General purpose readout board π LUP: overview and results
Authors:
Nico Giangiacomi,
Giuseppe Gebbia,
Fabrizio Alfonsi,
Gabriele d'Amen,
Gabriele Balbi,
Davide Falchieri,
Alessandro Gabrielli,
Giuliano Pellegrini,
Riccardo Travaglini
Abstract:
This work gives an overview of the PCI-Express board $π$LUP, focusing on the motivation that led to its development, the technological choices adopted and its performance. The $π$LUP card was designed by INFN and University of Bologna as a readout interface candidate to be used after the Phase-II upgrade of the Pixel Detector of the ATLAS and CMS experiments at LHC. The same team in Bologna is als…
▽ More
This work gives an overview of the PCI-Express board $π$LUP, focusing on the motivation that led to its development, the technological choices adopted and its performance. The $π$LUP card was designed by INFN and University of Bologna as a readout interface candidate to be used after the Phase-II upgrade of the Pixel Detector of the ATLAS and CMS experiments at LHC. The same team in Bologna is also responsible for the design and commissioning of the ReadOut Driver (ROD) board - currently implemented in all the four layers of the ATLAS Pixel Detector (Insertable B-Layer, B-Layer, Layer-1 and Layer-2) - and acquired in the past years expertise on the ATLAS readout chain and the problematics arising in such experiments. Although the $π$LUP was designed to fulfill a specific task, it is highly versatile and might fit a wide variety of applications, some of which will be discussed in this work. Two 7$^{th}$-generation Xilinx FPGAs are mounted on the board: a Zynq-7 with an embedded dual core ARM Processor and a Kintex-7. The latter features sixteen 12.5$\,$Gbps transceivers, allowing the board to interface easily to any other electronic board, either electrically and/or optically, at the current bandwidth of the experiments for LHC. Many data-transmission protocols have been tested at different speeds, results will be discussed later in this work. Two batches of $π$LUP boards have been fabricated and tested, two boards in the first batch (version 1.0) and four boards in the second batch (version 1.1), encapsulating all the patches and improvements required by the first version.
△ Less
Submitted 22 June, 2018;
originally announced June 2018.
