chap5.html

INDEX

5.1 Introduction

The system of sensors and control hardware that is used to point the observatory, maintain the stability, and provide data for determining where the observatory has been pointing is called the Pointing Control and Aspect Determination (PCAD) system. As Chandra detectors are essentially single-photon counters, an accurate post-facto history of the spacecraft pointing direction is sufficient to reconstruct an X-ray image.

This chapter briefly discusses the hardware that comprises the PCAD system, how it is used, and its flight performance. Further information can be found on the Aspect Information web page within the main CXC Science web site (https://cxc.harvard.edu/cal/ASPECT/).

5.2 Physical Configuration

The main components of the PCAD system are:

Aspect Camera Assembly (ACA) - 11.2 cm optical telescope, stray light shade, two CCD detectors (primary and redundant), and two sets of electronics
: Inertial Reference Units (IRU) - Two IRUs, each containing two 2-axis gyroscopes
: Fiducial Light Assemblies (FLA) - LEDs mounted near each science instrument (SI) detector which are imaged in the ACA via the FTS
: Fiducial Transfer System (FTS) - The FTS directs light from the fiducial lights to the ACA via the retroreflector collimator (RRC)-mounted at the HRMA center-and a periscope
: Coarse Sun Sensor (CSS) - Sun position sensor, all-sky coverage
: Fine Sun Sensor (FSS) - Sun position sensor, 50 deg field of view (FOV) and 0.02 deg accuracy
: Earth Sensor Assembly (ESA) - Conical scanning sensor, used during the orbital insertion phase of the mission
: Reaction Wheel Assembly (RWA) - 6 momentum wheels that change spacecraft attitude
: Momentum Unloading Propulsion System (MUPS) - Liquid fuel thrusters that allow RWA momentum unloading
: Reaction Control System (RCS) - Thrusters that change spacecraft attitude

Since data from the CSS, FSS, and ESA are not normally used in the processing of science observations, these are not discussed. However, in the unlikely event of a complete failure of the ACA, use of the CSS and FSS data would be attempted.

5.2.1 ACA

The aspect camera assembly (Figure 5.1) includes a sunshade ( ∼ 2.5 m long, ∼ 40 cm in diameter), an 11.2 cm, F/9 Ritchey-Chretien optical telescope, and light-sensitive CCD detectors. This assembly and its related components are mounted on the side of the HRMA. The camera's field of view is 1.4^° ×1.4^°, and the sunshade is designed to protect the instrument from the light from the Sun, Earth, and the Moon, with protection angles of 20^°, 6^°, and 6^°, respectively.

Figure 5.1: Aspect camera assembly

The aspect camera focal-plane detector is a 1024×1024 Tektronix CCD chip operating between −4.5 ^°C and −16 ^°C, with 24×24 micron (5 arcsec × 5 arcsec) pixels, covering the spectral band between 4000 and 9000 Å. The optics of the camera are defocused (point source FWHM = 9 arcsec) to spread the star images over several CCD pixels, both to increase accuracy of the centering algorithm and to reduce variation in the point response function over the field of view. There is a spare identical CCD chip, which can be illuminated by inserting a rotatable mirror.

The ACA electronics track a small pixel region (either 4 pixels ×4 pixels, 6 pixels ×6 pixels, or 8 pixels ×8 pixels) around each fiducial light and star image. There are a total of eight such image slots available for tracking. Typically five guide stars and three fiducial lights (Section 5.2.2) are tracked. The average background level is subtracted on-board, and image centroids are calculated by a weighted-mean algorithm. The image centroids and magnitudes are used on-board by the PCAD and are also telemetered to the ground along with the raw pixel data.

The spectral response of the CCD detector (Figure 5.2) is such that faint cool stars (e.g. type N0) with visual magnitudes much fainter than selected guide stars (i.e., 10.5 mag) can produce large numbers of counts. These so-called "spoiler stars" are effectively avoided in the mission planning stage.

Figure 5.2: Spectral response of the ACA CCD. The same signal-to-noise is achieved for a V=11.7 magnitude N0 star as for a V=10 magnitude G0V star. Also shown are the spectra and the standard visual response for the two stars.

5.2.2 Fiducial Lights and Fiducial Transfer System

Surrounding each of the SI detectors is a set of light emitting diodes, or "fiducial lights," which serve to register the SI focal plane laterally with respect to the ACA boresight. Each fiducial light produces a collimated beam at 635 nm that is imaged onto the ACA CCD via the retroreflector collimator, the periscope, and the fiducial transfer mirror (Figure 5.3).

Figure 5.3: Fiducial Transfer System. Fiducial lights are imaged onto the ACA CCD via the retroreflector collimator, the periscope, and the fiducial transfer mirror.

5.2.3 IRU

Two Inertial Reference Units (IRU) are located in the front of the observatory on the side of the HRMA. Each IRU contains two gyros, each of which measures an angular rate about 2 gyro axes. This gives a total of eight gyro channels. Data from four of the eight channels can be read out at one time. The gyros are arranged within the IRUs and the IRUs are oriented such that all 8 axes are in different directions and no three axes lie in the same plane. The gyros' output pulses represent incremental rotation angles. In high-rate mode, each pulse nominally represents 0.75 arcsec, while in low-rate mode (used during all normal spacecraft operations) each pulse represents nominally 0.02 arcsec.

5.2.4 Momentum Control - RWA and MUPS

Control of the spacecraft momentum is required both for maneuvers and to maintain stable attitude during science observations. Momentum control is primarily accomplished using six Teldix RDR-68 reaction wheel units mounted in a pyramidal configuration. During observing-with the spacecraft attitude constant apart from dither-external torques on the spacecraft (including gravity gradient and magnetic torques) will cause a buildup of momentum in the RWA. Momentum is then unloaded by firing the MUPS and simultaneously spinning down the reaction wheels.

5.3 Operating Principles

The Chandra aspect system serves two primary purposes: first, for on-board spacecraft pointing control and aspect determination and second, for post-facto ground aspect determination, used in X-ray image reconstruction and celestial location.

The PCAD system has nine operational modes (six normal and three safe modes) which use different combinations of sensor inputs and control mechanisms to control the spacecraft and ensure its safety. These modes are described in Section 5.7.1. In the normal science pointing mode, the PCAD system uses sensor data from the ACA and IRUs and control torques from the RWAs to keep the target attitude within ∼ 30 arcsec of the telescope boresight. This is done using a Kalman filter which optimally combines ACA star centroids (typically five) and angular displacement data from two 2-axis gyroscopes. On short time scales ( ∼ sec) the spacecraft motion solution is dominated by the gyroscope data, while, on longer timescales, it is the star centroids that determine the solution.

Post-facto aspect determination is done on the ground and uses more sophisticated processing and better calibration data to produce a more accurate aspect solution. The suite of CXC tools to perform this processing is called the aspect pipeline. The key improvements over PCAD aspect come from better image centroiding and using Kalman smoothing (which uses all available data over the observation period, as opposed to historical data). In addition, the aspect pipeline folds in the position of the focal-plane instrument as determined by the fiducial light data.

5.4 Performance

This section provides information about the aspect system performance, with an emphasis on parameters that impact science and/or observation planning. Table 5.1 provides a summary of results for four key aspect requirements (Sections 5.4.1 - 5.4.4) that originate in the Observatory Level-2 Project Requirements Document. In each case, the actual performance exceeds the requirements.

Table 5.1: Aspect System Requirements and Performance

Description	Requirement	Actual
Celestial location	1.0 arcsec (RMS radius)	0.57 arcsec
Image reconstruction	0.5 arcsec (RMS diameter)	0.39 arcsec
Absolute celestial pointing	30.0 arcsec (99.0%, radial)	8 arcsec
PCAD 10 sec pointing stability	0.12 arcsec (95% RMS)	0.12 arcsec (pitch)
		0.09 arcsec (yaw)

5.4.1 Celestial Location Accuracy

Celestial location accuracy measures the absolute accuracy of Chandra X-ray source locations. Based on observations of 315 point sources detected within 3 arcmin of the boresight and having accurately known coordinates, the 90% source location error circle has a radius of less than 0.82 arcsec overall, and less than 0.93 arcsec for each SI (Figure 5.4). Approximately 3% of sources are outside a 1 arcsec radius. The difference in astrometric accuracy for different SIs is a function of two factors: number of available data points for boresight calibration and accuracy of the fiducial light SIM-Z placement. These values apply for sources within 3 arcmin of the aimpoint and with the SIM-Z at the nominal detector value. Observations on ACIS or HRC-S at large off-nominal SIM-Z can suffer additional residual aspect offsets of up to 0.5 arcsec and, for HRC-I, this can be up to 3 arcsec. The plotted level of accuracy applies to observations after 2018-Oct-5, which were processed for this analysis with the most recent calibration data. Achieving this level of accuracy for archive analysis may require reprocessing with CALDB version 4.10.1 or later.

For further technical details about celestial location accuracy and performance, see https://cxc.cfa.harvard.edu/cal/ASPECT/celmon/.

Figure 5.4: Cumulative histogram of celestial accuracy for Chandra X-ray source locations for each SI using observations since 2018-Oct. Radial offset is the distance in arcsec between the optical coordinate, typically from the Tycho-2 catalog, and the Chandra position.

5.4.2 Image Reconstruction

Image reconstruction performance measures the effective blurring of the X-ray PSF due to aspect reconstruction. A direct measure of this parameter can be made by determining the time-dependent jitter in the centroid coordinates of a fixed celestial source. Any error in the aspect solution will be manifested as an apparent wobble in the source location. Unfortunately this method has limitations. ACIS data are count-rate limited and only produce an upper limit: aspect reconstruction effectively convolves the HRMA PSF with a Gaussian having FWHM of less than 0.25 arcsec. HRC observations can produce acceptably high count-rates, but the HRC photon positions (at the chip level) have systematic errors due to uncertainties in the HRC de-gap calibration (See "Position modeling, de-gap corrections, and event screening" in Section 7.16). These errors exactly mimic the expected dither-dependent signature of aspect reconstruction errors, so no such analysis with HRC data has been done. An indirect method of estimating aspect reconstruction blurring is to use the aspect solution to de-dither the ACA star images and measure the residual jitter. Performing this for 350 observations shows that 99% of the time the effective blurring is less than 0.20 arcsec (FWHM).

Another component of the image reconstruction error is error in the registration of the SI and the ACA, specifically registration error due to drift in the apparent fiducial light positions used to establish this registration. Drift is caused by changes in the alignment of the periscope (see Figure 5.3). Registration error from this drift is reduced in ground aspect processing by a periscope drift correction applied in the aspect pipeline, but residual error of up to 0.5 arcsec may persist. Users with long ( 50ks) observations of a bright, on-axis source, and a science goal that would benefit from sub-arcsec image reconstruction should see https://cxc.harvard.edu/ciao/ahelp/correct_periscope_drift.html for a correction tool.

5.4.3 Absolute Celestial Pointing and Aimpoint Stability

Absolute celestial pointing refers to the accuracy with which an X-ray source can be positioned at a specified location on the detector. Consideration of this accuracy is required in observation planning and is especially relevant for planning of observations sensitive to position on the detector, such as ACIS subarray or windowed observations. It is worth noting that the absolute celestial pointing accuracy discussed here does not have an impact on the accuracy of image reconstruction and celestial location in the processed X-ray data.

The absolute celestial pointing accuracy is determined by the precision to which the pointing system works as well as by the aimpoint stability and drift as discussed in the context of observation planning in Section 4.5. That section provides detailed information on the updated approach with default detector aimpoints and Table 4.5 and Figures 4.28-4.33 provide corresponding uncertainty boxes showing the precision to which a source can be placed on a given detector. These figures include detector-specific padding for dither and address concerns about placement in CCD pixel space.

The absolute celestial pointing performance is dependent on the stability of the aspect camera optical axis because the spacecraft pointing and ultimately the aspect reconstruction is referenced to the ACA frame. In this performance context, the absolute celestial pointing accuracy is defined as the 99% radial offset of actual aimpoints from the planned aimpoint location. This quantifies the ability to position a target near the intended aimpoint.

As of 2023-Oct the absolute celestial pointing accuracy is approximately 8 arcsec. This is comprised of two observable components, mean absolute pointing error (±5 arcsec, per-axis) and intra-observation drift (±3 arcsec, per-axis). These error component estimates are based on review of recent and historical data. Plots of the two components in recent data may be seen in Figures 5.5 and 5.6 respectively. Both the mean absolute pointing error and the intra-observation drift are correlated with thermal conditions on the spacecraft as the aspect camera optical axis alignment (relative to the HRMA alignment) is sensitive to the ACA housing temperature.

Figure 5.5: Difference between observed and planned aimpoint in CHIPX and CHIPY directions for recent observations with planned aimpoints within 100 arcsec of the nominal aimpoint.

Figure 5.6: Peak to peak aimpoint drift within individual science observations. This figure shows that total intra-observation drift (DY+DZ) up to 6 arcsec during observations is possible and, in rare cases, drift over 15 arcsec can occur.

For further technical details about aimpoint monitoring and performance see https://cxc.cfa.harvard.edu/mta/ASPECT/aimpoint_mon/.

5.4.4 PCAD 10-Second Pointing Stability

The PCAD 10-second pointing stability performance is measured by calculating the root mean squared (RMS) attitude control error (1-axis) within successive 10 second intervals. The attitude control error is simply the difference between the ideal (commanded) dither pattern and the actual measured attitude. Flight data show that 95% of the RMS error measurements are less than 0.12 arcsec (pitch) and 0.09 arcsec (yaw). Systematic offsets are not included in this term.

5.4.5 Relative Astrometric Accuracy

Relative astrometric accuracy refers to the residual astrometric offsets assuming that the X-ray coordinates have been registered using well-characterized counterparts of several X-ray sources in the field. The most comprehensive dataset for measuring relative astrometry is based on the 900 ksec ACIS-I observation of the Orion Nebula. The members of the COUP (Chandra Orion Ultradeep Project) provided data for over 1300 X-ray sources listing the offset from a 2 Micron All-Sky Survey (2MASS) counterpart and the off-axis angle. (Full details are available in . The 1152 sources with more than 50 counts were used for the analysis.) The left plot of Figure 5.7 shows a scatter plot of offset (arcsec) versus off-axis angle (arcmin). The right side of the figure shows cumulative histograms of the fraction of sources with relative offset below the specified value. This is broken into bins of off-axis angle as listed in the plot. In the "on-axis" (0 - 2 arcmin) bin, 90% of sources have offsets less than 0.22 arcsec. After accounting for the ∼ 0.08 arcsec RMS uncertainty in 2MASS coordinates, this implies the intrinsic 90% limit is 0.15 arcsec. See https://www.ipac.caltech.edu/2mass/releases/allsky/doc/.

Figure 5.7: Left: scatter plot of offset (arcsec) versus off-axis angle (arcmin) for sources in the ultra-deep ACIS-I Orion observation. Right: cumulative histograms of the fraction of sources with relative offset below the specified value.

5.4.6 On-Board Acquisition and Tracking

As described in Section 5.8, in normal operations the ACA is used to acquire and track stars and fiducial lights. Occasional failures in acquisition and difficulties in tracking are expected due to uncertainties in star position and magnitude, the presence of spoiler stars, CCD dark current noise (see Section 5.6.3), and other factors.

Table 5.2 summarizes success statistics for star acquisition and tracking during the last year. A star is "successfully" tracked if it spends less than 5% of the observation in the loss of track state.

In May 2023, the ACA on-board software was updated to improve background subtraction and mitigate the impact of increasing CCD dark current noise. This update has resulted in improved guide star tracking performance. The tracking success metrics in Table 5.2 include data from before and after the update was uplinked and activated. The improved background subtraction also allows stars fainter than 10.3 mag to be used as guide stars, but very few have been used.

Table 5.2: Star Acquisition and Tracking Success

Catalog Star Magnitude	Acquisition Success	Tracking Success
All stars	94%	99%
10.0 - 10.3 mag stars	78%	90%
10.3 - 10.9 mag stars	72%	-

5.5 Heritage

The Chandra aspect camera design is based on the Ball CT-601 star tracker, which was also used for the Rossi X-ray Timing Explorer. The Chandra IRUs are nearly identical to the space-qualified Kearfott IRU (SKIRU) V, some 70 of which have been built by the manufacturer, Kearfott. These IRUs are similar to those used on the Compton Gamma-Ray Observatory (CGRO).

5.6 Calibration

5.6.1 Pre-launch Calibration

IRU component testing at Kearfott provided calibration data necessary for accurate maneuvers and for deriving the aspect solution. The key parameters are the scale factor (arcsec/gyro pulse) and the drift rate stability parameters. The stability parameters specify how quickly the gyro readout random-walks away from the true angular displacement. These terms limit the aspect solution accuracy during gyro hold observations (described in further detail in Section 5.8.2).

ACA component testing at Ball provided calibration data necessary for on-orbit pointing control and for post-facto ground processing. On-orbit, the ACA uses CCD gain factors, the plate scale factor, and temperature dependent field distortion coefficients to provide the control system with star positions and brightnesses. In ground processing, the CXC aspect pipeline makes use of those calibration data as well as CCD read noise, flat-field maps, dark current maps, and the camera PSF to accurately determine star positions.

5.6.2 Orbital Activation and Checkout Calibration

Orbital activation and checkout (OAC) of the PCAD occurred during the first approximately 30 days of the Chandra mission. During the first phase of OAC, before the HRMA sunshade door was opened, it was possible to use the ACA to observe the fiducial lights; this is referred to as period 1. After the sunshade door was opened, in period 2, it was possible to fully check the aspect camera using star light.

Chandra activation produced the following aspect system calibration data:

Bias, alignment, and scale factor of the CSS and FSS (period 1)
Coarse gyro bias (period 1)
ACA CCD dark current map (period 1)
Fiducial light intensity, image, and centroid at nominal voltage (periods 1 and 2)
IRU bias, alignment, and scale factor (period 2)
ACA alignment and field distortion coefficients (period 2)

5.6.3 On-orbit Calibrations

During the Chandra science mission, aspect system components require on-orbit calibration to compensate for alignment or scale factor drifts and to evaluate ACA CCD degradation due to cosmic radiation.

The following ACA calibrations are performed, as-needed, based on the trending analyses of aspect solution data.

IRU Calibration

The IRU calibration coefficients are evaluated quarterly and updated, as needed, approximately once every two years. These are impacted by IRU configuration changes, including the IRU gyro swap on 2020-Jul-31.

Dark Current

Cosmic radiation damage will produce an increase in both the mean CCD dark current and the non-Gaussian tail of "warm" (damaged) pixels in the ACA CCD. This is illustrated in Figure 5.8, which shows the distribution of dark current shortly after launch and in 2023-Aug. The background non-uniformity caused by warm pixels (dark current > 100 e⁻/sec) is the main contributor to star centroiding error, although the effect is substantially reduced by code within the aspect pipeline that detects and removes most warm pixels.

Figure 5.8: Differential histogram of dark current distribution for the ACA CCD on 1999-Aug-11 and 2023-Aug-29.

The fraction of pixels that appear to be warm during an observation is dependent on the underlying damage to the CCD and the CCD temperature during the observation. As the ACA thermoelectric cooler no longer has sufficient power margin to maintain the CCD at −19 ^°C, temperatures fluctuate between −16 ^°C and −4.5 ^°C. The effective warm pixel fraction over this temperature range is expected to be between 20 and 55% during the next year.

Dark current calibration data is gathered from as many full-frame CCD readouts as possible during engineering observations during the weekly schedule. After at least 10 full-frame readouts are collected, a dark current product based on the combined data from those readouts is created. A new calibration product from this process is evaluated and released for operational use approximately every month.

Flickering Pixels

The dark current of radiation damaged pixels is observed to fluctuate by factors of up to 25% on timescales of 1 to 50 ksec. This behavior was studied using a series of ACA monitor windows commanded during perigee passes in 2002. In 2008 and 2009, similar monitor window data were acquired and analyzed, and the flickering pixel behavior was seen to be qualitatively unchanged from that seen in 2002. An important consequence of the flickering pixel phenomenon is that the dark current pixel values obtained during the dark current calibration may not be directly subtracted from observation pixel data in post-facto processing. Instead, users of monitor window data should use a warm pixel detection algorithm such as the one implemented in ground processing Cresitello-Dittmar et al., [2001].

Charge Transfer Inefficiency (CTI)

Radiation damage degrades the efficiency with which charge is transferred in the CCD by introducing dislocations in the semiconductor, trapping electrons and preventing their transfer. The most important consequence is a "streaking" or "trailing" of star images along the readout column(s), which can introduce systematic centroid shifts. These shifts depend primarily on CCD transfer distance to the readout and star magnitude.

The procedure for calibrating the mean CTI is to dither a faint star across the CCD quadrant boundary and observe the discontinuity in centroid (the CCD is divided electrically into four quadrants). In 2004, a total of 20 calibration observations were performed during perigee, each with a guide star dithering over a quadrant boundary. Despite significant concerns prior to launch, as well as notable CTI degradation in the ACIS front-illuminated chips, there is no evidence of increased CTI in the ACA CCD.

Field Distortion

The precise mapping from ACA CCD pixel position to angle relative to the ACA boresight is done with the "ACA field distortion polynomial." This includes plate scale factors up to third order as well as temperature-dependent terms. To verify that no mechanical shift in the ACA had occurred during launch, a field distortion calibration was performed during the orbital activation and checkout phase. The on-orbit calibration revealed no mechanical shift. Such a shift would have caused degraded celestial location accuracy.

The calibration was done by observing a dense field of stars with the spacecraft in normal pointing mode. Two reference stars were observed continuously, while sets of four stars each were observed for 100 sec. The calibration was completed after observing 64 stars over the ACA field of view, taking roughly 60 minutes. There are currently no plans to repeat this on-orbit calibration. Instead, the field distortion coefficients are monitored by long-term trending of observed star positions relative to their expected positions.

Responsivity

Contamination buildup on the CCD surface was predicted in pre-launch estimates to result in a mean throughput loss of 9% after five years on-orbit, although the calculation of this number has significant uncertainties. The buildup of contaminants is tracked by a trending analysis of magnitudes for stars which have been observed repeatedly throughout the mission (e.g. in the AR Lac field). To date, these trending analyses show no indication of contamination build-up. In the unlikely event that future contamination occurs and causes significant operational impact, "baking-out" the CCD on-orbit will be considered. In this procedure, the current to the CCD thermo-electric cooler is reversed so as to heat the device to approximately 30 ^°C for a period of several hours. After bake-out the CCD would be returned to its nominal operating temperature of −19 ^°C.

5.7 Operations

5.7.1 PCAD Modes

The PCAD system has nine operational modes (six normal and three safe) which use different combinations of sensor inputs and control mechanisms to control the spacecraft and ensure its safety. These modes are listed in Table 5.3. Normal science observations are carried out in Normal Pointing Mode (NPM), while slews between targets are done in Normal Maneuver Mode (NMM).

Table 5.3: PCAD modes

Mode	Sensors	Control	Description

Standby	-	-	OBC commands to RWA, RCS, and Solar Array Drive Assembly (SADA) disabled, for initial deployment, subsystem checkout, etc.
Normal Pointing	IRU, ACA	RWA	Point at science target, with optional dither
Normal Maneuver	IRU	RWA	Slew between targets at peak rate of 2^° per minute
Normal Sun	IRU, CSS, FSS	RWA	Acquire the Sun and hold spacecraft −Z axis and solar arrays to the Sun
Safe Sun	IRU, CSS, FSS	RWA	Safe mode: acquire the Sun and hold spacecraft −Z axis and solar arrays to the Sun
Derived Rate Safe Sun	IRU, CSS, FSS	RWA	Similar to Safe Sun Mode, but using only one gyro (two axes) plus Sun sensor data

Transfer orbit only - now disabled

Powered Flight	IRU	RCS	Control Chandra during Liquid Apogee Engine burns
RCS Maneuver	IRU	RCS	Control Chandra using the RCS
RCS Safe Sun	IRU, CSS, FSS	RCS	Same as Safe Sun Mode, but using RCS instead of RWA for control

5.7.2 Operational Constraints

The ACA will meet performance requirements when the ACA line-of-sight is separated from: the Sun by 45 deg or more; the limb of the bright Earth by 10 deg or more; and the dark Earth or the Moon by 6 deg or more. If these restrictions are violated, the star images may be swamped by scattered background light, with the result that added noise on the star position will exceed the 0.360 arcsec requirement (1σ, 1-axis).

5.7.3 Output Data

The important output data from the ACA are the scaled raw pixel intensities in regions (4 pixels ×4 pixels, 6 pixels ×6 pixels, or 8 pixels ×8 pixels) centered on each of the star and fiducial light images. These data are placed in the engineering portion of the telemetry stream, which is normally allocated 8 kbit/sec. During an ACA dark current calibration (Section 5.6.3), Chandra utilizes a 512 kbit/sec telemetry mode in real-time contact to enable read-out of the entire CCD (1024 pixels ×1024 pixels). The key data words in telemetry from the IRU are the 4 accumulated gyro counts (32 bits every 0.256 sec).

5.8 Performing an Observation

5.8.1 Star Acquisition

After maneuvering at a rate of up to 2^°/minute to a new celestial location using gyroscope data and the reaction wheels, Chandra begins the star acquisition sequence, a process that typically takes from 1 to 4 minutes. First, the OBC commands the ACA to search for up to 8 acquisition stars, which are selected to be as isolated from nearby stars as possible. The search region size is based on the expected uncertainty in attitude, which is a function of the angular size of the slew. If two or more acquisition stars are found, an attitude update is performed using the best (brightest) pair of stars. This provides pointing knowledge to 3 arcsec (3σ per axis). Next, the guide star search begins. Depending on the particular star field configuration, the star selection algorithm may choose guide stars which are the same as the acquisition stars. In this case, the guide star acquisition time is somewhat reduced. When at least two guide stars have been acquired and pointing control errors converge, the on-board Kalman filtering is activated and the transition to Normal Point Mode is made. At this point sensing of the fiducial lights begins.

5.8.2 Science Pointing Scenarios

The on-board PCAD system is flexible and allows several different Chandra science pointing scenarios, described in the following sections.

Normal Pointing Mode Dither

The large majority of observations are performed using Normal Point Mode, with dither selected. In this case, the Chandra line-of-sight will be commanded through a Lissajous pattern. Dithering distributes photons over many detector elements (microchannel pores or CCD pixels) and serves several purposes: dithering reduces uncertainty due to pixel to pixel variation in quantum efficiency (QE); dithering allows sub-sampling of the image; and, in the case of the HRC, dithering distributes the total exposure over many microchannel pores-which is useful since the QE of a pore degrades slowly with exposure to photons. The dither pattern parameters are amplitude, phase, and period, each set independently for two axes. Each of the six parameters is separately commandable and differ for the two different instruments (See Chapters 6 and 7). The default values for these parameters are given in Table 5.4. Dither can be disabled for ACIS observations, while the minimum dither rate required to maintain the health of the HRC is 0.02 arcsec sec⁻¹. The maximum dither rate, determined by PCAD stability requirements, is 0.22 arcsec sec⁻¹.

Table 5.4: Default dither parameters

Parameter	HRC	ACIS
Phase (pitch)	0.0 rad	0.0 rad
Phase (yaw)	0.0 rad	0.0 rad
Amplitude (pitch)	20.0 arcsec	16.0 arcsec
Amplitude (yaw)	20.0 arcsec	16.0 arcsec
Period (pitch)	768.6 sec	2000.0 sec
Period (yaw)	1087.0 sec	1414.0 sec

Normal Pointing Mode Steady

This mode is identical to Normal Pointing Mode Dither, but without the dither.

Pointing at Solar System Objects

Observations of moving solar system objects are done using a sequence of pointed observations, with the object moving through the field of view during each dwell period. Except in special circumstances, each pointing is selected so that the object remains within 5 arcmin of the Chandra line-of-sight. Most solar system objects move slowly enough that a single pointed observation will suffice.

Raster Scan

Survey scans of regions larger than the instrument field of view are specified simply with a grid, i.e., a list of target coordinates giving the field centers. The fields can optionally overlap, depending on the science requirements of the survey.

Offset and Gyro Hold

In special circumstances it will be necessary to perform observations without tracking guide stars. It may occur that a field has no suitable acquisition and guide stars, although this situation has not been encountered to date. A more likely situation is that a very bright object, such as Earth or the Moon, saturates the ACA CCD and precludes tracking stars. In this case, Chandra will first be maneuvered to a nearby pointing which has guide stars to establish fine attitude and a gyro bias estimate. A dwell time of approximately 20 minutes is needed to calibrate the bias estimate, which is the dominant term in the drift equation below. Chandra will then be maneuvered to the target. The default automatic transition to NPM will be disabled, and the spacecraft will hold on the target attitude in NMM.

After a maximum of 3.6 ksec, Chandra will be maneuvered back to the nearby field with guide stars to re-establish fine attitude and update the gyro drift rate. While holding on gyros only, the spacecraft attitude will drift due to noise in the gyros, which results in an aspect solution error. The variance of the angle drift for each gyro axis, in time t, is given by

σ² = σ_b² t² + σ_v² t + σ_u² t³ / 3

Ground test data for gyro noise parameters indicate worst case values of σ_u = 1.5×10⁻⁵ arcsec sec^−3/2 and σ_v = 0.026 arcsec sec^−1/2. Analysis of the residual Kalman filter bias estimate gives σ_b = 0.002 arcsec sec⁻¹. This results in 1σ angle drift errors of: 0.3 arcsec for 0.1 ksec; 2.2 arcsec for 1 ksec; 11 arcsec for 5 ksec; and 22 arcsec for 10 ksec.

5.8.3 PCAD Capabilities (Advanced)

Monitor Star Photometry

The ACA has the capability to devote one of the eight image slots to "monitor" a particular sky location. This allows simultaneous optical photometry of a target in the ACA field of view. However, since there are a fixed number of image slots, devoting a slot to photometry instead of tracking a guide star results in a degradation of the image reconstruction and celestial location accuracy (Section 5.4). Using a monitor slot represents a 15 - 25% increase in the aspect image reconstruction RMS diameter, depending on the particular guide star configuration. The photometric accuracy which can be achieved depends primarily on the star magnitude, integration time, CCD dark current, CCD read noise, sky background, and CCD dark current uncertainty.

The ACA capability to collect data for optical photometry does not have utility for optical sources brighter than m_ACA=5.2 mag or sources fainter than m_ACA=10.3 mag. The conversion from V and B magnitude to ACA instrument magnitude, based on flight data, is given approximately by

m_ACA = V + 0.426 − 1.06(B−V) + 0.617(B−V)² − 0.307(B−V)³

Dark current uncertainty ultimately limits the photometric accuracy at the faint end; this uncertainty results from uncalibrated pixel-to-pixel changes in dark current due to radiation damage. This includes both changing background pixels as Chandra dithers as well as intrinsic flickering in the radiation-damaged CCD pixels. This flickering, which occurs on time scales from less than 1 ksec to more than 10 ksec, poses fundamental problems for accurate photometry since the background dark current is a strong random function of time. With straightforward data processing, the noise introduced by the dark current variations (both spatial and temporal) is approximately 1450 e⁻ sec⁻¹. A star with an ACA magnitude of 10.2 mag produces about 5800 e⁻ sec⁻¹, giving a signal-to-noise ratio (S/N) of 4.0. This represents the practical faint limit for ACA monitor star photometry. Somewhat improved S/N could be obtained with a more sophisticated analysis that tracks the time-dependent dark current of each pixel. Users interested in processing ACA monitor window data should see https://cxc.harvard.edu/ciao/threads/monitor_photom/.

5.9 Ground Processing

For each science observation, the aspect system data described in Section 5.7.3 are telemetered to the ground to allow post-facto aspect determination by the CXC aspect pipeline as part of the standard CXC data processing pipeline. The important components of the pipeline are:

Gyro process: Filter gyro data, gap-fill, and calculate raw spacecraft angular rate

ACA process: Filter bad pixels, make CCD-level corrections (e.g. dark current), find spoiler stars, centroid, make camera-level corrections, convert to angle

Kalman filter and smooth: Optimally combine ACA and gyro data to determine ACA celestial location and image motion

Combine ACA and fids: Derive fiducial light solution and combine with ACA solution to generate image motion and celestial location at the focal-plane science instrument

5.9.1 Data Products

The data products which are produced by the aspect solution pipeline are listed in Table 5.5. Key data elements include: IRU accumulated counts; raw pixel data for 8 images; observed magnitudes and pixel positions of the aspect stars and fiducial lights versus time; and aspect solution versus time. The star data are used to determine the right ascension, declination, and roll (and corresponding uncertainties) of the HRMA axis as a function of time. The fiducial light images are used to track any drift of the SIM away from the nominal position. One cause of such drift is thermal warping of the optical bench assembly. The Kalman filtering routines also calculate an optimal estimate of the gyro bias rate as a function of time.

Table 5.5: Aspect pipeline data products

Product	Description
ASPSOL	Final aspect solution with errors
ASPQUAL	Aspect solution quality indicators
AIPROPS	Aspect Intervals
ACACAL	ACA calibration data from the optical database (ODB) and CALDB
GSPROPS	Guide star properties, both from the AXAF Guide and Acquisition Star Catalog and as actually observed with the ACA
FIDPROPS	Fiducial light properties, as commanded and as observed
ACADATA	Aspect camera telemetry (including ACA housekeeping) and images after CCD-level correction
ACACENT	Image centroids and associated fit statistics
GYROCAL	Gyro calibration data from ODB and CALDB
GYRODATA	Gyro raw data and gap-filled, filtered data
KALMAN	Intermediate and final data in Kalman filter and smoother

5.9.2 Star Catalog

The Aspect system uses the Advanced X-Ray Astrophysics Facility (AXAF) Guide and Aspect Star Catalog (AGASC) version 1.7. Further information about the AGASC, as well as access to catalog data, can be found on the CXC AGASC web page https://cxc.harvard.edu/agasc). The AGASC was prepared by the CXC Mission Planning and Operations & Science Support groups, and is a compilation of the Hubble Guide Star Catalog, the Positions and Proper Motion Catalog, and the Tycho Output Catalog.

5.10 References

Cresitello-Dittmar, M., Aldcroft, T. L., & Morris, D. 2001, ADASS X, 238, 439

Getman, K. V., Flaccomio, E., Broos, P. S., et al. 2005, ApJS, 160, 319