ACIS-I Fid #6 Anomaly

Background

Over the mission we have seen several related instances of large centroids deviations in fid ACIS-I-6. These deviations are step functions in time and are correlated directly to the dither period. Below are two log-scale movies of fidlight images from obsid 2835, sampled every 100 readouts (~400 seconds). The left animation is from a well-behaved fid, while the right animation shows ACIS-I-6 having the symptoms of this issue. The flickering pixel in the second row from the bottom is the feature of interest.

    

Below are plots of the rms offsets in ACIS-I-6 for obsid 5149 and obsid 2835. 5149 shows a minimal deviation, with offsets better than the V&V rejection criteria of .2 arc-seconds max in y and no visible change in any other parameters. 2835 is the 'worst case scenario'. Offsets in y propagate into z and even magnitude. Re-processing without fid #6 was required to correct the aspect solution. When a guide star within the green band has been clocked to the serial register, then the fid light has been clocked to the position of the trap.

 

Geometry


These deviations are caused by an electron trap in the CCD that lies in the same readout column as the typical ACIS-I-6 position. When a guide star image is digitized as the fid clocks over the trap, the fid image becomes distorted. See Tom's 2001 analysis report for the derivation of this solution. The diagram below illustrates the basic geometry. When a guide star within the green band has been clocked to the serial register, then the fid light has been clocked to the position of the trap.

The distance D1 between ACIS-I-6 (blue dot) and the trap (red square) is 300+/-1 pixels. If D2, the distance from the guide star to the register is near (due to dither) equal to D1, then image distortions are likely.

In the catalog for obsid 5149 above, the guide star in slot #7 is the culprit. It is located at row 208.4, which puts it 303.1 pixels from the register. In the catalog for obsid 2835 below, slot #4 is the problem. This guide star is located 305 pixels from the register at row 206.5.

Forty seven catalogs over the mission have used guide/fid combinations that meet the geometry requirements above. Using products generated by the 'vv_asp' tool, the rms y-offset of ACIS-I-6 was measured for each of these and are shown in the plot below.

The 'Trap Separation' is the difference (in pixels) between the trap/fid separation and the guide/register separation. So the distance from the register to the guide star is 300 plus the 'Trap Separation'. This plot clearly shows where the effect of the trap is seen. Red lines are drawn to mark the proposed 'exclusion zone' discussed below for guide stars when using ACIS-I-6. Some effect is seen in areas outside of these zones, but the magnitude is below the V&V rejection criteria. The reason for some guide stars within the exclusion zones not causing an image shift requires further work as does understanding of the central 'quiet' zone.

Recommendations


There are two proposed solutions. The first is to implement a check at the SAUSAGE level for guide stars that fall into the exclusion zones. Due to the size of the potential effect, it is probably best to either select these stars in the third (or fourth?) pass or to remove them altogether at the candidate level. The other suggestion is to try to fill the trap with charge by brightening the fids. The brightening factor needed to mitigate the trap effect is unknown, but it is within safe limits to brighten the fids substantially ( See Report).



Aspect Information main page



Comments or questions: Aspect Help

Last modified:12/27/13