Erlangen 2026 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

HK: Fachverband Physik der Hadronen und Kerne

HK 35: Instrumentation VI

HK 35.3: Talk

Thursday, March 19, 2026, 14:30–14:45, PHIL A 301

High throughput cluster finding on the readout FPGA for the CBM-TRD — •David Schledt for the CBM collaboration — Goethe University Frankfurt am Main, Germany

The CBM experiment at FAIR/GSI in Darmstadt, will measure rare diagnostic probes of the QCD phase diagram at interaction rates of up to 10 MHz with high sensitivity and statistics. The CBM DAQ is based on self-triggered readout electronics resulting in large amount of data, which has to be processed in real time. The SPADIC readout ASIC transmits the full pulse shape information and features forced neighbor readout. Therefore, the initially produced data volume is relatively high, which lends itself for processing on the readout FPGA. The first processing step to reduce the data volume is to extract the information encoded in the pulse shape, ie. the energy and time. The next step to further reduce the data volume before the online processing is to combine adjacent channel hits into clusters. As CBM runs without a trigger system the hits need to be grouped in time and space in real time. Therefore, a fast high throughput cluster finding algorithm is necessary to process the data without data losses. This work will present how such an algorithm can be implemented on the readout FPGA, with a particular focus on the implementation with high-level-synthesis (HLS).

Supported by: German BMFTR-grants 05P24PM1 and 05P24RF2

Keywords: DAQ; FPGA; HLS; data processing