Location accuracy limitations for CCD cameras

The accurate measurement of the position of celestial objects is a fundamental step for several astrophysical investigations. For ground based instruments, the atmosphere is considered the basic limiting factor; in space, the knowledge of the instrumental parameters and/or of their stability define the performance limits, but CCD cameras operated in time delay integration may take advantage of their operating mode to reduce significantly the calibration problem. We implemented a low-cost laboratory experiment aimed at assessing the precision achievable in the location determination with a CCD camera, by evaluating the measurement repeatability throughout a set of images of a simulated stellar field. Our experiment provides an initial location dispersion of the order of 1/100 of the CCD pixel, with clear evidence of dominant common mode effects. After removing such terms with straightforward numerical procedures, we achieve a final location precision of 1/700 pixel on individual images, or 1/1300 pixel on co-added images. The scaling of precision with target magnitude is in quite good agreement with theoretical expectations. The initial common mode systematics appear to be induced by the thermal control of the CCD camera head, which degrades the structural stability. In actual implementations, such problems can be greatly reduced by proper design. Finally, our results show that residual effects, which could hamper the final astrometric accuracy, can be calibrated out with simple procedures.