A Multifunctional Processing Board for the Fast Track Trigger of the H1 Experiment

TL;DR

This paper describes a multifunctional processing board designed for the second level of the Fast Track Trigger system in the H1 experiment, enabling rapid 3D track reconstruction in high-energy physics experiments.

Contribution

It introduces a novel hardware implementation combining FPGA and DSPs with integrated CAM for fast track linking and fitting in trigger systems.

Findings

Achieved 48 track reconstructions within 23 microseconds.

Successfully integrated FPGA and DSPs for high-speed processing.

Demonstrated effective use of CAM for fast track segment linking.

Abstract

The electron-proton collider HERA is being upgraded to provide higher luminosity from the end of the year 2001. In order to enhance the selectivity on exclusive processes a Fast Track Trigger (FTT) with high momentum resolution is being built for the H1 Collaboration. The FTT will perform a 3-dimensional reconstruction of curved tracks in a magnetic field of 1.1 Tesla down to 100 MeV in transverse momentum. It is able to reconstruct up to 48 tracks within 23 mus in a high track multiplicity environment. The FTT consists of two hardware levels L1, L2 and a third software level. Analog signals of 450 wires are digitized at the first level stage followed by a quick lookup of valid track segment patterns. For the main processing tasks at the second level such as linking, fitting and deciding, a multifunctional processing board has been developed by the ETH Zurich in collaboration with Supercomputing Systems (Zurich). It integrates a high-density FPGA (Altera APEX 20K600E) and four floating point DSPs (Texas Instruments TMS320C6701). This presentation will mainly concentrate on second trigger level hardware aspects and on the implementation of the algorithms used for linking and fitting. Emphasis is especially put on the integrated CAM (content addressable memory) functionality of the FPGA, which is ideally suited for implementing fast search tasks like track segment linking.