The Remote Analog to Digital Conversion DAQ System for the TRISTAN Detector Upgrade

This paper presents the conceptual design and implementation of a remote analog-to-digital conversion (RADC) data acquisition system developed to handle the high-rate, precision electron spectroscopy required for the TRISTAN detector upgrade in the search for keV-scale sterile neutrinos.

The Gist

The "High-Speed Digital Camera" for Tiny Particles: A Simple Guide

Imagine you are trying to photograph a swarm of millions of tiny, hyperactive gnats flying through a dark room. These gnats are moving so fast that if you use a regular camera, they just look like a blurry mess. Even worse, the room is filled with massive, powerful magnets and high-voltage electricity that would fry a normal camera instantly.

This is essentially the problem scientists are facing with the TRISTAN detector upgrade. They are looking for "sterile neutrinos"—ghostly, invisible particles that could explain what Dark Matter is. To find them, they need to watch how tritium (a type of radioactive hydrogen) decays, which releases electrons.

Here is how they built a "super-camera" to solve this.

---

1. The Problem: The "Electric Storm" Environment

The experiment takes place inside the KATRIN experiment, a massive setup that is basically an electric storm. There are huge magnetic fields and incredibly high voltages.

If you tried to put a standard computer or a high-tech sensor right next to the particles, the electronics would "pop" like a lightbulb in a lightning storm.

2. The Solution: The "Remote Control" Strategy (RADC)

The researchers invented something called RADC (Remote Analog to Digital Conversion).

Think of it like this: Instead of putting a whole expensive computer inside the lightning storm, they only put a very simple, "tough" sensor (the Front-End) right next to the particles. This sensor is like a rugged, waterproof GoPro. It doesn't "think"; it just captures the raw electrical "shape" of the particle's arrival and immediately turns it into a digital signal.

Then, they send that signal through fiber-optic cables (which are like high-speed glass tunnels) to a "Brain" (the Back-End) located far away in a safe, quiet, air-conditioned room. This way, the "Brain" can do all the heavy thinking without ever getting hit by the magnetic storm.

3. The "Brain": Sorting the Chaos

Once the data reaches the Back-End, it’s a firehose of information. If they tried to save every single tiny detail of every single particle for four months, they would run out of hard drive space—it would be like trying to record every single raindrop in a hurricane.

To fix this, the "Brain" (using powerful chips called FPGAs) uses three different "recording modes":

• The "Slow Motion" Mode (Waveform): This records every tiny detail of a single event. It’s like a high-speed video. It’s too much data to use all the time, so they only use it for a few seconds to check if the "camera" is working correctly.
• The "List" Mode (ListMode): This is like a spreadsheet. Instead of a video, it just writes down: "At 12:01:05, a particle hit with X energy." This is much smaller, but still huge.
• The "Summary" Mode (Histogram): This is the most efficient. Instead of recording every individual particle, the system just keeps a "tally." It’s like a person standing at a stadium gate with a clicker, just counting how many people enter every minute. This turns a mountain of data into a small, manageable pile of numbers.

4. Dealing with "Pileup" (The Traffic Jam)

Because the particles are coming so fast, sometimes two or three hit the sensor at almost the exact same time. This is called "pileup."

Imagine two cars crashing into each other at a toll booth. If you only look at the wreckage, you might think it was one giant, super-fast car, rather than two normal cars hitting at once. The TRISTAN system is smart enough to recognize these "crashes" and flag them so they don't mess up the scientific results.

Summary

In short, the researchers have built a rugged, high-speed, remote-controlled observation system. It can survive a magnetic storm, capture incredibly fast "snapshots" of invisible particles, and intelligently decide which data is worth keeping and which can be summarized, ensuring they don't drown in a sea of digital information while hunting for the secrets of the universe.

Technical Summary

Technical Summary: The Remote Analog to Digital Conversion (RADC) DAQ System for the TRISTAN Detector Upgrade

1. Problem Statement

The TRISTAN detector is a high-precision upgrade to the KATRIN experiment, designed to search for keV-scale sterile neutrinos by measuring the tritium $\beta$-decay spectrum. This measurement presents extreme data acquisition (DAQ) challenges:

• High Event Rates: The system must handle over 1,000 silicon drift detector (SDD) pixels, each operating at rates of $\sim 10^5$ counts per second (kcps).
• Harsh Environment: The front-end electronics must operate in close proximity to the detector, exposed to high voltages (up to 25 kV) and strong magnetic fields ($\sim 100$ mT).
• Data Volume: Raw waveform streaming for all channels would generate massive data rates (up to 186.7 GB/s), making long-term storage and analysis of a multi-month measurement campaign impossible.
• Signal Integrity: High-rate operation risks significant energy reconstruction bias due to pileup, crosstalk, and ADC non-linearities.

2. Methodology

The authors propose a Remote Analog to Digital Conversion (RADC) architecture. This design physically and logically separates the digitization process from the signal processing to optimize for both environmental resilience and data throughput.

A. Front-End System (The Tile Main Board - TMB):

• Location: Mounted directly on the vacuum chamber (in-vacuum/near-vacuum).
• Design: To survive the environment, the TMB is non-ferromagnetic and uses isolation transformers. It utilizes a modular "plug-on" design to simplify maintenance.
• Digitization: Signals from the ETTORE ASICs are processed through 21 ADC boards. Each board uses an 8-channel 14-bit ADC (AD9257) with a 5th-order Bessel low-pass filter to minimize pulse distortion.
• Signal Management: A unique "Gatti Slider Board" provides an adjustable DC offset to the signals before digitization, allowing the system to map detector outputs into the optimal ADC range and mitigate non-linearity effects.
• Transmission: Data is serialized and sent via high-speed fiber-optic links (Samtec FireFly™) to the back-end, ensuring electrical isolation and immunity to electromagnetic interference.

B. Back-End System (Serenity-S1 FPGA):

• Hardware: Uses the Serenity-S1 ATCA-card (based on AMD Virtex UltraScale+ FPGAs), originally developed for the CMS experiment at the LHC.
• Digital Signal Processing (DSP): The FPGA implements two specialized filters:
  1. Fast Trigger Filter: A triangular filter used to identify event timing and distinguish coincident events.
  2. Energy Filter: An exponential deconvolution followed by a trapezoidal filter to reconstruct precise energy values.
• Event Logic: The system includes sophisticated "trigger maps" to flag pileup, charge sharing between neighboring pixels, and crosstalk.

C. Data Reduction Strategy:
To manage the data deluge, the system implements four readout modes:

1. Waveform/ListWave: Full waveform capture (used for calibration).
2. ListMode: Triggered event lists (used for systematic studies).
3. Histogram Mode: Real-time population of energy histograms (the primary mode for physics data taking).

3. Key Contributions

• RADC Architecture: A novel separation of "digitization-at-source" (front-end) and "processing-at-distance" (back-end) that solves the conflict between high-rate requirements and harsh environmental constraints.
• Modular TMB Design: A highly specialized, radiation/magnetic-field-hardened PCB architecture capable of providing bias, power, and high-speed readout for 166 pixels per module.
• Advanced FPGA-based Filtering: Implementation of real-time exponential deconvolution and trapezoidal shaping on high-end FPGAs to enable high-precision spectroscopy at high rates.
• Multi-Tiered Data Management: A scalable readout hierarchy that reduces data volume by up to five orders of magnitude through intelligent histogramming.

4. Results and Significance

• Data Rate Optimization: The system successfully reduces the projected data volume from an unmanageable $\sim 186$ GB/s (Waveform mode) to a manageable $\sim 10^{-100}$ TB range for a four-month measurement campaign by utilizing Histogram mode.
• Precision and Reliability: The architecture allows for $>99\%$ trigger efficiency for 2 keV energy depositions and provides the tools (Gatti offset, pileup flagging) necessary to maintain the energy resolution required for sterile neutrino searches.
• Scientific Impact: This DAQ system provides the technological foundation for the TRISTAN upgrade, enabling the KATRIN experiment to reach a sensitivity of $\sin^2(\theta) < 10^{-6}$ for keV-scale sterile neutrinos, potentially providing a solution to dark matter mysteries.