TL;DR
The MicroBooNE experiment has successfully characterized and mitigated various sources of electronic noise in its liquid argon time projection chamber (LArTPC). By implementing a combination of offline filtering techniques and hardware upgrades, they've achieved significantly lower noise levels than previous experiments, crucial for precise reconstruction of neutrino interactions.
What Problem Does This Paper Address?
Liquid argon time projection chambers (LArTPCs) are powerful tools for detecting neutrinos and searching for rare particle decays due to their excellent spatial resolution. However, extracting meaningful physics signals from these detectors hinges on the ability to accurately measure the ionization charge deposited on the wire planes. This measurement is severely hampered by electronic noise, which can obscure or distort the faint signals from particle interactions. The MicroBooNE experiment, a large LArTPC, encountered unexpected levels of "excess noise" beyond the inherent electronic noise, necessitating a thorough investigation and mitigation strategy to ensure its physics goals could be met.
Methods
The paper details a multi-pronged approach to tackle the noise problem:
- Noise Characterization: The authors meticulously analyzed data to identify and categorize different types of excess noise. This involved examining frequency spectra, time-domain waveforms, and correlating noise with specific detector components and operating conditions.
- Offline Noise Filtering: A sophisticated offline noise filter was developed. This filter uses techniques like correlated low-frequency noise subtraction and frequency domain masking to remove identified noise sources while minimizing impact on genuine signals.
- Hardware Upgrades: Two significant hardware upgrades were implemented:
- Replacement of low-voltage regulators with lower-noise models.
- Installation of additional filtering in the high-voltage system.
- Performance Metrics: The effectiveness of the noise mitigation was quantified using metrics like Equivalent Noise Charge (ENC) and Peak-Signal-to-Noise Ratio (PSNR).
Datasets
The study utilizes data collected during the first year of MicroBooNE operations (October 2015 - July 2016). This includes raw data from the TPC readout channels, as well as data from pulser calibration signals. No new benchmark datasets were introduced; the focus was on analyzing real experimental data.
Key Insights and Ideas
- Multiple Noise Sources: The excess noise wasn't a single culprit but a combination of factors, including issues with low-voltage regulators, the high-voltage power supply, and even unexpected ASIC behavior (like periodic saturation).
- Cold Electronics Advantage: Operating the front-end electronics immersed in liquid argon at cryogenic temperatures significantly reduces inherent electronic noise compared to room-temperature systems.
- Wire Length Matters: Noise levels are directly correlated with wire length, with longer wires exhibiting higher noise. This is a critical consideration for detector design and analysis.
- Hardware Fixes Complement Filtering: While offline filtering is effective, hardware upgrades proved crucial for substantially reducing the magnitude of certain noise sources, simplifying the offline processing.
- Induction vs. Collection Planes: Induction planes, which detect induced currents as charge drifts, are inherently more susceptible to noise and exhibit lower signal-to-noise ratios compared to collection planes.
Key Results
- Significant Noise Reduction: After applying offline filtering and hardware upgrades, the residual Equivalent Noise Charge (ENC) for the longest wires (4.7 m) was reduced to below 400 electrons.
- Improved PSNR: The Peak-Signal-to-Noise Ratio (PSNR) for standard candle signals (ideal MIP tracks) saw significant improvements, with values increasing from around 6-19 to 16-38 across different planes after noise filtering.
- Reduced Non-Functioning Channels: Hardware upgrades and improved understanding led to a reduction in the number of non-functioning channels, though a small percentage remain unusable.
- Lower Noise than Previous Experiments: The achieved noise levels are significantly lower than those reported in previous LArTPC experiments that used warm front-end electronics.
Limitations and Future Work
- Unidentified 900 kHz Burst Noise: The exact source of the 900 kHz burst noise remains unconfirmed, though it's suspected to be related to the PMT high-voltage supply or interlock system. While currently attenuated to acceptable levels, further investigation might be needed if it increases.
- Remaining Coherent Noise: Despite hardware upgrades, a small amount of low-frequency coherent noise from voltage regulators persists, requiring continued offline filtering.
- Future ASIC Designs: The experience gained has informed the design of next-generation ASICs for future experiments (like DUNE and SBN), incorporating features like improved bias current settings and electrostatic discharge protection.
- Digital Signal Processing: Further digital signal processing, such as deconvolution of electronics shaping and wire field response, is needed for optimal charge resolution, which is beyond the scope of this paper.
Summary updated