Zero Suppression and Data Compression for SDD Output in the ALICE Experiment

We describe a proposal for the zero suppression and data compression for the Silicon Drift Detectors in the ALICE experiment. The proposal seeks to maintain maximum precision within the limits of data transmission bandwidth, to retain two-dimensional cluster reconstructability and to monitor statistically the background. Two thresholds (high and low) are employed to facilitate understanding of the cluster neighbourhoods. This choice also helps to suppress single high background peaks and provides a statistically cleaner sample for background monitoring. Background average and standard deviation are monitored by counting the zero signal (due to negative inputs to the ADCs) and the signals above the thresholds, then using a minimisation algorithm. Background counts which overflow the small counter ranges are discarded to avoid wasting bits and then corrected statistically offline.<BR>First the 10-bit output of the ADCs is compressed to 8 bits using a quasi-parabolic monotonic characteristic which requires no conversion table, no clock cycles and in which all codes are utilised. The expansion to the original values for offline analysis includes a compensation for the truncated signals which is statistically without bias. The main algorithm is purely combinatorial and thus requires only one clock cycle to code the signals and send them on through a circular twisted barrel shifter to the output FIFO. Simulations of the circuits using FPGAs and of the compression algorithm in software are in proress.