Chandra Aspect Operations

Chandra Aimpoint Trending (static version)

The observation aimpoint is defined as the position on the focal plane where an on-axis target is located, assuming that the SIM offset is zero and both Y Offset and Z Offset (aimpoint offset) are zero. Over the course of the mission the mean aimpoint has drifted over 30 arcsec, due primarily due to the steadily increasing temperature of the Aspect Camera Assembly (ACA). See the Details and References sections for further information.

In addition to a mean drift there has been a steady increase in the scatter on time scales of days. This term is directly related to the absolute pointing accuracy which describes the positional accuracy with which a target can be placed on the detector. Starting around 2015 the amplitude of this scatter has been large enough to potentially impact "pointing-sensitive" observations in which an absolute pointing error might affect science or instrument safety. This includes ACIS windowed or subarray observations, or grating observations of unusually bright sources that must be kept just off the detector.

The mean aimpoint drift adds to pointing uncertainty by introducing a discrepancy between the predicted median aimpoint (used in the planning process) and the actual median aimpoint around the time of observation. Currently the predicted aimpoint for each detector is updated once per year, typically in late November prior to issuing an update to the Proposers' Observatory Guide for the subsequent Call for Proposals. Once the yearly update is made aimpoint is assumed to be static (for planning purposes) until the next update. As of 2015 this approximation is not adequate for pointing-sensitive observations.

Webpage contents and usage

This webpage serves as a reference for both the long-term and current trend in the Chandra aimpoint. It is updated daily and gives the effective bounding box of observed aimpoint positions in chip coordinates over the last 6 months. At this time only ACIS-S and ACIS-I are included because there have been no pointing-sensitive observations on HRC to date.

Many users with pointing-sensitive observations will only need to reference the 6-month aimpoint bounds and median and compare with the latest predicted aimpoint values. This will provide guidance for adjusting the target offset with assistance from your USINT contact.

Going deeper

You are viewing the static version of the aimpoint trending page which includes static PNG figures. If you would like to dig a little deeper and understand how this issue has evolved then return to the dynamic version.

Observed aimpoint differences trend

The following plot shows the difference in CHIPX and CHIPY between the planned observation aimpoint and the actual aimpoint. The planned aimpoint is computed using the planned aimpoint chip coordinates (CHIPX/Y) and observer target offsets and the SIM-Z position. The actual aimpoint is computed using dmcoords and keyword values from the CXC archive L2 X-ray event file. The plot shows up to 6 months of data starting from when dynamic aimpoints were initially put into use (AUG2916 schedule). The data values are stored in the observed aimpoints table (HTML or ASCII).

Intra-observation aimpoint drift

During an observation the aimpoint can drift, and this is illustrated in the plot below. However, from the perspective of planning observations this need not be considered because it is already included in the Aimpoint Trending plots. This is because those plots sample from 1 ksec intervals within every science observation (instead of per-observation means), thus picking up the extremes. The plots below show a representative sampling of aimpoint positions during Chandra science observations. This includes points corresponding to the minimum, 10th percentile, median, 90th percentile, and maximum during one-month bins. This sampling is complete with regard to outliers and does not distinguish between individual observations. See the Details section for a full description of this process and links to the actual code.

In each plot there is a box which highlights the maximum range of aimpoint CHIPX and CHIPY within the last 6 months. In addition there is a red star which shows the current aimpoint used for planning.

The numerical table values shown in this page are available in machine-readable JSON format as info.json.

ACIS-S

A key point to highlight for ACIS-S is that the planning aimpoint is at the extreme corner of the observed aimpoint extent. This means that the aimpoints for some observations in the last 6 months have been offset by nearly 40 pixels in CHIPX and over 20 pixels in CHIPY. Another important point is that there is a strong correlation between CHIPX and CHIPY which is most apparent when viewing the data dynamically using the linked brush, but is also visible noting that the darkest red points are all from the last year of observations.

The coordinates of the 6-month bounding box (shaded box in the plot below) and the 2015.0 planning aimpoint from the POG (red star) are:

Min Midpoint Max POG
CHIPX 204.6 242.1 279.6 200.7
CHIPY 464.8 483.1 501.4 476.9

To account for the difference between the current POG value and the current aimpoint, ADD the following values to the existing observation target offsets DY and DZ:

DY +20.3 arcsec+0.339 arcmin
DZ +3.0 arcsec+0.051 arcmin

ACIS-I

Although not as extreme as ACIS-S, the planning aimpoint for ACIS-I is offset by about 20 pixels from the center of the 6-month box.

The coordinates of the 6-month bounding box (shaded box in the plot below) and the 2015.0 planning aimpoint from the POG (red star) are:

Min Midpoint Max POG
CHIPX 916.7 935.0 953.3 930.2
CHIPY 935.7 973.3 1010.8 1009.6

To account for the difference between the current POG value and the current aimpoint, ADD the following values to the existing observation target offsets DY and DZ:

DY +17.9 arcsec+0.298 arcmin
DZ +2.4 arcsec+0.039 arcmin

Details

Background

The relationship at the foundation of this work is the one-to-one correspondence between the aspect solution SIM offset values (DY, DZ) and the aimpoint offset. Those values are derived primarily by tracking positions of the fiducial lights which are located in the instrument focal plane and are imaged in the ACA by means of the Fiducial Transfer System optics. The one-to-one correspondence relies on our understanding that the alignment of the HRMA optical axis relative to the science instruments is relatively stable (HRMA Optical Axis and Telescope Aimpoint). This implies that any apparent drift in the fid positions is due to change in the ACA to HRMA alignment, and not distortion of the Optical Bench Assembly. Prior to launch it had been surmised that the latter term would dominate.

The ACA is used by the on-board Pointing Control and Attitude Determination system to maintain pointing. If the ACA alignment is shifting with respect to the HRMA - SIM system, then an aimpoint shift is observed. Effectively the ACA points precisely at the target but the rest of the spacecraft hangs off the ACA and wanders somewhat in alignment.

The trend of ACA housing temperature is shown below, and one easily sees the qualitative similarity with the aimpoint trending data.

Processing

Aspect level-1 processing of each science observations produces an aspect solution file that is sampled each 0.25625 seconds and contains the (DY, DZ) values corresponding to the ACA alignment shift. The minimum time scale for significant shifts is about 10 ksec, so our ingest code takes a sample every 1.0 ksec, with a minimum of 2 samples for each observation. This was done for all observations since 2000:001 and the data stored in a single HDF5 file. A daily cron job adds new aspect solution data as they become available.

The raw data are available in FITS format in aimpoint_asol_values.fits.

Converting from (DY, DZ) to (CHIPX, CHIPY) is a simple linear transformation if only ~arcsec accuracy is required. In this case that is sufficient. Deriving and validating this tranformation is not entirely trivial, and is documented in this IPython notebook on Absolute Pointing Uncertainty. (This notebook is currently rough and poorly documented, but I plan to address this.) The resultant transformations from aspect solution DY and DZ to CHIPX and CHIPY are shown below. Note that in the IPython notebook the DY and DZ values refer to the dmcoords inputs which have the sign reversed from the ASOL values.

ACIS-S ACIS-I
CHIPX = 252.25 - 41.69 * DY
CHIPY = 519.95 - 41.67 * DZ
CHIPX = 971.91 - 41.74 * DZ
CHIPY = 963.07 + 41.74 * DY

The step of generating presentation plots to visualize and understand the data takes some care. In order to see the expanding extent with TIME while also getting a sense of the CHIPX - CHIPY correlation, we need to visualize all three axes. A 3-d plot might work, but static 2-d representations of 3-d data can be unsatisfying. Instead the strategy is to use linking and brushing and a Javascript-powered live plot in the web page. In order to do this and have manageable page load times, we need to sample the data while capturing behavior of the extrema which matter here.

The approach taken is to bin the data in one month intervals. Within each bin find the row that contains the minimum, 10th percentile, 50th percentile (median), 90th percentile, and maximum of CHIPX. For each of those rows sample CHIPX and CHIPY. Now repeat that process for the rows containing percentile values of CHIPY. This gives 10 CHIPX, CHIPY pairs for each month. This fills out the sampling reasonably well.

Analysis code

The scripts used in this analysis are in available on GitHub in the aimpoint_mon repository.

References


Last modified:03/28/24



The Chandra X-Ray Center (CXC) is operated for NASA by the Smithsonian Astrophysical Observatory.
60 Garden Street, Cambridge, MA 02138 USA.    Email: aspect_help@head.cfa.harvard.edu
Smithsonian Institution, Copyright © 1998-2004. All rights reserved.