INSTR'20

The use of FPGA in drift chambers for data transfer rate reduction

Not scheduled

20m

Budker Institute of Nuclear Physics

Board: 93

Poster Electronics, Trigger and Data Acquisition

Speaker

Mr Gianluigi Chiarello (INFN Roma1)

Description

The widespread use in drift chambers of light helium-based gas mixtures is aimed at minimizing the multiple scattering contribution to the momentum measurement for low momentum particles. However, because of the limited number of ionization clusters produced, these gas mixtures introduce a substantial bias in the impact parameter estimate, particularly for short impact parameters and small drift cells. Recently, an alternative impact parameter reconstruction technique (Cluster Timing) has been proposed, which consists in using, with statistical considerations, the distribution of the drift times of all individual ionization clusters, to reduce the bias and, consequently, to improve the spatial resolution. An efficient application of Cluster Timing techniques requires converting the sense wire signals from analog to digital, with at least 10 to 12 bits resolutions at sampling frequencies of at least 1-2 GSa/s. These constraints, together with the maximum drift time, of the order of several hundred nanoseconds, and with the large number of acquisition channels, typically of the order of tens of thousand in modern drift chambers at colliders, may require data transfer rates of the order of TB/s and, therefore, impose sizeable data reduction techniques. The amplitudes and the drift times of the peaks associated to the individual ionization clusters identified in the wire signal constitute the essential amount of data to be processed, transferred and stored and, to this purpose, fast readout algorithms, implemented within FPGA’s and executed in real time represent an attractive solution. The CluTim algorithm, described here, identifies in real-time the peaks corresponding to the different ionization cluster, stores each peak amplitude and time in an internal memory and sends the stored data to an external device when a specific trigger signal occurs. Such a quasi-on-line procedure results in data reduction factors of almost two orders of magnitude with respect to the raw digitized data. A hardware test of such an application is illustrated with the details of the applied algorithm.