Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/~INS/2010CalWorkshop/penton.pdf Дата изменения: Mon Apr 23 22:58:47 2012 Дата индексирования: Sat Mar 1 14:26:11 2014 Кодировка: Поисковые слова: green flash

The 2010 STScI Calibration Workshop Space Telescope Science Institute, 2010 Susana Deustua and Cristina Oliveira, eds.

Cosmic Origins Spectrograph: On-Orbit Performance of Target Acquisitions
Steven V. Penton Center for Astrophysics and Space Astronomy - Astrophysics Research Lab, University of Colorado, Boulder, CO, 80309 Abstract. COS is a slit-less sp ectrograph with a very small ap erture (R=1.25 ). To achieve the desired wavelength accuracies, HST+COS must center the target to within 0.1 of the center of the ap erture for the FUV channel, and 0.04 for NUV. During SMOV and early Cycle 17 we fine-tuned the COS target acquisition (TA) procedures to exceed this accuracy for all three COS TA modes; NUV imaging, NUV sp ectroscopic, and FUV sp ectroscopic. In Cycle 17, we also adjusted the COSto-FGS offsets in the SIAF file. This allows us to recommend skipping the time consuming ACQ/SEARCH in cases where the target coordinates are well known. Here we will compare the on-orbit p erformance of all COS TA modes in terms of centering accuracy, efficiency, and required signal-to-noise (S/N). 1. Introduction

There are four COS Target Acquisition (TA) modes: · ACQ/SEARCH: p erforms a search in a spiral pattern by executing individual exp osures at each p oint in a square grid. This mode can use either disp ersed-light or imaging exp osures. · ACQ/IMAGE: obtains an NUV image of the target field, moves the telescop e to center the ob ject, and obtains a second NUV image as a confirmation of the centering. · ACQ/PEAKXD:determines the cross-disp ersion (XD) centroid of a sp ectrum in the direction p erp endicular to disp ersion and moves the telescop e to center the XD direction. · ACQ/PEAKD: centers the target sp ectrum in the along-disp ersion (AD) direction by executing individual exp osures at each p oint in a linear pattern along the AD axis. Coordinate accuracy and target brightness will determine your choice of TA strategy and optional parameters. ACQ/IMAGE is often the fastest and most accurate centering mode, but, dep ending on target brightness, can take longer than the sp ectroscopic TA sequence of ACQ/PEAKXD+PEAKD. TAs on targets with p ositional uncertainties larger than 0.4 need to b egin with an ACQ/SEARCH to ensure the target is in the COS ap erture. This document summarizes two more detailed COS TA documents: COS TIR 201003(v1) "COS Target Acquisition Guidelines, Recommendations, and Interpretation", and COS TIR 2010-14(v1)) "On-Orbit Target Acquisitions with HST+COS". Interested observers should use these documents, and the COS Instrument Handb ook, as their primary sources of information on COS TA. 2. Centering Requirements Wavelengths assigned to COS data are required to have an absolute uncertainty of less than ±15 k/ms in the medium resolution modes, ±150 km/s in G140L mode and ± 175 km/s in G230L mode. In the XD direction, the requirement is to b e centered to within ±0.3 , however, our goal is ±0.1 for FUV flat-fielding purp oses. Since the AD requirement is in 400


On-Orbit Performance of COS TAs

401

units of km/s, it is detector and wavelength dep endent as defined and shown in Equation 1. AD(ar csec) = Dvelocity requirement в c в disp ersion (° p) в platescale (p/ ) A/ (1)

Assuming that the wavelength error budget is split evenly b etween the COS TA and wavelength scale accuracy, the allowable TA centering errors, in arcseconds ( ), are given in Table 1. The strictest requirement for each grating is highlighted in bold. Table 1: COS Centering Requirements
Grating G185M G225M G285M G230L G130M G160M G140L Total Error Budget NUV 0.058 0.076 0.084 0.086 FUV 0.150 0.153 0.247 TA Requirement 0.041 0.054 0.059 0.061 0.106 0.108 0.175

3.

HST Blind Pointing

By reverse engineering the actions p erformed by COS TAs during SMOV, we determined a bias in the blind p ointing of HST+COS (left panel of Figure 1). The mean blind p ointing offset in [AD,XD] coordinates was [-0.22±0.47, -0.17±0.49] , indicating a offset in the FGS-to-COS alignment. The outer black dashed circles in the Figure 1 represents the R=1.25 ap erture, the red dotted circle is the centering limit to transmit maximum flux, the blue dashed circle is the FUV centering goal (0.106 ), and the small black solid circle is the NUV centering goal (0.41 ). In March 1, 2010 an adjustment was made to the Science Instrument Ap erture File (SIAF) to correct for this blind-p ointing bias. The right panel of Figure 1 shows the blind p ointing estimate for observations taken after March 1, 2010. As exp ected, the HST blind p ointing accuracy is not a function of TA mode employed (the blind p ointing accuracies of each mode is given in Figure 1.), dominant FGS (red squares), or whether the target was a guest observer (GO), or calibration (CAL) program. 4. ACQ/SEARCH

As envisioned prior to launch, ACQ/SEARCHs should b e able to center a target to within 0.1-0.2 in b oth AD and XD given that the target was within the b ox on the sky contained by the outer dwell p oints. This exp ectation was based on simulations based up on the predicted, and not the observed PSF. The observed on-orbit PSF is noticeably asymmetric and contains a much larger p ercentage of the light in the extended wings (see Figure 2). The PSF asymmetries and extended wings, along with the extended transmission function of the ap erture tend to feed incorrect information into the ACQ/SEARCH centering algorithm. Based up on the centerings required by TA procedures following ACQ/SEARCHs, the initial p ointing accuracies of ACQ/SEARCH can b e deduced. As detailed in COS TIR2010-03(v1), imaging ACQ/SEARCHs show average [AD,XD] centering errors of magnitude [0.29,0.24] , while NUV and FUV sp ectroscopic ACQ/SEARCHs show average centering errors of [0.10,0.07] and [0.19,0.15] , resp ectively. On-orbit, a single ACQ/SEARCH should, therefore, not b e exp ected to center the target to b etter than 0.3 in either AD and XD, and dep ending on where the target falls in the pattern, it could easily b e 0.4 off-center. This is why we recommend that all COS TAs follow up their initial ACQ/SEARCH with either an ACQ/IMAGE, ACQ/PEAKXD+PEAKD, or a second 2 в 2 ACQ/SEARCH. These p ointing statistics


402

Penton

Figure 1: HST Blind Pointing: GO observations are highlighted by green circles, calibration exp osures by red circles. If the dominant FGS was #3 a red box is added. The left panel is b efore the March 1, 2010 SIAF adjustment, the right panel is after the COS-to-FGS correction. include only those ACQ/SEARCHs with S/N > 25 in the brightest dwell p oint. This is the minimum recommended S/N for ACQ/SEARCH, and the suggested S/N is 40. Higher S/Ns do not improve the final target centering as ACQ/SEARCH is always followed by additional TA procedures. Because of the improvement in the HST 'blind p ointing' due to the FGS-to-COS adjustment on March 1, 2010, targets with very well known coordinates are no longer required to precede ACQ/IMAGEs or sp ectroscopic ACQ/PEAKXD+ACQ/PEAKD sequences with ACQ/SEARCHs. Table 2 shows our up dated ACQ/SEARCH usage guidelines. SCAN-SIZE equal to zero implies that the observer is bypassing ACQ/SEARCH altogether and proceeding directly to an ACQ/IMAGE or ACQ/PEAKXD+ ACQ/PEAKD. SCAN-SIZE 0 2 3 4 5 Target Co 0.4 0.4 0.7 < 1.0 < 1.3 < ordinate Uncertainty ( ) 0.7 1.0 1.3 1.6 Table 2:
a b

STEP-SIZE 1.767 1.767 1.767 1.767
a b b b

CENTER method FW FWF FWF FWF

Default STEP-SIZE value and the largest value which samples the search area without gaps. If the target coordinate uncertainty is on the low edge of the given range, you can reduce the STEP-SIZE slightly (e.g., 1.5 ) to improve centering accuracy by sacrificing the total area of the sky covered by the search pattern.


On-Orbit Performance of COS TAs

403

Figure 2: The left estimate is on the r has a central p eak symmetric and has 5. ACQ/IMAGE

figure shows the on-orbit PSF at the COS ap er ight . The red circles are the size of the ap erture. and a symmetric outer ring of radius 0.75 . The considerable p ower (> 5%) in extended wings and

ture, the pre-launch The pre-launch PSF on-orbit PSF in not rings.

First, a wavelength calibration lamp (WCA) exp osure determines the location of the ap ertures on the NUV detector, then an image of the sky is taken. After measuring the p ositions of the target and the WCA, the slew required to center the target is determined and p erformed. Dep ending on target flux, this can b e p erformed with (PSA or BOA) + (MIRRORA or MIRRORB). PSA/BOA images land on ab out the same location on the detector, but MIRRORA and MIRRORB (a reflection off the order sorter in front of MIRRORA) images land at different locations. Figure 3 b elow shows the on-orbit WCA, initial target locations, and the TA subarrays used in determining sp ot locations. The dashed b oxes are prop osed subarrays for an up coming adjustment. This adjustment is required due to the rising NUV background rate. In this figure, only ACQ/IMAGEs after the FGS-to-COS adjustment of March 1, 2010 are shown. Comparing the final target locations in the confirmation images to those commanded, we can measure the on-orbit centering accuracy of ACQ/IMAGE. The final [AD,XD] centering accuracy of all S/N > 25 ACQ/IMAGE to date is an impressive [-0.007 ± 0.016 , -0.009 ± 0.012 ]. This is the minimum recommended S/N for ACQ/IMAGE, and the suggested S/N is 40 for PSA TAs and S/N of 60 for BOA TAs. Higher S/Ns will improve the final target centerings. 6. ACQ/PEAKXD

ACQ/PEAKXD sequences flash the wavelength calibration lamp (WCA), followed by a brief target sp ectrum. The WCA p osition determines the desired XD target location. The slew required to put the sp ectrum at this XD location is calculated and p erformed. For NUV TAs, one sp ecifies which strip e to use (STRIPE=A, B, or C). For FUV TAs, one must sp ecify which segment(s) to use (SEG=A, B, or BOTH). To limit detector and Geocoronal backgrounds from contributing to the XD location determinations, subarrays are used; one for the NUV and one or two for each of the FUV segments (as shown in Figure 4). FUV ACQ/PEAKXDs are complicated by the: 1. distortions present in the raw coordinate frame (no thermal or geometric correction) 2. wavelength dep endent XD profiles (different targets are centered differently)


404

Penton

Figure 3: Detector coordinate locations of NUV Imaging TA subarrays. Solid lines are current subarrays, dashed b oxes are prop osed Cycle 19 replacements. The two distributions on the left are the WCA locations for MIRRORA and MIRRORB images. The two distributions on the right are the initial target locations for PSA/BOA + MIRRORA and MIRRORB images 3. mapping of segment-B sp ectra onto the segment-A coordinate system (each segment has different raw digital element XD sizes and zero p oint offsets). These complications can cause ob jects with different sp ectral energy distributions (SEDs) to b e centered differently. This offset can b e as much as 0.3 , and is the most problematic for G160M TAs. NUV ACQ/PEAKXDs are complicated by: 1. an additional error is present if STRIPE is not equal to B (±0.06 ) 2. the strip es are tilted on the detector (different targets are centered differently (±0.05 ) 3. each grating has a unique plate scale, but the average is currently used (±0.05 ). These NUV effects are smaller than the those seen in FUV TAs. The minimum recommended S/N for ACQ/PEAKXD is 40 for b oth BOA and PSA TAs. Higher S/Ns will improve the final target XD centerings. 7. ACQ/PEAKD

ACQ/PEAKD is a 1D along-disp ersion (AD) version of ACQ/SEARCH. The transmission of the COS ap ertures is essentially flat within the central 0.5 then tails off in a non-linear, but approximately symmetric profile. The upp er left panel of the left figure of Figure 5 shows the normalized ACQ/PEAKD results. The upp er right panel of this figure shows the normalized counts after alignment based up on the chosen centering method. The lower left panel highlights the NUV results, while the lower right panel of the left figure gives the FUV results. The centering results for all appropriate observations were merged to a comp osite profile representing the average on-orbit AD profile. The average correcting maneuver was 0.018± 0.127 . This profile was used to create simulations for all ranges of the parameters SCAN-SIZE and STEP-SIZE, with a detector background appropriate for a 15 second observation. As shown in the right


On-Orbit Performance of COS TAs

405

!

"

Figure 4: FUV TA subarrays sup erimp osed on FUV raw detector coordinate images. Segment-A is on top, segment-B in on the b ottom. Note the significant XD geometric distortion "kink" near X/10=400­500. figure of Figure 5, there are several ACQ/PEAKD centering options that are exceeding the FUV (0.106 ) and NUV (0.041 ) centering requirements. 8. Detector Backgrounds/SAA

The NUV detector background has b een rising since launch, and is now 300 counts/sec for the entire detector. NUV ACQ/SEARCH and WCA ACQ/IMAGE subarrays are b eing optimized to minimize impact on COS TA. The FUV detector background has been steady since launch at 2-3 counts/sec/segment. No FUV subarrays currently need to b e optimized for COS TA due to background effects. Both detectors show large background rate excursions near the SAA. The COS SAA defining contours were modified in May 2010 to prevent COS TAs from occurring near the SAA. 9. COS Fo cus

By monitoring the FWHM of the confirmation ACQ/IMAGEs we are tracking the imaging performance of COS, and hence, the stability of the HST+COS focus. The FWHM of the confirmation images from all COS ACQ/IMAGEs are shown in Figure 6. Non-p oint sources show larger FWHMs, so the focus is monitored by the narrowest measurements. No significant change in imaging p erformance has b een detected for either the PSA or BOA in either the along-disp ersion (AD) or cross-disp ersion (XD) directions 10. Conclusions and Recommendations

By evaluating the on-orbit actions p erformed by the COS TA flight software (FSW), we have been able to demonstrate the excellent performance of all COS TA modes during SMOV and Cycle 17. In this article and in COS ISR 2010-14(v1), we have up dated Cycle 18 COS TA parameter choices and strategy recommendations. During SMOV and Cycle 17 many COS FSW TA parameter values were adjusted to our b est determinations of the values for early op erations. As outlined in COS TIR 2010-03(v1), there are still some outstanding TA issues and further TA optimizations are planned for Cycle 18 and b eyond.


406

Penton

Figure 5: ACQ/PEAKD observations and simulation results. The upp er left panel of the left figure shows the normalized ACQ/PEAKD results. The upp er right panel of this figure shows the normalized counts after alignment based up on the chosen centering method. The lower left panel highlights the NUV results, while the lower right panel of the left figure gives the FUV results. The centering results for all appropriate observations were merged to a comp osite profile representing the average on-orbit AD profile. This profile was used to create simulations for all ranges of the parameters SCAN-SIZE and STEP-SIZE, with a detector background appropriate for a 15 second observation. The right figure shows that there are several ACQ/PEAKD centering options that are exceeding the FUV (0.106 ) and NUV (0.041 ) centering requirements. ACQ/PEAKDs with CENTER=RTB (return-to-brightest) never achieve these requirements.

Figure 6: COS focus monitoring. Neither the BOA ACQ/IMAGE images (top, circles) or the PSA images (b ottom, triangles) show any change in AD (blue) or XD (red) FWHM on-orbit.


On-Orbit Performance of COS TAs References

407

Keyes, C. & Penton, S. 2010, COS ISR 2010-14(v1), "COS Target Acquisition Guidelines, Recommendations, and Interpretation" Penton, S. & Keyes, C. 2010, COS TIR 2010-03(v1), "On-Orbit Target Acquisitions with HST+COS"