Speaker
Description
Future high-energy physics (HEP) and nuclear physics (NP) experiments will depend on increasingly granular, low-mass tracking and vertex detectors to achieve unprecedented spatial and temporal resolution. This segmentation trend imposes strict requirements on readout integrated circuits (ROICs): amplification, filtering, amplitude and timing extraction, and higher-level feature analysis must fit within minimal silicon area, operate with ultra-low power dissipation, and minimize interference.
Traditional frame-based or polling-based readout architectures introduce dead time, waste bandwidth on empty or redundant data, and increase power consumption—limiting their scalability for next-generation systems. The also disallow intrinsic timing resolution that instead of nanoseconds is shifted into microseconds scale in the frame-based readout systems.
Event-Driven Readout (EDR) architectures address these challenges by transmitting only zero-suppressed, time-stamped event data when needed. This eliminates idle bandwidth use, prevents collisions through dynamic arbitration, and ensures latency-free operation. The concept mirrors the biological efficiency of the human eye, which transmits only essential encoded signals from the retina to the brain for parallel processing.
Building on the Segmented, Ionizing Radiation with Event-Notified Acquisition (SIRENA) detector platform and its patented EDWARD (Event-Driven with Access and Reset Decoder) protocol, we present a scalable, collision-free architecture supporting both greedy and non-greedy arbitration schemes. EDWARD requires no pixel-level geo-priority, remains inactive in the absence of events, and preserves data integrity even under high hit rates. Its selective, on-demand readout substantially reduces the power–bandwidth footprint while enabling deployment in large-area detectors.
The architecture supports a range of Address-Event Representation (AER) schemes—from framed readouts to frameless, neuromorphic-inspired designs—allowing adaptation across collider detectors, spectroscopic X-ray systems, nuclear safeguards instrumentation, and distributed radiation monitoring.
We will trace the evolution of readout methods from early X–Y coordinate scanning and token-passing arbitration (e.g., token rings in current LHC detectors) to dynamic frameless implementations. The discussion will integrate lessons from photon science applications and Monolithic Active Pixel Sensor (MAPS)-based collider detectors, where scaling and efficiency demands mirror those of upcoming HEP and NP experiments.
Key advantages of this approach include:
Eliminated continuous clock or strobe distribution, activating readout only in response to actual radiation-induced events.
No built-in pixel-level geo-priority, and zero activity in the absence of events.
Simplified CAD/EDA integration, leveraging standard design flows and industry-proven tools.
Reduced infrastructure overhead, with minimized clock distribution and peripheral routing.
Compatibility with AI-driven, near-sensor processing, enabling real-time pattern recognition and adaptive data reduction.
These attributes make event-driven architectures a critical enabler for the DOE mission in HEP and NP. They also provide a technology transfer bridge to commercial applications in radiation detection, nuclear monitoring, and precision materials characterization. By aligning detector innovation with scalable, low-power electronics, EDR not only advances experimental capabilities but also accelerates the path from lab-developed intellectual property to industrial adoption.
The presentation will be illustrated by multiple animations allowing to absorb better the developed concept.
