Erlangen 2026 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

T: Fachverband Teilchenphysik

T 6: Electronics, Trigger, DAQ I

T 6.2: Talk

Monday, March 16, 2026, 16:30–16:45, KH 00.023

Firmware developments and hardware tests for the HL-LHC ATLAS Liquid Argon Signal Processor system — Peter Fischer, •Markus Helbig, Rainer Hentges, Arno Straessner, Johann C. Voigt, and Philipp Welle — Institut für Kern- und Teilchenphysik, TU Dresden

In the HL-LHC era, the ATLAS Liquid Argon Calorimeters will be equipped with the new LAr Signal Processors (LASP). This off-detector processing system based on Altera Agilex 7 FPGAs will perform energy reconstruction for all 182 468 detector cells at every bunch crossing in real time and provide them to the new trigger systems Global Event Processor (GEP) and the Forward Feature Extractor (fFEX) as well as the main readout.

To cope with the higher pile-up, the use of 1-dimensional Convolutional Neural Networks (CNNs) with up to 400 parameters for energy reconstruction has been studied as an alternative to the current Optimal Filter approach. Using efficient FPGA resource allocations, a sufficient number of CNN instances to process up to 384 cells are fit on a single FPGA, leaving enough resources for other components. Their implementation has been verified in standalone hardware setups based on FPGA development kits and LASP testboards.

Furthermore, the first Agilex-based prototype board has recently become available. On this board, the interfaces from the front-end and towards the trigger systems are implemented as optical transceivers using Samtec FireFly modules. Their performance is studied to guarantee reliable data transmission.

Keywords: Liquid Argon; Calorimeter; Firmware; optical transceiver; neural network