Abstract
The Barium Fluoride (BaF2) detector is a critical experimental facility at the China Spallation Neutron Source (CSNS) White Neutron Source (WNS), designed for precise measurements of (n, γ) cross-sections. With 92 crystal elements providing complete 4π steradian coverage, the detector requires a high-performance readout system. This proposal details a distributed architecture using four PXIe crates, employing 1 GSps waveform digitizing with 12-bit resolution. The system effectively manages massive data rates through parallel transmission over Gigabit Ethernet and ensures sub-nanosecond synchronization via a high-precision distribution network. Prototype tests demonstrate excellent signal integrity and robust performance.
The Challenge: Capturing the "Data Firehose"
Measuring the probability of a nucleus capturing a neutron—the (n, γ) cross-section—is fundamental to understanding stellar nucleosynthesis and nuclear waste management. The CSNS-WNS detector uses a spherical array of 92 BaF2 crystals to capture these events. However, traditional electronics that only record peak values lose vital information needed for Pulse Shape Discrimination (PSD), which is essential for distinguishing gamma rays from background alpha particles.
Capturing full waveforms at 1 GSps with 12-bit resolution creates a massive engineering hurdle: each channel generates approximately 12 Gbps of raw data. The system must digitize, synchronize, and process this data in real-time while maintaining the picosecond-level timing accuracy required for neutron time-of-flight measurements.
Distributed Architecture and Methods
To address these requirements, researchers proposed a distributed readout system based on the PXI Express (PXIe) standard. The architecture is organized into four synchronized crates, each housing multiple Field Digitization Modules (FDMs).
Key technical features include:
- Waveform Digitizing: Utilizing 12-bit ADCs sampling at 1 GSps to capture the complete physical profile of every pulse.
- FPGA-Centric Processing: Each FDM utilizes a Xilinx Kintex-7 FPGA for real-time data handling. To manage bandwidth, the FPGAs implement zero-channel compression (discarding baseline data between pulses) and hardware-based trigger logic.
- Synchronous Distribution: A two-stage network using Clock Data Recovery (CDR) techniques ensures all 92 channels operate on a unified clock. This setup achieves incredibly low jitter across the entire distributed system.
- Data Transmission: Digitized signals are concentrated to the PXIe crate controllers via the PCIe backplane and transmitted to the Data Acquisition (DAQ) system over parallel Gigabit Ethernet links.
Key Results and Performance
Prototype testing validated the system's ability to meet the stringent requirements of the CSNS-WNS facility. The system achieved high-fidelity pulse capture, essential for accurate physics measurements.
Benchmarks from the prototype evaluation include:
- Signal Integrity: The system achieved an Effective Number of Bits (ENOB) of 9.03 for signals up to 100 MHz, ensuring that the digitized waveforms accurately represent the original analog pulses.
- Throughput: Data transmission reached 942 Mbps per Gigabit Ethernet link, representing over 90% utilization of theoretical bandwidth.
- Timing Precision: The clock distribution network showed a skew of only 200 ps between slots and a terminal jitter of approximately 12 ps, far exceeding the sub-nanosecond requirement.
- Experimental Validation: Laboratory tests using cosmic rays and alpha sources confirmed that the electronics could successfully distinguish between different radiation types using Pulse Shape Discrimination.
Conclusion
The proposed readout electronics system successfully addresses the challenges of high-speed, high-resolution data acquisition for the 92-element BaF2 detector. By combining a distributed PXIe architecture with advanced FPGA processing and high-precision synchronization, the system provides a robust platform for future nuclear physics experiments at CSNS. Future work will focus on full-scale deployment and the integration of automated calibration routines for the complete 92-channel array.
Summary updated