Measuring ACA CTI by dithering a star across quadrant boundaries
Introduction
CTI (charge transfer inefficiency) will have the effect of leaving trailing charge behind as a star or fid image is clocked out during readout. This reduces the observed image brightness as well as shifting the image centroid away from the readout clocking direction. In a series of engineering tests we are measuring the size of the CTI-induced centroid shift by dithering a star across a quadrant boundary, effectively reversing the sign of the centroid shift over a dither period. This applies for both parallel and serial clocking directions.Data
The following table summarizes the available CTI dither data. The listed delta Y angle and Z angle values are peak-to-peak, while the dither amplitude is the sinusoidal amplitude. Clicking on the obsid will show the centroid residuals (with respect to the mean ground aspect solution) for the stars.| Date | Obsid | Star mag | Median row (pixels) | Median col (pixels) | dYag (arcsec) | dZag (arcsec) | dMag (mag) | Dither (arcsec) | Comment |
|---|---|---|---|---|---|---|---|---|---|
| 2004-02-21 | 60364 | 10.2 | -1.5 | -211 | 0.3 | < 0.1 | 0.08 | 8 | Dither ampl insufficient |
| 2004-03-21 | 60328 | 9.1 | -1.0 | 367 | 0.2 | < 0.1 | < 0.02 | 20 |
Comments
- The 60364 data show a large periodic magnitude variation, which could be due to a gain differential between the two quadrants. However, the value of 0.08 mags corresponds to about an 8% error in gain, which would be surprising. The next step is to look for such an effect in other data, for instance dark cal data.
- The period of centroid variations in both datasets (~1000 seconds) matches the dither period in Y. This confirms that the effect is related to the quadrant boundary crossing and that it is not due to the aforementioned gain variation. This latter effect would result in a pattern that looks more like abs(sin(t)), which in the presense of noise would appear to have twice the frequency of the dither pattern.
- The size of centroid variations for obsid 60328 (where the image dithered fully across the QB) is less than in obsid 60364 (0.2 compared to 0.3 arcsec). This is consistent with the expectation for a CTI effect in which defects can trap a fixed amound of charge, so that a fainter star has a larger relative charge loss. More data covering a range of magnitudes are needed.
- In obsid 60328, the CCD row of the lowermost image row ranged from -9 to -1. This is adequate for the test, but the range from -8 to 0 would be ideal. The goal should be to place the image center at a row/col value of -0.5.
Analysis directory
The files for this analysis are located in/proj/sot/ska/analysis/cti_dither
Aspect Information main page
Comments or questions: Aspect Help
Last modified:12/27/13