Upgrade of the Trigger and Data Acquisition System for Continuous Imaging and Multi-Camera Operation in CYGNO

This paper presents an upgraded Trigger and Data Acquisition system for the CYGNO experiment that enables continuous imaging, robust global time-tagging, and synchronized multi-camera operation, establishing a scalable foundation for the upcoming CYGNO-04 phase and future large-scale optical TPC detectors.

The Gist

The Big Picture: Catching Ghosts in a Foggy Room

Imagine the CYGNO experiment is a giant, high-tech room filled with a special, invisible fog (a gas mixture). Scientists want to catch "ghosts" (rare particles) that occasionally float through this room. When a ghost hits the fog, it leaves a tiny, glowing trail, like a firefly blinking in the dark.

To see these trails, the scientists use two tools:

1. Super-fast eyes (PMTs): These are like high-speed cameras that can see a flash of light in a nanosecond (a billionth of a second). They tell you exactly when something happened.
2. Big, slow cameras: These take huge, detailed photos of the whole room to show you what the trail looks like.

The problem? The "fast eyes" work in the blink of an eye, but the "big cameras" are slow. They take a long time to snap a photo, develop it, and get ready for the next one.

The Old Problem: The "Stop-and-Go" Traffic Jam

In the previous version of their system (called LIME), the big cameras worked like a traffic light that was stuck on red for too long.

• How it worked: The camera would take a photo, then spend a long time "resetting" and getting ready for the next one. During this reset time (about 38% of the time), the camera was blind.
• The Analogy: Imagine a security guard taking a photo of a hallway. He takes a picture, then spends 4 seconds looking at the photo, putting it in a folder, and clearing his mind before he can look down the hall again. If a thief walks by while he's looking at the photo, the guard misses them.
• The Result: The system was missing a lot of data because the camera was "dead" (blind) for too long. Also, if a long particle trail started while the camera was resetting, the photo would get cut off, like a movie that starts halfway through.

The New Upgrade: The "Continuous Stream"

The paper describes a major upgrade to the Trigger and Data Acquisition (T-DAQ) system (the brain that controls the cameras and computers) to fix these issues. They tested this on a smaller prototype called MANGO to get ready for the big CYGNO-04 machine.

Here are the three main improvements:

1. The "Rolling Shutter" (Continuous Imaging)

Instead of stopping to reset, the new system makes the camera work like a video camera rather than a still camera.

• The Change: The camera starts taking photos and keeps going without stopping. It reads the bottom row of the image while the top row is still being exposed to light.
• The Analogy: Instead of the guard stopping to file his photo, he now has a conveyor belt. He snaps a photo, drops it on the belt, and immediately looks down the hall again. The "dead time" drops from 38% to almost 0%. He never misses a thief.

2. The "Universal Clock" (Extended Time-Tagging)

Because the camera is now running continuously, it gets confusing to match the "fast eyes" (PMTs) with the "big camera" photos.

• The Problem: The old system used a small clock that reset every few seconds. If a particle hit the room 10 seconds after the camera started, the clock would have "rolled over" and forgotten the time.
• The Fix: They upgraded the software to use a 60-bit clock (a super-long counter).
• The Analogy: Imagine the old clock was a stopwatch that reset after 9 seconds. If you were timing a marathon, you'd lose track of the time. The new clock is like a stopwatch that can count for thousands of years without resetting. Now, every flash of light from the "fast eyes" is stamped with a precise time that matches perfectly with the continuous video stream.

3. The "Concert Conductor" (Multi-Camera Sync)

The next big machine (CYGNO-04) won't just have one camera; it will have six cameras looking at the room from different angles.

• The Challenge: If you have six cameras, they need to snap their photos at the exact same millisecond. If Camera A snaps a photo a split-second before Camera B, the 3D picture of the particle trail will be blurry and useless.
• The Solution: They built a system where all cameras listen to the same "conductor" (a timing signal).
• The Analogy: Think of a choir. In the old system, each singer tried to guess when to start singing. In the new system, a conductor waves a baton, and everyone starts singing on the exact same beat. There is no "Master Camera" bossing everyone else around; any camera can be the conductor, making the system flexible and tough. If one camera breaks, the others keep singing in perfect sync.

Why Does This Matter?

This upgrade is the foundation for the CYGNO-04 experiment.

• Efficiency: They won't waste time waiting for cameras to reset.
• Precision: They can perfectly match the "when" (from the fast eyes) with the "what" (from the big cameras).
• Scalability: They can now add more cameras without the system falling apart.

The Bottom Line:
The scientists took a system that was like a slow, stop-and-go security guard and turned it into a high-speed, continuous surveillance team with a perfect clock and a synchronized choir. This allows them to catch rare, fleeting particles with much higher confidence, paving the way for discovering new physics in the future.

Technical Summary

1. Problem Statement

The CYGNO experiment utilizes an optical readout Time Projection Chamber (TPC) to detect rare events (such as dark matter or neutrino interactions) by imaging ionization tracks in a gaseous medium (He/CF$_4$). The system relies on two heterogeneous data streams:

• Photomultiplier Tubes (PMTs): Provide prompt timing signals on the nanosecond scale.
• Scientific Cameras: Provide high-spatial-resolution images (megapixel scale) with exposure times in the tens to hundreds of milliseconds.

The previous data acquisition system (DAQ), based on the LIME prototype, suffered from critical limitations:

• High Dead Time: The frame-based acquisition scheme resulted in a dead time of approximately 38%. This was caused by the camera's duty cycle (sensor row activation time + exposure time) and the requirement to wait for digitizer readout completion before issuing the next trigger.
• Event Loss: Extended particle tracks could be truncated or entirely lost if they occurred during the camera's inactive intervals (row readout or inter-frame gaps).
• Scalability Issues: The system was designed for a single camera. The upcoming CYGNO-04 phase requires the simultaneous, synchronized operation of multiple cameras (up to six), which the legacy architecture could not support without a master-slave dependency or complex timing synchronization.
• Time-Tagging Ambiguity: The existing 30-bit trigger time tag (TTT) had a limited range (~9.1 seconds), making it insufficient for long-duration runs and difficult to correlate PMT signals with specific camera frames in continuous mode.

2. Methodology and System Design

The authors developed a comprehensive upgrade to the Trigger and Data Acquisition (T-DAQ) system, validated on the MANGO prototype. The upgrade focuses on three main pillars:

A. Continuous Imaging Mode

• Concept: The camera is decoupled from the per-frame trigger logic. Instead of waiting for a software trigger for every frame, a single software trigger initiates a "START TRIGGER" mode at the beginning of a run.
• Implementation: The camera acquires frames continuously (back-to-back). The exposure signal (CAMEXP) is only asserted for the first frame.
• Dead Time Reduction: By eliminating the inter-frame gap and relying on the minimal readout time of a single sensor row ($t_{row} \approx 86.4 \mu s$) rather than the full frame cycle, the dead time fraction is reduced from 38% to **0.03%**.

B. Extended Group Trigger Time Tag (EGTTT)

• Problem: In continuous mode, a single physical event might occur between frames, and the standard TTT resets per frame, causing ambiguity.
• Solution: A firmware upgrade was implemented on the CAEN V1742 digitizer to create a 60-bit time tag.
  • The lower 30 bits correspond to the original TTT.
  • The upper 30 bits extend the timestamp to cover the entire duration of a run (minutes to hours).
• Reference: PMT events are time-stamped relative to the start of the run, ensuring unambiguous temporal association with camera images.

C. Multi-Camera Synchronous Architecture

• Strategy: The system operates without a "master" camera. Any camera can provide synchronization signals, enhancing fault tolerance.
• Hardware Logic:
  • A pulser module generates a periodic clock signal distributed to all cameras via TTL logic.
  • Cameras operate in synchronous readout trigger mode, where the first pulse starts the exposure, and subsequent frames advance based on a predefined number of pulses.
  • A SYNC-IN signal for the PMT digitizer is generated by logically combining a camera-derived exposure signal (defining the rising edge) and a DAQ-server gate signal (defining the falling edge).
• Time Coordination: To align camera and PMT data, the Global Exposure (GE) signal from one camera is converted into a short logic pulse (~100 ns) and routed to the PMT digitizer. This creates a distinct waveform marker for every camera frame within the PMT data stream, allowing precise offline correlation.

3. Key Contributions

1. Continuous Imaging Paradigm: Successfully transitioned the CYGNO DAQ from a frame-triggered to a continuous acquisition mode, drastically reducing dead time.
2. 60-bit Time-Tagging: Developed and validated a firmware solution to extend the dynamic range of trigger time tags, solving the synchronization issue for long runs.
3. Masterless Multi-Camera DAQ: Designed a scalable architecture where multiple cameras (tested up to 3, planned for 6) operate synchronously without a designated master unit, using a distributed clock and logical gating.
4. Unified Time Reference: Established a robust method to correlate optical images with PMT timing data using a shared digitizer and specific logic pulses, ensuring precise event reconstruction.

4. Results and Performance Validation

The upgraded system was rigorously tested on the MANGO prototype:

• Stability: A test run of 1,039 consecutive frames (approx. 6 min 46 s) showed zero skipped or duplicated frames.
• Dead Time: The measured dead time fraction dropped to 0.03%, confirming the theoretical reduction from 38%.
• Timing Precision:
  • Inter-camera timestamp differences showed a spread of ~30 $\mu s$, attributed to intrinsic sensor re-exposure delays.
  • Inter-camera timing jitter was measured at ~10 $\mu s$, consistent with Hamamatsu QUEST 2 camera specifications.
• Synchronization: The system successfully synchronized 2 and 3 cameras. The frame indices increased monotonically, and the SYNC-IN logic successfully generated unique markers for every frame in the PMT data stream.
• Scalability: The architecture is modular and designed to support the full CYGNO-04 configuration (6 cameras).

5. Significance and Implications for CYGNO-04

This work establishes the foundational infrastructure for the CYGNO-04 demonstrator, the next major phase of the experiment.

• Operational Efficiency: The reduction in dead time ensures that low-energy and localized events are not lost, significantly improving the detector's sensitivity.
• Scalability: The multi-camera architecture enables the simultaneous operation of multiple optical readout units, a prerequisite for the larger volume of the CYGNO-04 detector.
• Data Challenges: The authors note that continuous imaging with 6 cameras will generate a raw data rate of ~340 MB/s. They highlight the necessity of developing online data-reduction strategies (e.g., machine learning-based Region of Interest identification) to filter noise and reduce storage requirements, which is currently under development.
• Broader Impact: The solutions presented (continuous imaging, extended time-tagging, and masterless synchronization) provide a scalable model for future large-scale optical TPC detectors in rare-event searches.

In conclusion, the paper demonstrates a successful transition from a prototype-level DAQ to a robust, high-efficiency system capable of supporting the complex, multi-camera requirements of the next generation of CYGNO experiments.