Hardware and software structure of a neural network trigger based on the ETANN and MA16 chips

A neural-network trigger has been developed for secondary vertex detection in the WA92 experiment at CERN, looking for the production of beauty particles. The module includes the Intel analog ETANN chip and the Siemens digital (16-bit I/O) MA 16 microprocessor, working independently. The experimental trigger results have already been presented elsewhere. Since this represents the first application of a neural trigger in an high energy physics experiment, we describe here in some detail its hardware and software structure. Specifics are given on the utilization of the ETANN chip, caring about the stability problems associated with its analog nature, and of the MA 16 chip, for which this also represents the first stand-alone application.