-003 Issue 1.0 page 17) and to the direction of the chopper movements. We have executed three 9 9 rasters, with a hole of 1.5 mm and an external Black Body of C, raster step equal to mm (i.e. 1/4 th of spectrometer pixel). The rasters were all centered on module 12 th in order to fully sample the PSF in this module. In fact, in FMILT 2, we executed a 5 5 raster centered on module 12, and this was not su cient to get the full PSF. The three rasters were executed at the same grating position (corresponding to 76.9 and µm) varying the chopper positions to , 650, and (command units) which correspond to the -Large, (Optical) zero, and +Large chopper throws o ered in HSPOT, which correspond to -3, 0 and +3 arcmin in the sky. These sets of data are meant to investigate the spatial distortion introduced by the chopper movements and the orientation of the detector with respect to the chopper movement direction. The three 9 9 rasters, were executed at grating positions , and all with the chopper in the Optical zero position. These data sets are meant to investigate the distortion induced by di erent grating position and to study the relative orientation of the chopper movement direction, the detector and the grating. Table 3 summarizes all measurements. All data sets together allow us to study the PSF in each module, and in each spectral pixel, and the distortion induced by the optics and/or by chopper and grating movements. Table 3: Raster size Grating Wavelength Filter Chopper Angle on the sky Throw (c.u.) µm (c.u.) arcmin (HSPOT) 27 x /153.8 B Large 27 x /153.8 B Zero 27 x /153.8 B Large 9x /213.8 A Zero 9x /183.9 A Zero 9x /123.1 A Zero 4.1.1; 4.1.2; E. Analysis After standard decompression, wavelength calibration (SPG module:wavecalc.py using FMILT 1 wavelength calibration from H.F.), we resample the XY stage coordinates informations, which are sampled every 2 seconds and delivered in the normal HK, to the Frames granularity (SPG module convxystage2pointing.py, after correction for the time problems found by D.L.). In this way, we can average together the Frames taken at the same raster positions. The background subtraction is a tricky issue. In FMILT 3 we have used a better method than in FMILT 2 (see reports on Spectral Spatial Calibration of FMILT 2 analysis). In fact, we realized that the old background subtraction, done fitting a polynomial on the whole pixel s history, was not good enough because it included also the part of the data where the pixel sees the source. This led to significant residuals which showed up as

5 PACS Test Analysis Report FM-ILT Page 5 Figure 1: Conversion between the XYstage coordinates direction and the XY detector array direction. Also the chopper moovement direction is indicated. quite strong structures in the background. In this new version we have performed the same fit but excluding for every spectral pixel of each module, the time range where the pixels see the source. This is quite laborious because currently it is not possible to perform a fit only on ranges of data, asking the fit to ignore some part of the data. So, this has been done manually, excluding for each module (but the same for all spectral pixels of that module) that part of data containing the source and substituting these data with a straight line. Then a polynomial fit on these data has been done. The improvement is quite impressive expecially for the BLUE (higher S/N than the RED channel) as one can see from figure 2. Long but e cient, this method really gives a flat background centered on 0. Then, for every spectral pixel, and every module, we produce a image, where each pixel corresponds to an XY Stage raster coordinate. In the raster, the size is su ciently big to ensure each module has a full well sampled PSF. The file FILT x27 chop pl 01.tm is truncated because of a telemetry drop problem happened during the test execution. There was no time to repeat this raster and it covers most of the FOV anyway. In the 9 9 raster, we cannot subtract the background as explained above, simply because these measurements are too short to ensure a good fit. We executed these mini raster including an o position at the beginning of the recorded file, that we subtract then to the data.

6 PACS Test Analysis Report FM-ILT Page 6 Figure 2: Example of the two methods used for the background subtracion. Method 1 (top panel), performs plynomial fit on the whole data, method 2 (bottom panel) performs plynomial fit only on that part of the data which do not show the source. Further analysis depends on the goals of the investigation and will be reported along with the di erent results ; 4.1.2; F. Results 4.1.1; 4.1.2; F.1. PACS spectrometer FOV We first produce the 27x27 raster for each module averaging all science spectral pixels. Then we did the same for the open and dummy channels only, which should not show any signal. Results are shown in figures from 3 to 8. Figures 3 (BLUE) and 4 (RED) show the 5 5 FOV where each module shows the PSF obtained averaging all science spectral pixels. Data are for the same grating position (i.e command units equal to 76.9 and µm), and chopper position going from -Large to +Large from left to right in the figures.

7 PACS Test Analysis Report FM-ILT Page 7 Figure 5 shows the comparison for the BLUE channel of FMILT 1, FMILT 2 anf FMILT 3, at chopper position zero. From these figures we can notice the following results: The right column of the FOV, is not as week as it was in FMILT 2. We have basically recovered the entire spatial FOV. The bright spot which was visible on the left side of module 4 in FMILT 1 and FMILT 2, is gone completely. Every module of the RED channel shows a faint signal in the position of the PSF of module 11, which indicates cross talks. Figures 6 (BLUE) and 7 (RED) show the 5x5 FOV obtained for the DUMMY channel only. We see that the RED channel shows a source in all modules at all chopper positions. This PSF is always in the same position and corresponds to the position of the PSF in module 11. This indicates strong cross talk from this module to the dummy channel. In the data of FMILT 2 only cross talk in the supply group 2 were detected, although this might be due to a poorer S/N with respect to the data taken in FMILT 3. We checked that this is not due to saturation e ects. All data sets have signals well below 8 V/sec. Some hints of crosstalk from module 11 is also visible in the blue, but it is much less strong than in the RED channel. Figures 8 (BLUE) and 9 (RED) show the 5x5 FOV obtained for the OPEN channel only. Crosstalks presumably from module 11 are visible in supply group 4 (BLUE) as negative crosstalk at chopper position - Large. Some hints of crosstalks are also visible in the OPEN BLUE channel in all modules at chopper position 0. The crosstalks are very strong in supply group 2 of the OPEN RED channel at all chopper positions.

8 PACS Test Analysis Report FM-ILT Page 8 Figure 3: 5x5 spatial FOV of the PSF in the BLUE obtained averaging all spectral pixels, at grating position equal to command units, and chopper positions equal to , +650, and command units. The raster corresponding to chopper position +Large has not been executed completely (see text for details).

9 PACS Test Analysis Report FM-ILT Page 9 Figure 4: 5x5 spatial FOV of the PSF in the RED obtained averaging all spectral pixels, at grating position equal to command units, and chopper positions equal to , +650, and command units. The raster corresponding to chopper position +Large has not been executed completely (see text for details)

10 PACS Test Analysis Report FM-ILT Page 10 FMILT 1 FMILT 2 FMILT 3

11 PACS Test Analysis Report FM-ILT Page 11 Figure 6: 5x5 spatial FOV of the PSF in the BLUE for the DUMMY channel, at grating position equal to command units, and chopper positions equal to , +650, and command units.

12 PACS Test Analysis Report FM-ILT Page 12 Figure 7: 5x5 spatial FOV of the PSF in the RED for the DUMMY channel, at grating position equal to command units, and chopper positions equal to , +650, and command units.

13 PACS Test Analysis Report FM-ILT Page 13 Figure 8: 5x5 spatial FOV of the PSF in the BLUE for the OPEN channel, at grating position equal to command units, and chopper positions equal to , +650, and command units.

14 PACS Test Analysis Report FM-ILT Page 14 Figure 9: 5x5 spatial FOV of the PSF in the RED for the OPEN channel, at grating position equal to command units, and chopper positions equal to , +650, and command units.

15 PACS Test Analysis Report FM-ILT Page ; 4.1.2; F.2. Comparison between the theoretical and the observed PSF Figures 10 and 11shows the comparison between the theoretical PSFs at 76.9 and µm, obtained convolving the PSF at these wavelengths with a hole of 1.5 mm, and the observed PSFs at the same wavelengths. We show the comparison for the BLUE and the RED channels, only for module 0, 12 and 24, although the results are similar for all others modules. These figures clearly show that the agreements between the theoretical and observed PSF is excellent and we can therefore conclude that the PSFs is as expected in all modules at the sampled wavelengths. Tables 4 to 6 list the 2D gaussian parameters resluting from the fit on the 5 5 PSFs obtained averaging all spectral pixels, for the chopper position at the minus large (table 4 ), optical Zero (table 5 ) and plus large (table 6). Table 4: 2D fit parameters results for the BLUE and RED spectrometer at chopper position equal to minus Large. the PSF peak position is in XY stage coordinates and the sigma in the same coordinate system are expressed in spectrometer pixel units. module x peak y peak x y x peak y peak x y x stage y stage (p.u.) (p.u.) x stage y stage (p.u.) (p.u.) BLUE BLUE BLUE BLUE RED RED RED RED ; 4.1.2; G. Analysis 4.1.1; 4.1.2; G.1. Distortion

16 PACS Test Analysis Report FM-ILT Page 16 Table 5: 2D fit parameter results for the BLUE and RED spectrometer at chopper position equal to Optical Zero. the PSF peak position is in XY stage coordinates and the sigma in the same coordinate system are expressed in spectrometer pixel units. module x peak y peak x y x peak y peak x y x stage y stage (p.u.) (p.u.) x stage y stage (p.u.) (p.u.) BLUE BLUE BLUE BLUE RED RED RED RED For each of the raster, we fit the PSF obtained in each spatial pixel, either averaging all spectral pixels, or in each spectral channel, with a 2d Gaussian, obtaining the and the x, and y position on the array of the peaks. We then transform each peak position into XY stage coordinates. The results are shown in figure 12. It shows the XY stage coordinates of the PSF peak position for the RED and the BLUE overplotted, for each chopper position, and at grating commanded units equal to The crosses represent the commanded XY stage coordinates during the tests. We note that: the spatial FOV is rotated w.r.t. the XY stage. The angle of rotation is a function of the chopper which might indicate that the array is rotated also w.r.t. the chopper. This is illustrated in figure 13 where the rotation in degree of the RED and BLUE arrays has been calculated from the XY stage position of the last row (module from 20 to 24). the displacement of modules from 5 to 9, is still there, as expected, since the mirror movements done before FMILT 3 test campaign begun, was not supposed to modify this aspect. no significant di erences of the FOV fingerprint between the di erent chopper positions are noticeable. Figure 13 shows the rotation of the FOV with respect to the XY stage coordinates, as function of the chopper position. We see that this rotation is significant and decreases as the chopper throw increases. The rotation

17 PACS Test Analysis Report FM-ILT Page 17 Table 6: 2D fit parameter results for the BLUE and RED spectrometer at chopper position equal to Plus Large. the PSF peak position is in XY stage coordinates and the sigma in the same coordinate system are expressed in spectrometer pixel units. module x peak y peak x y x peak y peak x y x stage y stage (p.u.) (p.u.) x stage y stage (p.u.) (p.u.) BLUE BLUE BLUE BLUE RED RED RED RED goes from to degrees going from -Large to +Large chopper throw. Also important the rotation seems to be the same for the RED and the BLUE array. In this figure we have also included the same measurements done for FMILT 2 data, where we had two more chopper positions (at ± Medium chopper throw). The overall agreement is good, but the dispersion of the data between the RED and BLUE points is much less in FMILT 3 than in FMILT 2. This is probably due to the higher S/N the data taken in FMILT 3 have w.r.t. the data taken in FMIL 2 and the better background subtraction we apply to the FMILT 3 data.

18 Figure 10: Cayan: observed monodimensional profiles of the PSFs obtained averaging all spectral pixels for PACS PACS Test Analysis Report FM-ILT Page 18

19 Figure 11: Cayan: observed monodimensional profiles of the PSFs obtained averaging all spectral pixels for PACS PACS Test Analysis Report FM-ILT Page 19

20 PACS Test Analysis Report FM-ILT Page 20 Figures 14 and 15 shows the fitted PSF peak XY coordinates for the BLUE and the RED respectively, obtained in FMILT 2 and FMILT 3. There is a displacement of about one half spectral pixels between the two. It is also noticeable in the RED channel (figure 15) that the last slice (from 20 to 24) position in X stage coordinate, is di erent from what obtained in FMILT 2. Currently its distance from the next slice (from 15 to 19) is much more similar to the distance the other slices have. This means that the RED channel is less distorted than what found in FMILT 2. Figures 16 (BLUE) and 17 (RED) show the position of each PSFs peak for each spectral pixel. As found already in FMILT 2, these figures clearly indicate that there is a displacement of the PSF peak in the spectral domain both in the BLUE and in the RED channels. The displacement becomes more and more severe going from module 0 to module 24. The maximum displacement is of about mm i.e. one half of pixel, in both channels for module 24.

21 PACS Test Analysis Report FM-ILT Page 21 Large Opt. 0 + Large Figure 12: PSF peak position of each module on the XY stage coordinates (i.e. sky coordinates) for all chopper positions and grating position equal to command units.

22 PACS Test Analysis Report FM-ILT Page 22 Figure 13: Calculated rotation between the spatial FOV and the XY stage as function of chopper. Also shown are the rotation angles calculated in FMILT 2 for comparison.

23 PACS Test Analysis Report FM-ILT Page 23 Figure 14: FMILT 2 - FMILT 3 BLUE PSF peak position comparison, of each module on the XY stage coordinates (i.e. sky coordinates) for chopper positions at Optical zero, grating position equal to command units.

24 PACS Test Analysis Report FM-ILT Page 24 Figure 15: FMILT 2 - FMILT 3 RED PSF peak position comparison, of each module on the XY stage coordinates (i.e. sky coordinates) for chopper positions at Optical zero, grating position equal to command units.

25 PACS Test Analysis Report FM-ILT Page 25 Large Opt. 0 + Large Figure 16: BLUE PSF peak position of each module on the XY stage coordinates (i.e. sky coordinates) for all chopper positions and grating position equal to command units and for every spectral pixel

26 PACS Test Analysis Report FM-ILT Page 26 Large Opt. 0 + Large Figure 17: RED PSF peak position of each module on the XY stage coordinates (i.e. sky coordinates) for all chopper positions and grating position equal to command units and for every spectral pixel

27 PACS Test Analysis Report FM-ILT Page ; 4.1.2; G.2. Spatial FOV in the sky as function of chopper position Since we have calculated the position of each spatial module is in XY stage coordinates (. i.e. the Y stage coordinate.) at each chopper position, we can how this this varies as function of chopper position. Figures 18 and 19 show the fitted PSF position of each module on the XY stage coordinates, for all chopper position at once for BLUE and RED respectively. Beside the rotation of the FOV w.r.t. the XY stage coordinates we have already analyzed, (figure 13) there is also a distortion which depends on the chopper positions. It does not seem to be linear. It can be fitted with a 2 nd degree polynomial, but with more chopper poistions, this distortion may come out to be more complicated that this. Figures 20 and 21 show the comparison between FMILT 2 and FMILT 3, where it is clear the di erence in the RED position of the row composed by modules from 20 to 24. This di erence, between FMILT 2 and FMILT 3 becomes more significant going from chopper position -Large to +Large ; 4.1.2; G.3. mm to arcsec scale Assuming that the chopper positions corresponding to ±Large, are exactly ±3 arcminutes in the sky, we can derive the scale corresponding to 1 arcsec in mm in the XY stage coordinates. We can do this for each module and for the scale derived from the position of the modules on the XY stage between the -Large and optical zero position and between +Large and optical zero position. The results are displayed in figures 22 and 23 for BLUE and RED channel respectively. The expected conversion is : mm equal to 1 arcsec. We see that the derived scale for each modules is very very close to the expected values for all modules and for both RED and BLUE channels ; 4.1.2; G.4. Spectrometer FOV in sky coordinates From PACS-ME-DS-003 page three, we learn that f=37154 mm. This correspnds to the following conversion factor: 1 = 0.18 mm Using this conversion factor, we can therefor transform the XY stage PSF position peak in each module and per each chopper position, in sky coordinates. We can calculate the distance in arcsec of each pixel w.r.t. a reference pixel. As reference we choose the PSF peak of module 12 in the BLUE channel at chopper angle equal Optical Zero and grating position equal to command units. The result is illustrated in figure 24 for the RED and BLUE channels. It is interesting to note, that the distance of module 12 in arcsec between chopper position ±Large, is and arcsec, in excellent agreement with what they should be (± 3 ) ; 4.1.2; G.5. Grating and detector alignement The 9 9 rasters were executed moving the source on a grid centered on module 12. The aim was to get a PSF sampled at 1/4 th of a spectral pixel of module 12 at fixed chopper position but moving the grating. The data

28 PACS Test Analysis Report FM-ILT Page 28 were reduced and analyzed as the big rasters (see Section E). We then fit the PSF peak position at each grating position on module 12 and transform these coordinates in XY stage coordinates, exactly as we did for the raster for all modules. Figurew 25 and 26 show the derived peaks for each grating position, including the one obtained for module 12 in the rasters, on the XY stage coordinates. We can see that the derived XY coordinates of the PSF peaks at various wavelengths agree very well, with a dispersion less than 1/4 th of the pixels. This means that the grating movements do not introduce spatial distortion in both channels. The PSF at µm does not have enough S/N for the Gaussian 2D fit to converge. This is why its corresponding symbol is missing in figure ; 4.1.2; G.6. Conservation of the flux In order to check whether each module receives some signals also from other modules or looses some signal when the source is in between the slices, we have created a sort of total map. We have first normalized the background subtracted PSFs in each module with their peak, in the rasters. If there are no additional or lost of signals while the source is moving, the result should be an uniform wide region everywhere but at the edges where the source is out of the spatial FOV. The results are shown in figures from 27 and 32. Each of these figure displays for one channel (BLUE or RED) and one chopper position, in the top panel with the full dynamical range, in the bottom panel, cutting the highest values in order to quantify better the contribution of the discontinuities, if presents, over the background. The big crosses in each figure represents the fitted peak positions of the PSFs. The small crosses represent the coordinates of the raster executed. From these figures we can see the overall shapes of the spectrometer FOV, with its characteristic of having the second slice displaced with respect to the others and the whole FOV being rotated with respect to the XY stage. In general we can also see that the BLUE channel is more homogenous than the RED channel. The variation above the median, although di cult to determine with precision without a proper flat fielding, is around 20% in the BLUE (but varies around this value). The RED channel has a big and intense signal in a position which is close to module 11. This is due to the crosstalk from module 11 visible in all modules. Its contribution can be as high as a factor of 3 to 4. Its position however, does not corresponds exactly to the peak position of module 11, although it is displaced towards the bottom of just 1/4 th of a spectral pixel. It is di cult to assess with these data whether this displacement is real or not, and if real, what does it tell us. It is clear that the presence of this significant cross talk from module 11 in the RED spectrometer, prevents us to assess how homogeneous is the FOV when all signals are summed up. In order to investigate better this issue we have tried to correct the signal arising form this cross talk in each image, i.e. for each module. We proceeded in the following way. After visual inspection of the signal from each module far enough from the position of the PSF in module 11, we have individuated a template signal which we assume represents the contribution of this cross talk for all modules. This is better shown in figure 33 where the signal from module 11 is overplotted to the signal of module 1, judged to be the best cross talk template. The signal from module 1 in fact, show significant dips at the position of the source in module 11, indicated in the figure with red boxes. So, in each position included in the boxes, we have subtracted the signal of module 1 from all modules. We show the result for module 4 in figure 34. Then we normalized and summed up all 25 signals as before, and created the map of the conservation of the flux shown in figure 35 (for chop at Optical Zero only). Here we can see

29 PACS Test Analysis Report FM-ILT Page 29 that the FOV is more homogeneous than previously obtained summing up the signal of each module without the attempt to mask out the cross talk (see figure 27). Despite the fact that the explained procedure does not insure a perfect cross talk correction in all module, therefor some degree of inhomogeneity in the map shown in figure 35 is still due to the cross talk contribution, it looks like the RED FOV is e ectively less homogeneous than the BLUE FOV ; 4.1.2; G.7. Raster Maps In this section raster maps of the PSF are shown. These maps have been done just shifting each raster image of each module on top of each other using the fitted PSF position. Then, each pixel is devided by an integer equal to the times it has been summed up in the reconstructed map. This execrcise should be done after a proper RSRF correction, which is not available at the time this report is written. This implies that, although the background subtraction works well, some structure still remains and they are well visible in the final maps. The maps are shown in figures 36 and 37 for the BLUE and the RED channels respectively and for the chopper position equal to the Optical 0.

30 PACS Test Analysis Report FM-ILT Page 30 Figure 18: BLUE PSF peak position of each module on the XY stage coordinates (i.e. sky coordinates) for all chopper positions and grating position equal to command units.

31 PACS Test Analysis Report FM-ILT Page 31 Figure 19: RED PSF peak position of each module on the XY stage coordinates (i.e. sky coordinates) for all chopper positions and grating position equal to command units.

32 PACS Test Analysis Report FM-ILT Page 32 Figure 20: BLUE PSF peak position of each module on the XY stage coordinates (i.e. sky coordinates) for all chopper positions and grating position equal to command units for FMILT 2 (crosses) and FMILT 3 (solid circles).

33 PACS Test Analysis Report FM-ILT Page 33 Figure 21: RED PSF peak position of each module on the XY stage coordinates (i.e. sky coordinates) for all chopper positions and grating position equal to command units for FMILT 2 (crosses) and FMILT 3 (solid circles).

34 PACS Test Analysis Report FM-ILT Page 34 Figure 22: Conversion between mm and arcesec for each module of the BLUE detector as sees from the data assuming that chopper at +Large, 0 and -Large corresponds to +3, 0 and -3 arcminutes in the sky

35 PACS Test Analysis Report FM-ILT Page 35 Figure 23: Conversion between mm and arcesec for each module of the RED detector as sees from the data assuming that chopper at +Large, 0 and -Large corresponds to +3, 0 and -3 arcminutes in the sky

36 PACS Test Analysis Report FM-ILT Page 36 Figure 24: Posistion in the sky of the Spectrometer FOV in the RED and BLUE channels, per each chopper position. The reference position (0,0) corresponds to module 12 in the BLUE channel at chopper equal to optical Zero.

37 PACS Test Analysis Report FM-ILT Page 37 Figure 25: 5 5 BLUE spatial FOV reconstructed from the 9 9 rasters around module 12 for grating position at (top-left), (top-tight) and (bottom).

38 PACS Test Analysis Report FM-ILT Page 38 Figure 26: 5 5 RED spatial FOV reconstructed from the 9 9 rasters around module 12 for grating position at (top-left), (top-tight) and (bottom).

39 PACS Test Analysis Report FM-ILT Page 39 + Large Figure 27: Summ of the signal for the BLUE at chopper position + Large in its full dynamic range (top panel)

40 PACS Test Analysis Report FM-ILT Page 40 + Large Figure 28: Summ of the signal for the RED at chopper position + Large in its full dynamic range (top panel)

41 PACS Test Analysis Report FM-ILT Page 41 + Optical 0 Figure 29: Summ of the signal for the BLUE at chopper position Optical Zero in its full dynamic range )top

42 PACS Test Analysis Report FM-ILT Page 42 + Optical 0 Figure 30: Summ of the signal for the RED at chopper position Optical Zero in its full dynamic range (top

43 PACS Test Analysis Report FM-ILT Page 43 Large Figure 31: Summ of the signal for the BLUE at chopper position - Large in its full dynamic range )top panel)

44 PACS Test Analysis Report FM-ILT Page 44 Large Figure 32: Summ of the signal for the RED at chopper position - Large in its full dynamic range )top panel)

45 PACS Test Analysis Report FM-ILT Page 45 Figure 33: Signal from module 1 and module 11 overplotted. Red boxes indicate the position where cross talk from module 11 is visible in module 1.

46 PACS Test Analysis Report FM-ILT Page 46 Figure 34: Comparison between the singal of module 4 before and after masking the cross talk contribution. See text for details.

47 Figure 35: Summ of the signal for the RED at chopper position Optical Zero after cross talk from module 11 correction, in its full dynamic range (top panel) and in its highest countour values. To be compared with figure PACS PACS Test Analysis Report FM-ILT Page 47

48 PACS Test Analysis Report FM-ILT Page 48 Figure 36: Reconstructed raster map in the BLUE, at 76.9 µm and chopper position equal to Optical Zero. We can clearly see that the resulting background has structures due to the fact that we have not flat fielded the images before combining them into the map. Nevertheless, in the BLUE channel two ghosts appear clearly: one on each side of the reconstructed PSF. In the RED channel the PSF is elongated on the left, and this is due to the crosstalk from module 11 present in each single image used to reconstruct the map. A crude estimation of the most intense ghost reveals that it is negligible (0.1% level) with respect to point source photometry. But its contribution is presumably much more significant with respect to the more extended faint emission. With the current data it is not possible to quantify this contribution. The estimated rotation angle between the direction defined by the ghosts and the PSF and the XY stage coordinates is The distance of each ghost from the PSF is 2.97 and 2.77 spetral pixel for the left and right ghost respectively.

49 PACS Test Analysis Report FM-ILT Page ; 4.1.2; G.8. Ghosts in the BLUE The results shown in the previous section induced us to go back and check again the single PSFs images in each module. The ghosts are clearly detected when one stretches the dynamical range of figures 3, as it is shown in figure 38. The path followed by the ghost is regular: ghosts in the first two columns of the spatial FOV are on the right of the PSF; ghosts of the last two columns are on the left of the PSFs. This could also mean that there are always two ghosts, one on each side of the PSFs, but only one at the time is visible since the amplitude of the raster is not su cient to include both at the same time. Although not confirmed, this hypothesis could be strengthen by the fact that in the central column the ghost appears sometime on the right sometime on the left of the PSF. At the moment no explanation has been found for the origin of these ghosts. We do not know whether these originate in the external window, or inside PACS. It looks like their intensities vary as function of chopper position. Form figure 38 it looks like they are stronger when the chopper is at its Optical zero. But we believe this e ect is more due to the goodness of the background subtraction, which is better for this chopper position than for the others, and therefore make the ghosts appear stronger. No firm conclusions on this point are possible without a proper flatfield ; 4.1.2; H. Conclusion The main results of the FMILT 3 spectral spatial calibration campaign are the following: We have recovered the entire spatial FOV, both in the BLUE and RED channels. The observed PSFs are in excellent agreement with the theoretical PSF Strong crosstalks with module 11 are detected in the RED channel for science and dummy pixels in all modules; in the RED open channel only supply group 2 shows such strong crosstalks; in the BLUE channel crosstalks are much fainter but they probably exist as well. The spatial FOV is rotated w.r.t. the XY stage. The rotation angle is a function of the chopper position and it varies from to from -Large to +Large chopper throws. The position of each module on the sky varies for each spectral pixel. This variation increases smoothly from module 0 to module 24 where the spread on the spectral domain is maximum and corresponds to 1/2 of spectral pixel for both channels the mm/arcsec scale determined from the data, assuming that the chopper throws equal to ±Large and Optical zero exactly correspond to ±3 arcminutes and 0 arcminutes in the sky, is as expected(, i.e mm = 1 arcsec), and the dispersion around this value for each module is very small The grating movements do not introduce significant spatial distortion Ghosts are visible in the BLUE channel, at all chopper position. Their origin is still unknown. Their contribution to the point source obtained at 76.9 µm with a Black Body of C is negligible but it can be significant for fainter emission.

50 PACS Test Analysis Report FM-ILT Page ; 4.1.2; I. Calibration 4.1.1; 4.1.2; I.1. Spectrometer spatial calibration files: distortion In this section we describe how the spatial calibration files for the spectroemter have been derived. A description on how the distortion files should look like is in PICC-ME-TN-019. In this definition care has been taken as to make the photometer and spectrometer distortion calibration file structures as similar as possible. We therefore will refer to the Bolometer Spatial Calibration FM-ILT report, in this section, to enphatize the analogies and di erences on how the spatial calibration files for the two instruments were derived. These spatial calibration files are essentially of two types: One file describes the conversion between array pixel coordinates (p,q) into coordiantes u,v in the focal plane (in mm). For the spectrometer, instead of defining a 5 5 array coordinates p,q, we define 1 dimensional array each representing one module. The first step of the bolometer spatial calibration is to describe the distortion of each subarray matrix w.r.t. a the center of a reference matrix, in u,v coordiantes, assuming that the psysical size of the bolometer pixel is 0.75 mm. This calibration file has been called SubarrayArray for the bolometer. We call the analog for the spectrometer ModuleArray. For the spectrometer we describe each module position w.r.t. module 12. Each position is a mean of the position of the measurements taken at the three chopper positions. We have in fact verified that these three measurements do not di er significatly in terms of each module position w.r.t. module 12 (see figure 39). To convert the pixel (module) coordinates to u,v array coordinates on the focal plane, we have assumed that the pixel size is mm (A.Polg. private communication). So, we scale the whole spectrometer fingerprint by the ratio between the mean of the distrance of all adjacent pixels within a slice to 3.6, such that in the u,v representation the mean distancies are very close to 3.6 mm. We then align the array to the chopper direction (v). The angle between the chopper direction and the XY stage, has been calculated by running a first version of ModuleArray calibration file into the procedure that calculates the global distortion (see below). The calibration file then contains, for each module, the center position coordinate u center,v center w.r.t. module 12, and the corner pixel coordinates obtained assuming 3.6 mm as pixel size and perfectly squared pixels. This calibration file, called PacsCal SpecModuleArray FM 1 0.fits, contains u,v, corodinates for centers and corners, for the blue and the red, in complete analogy with the corresponding files produced for the bolometer. The second calibration file describes the global optical distortion on the focal plane, ı.e. the distortion as function of chopper position. We apply the exact same fitting procedure applied on the bolometer and described in detailed in section E3 of the Bolometer Spatial calibration report and in PICC-ME-TN-019. Since in the spectrometer case, we have much less points, both for the limited number of pixels, and for only three chopper positions, we tried first polynomial order equal to 1. The fit was not good, and at the end we find a very good fit using N=M=O equal to 2, as for the photometer. The results are shown in figures 40 and 41 for the BLUE and the RED detectors. We obtained and r.m.s. residues of both in y and x direction in the BLUE, and and in y and x direction respectively in the RED. The calibration file, PacsCal SpecArrayInstrument FM 1 0.fits contains the polynomial orders and coefficients for the x and y direction of the RED and BLUE detectors, in complete analogy with what derived for the bolometer.

51 PACS Test Analysis Report FM-ILT Page 51 For completeness we report in Table 7 the chopper angles in degrees obtained using the Pos2FpuAnglePos and FM2 2 version in DP. In table 8 we list the various angles obtained assumig that the first order coe cient a 001 of the chopper position obtained from the fit to the global distortion, is the angle between the chopper and the XY stage. This is also the angle that has been used to align the detector to the chopper direction. Chopper c.u. Table 7: Chopper angles in command units and in degrees Chopper angle Table 8: Misalignement angles caclualted as explained in the text. Analogous to Table 10 in the Bolomenter Spatial Calibration report. Angle Value degree Comments Chopper to XY stage 2.1 Average between a 001 coe. of red and blue Chopper to blue array 0.8 XY to blue array - Chopper to XY stage angle Chopper to red array 0.8 XY to blue array - Chopper to XY stage angle XY to blue array 2.94 Mean of array to XY stage tilt of all slices XY to red array 2.89 Mean of array to XY stage tilt of all slices 4.1.1; 4.1.2; I.2. Software tool used for the analysis The standard decompression, application of some existing SPG modules, background subtraction and PSF reconstruction form the rasters, have been done within the DP software. Further analysis which included robust 2D Gauss fitting, and writing and reading of tables, has been entirely done in IDL, because of lack of functionalities in DP and much more e ciency in IDL than in DP. The Calibration files were derived in IDL and then transfoermed in fits files within IA where we defined also the class and associated methods for each calibration file ; 4.1.2; I.3. External BB flux density With the aim of preparing the PV phase, and particularly to understand the flux level of the point sources we need for the Spatial Calibration in order to get an acceptable S/N, we have tried to estimate the external BB flux in Jy. The calibration between ADU and Jy was done by H. Feuchtgruber based on a full spectrum obtained on cold OGSE BB#1, i.e. on an extended source. (The file used is: FILT Batch PACS Spec Rsrf OBS 86.tm) For this measurement he found the following numbers: BLUE detector: GPR = (same grating position as the XY Stage raster measurements, see Table 3.) Signal (dark)= -12 (ADU) Signal (BB@28K)= (ADU) Capacitor = 0.1 pf. RED detector: GPR = (same grating position as the XY Stage raster measurements, see Table 3.)

52 PACS Test Analysis Report FM-ILT Page 52 Signal (dark)= -704 (ADU) Signal (ADU) Capacitor = 0.1 pf. The raster measurement at chopper equal to Optical Zero gives the following corresponding numbers: BLUE detector: GPR = (same grating position as the XY Stage 27 raster measurements, see Table 3.) Signal peak point source= (ADU) Signal background= (ADU) source=828 Capacitor = 0.1 pf. RED detector: GPR = (same grating position as the XY Stage 27 raster measurements, see Table 3.) Signal peak point source= (ADU) Signal background= (ADU) source=169 Capacitor = 0.4 pf. Taking into account that signal(@04 pf)=signal(@0.1 pf)/3.21, the RED source signal is equal to ADU. The ADU to Jy conversion (H.F. private communication) are the following: Conversion BLUE (@ grating position = ) = ADU/Jy Conversion RED (@ grating position = ) = ADU/Jy Note that these numbers apply only to signals detected with a capacitor equal to 0.1 pf. For any other capacitor setting, one should scale the signal to this capacitor before applying these conversions. With these conversion numbers we obtain: Source BLUE = 828 digits = 103 Jy Source RED = digits = 22 Jy These numbers have to be corrected for the point source correction (the ADU/Jy correction has been derived from data taken on an extended source while in the XY Stage rasters we looked at a point source) and for the air transmission correction. These are (H.F. private communication):

53 PACS Test Analysis Report FM-ILT Page 53 Point source 77 µm = 0.85 Point source 154 µm = 0.53 Air transmission 77 mum = 0.94 Air transmission 154 mum = 0.98 Therefor, the external BB at 1000 K corresponds to a flux density of: 103/point source correction= 121 Jy/airtransmission correction= µm 22/point source correction= 41.5 Jy/airtransmission correction= µm The theoretical ratio between the µm and the 154 µm, for a pure BB at 1000 K is 3.84, the observed ratio is The di erence might be due to the window tranmission, not well known, that is likely to change as function of wavelengths ; 4.1.2; I.4. The background level in the raster maps In this section, we estimate the background level we had during the XY raster tests, i.e. while looking to the outside world through the cryostat window, at 77 and 154 µm. We then compare this background to the typical expected background. We evaluated the ratio between the background during the XY Stage tests and the expected background both in V/s and in Jy. For the RED detector we proceeded as explained below. We started from the following assumption taken from PICC-MA-TR-29: 1) the signal per pixel from the CS2 at 55 K, taken with a capacitor equal to 0.1 pf and at µm is 1.6 V/s Since, in the XY stage measurements, in the RED filter the capacitor was set to 0.4 pf, we have to scale the above signal value to the capacitor ratio (equal to 3.21 see Wiki page) in order to be able to compare the two measurements. Therefore, the signal arising from a CS2 at 55 K, with a capacitor equal to 0.4 pf at µm is 0.5 V/S. The signal from the CS2 at 55 µm at µm is not quite the expected background at the same wavelength, as it is shown in figure 42, but a factor 2.4 higher. Therefore the background at µm, with capacitor equal to 0.4 pf should correspond to 0.5/2.5=0.2 V/s. In order to compare this value with the background signal detected in the XY stage tests, we should scale the signal from µm to 154 µm, i.e. the long wavelength where the big raster measurements were done. This is done using the RSRF shown in figure 43. From this figure we can see that the RSRF at µm is 0.47 times the RSRF at 154 µm. Therefor, the background at 154 µm with a capacitor equal to 0.4 pf corresponds to a signal equal to 0.2/0.47 =0.42 V/s.. The background signal in the same conditions and wavelength in the XY-stage raster measurements at chopper position equal to Opt. 0 is 4.9 V/s. This corresponds to a XY stage raster background 11 times higher than the expected background.

54 PACS Test Analysis Report FM-ILT Page 54 We can cross check this number knowing that the background in the XY-stage raster measurements corresponds also to 1700 Jy (H.F. private communication) and that the background at 154 µm is 175 Jy. Their ratio is equal to 9.7, very close, taking into account the uncertainties, to the ratio measured in V/s. For the BLUE detector we proceeded as for the RED, taking into account that in this case the capacitor was set at 0.1 pf and that the emission from a CS2 at 55 K is very close to the background at 77 µm (see fig 42). The resulting ratio in V/s is 6.6, while the ratio in Jy is 6.7. Taking into account the uncertainties, most likely around 30%, mostly due to the fact that we read the various factors from figures, and also that we applied an unique capacitor conversion that in reality varies from pixel to pixel, we can conclude that we have got an excellent agreement between the estimates done in V/S and in Jy. The background levels for 0.1 pf, at 77 and 154 µm are, 3000Jy and 1700 Jy respectively. The Signal to Noise reached in the XY stage raster measurements, at chopper position equal to Optical Zero, for a 12 sec dwell time, and an external source equal to 129 and 42 Jy respectively at 77 and 154 µm are 100 and 12. Table 9 summirizes all the relevant numbers we need in order to plan PV phase. Table 9: Important numbers to consider as input to the Spectral Spatial calibration PV planning Description Value Observed Ext. BB fux 77 µm 129 Jy Observed Ext. BB fux 154 µm 42 Jy Observed background level in XY 77 µm 3000 Jy Observed background level in XY 154 µm 1700 Jy Observed Background / 77 µm 6.6 Observed Background / 154 µm 10 Observed S/N at 77 µm with 12 sec in each raster 100 Observed S/N at 164 µm with 12 sec in each raster ; 4.1.2; I.5. Implication for IST and PV phase The PCD (PACS-MA-GS-001) requirement for the Spectral Spatial calibration, separates 5 issues: Spec. central pointing and grating alignemens (req ) Spec. FOV Distortion, req Spec. PSF, req Spec. ghosts, req Spec. straylight, req Allthaugh in the PCD di erent procedures to be executed in FMILT were defined to tackle each of this measurement, in real time, due to the high time pressure, we actually executed only a couple of measurement types: i.e. small and big raster, and a visual central pointing of the source in the center of the spectrometer FOV.

55 PACS Test Analysis Report FM-ILT Page 55 Nevertheless, the FMILT 3 test campaign has produced very nice data sets, suitable for addressing all main issues related to the spatial calibration of the spectrometer. In principle, the same test sequence is therefor suitable for being repeated in PV phase, with the inclusion of full raster at chopper throws equal to ± Medium that we did not perform in FMILT 3 for lack of time, (priority I), and if possible full rasters at same chopper throw but di erent grating position to check whether all spatial pixels are not influenced by the grating movements as we could currently check only for module 12 (priority II). In PV phase, additional test execution are envisaged as explained in the PCD. The test types and the requiremets on the calibration sources necessary in terms of brightness and morphology for the first 3 requirements listed above, are listed in Table 10. In addition, both FS and PV should include tests time to investigate the origin of the BLUE ghosts. If they will be still present, meaning that they originates in PACS, we should map them carefully, and determine their contribution at di erent source flux levels and at di erent wavelengths. This could imply an additional calibration e ort during operation, to map ghosts periodically in time at certain key wavelengths and nominal chopper throws, in order to be able to correct these out in the final images ; 4.1.2; J. Flat Field extraction (preliminary From the XY stage raster measurements it is also possible to extract a Flat Field (FF). In principle one should first apply the RSRF before extracting the FF from these measurements. Since this is not currently yet possible due to the lack of the FM RSRF, the results shown here are still not correct, although the procedure has been established. Note that, extracting the Flat Filed from such raster measurements, is what is currently foreseen for PV. As explained previously, we have built 5 5 PSF for all spectral pixels averaged together, but also for each spectral pixel. For each of these 5 5 PSFs, that flat field is equal to the ratio between the PSF peak of each module w.r.t. the peak of module 12. The biggest uncertainty in this derivation is the background subtraction, which di er from di erent modules. This uncertainty increases with the decrease of the Point Source S/N. For what we have estimated in terms of flux necessary for high S/N, we conclude that we need very bright sources possibly at more than one key wavelengths.

56 PACS Test Analysis Report FM-ILT Page 56 Table 10: Main measurements type requested for the spectral spatial clibration in PV phase. REQ. type of measurement chop grat type of source tint open Central Pointing Pos. small choopped ( FOV) (7x7) Req T.B.D. 20 Jy blue (S/N 20) 20s wave raster, 1/3 step 1 set Faint haloes OK On array chop (1,1.5 pix throws) Req T.B.D. 10 Jy blue (S/N 20) 60s wave nod ortho chop FOV (o ) 1 set Faint haloes OK Repeat pix nod ortho chop FOV distortion 27x27 chopped ( FOV) raster, At least: T.B.D. 20 Jy blue (S/N 20) 20s wave 1/4 step. NOD to OFF pos every ± L,±M,0 >1 set No faint haloes better more chop? TBD time chopper( FOV) scan; 30 legs At least: T.B.D. 20 Jy blue 2, spaced 2, 1 /sec. ± L,0 (if >1 set No faint haloes better N.A. wave 2 ortho scans prev. done) Repeat pix nod ortho chop PSF 27x27 chopped ( FOV) raster, At least: T.B.D. 100 Jy blue (S/N 20) 20s wave 1/4 step. NOD to OFF pos every 0 >1 set No faint haloes more chop? TBD time Chopped AOR with NOD FOV At least T.B.D. 100 Jy blue (S/N 20) 20s wave 0 >1 set No faint haloes more chop?

57 PACS Test Analysis Report FM-ILT Page 57 Figure 37: Reconstructed raster map in the RED, at µm and chopper position equal to Optical Zero.

58 PACS Test Analysis Report FM-ILT Page 58 Large Opt. 0 + Large Figure 38: Same Figure as Figure 3, but with the dynamical range stretched such the ghosts are clearily visible.

59 PACS Test Analysis Report FM-ILT Page 59 Figure 39: position of each spectrometer module w.r.t. module 0 for the three chopper positions.

60 PACS Test Analysis Report FM-ILT Page 60 Figure 40: Global distortion fit for the BLUE: black crosses are the measurements, red the fitted pixels.

61 PACS Test Analysis Report FM-ILT Page 61 Figure 41: Global distortion fit for the BLUE: black crosses are the measurements, red the fitted pixels.

62 PACS Test Analysis Report FM-ILT Page 62 Figure 42: Ratio between the fluxes emitted by the CSs at various temperatures and the expected telescope background (H.F. private communication) in all PACS wavelength range.