Jihong Ren, Senior Member of Technical Staff, Rambus, Inc.
Haechang Lee, Senior Member of Technical Staff, Rambus, Inc.
Ruwan Ratnayake, Ph.D. Student, Engineering Intern, Harvard University
Brian Leibowitz, Senior Member of Technical Staff, Rambus, Inc.
Qi Lin, Senior Member of Technical Staff, Rambus, Inc.
Kyle Kelley, Ph.D. Student, Stanford University
Dan Oh, Engineering Manager, Rambus, Inc.
Vladimir Stojanovic, Assistant Professor, MIT
Jared Zerbe, Technical Director, Rambus, Inc.
Nhat M. Nguyen, Senior Design Manager, Rambus, Inc.

To achieve high data rates over backplane channels, sophisticated equalization techniques must be used. Conventional equalization techniques focus on improving voltage margin at the eye center. Recently, edge-based equalization techniques, which minimize voltage noise at data transitions, have been proposed as an alternative to data-based equalization. This paper compares edge-based equalization techniques proposed by Texas Instruments at DesignCon 2005 and Xilinx at DesignCon 2004 with conventional data-based equalization for high-speed serial link applications. Our simulation and lab data show that for most of the channels we examined, data-based equalization gives better performance than edge-based equalization. Finally, we will also investigate other possible edge-based equalization techniques.