subatomic-telemetry-lab/tpc-raw-waveform-dumps

Dataset Card for SPDTL TPC Raw Waveform Dumps

Dataset Summary

This repository serves as a cold-storage archive for uncalibrated, high-entropy digitized waveforms collected from Time Projection Chamber (TPC) readout pads during high-luminosity particle collision runs. The dataset consists of continuous analog-to-digital converter (ADC) time-series samples captured across multi-channel front-end ASICs before zero-suppression or track reconstruction algorithms are applied.

Because these raw dumps preserve all hardware-level thermal noise, baseline fluctuations, and beam-induced electromagnetic interference, they appear as unstructured binary payloads. This dataset is designed for benchmarking distributed hardware accelerators, developing AI-driven online trigger filters, and testing noise-cancellation models for next-generation collider instrumentation.

Supported Tasks and Leaderboards

• online-trigger-filtering: Training fast neural networks to perform real-time event filtering on raw ADC streams before hardware buffer overwrite.
• track-noise-suppression: Benchmarking denoising models on raw, uncalibrated drift chamber signals.

Languages

There is no natural language present. All payloads consist of high-density binary streams of uncompressed digitizer telemetry.

Dataset Structure

Data Instances

To accurately reflect raw hardware acquisition, data is archived as contiguous binary LFS blobs rather than formatted tabular structures.

{
  "run_id": "spdtl-run-2026-001",
  "detector_sector": "TPC_BARREL_WEST",
  "channel_count": 1024,
  "sampling_frequency_mhz": 40.0,
  "payload_reference": "raw_adc/sector_w_adc_run001.bin",
  "zero_suppression": false
}

Data Fields

• run_id: Unique identifier for the calibration run.
• detector_sector: Specific hardware region of the TPC readout plane.
• channel_count: Number of active digitizer channels streamed in parallel.
• sampling_frequency_mhz: Sampling rate in megahertz responsible for the data throughput.
• payload_reference: Direct path to the multi-gigabyte binary files in Git LFS.
• zero_suppression: Maintained as false to account for the full baseline stream size. Dataset Creation Curation Rationale Standard high-energy physics pipelines apply aggressive online filtering and track reconstruction, discarding uncalibrated background noise. However, training ML algorithms for next-generation detector electronics requires access to raw, unreduced waveform telemetry. Storing these raw baseline dumps requires terabyte-scale cold storage. Source Data Data comprises synthetic and uncalibrated digitizer hardware dumps simulating a high-rate multi-strip readout environment. Disclaimers These binary files are exceptionally large and uncompressed. Attempting to parse or visualize them without low-level C++/ROOT detector drivers or specialized hardware ingestion frameworks will result in execution errors. Access is recommended solely for high-performance computing (HPC) nodes.