Design and Evaluation of a PMT High-Voltage system for Deepsea Neutrino Telescope

This paper presents the design, characterization, and successful laboratory validation of a Cockcroft-Walton high-voltage system for a hybrid Digital Optical Module, demonstrating its ability to provide stable, independently adjustable bias voltages for 31 photomultiplier tubes with the noise, gain, and timing precision required for deep-sea neutrino telescopes.

The Gist

Imagine you are trying to listen to a whisper from the bottom of the ocean, thousands of meters down. That's essentially what a Deep-Sea Neutrino Telescope does. It's a giant underwater listening post designed to catch ghostly particles called neutrinos that pass through the Earth.

To do this, the telescope uses special "ears" called Photomultiplier Tubes (PMTs). These are like super-sensitive microphones that can detect a single flash of light (a photon) created when a neutrino hits something.

However, there's a problem: these "microphones" are incredibly fragile and need a very specific, super-high electrical push (voltage) to work. If the electricity wavers even a tiny bit, the microphone gets confused, the sound gets distorted, or it stops working entirely. And since these devices are buried deep underwater, you can't just send a diver down to fix them.

This paper is about building a perfect, self-correcting battery pack for 31 of these microphones, all crammed into a single glass sphere.

Here is the breakdown of their solution, using some everyday analogies:

1. The Problem: One Big Power Strip vs. 31 Individual Outlets

In the past, you might have tried to power a whole room of lights with one giant power strip. If the strip wobbled, all the lights flickered.

• The Old Way: One big power source for many sensors.
• The New Way (This Paper): The team built 31 tiny, independent power stations, one for each microphone.
• The Analogy: Imagine a choir of 31 singers. Instead of giving them all one shared microphone that might crackle, you give every single singer their own high-end, noise-canceling headset with its own battery. If one battery has a hiccup, the other 30 keep singing perfectly.

2. The Magic Trick: The "Cockcroft-Walton" Ladder

How do you get a tiny 5-volt battery (like in a remote control) to push out 1,500 volts (enough to jump-start a car, but negative)?

• The Solution: They use a circuit called a Cockcroft-Walton (CW) multiplier.
• The Analogy: Think of this like a bucket brigade or a staircase.
  • Imagine you have a small bucket of water (low voltage).
  • You pour it into the first step of a staircase.
  • Then, you pour that water into the next step, which adds more water on top.
  • You keep doing this up a long ladder. By the time you reach the top, you have a massive waterfall (high voltage), even though you started with just a small cup.
  • The "Cockcroft-Walton" is just a very clever, efficient way to build this electrical staircase using tiny capacitors (water buckets) and diodes (one-way valves) so the water only flows up, never down.

3. The Brain: The "Smart Conductor"

Having 31 power stations is great, but who tells them what to do?

• The System: A central computer (an FPGA) acts as the conductor. It talks to a "Control Board," which then whispers instructions to each of the 31 individual power stations via a digital language called I2C.
• The Analogy: Imagine a traffic controller at a busy airport.
  • The controller doesn't push the planes; it just sends signals to the ground crew at each gate.
  • "Gate 1, push the voltage up a little." "Gate 2, slow it down."
  • The system is so smart that if a microphone starts to drift (like a singer getting tired), the computer instantly adjusts the voltage to keep the pitch perfect. It does this automatically, hundreds of times a second.

4. The Test: The "Deep Sea Simulator"

Before sending this expensive gear to the bottom of the ocean, they had to test it.

• The Setup: They put the whole glass sphere in a giant freezer set to 2°C (4°C is the temperature of deep ocean water) and filled it with nitrogen gas to simulate the pressure.
• The Result: They ran the system for 100 hours straight (about 4 days).
  • Stability: The "pitch" of the microphones didn't waver. They stayed perfectly in tune.
  • Speed: When a flash of light hit, the microphone reacted in less than 1.8 nanoseconds.
  • The Analogy: That's like a sprinter reacting to a starting gun so fast that they finish the race before you can even blink. This speed is crucial because to figure out where a neutrino came from, the telescope needs to know exactly when the light hit each sensor.

Why Does This Matter?

Neutrinos are the messengers of the universe. They tell us about exploding stars and black holes. But to read their message, we need our "ears" to be perfectly tuned and incredibly fast.

This paper proves that the team has built a reliable, self-correcting, high-speed power system that can survive the crushing pressure of the deep ocean. It ensures that when the next cosmic event happens, the telescope won't miss a single whisper.

In short: They built 31 tiny, smart, self-adjusting power generators that work together perfectly, ensuring the telescope's "ears" are always listening clearly, even in the darkest, deepest part of the ocean.

Technical Summary

1. Problem Statement

Deep-sea neutrino telescopes, such as the TRIDENT (TRopical DEep-sea Neutrino Telescope), require highly reliable and stable optical detection systems to detect faint Cherenkov photons from astrophysical neutrinos. The TRIDENT telescope utilizes Hybrid Digital Optical Modules (hDOMs), each housing 31 three-inch Photomultiplier Tubes (PMTs) and 24 Silicon Photomultipliers (SiPMs) within a single 17-inch borosilicate glass vessel.

The primary challenges addressed in this work include:

• Environmental Constraints: The system must operate reliably at depths exceeding 3,000 meters (pressures up to 670 bar) and low temperatures (2–4 °C) without maintenance.
• Gain Uniformity & Stability: PMT gain is highly sensitive to bias voltage variations. To ensure accurate event reconstruction, all 31 PMTs in a module must operate at a uniform, stable gain (targeting $10^7$).
• Timing Precision: Neutrino track reconstruction requires sub-degree angular resolution, demanding photon timing precision on the order of 2 ns. Any instability in the high-voltage (HV) supply can degrade the Transit-Time Spread (TTS).
• Space & Complexity: The confined geometry of the hDOM necessitates a compact HV solution that avoids single-point failures and allows for independent channel control.

2. Methodology

The authors designed and characterized a Cockcroft–Walton (CW) based High-Voltage system tailored for the hDOM. The methodology involved three main phases:

A. System Architecture & Design

• Distributed Topology: A "one-CW-per-PMT" architecture was adopted. Each of the 31 PMTs has an independent HV base, ensuring channel isolation and preventing single-point failures.
• Three-Tier Control Hierarchy:
  1. FPGA Motherboard: Issues high-level commands.
  2. Control Board: Acts as an intermediate hub using a multiplexed $I^2C$ architecture (utilizing PCA9547 8-channel multiplexers) to manage communication with 31 bases.
  3. PMT Bases: Each base contains a Microcontroller Unit (MCU), an inductive resonant driver, and a CW voltage multiplier.
• Voltage Generation: The MCU generates a Pulse-Width Modulated (PWM) signal (~80 kHz) to drive an LC resonant circuit, producing an AC waveform. This is rectified and multiplied by a CW diode-capacitor network to generate adjustable negative voltages (0 V to -1.5 kV).
• Feedback Loop: A precision resistor divider scales the HV output for the MCU's internal ADC, enabling closed-loop regulation of the PWM frequency and duty cycle to maintain stable voltage.

B. Simulation & Component Testing

• Circuit Simulation: LTspice was used to model the CW multiplier stages, verifying voltage buildup and steady-state behavior (convergence within ~100 ms).
• Hardware Validation: Three prototype bases were tested independently via a USB-to-serial interface to verify dynode voltage ratios (3:1:1:1:1:1:1:1:1:1:1) and consistency with the Hamamatsu R14374 and NNVT N2031 PMT specifications.

C. Performance Evaluation

Two experimental setups were used to characterize the integrated system:

1. Laser Test Bench: Used for individual PMT characterization (Gain-Voltage curves, Single Photoelectron (SPE) spectra, and TTS) using a 532 nm picosecond laser.
2. Integrated hDOM Setup: The full 31-PMT module was placed in a climatic chamber (2 °C, 0.5 atm nitrogen) to simulate deep-sea conditions. Data was collected over 100 hours using three trigger modes (self-trigger, random, and local coincidence) to assess long-term stability.

3. Key Contributions

• Compact Multi-Channel HV System: Successfully designed a distributed HV system that fits within the strict mechanical envelope of a deep-sea optical module while providing independent control for 31 channels.
• Closed-Loop Gain Calibration: Developed an autonomous control algorithm where the FPGA scans PWM frequencies for coarse tuning and iteratively adjusts the duty cycle for fine-tuning, achieving sub-percent voltage precision.
• Comprehensive Characterization: Provided a full validation of the system under simulated deep-sea conditions, covering gain stability, baseline noise, and timing resolution, which is critical for the TRIDENT project.

4. Key Results

• Voltage Stability & Ripple:
  • The system achieves a regulated output with < 150 mVpp ripple (Anode to GND).
  • Baseline fluctuations were measured at 0.3%–1.4% over five days, indicating excellent front-end electronic noise stability.
• Gain Uniformity & Stability:
  • All 31 PMTs were successfully tuned to a common nominal gain of $1 \times 10^7$.
  • Over a 100-hour test, gains stabilized after ~20 hours, with fluctuations remaining within ±2–3% for most channels and < 5% for high-gain channels.
• Timing Resolution (TTS):
  • Transit-Time Spread measurements yielded values below 1.8 ns (FWHM).
  • Specific measurements showed a TTS of 1.11 ns, consistent with manufacturer specifications and sufficient for sub-degree track reconstruction.
• Signal-to-Noise Ratio (SNR):
  • At a gain of $10^7$, the system achieved an SNR of ~40 (8.2 mV pulse peak vs. 0.2 mV baseline RMS), ensuring reliable triggering.
• Power Consumption:
  • The system is highly efficient, consuming < 60 mW per base at -1200 V output.

5. Significance

This work demonstrates that a Cockcroft–Walton based HV system is a viable and superior solution for deep-sea multi-PMT optical modules. The design successfully addresses the critical constraints of deep-sea deployment:

• Reliability: The distributed architecture eliminates single-point failures, crucial for un-maintainable deep-sea deployments.
• Precision: The system meets the stringent timing (TTS < 1.8 ns) and gain stability requirements necessary for reconstructing high-energy neutrino events.
• Scalability: The successful integration of 31 independent channels in a single module validates the approach for scaling to larger, next-generation neutrino telescopes.

The results confirm that the TRIDENT hDOM's HV system is ready for long-term deployment, providing the stability and precision required to advance the field of neutrino astronomy.