• DocumentCode
    628671
  • Title

    Pre-emphasis parameter optimization for high speed channels using de-convolution approach

  • Author

    Jifeng Mao ; Chandra, U.

  • Author_Institution
    Dell Force 10, San Jose, CA, USA
  • fYear
    2013
  • fDate
    28-31 May 2013
  • Firstpage
    2069
  • Lastpage
    2074
  • Abstract
    This paper covers a novel technique for quickly determining optimum settings for pre-emphasis parameters used in high speed SerDes links. Time domain de-convolution is used for determining the optimum pre-emphasis settings for the high speed differential channel. The optimization is done in the frequency domain through FFT and method of exhaustion. The results from the analysis are applied on 10Gbps SerDes links of a top-of-rack Ethernet switch. Measurement results from a vector network analyzer and oscilloscope are shown as part of the verification. This paper is especially useful for optimizing the SerDes channels and it helps the engineers tune the transmitter pre-emphasis without heavily depending on AMI models, and can be treated as a complementary method to the conventional simulation techniques using IBIS models. The proposed method can be applied at the early and troubleshooting stage of the high speed link design.
  • Keywords
    deconvolution; equalisers; fast Fourier transforms; frequency-domain analysis; local area networks; optimisation; time-domain analysis; AMI models; FFT; IBIS models; SerDes channels; bit rate 10 Gbit/s; frequency domain; high speed SerDes links; high speed differential channel; method of exhaustion; optimum preemphasis settings; preemphasis parameters; time domain deconvolution; top-of-rack Ethernet switch; transmitter preemphasis; Frequency-domain analysis; Receivers; Scattering parameters; Time-domain analysis; Transfer functions; Transmission line measurements; Transmitters;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Electronic Components and Technology Conference (ECTC), 2013 IEEE 63rd
  • Conference_Location
    Las Vegas, NV
  • ISSN
    0569-5503
  • Print_ISBN
    978-1-4799-0233-0
  • Type

    conf

  • DOI
    10.1109/ECTC.2013.6575864
  • Filename
    6575864