Previous
abstract 
Next
abstract
We present the results of algorithm development for on-board CCD bias (or meanframe) determination for the AXAF CCD Imaging Spectrometer (ACIS). The ACIS instrument will have the capability of storing a bias value for each pixel in a frame (e.g. 1024 x 1024). This will allow us to make pixel by pixel corrections on-board. This capability will reduce the effects of hot pixels and prevent the erroneous shifting of mean values that has been encountered with the ASCA SIS instruments.
The bias frame is generated from data frames collected prior to the observation. The algorithm must be optimized to create an accurate bias frame with minimum loss of observing time, i.e. in the fewest number of frames. This optimization is governed by the processing capabilities of the satellite which is limited in its ability to store, process and ultimately, telemeter data to the ground. Success was determined by comparing data processed with the `on-board' bias to that of data processed with a typical laboratory meanframe. Two main aspects of the processing were examined for the comparison. The first is event recognition, the ability to detect and correctly categorize events. The second is spectral energy resolution, the measurement of the full width half maximum (FWHM) of spectral lines (typically $^{55}$Fe).
We have chosen an algorithm that involves a two-step process. In the first step a candidate bias frame is developed from a combination of the first few initial data frames, rejecting contaminating events above a specified threshold. The second step refines the estimate with a more severe exclusion of contaminating events. Non-event values, those falling within thresholds adjusted in units of sigma around the mean, are weighted and added to the candidate bias for successive frames until a final bias frame has been created.
