Do's and don'ts with the agilent's G-link chipset

  • Authors:
  • Alberto Aloisio;Francesco Cevenini;Vincenzo Izzo

  • Affiliations:
  • I.N.F.N., Sezione di Napoli and Università di Napoli "Federico II", Dipartimento di Scienze Fisiche, Napoli, Italy;I.N.F.N., Sezione di Napoli and Università di Napoli "Federico II", Dipartimento di Scienze Fisiche, Napoli, Italy;I.N.F.N., Sezione di Napoli and Università di Napoli "Federico II", Dipartimento di Scienze Fisiche, Napoli, Italy

  • Venue:
  • RTC'05 Proceedings of the 14th IEEE-NPSS conference on Real time
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

The Agilent's G-Link serializers-deserializers are long lasting and widely used devices to implement serial links in the range of ≃1Gbit/s on copper and fibres. They feature a wide clocking window spanning the 13 to 70 MHz range, the ability to compensate for a constant phase delay in the clock distribution between the Tx and Rx node, low and deterministic link latency. All that makes this old-fashioned, power-hungry bipolar chipset still a popular choice in the design of nowadays trigger and data acquisition systems. The last entry in the family, named HDMP-103xA, has been released in second half of 2001, as a drop-in replacement for the previous HDMP-103x silicon version. Despite the recent upgrade, the latest GLink chip-set still suffers from subtle misbehaviors and undocumented bugs which may jeopardize also a very conservative design. In this paper we report our experience with G-Link during the design and test ofthe optical link for Level-1 Muon Trigger of the ATLAS experiment.