From hermanm@eureka Mon Jan 6 03:20:37 1997 Received: from eureka.mit.edu by wiwaxia AA09989; Mon, 6 Jan 97 03:20:35 EST Received: by eureka AA20867; Mon, 6 Jan 97 03:20:33 EST Date: Mon, 6 Jan 97 03:20:33 EST From: Herman L. Marshall To: bertb@sron.ruu.nl, dd@eureka, frits@astro.columbia.edu, jdrake@head-cfa.harvard.edu, jiahong@head-cfa.harvard.edu, juda@cfa.harvard.edu, kaastra@rulrol.LEIDENUNIV.NL, kaf@eureka, prp@mpeu02.rosat.mpe-garching.mpg.de, theog@sron.ruu.nl Subject: Preliminary report on Grating shutter focus tests Status: R Sorry for the length of the report.... Herman _______ To: Grating folks From: HLM Subject: Grating focus and MEG +3 anomalous images SUMMARY I have examined the HSI shutter focus data for several tests. The first objective was to determine the focus location for the grating tests and the second objective was to measure certain anomalies in the dispersed images. Results are incomplete and poorly understood at present but here is some information that should help to understand what is happening. The defocus measurements seem to indicate that the HSI is about 180 microns ahead of focus for all gratings (order 0). Defocus values at other orders have not been measured yet. Images of each MEG +3 order show anomalous additional spots from 20-60 pixels (100-400 micron) from the dominant image. The fractional light in the spots varies from 2% to 6% of the dominant image. The spot at the largest distance also appears in the +1 image as well, which may indicate that an MEG grating facet (in the N shutter set) is misaligned (rotated about +x) by 24 arcmin. The other spots would only indicate a misalignment of 6-8'. Tests have been planned to diagnose the problem further. DETAILS OF ANALYSES 0th order shutter focus tests The Shutter Focus tests using the gratings were IDs D-{L,H}XH-SF-1.00{1-11}. The first 4 of the series were images of order 0 for the LETG, MEG, and HEG, in order. The SF tests normally require 8 separate HSI images: one for each quadrant and two iterations. The first iteration is used to estimate the focus position and the second is used to verify the computed defocus by shifting to the new focus position before the measurement. For the LETG, the 2nd floor SF algorithm failed because it was being run in automatic mode, which does not account for grating dispersion and cross dispersion images. Therefore, the LETG and HETG MEG SF tests did not have a second iteration. It was decided on the floor that the second iteration would be ok for the HEG (there was only one dominant image) and there would be no focus value update. Therefore, I included the data from this second iteration in my analysis. I examined the images and obtained centroids of them by two different methods (usually). The first method used the ASCDS routine srccentroid, which I used in phase 1R and fits 2D elliptical Gaussian profiles to the images to get centroids, ellipticities, and position angles. I routinely ignored the ellipticities and position angles after the first few measurements since the ellipticities were always high (which is reasonable for single quadrant images) and the position angles were not terribly interesting. The second method of determining the positions of the images used an ASCDS routine called evtimg, which simply determines the centroid of the photons in a specified region. The same region was used for both methods. The two methods gave almost identical answers for the position differences, so I will quote values from the srccentroid method only. It is also useful to comment on the images. All images but the HEG top and bottom quadrants were shaped pretty much as expected: the north and south images appeared like bowties oriented horizontally and the top and bottom quadrant images appeared like hourglasses. The HEG top and bottom quadrant images had banana shapes with positive curvature to the left and brighter at the bottom than at the top. It is not obvious why these images appear this way. The data for these images were reduced in the same manner as the rest, however. The analysis region was set to a circle of 27 pix (175 micron) radius that enclosed the bulk of the events (probably better than 99% based on the lowest level detected). Almost all of the regions were centered at pixel 13314,20091 so the same region was used for almost all analyses. The following table summarizes the results after taking differences. Grating RunIDs DelX-NS DelY-TB DelX-NS DelY-TB AveDel Defocus (in pixels) (in microns) (in microns) LETG 106296-300 -3.66 3.32 23.5 21.3 22.4 190.7 MEG 107301-4 -4.09 2.13 26.7 13.9 20.3 172.6 HEG 107305,7,8,9 -2.71 3.37 17.4 21.7 20.0 170.0 HEG 107310-3 -0.12 0.81 0.8 5.2 3.0 25.4 I used 6.43 microns per HSI pixel and a focal ratio of 8.5 for the telescope in the conversions. The images had at least 2000 counts each and were generally 10 pixels across, so I estimate that the uncertainties in the position differences is about 0.2 pix, converting to about 10 micron in the defocus direction. Systematic uncertainties probably dominate, however, because of the odd and asymmetric image shapes, so using 20-25 micron uncertainty may be appropriate. The shutter focus tests indicate that the HSI has been placed about 180 micron forward of focus. Note that the sign of the differences agrees between the NS and TB differences, both indicating forward of focus. After the first iteration of the HEG shutter focus, the amount of the defocus is significantly reduced, as desired. For a fair comparison, I should measure the HRMA SF data as well, which has been delayed for lack of data. MEG +3 order shutter focus tests I have examined a subset of the MEG focus check data, notably the +3 images that showed obvious additional features. Data for the N, B, and S quadrants made it through the pipeline, so I measured the power in the images and the separations of the additional spots. The table summarizes the results. Xm,Ym is the position of the main image. The K-alpha "satellite" line is well separated from the main line by about 55 pixels to lower dispersion. The flux in this line was about 11% of the main line in the south image. RunID Shutter Main Spot1 Spot2 counts counts X-Xm Y-Ym count X-Xm Y-Ym ____ _______ ______ ____________________ ____________________ 10764 North 8987 508 1.5 -20.5 207 -6.5 59.5 10765 Bottom 12970 389 -1.0 20.0 N/A 107466 South 21500 850 +5 -15 N/A Three of the spots are about 18 pixels from the center of the main image. This converts to 116 microns. If these offsets are due to the misalignments of individual grating facets, then the rotation angle would be about 7 arcmin. Similarly, for Spot2 of the north shutter, the distance is 390 microns and the implied rotation angle is 24 arcmin. These angles are much greater than the measured rotations of the actual MEGa, either before or after a shake test. The rms rotation angle is about 0.4 arcmin. The spots are rather "contained", in that it the spots to not appear to have significant structure, which might occur if there were several gratings involved per spot or if the gratings had significant rotation *across* the facet (which, of course, is physically very difficult). If all the gratings had the same efficiency then we'd expect that any one grating would contribute 2% to the total, since there are 192 MEG gratings divided into 4 quadrants for 48 per quadrant. The observed ratios are 5.7%, 3.0%, and 4% for the first spots in the N, B, and S quadrants, respectively and 2.3% for the second spot in the N image. Although there may well be significant facet-facet variation of the efficiency, it would appear a little odd at the outset that all of the images would be brighter than average. This might be understood by recognizing that there may be a very broad distribution of efficiencies at 3rd order (not measured previously and not constrained much by any lab data) and that many gratings might have efficiencies well below 1%. If so, then some of these facets may well be misaligned as well but may be undetectable. The appearances of the main image are somewhat puzzling. Both the north and south quadrant images appear more like arrowheads than bowties, both pointing to -X (on the detector, or toward South). The bottom shutter doesn't seem similar to the other two and also doesn't have the expected hourglass shape and appears more like a half-moon (flat side to north). Raytraces using rays from the best HRMA model that includes scattering and 1g distortions may be needed to interpret these images. The HSI pore efficiency variation with incident angle may be a factor as well. Finally, the image of the N quadrant of the MEG +1 image was examined (runid 107447). With Peter Predehl, a spot was found about 20 pixels above the main Al-Kalpha image. This is almost exactly 1/3 of the distance to spot2 of the north image at order +3 and it is on the correct side of the dispersion direction, so it would appear to confirm the hypothesis that a grating is rotated about the +x axis relative to the bulk of the facets in the N quadrant. HEG Images at 2nd Order These images were not reduced in any detail but they do not show spots similar to those found in the MEG. The main images are shaped somewhat like bananas or half moons, very similar to the top and bottom quadrants of the 0th order images. Oddly, there appears to be a "dead stripe" bifurcating the dispersed spectrum lengthwise in the south and north images. LETG Images from Shutter focus These focus check images were taken at both +3 and -3, where the expected deviation from the Rowland circle is about 0.1 mm and the dispersion distance is about 22 mm. Peter Predehl has taken a preliminary examination of these images and finds no significant anomalies. His preliminary estimates indicate the images are nearly in focus. Precise measurement of the K-alpha line is made difficult however, due to the presence of the K-alpha "satellite" line. This line is 55 pixels less dispersed than the MEG +3 order, so it is only 22 pixels from the LETG +3 line. Higher dispersion may be required or the entire K-alpha and satellite line might be included in the analysis. WORK IN PROGRESS We have yet to measure the positions of the images in the focus check tests at high orders in order to ensure that the detector is placed on the Rowland circle for later measurements. Other images haven't been examined or measured: a) the top quadrant of the MEG at +3, b) the -1 order of the MEG (all quadrants), c) the HSI shutter focus data for the non-grating case, and d) the HEG image taken at +2 and 65 mm out of focus (at "ring focus"). Image a may show us further MEG anomalies, the N images in set b would be expected to show a spot corresponding to the -1 version of spot 2 (N quadrant), the HSI shutter focus could confirm the 180 micron defocus, and the ring focus images may indicate if there are problems with any particular facet in the HEG set. We are planning measurements that may diagnose these anomalies, such as a ring focus with the MEG only at +3 in Al-K and a shutter focus test for both MEG shells independently. The latter should resolve each MEG facet so that misaligned ones would deviate from the expected ring and a shell by shell shutter focus would isolate the shell of a misaligned facet.