Timing techniques with picosecond-order accuracy for novel gaseous detectors

TL;DR

This paper develops a simulation-based neural network approach to achieve picosecond-level timing accuracy for PICOSEC Micromegas detectors, enabling fast online signal timing with minimal data storage.

Contribution

It introduces a novel simulation and neural network method for precise, real-time timing of detector signals, reducing data requirements compared to traditional techniques.

Findings

Achieved 18.3 ps timing precision with ANN

Validated ANN performance against offline processing

Demonstrated effective online timing with minimal data

Abstract

A simulation model is developed to train Artificial Neural Networks (ANN), for precise timing of PICOSEC Micromegas detector signals. The aim is to develop fast, online timing algorithms as well as minimising the information to be saved during data acquisition. PICOSEC waveforms were collected and digitised by a fast oscilloscope during a femptosecond-laser test beam run. A data set comprising waveforms collected with attenuated laser beam intensity, eradicating the emission of more than one photoelectron per light pulse from the PICOSEC photocathode, was utilised by a simulation algorithm to generate waveforms to train an ANN. A second data set of multi-photoelectron waveforms was used to evaluate the ANN performance in determining the PICOSEC Signal Arrival Time, relative to a fast photodiode time-reference. The ANN timing performance is the same as the results of a full offline signal processing, achieving a timing precision of 18.3$\pm$0.6 ps.