Neural network-based timewalk correction for the Timepix4 ASIC
Abstract
Timewalk corrections of hybrid pixel detectors conventionally employ heuristic fits performed independently in bins of the relevant input variables. This requires both an explicit functional ansatz and binning scheme, as well as densely populated calibration data whose required size grows rapidly as further input variables are added. This paper presents a timewalk correction for the Timepix4 readout chip based on a compact feed-forward neural network (NN), which learns the dependence directly from the input variables without requiring an explicit functional ansatz or binning. The NN is compared against the fit-based baseline using data from the Timepix4 beam telescope at the CERN SPS in 2025, with 100 μm and 300 μm thick planar sensors. Both methods parameterise the timewalk correction as a function of the measured charge and reconstructed intrapixel impact position. With only 20 % of the calibration data, the NN already achieves the same asymptotic time resolutions as the fit-based baseline: 180 ps and 300 ps for the two thicknesses, respectively.
Related
- A flexible electronics and DAQ system for the Timepix4 ASIC
- A flexible DAQ system for the Timepix4 ASIC in the 4DPHOTON project
- Graph Neural Post-selection for Quantum Error Correction
- Scalable neural decoders for practical real-time quantum error correction
- PyDoseRT Proton: A GPU Pencil-Beam Engine with a Convolutional Residual-Correction Network for Fast Proton Dose Calculation
BibTeX
@misc{bacher2026neural,
title = {{Neural network-based timewalk correction for the Timepix4 ASIC}},
author = {David Bacher and Kazu Akiba and Martin van Beuzekom and Victor Coco and Raphael Dumps and Tim Evans and Kevin Heijhoff and Malcolm John and Edgar Lemos Cid and Tommaso Pajero},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.10328},
eprint = {2610.10328},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.10328},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Neural network-based timewalk correction for the Timepix4 ASIC
AU - David Bacher
AU - Kazu Akiba
AU - Martin van Beuzekom
AU - Victor Coco
AU - Raphael Dumps
AU - Tim Evans
AU - Kevin Heijhoff
AU - Malcolm John
AU - Edgar Lemos Cid
AU - Tommaso Pajero
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.10328
UR - https://arxiv.org/abs/2610.10328
AB - Timewalk corrections of hybrid pixel detectors conventionally employ heuristic fits performed independently in bins of the relevant input variables. This requires both an explicit functional ansatz and binning scheme, as well as densely populated calibration data whose required size grows rapidly as further input variables are added. This paper presents a timewalk correction for the Timepix4 readout chip based on a compact feed-forward neural network (NN), which learns the dependence directly from the input variables without requiring an explicit functional ansatz or binning. The NN is compared against the fit-based baseline using data from the Timepix4 beam telescope at the CERN SPS in 2025, with 100 μm and 300 μm thick planar sensors. Both methods parameterise the timewalk correction as a function of the measured charge and reconstructed intrapixel impact position. With only 20 % of the calibration data, the NN already achieves the same asymptotic time resolutions as the fit-based baseline: 180 ps and 300 ps for the two thicknesses, respectively.
ER -