P R E F A C E NOTE FROM THE AUTHOR Welcome, and thank you for your interest in High-Speed Signal Propagation: AdvancedBlack Magic. This is an advanced-level reference text for experienced digital designers whowant to press their designs to the upper limits of speed and distance.If you need to transmit faster and further than ever before, this book is here to help.You'll find it packed with practical advice. The material in this book has been honed during my many years of work as chieftechnical editor of standards for both Fast Ethernet and Gigabit Ethernet-;projects which, Ihope, have touched your life in a favorable way. During those and many other projects, themodels and concepts described here have been of invaluable service to me. Now I'd like topass them on to you. When you are done reading, share your knowledge with those around you as mytechnical mentor, Martin Graham, has done with me. Educate your coworkers. Educate yourmanagement. Above all, continue to educate yourself. If this book inspires you to advanceyour understanding with even one laboratory measurement, then I will know you are on theright track. I would also like to say it has been a great pleasure teaching and working with manyof you through my classes and lectures. Above all, I appreciate those who take the time toshare with me their thoughts, their concerns, their dreams, and their problems. It alwaysinterests me to hear about real experiences from real engineers.I wish you the best of luck on your next design. See you on the Internet, Dr. Howard Johnson www.sigcon.com TOPICS COVERED Printed circuit traces:Limits to attainable speed and distanceRC and LC mode propagationSkin effect and dielectric loss design charts and equationsProximity effectSurface roughnessNon-TEM mode of propagationStep responseEffect of vias Differentialsignaling:Edge-coupled and broadside-coupled differential pairsEffect of bendsIntrapair skewDifferential trace geometry impedanceCrosstalkRadiation Inter-cabinetconnections:Coaxial cablesTwisted-pair cablesFiber opticsEqualizersGeneral building wiring for LAN applications Clock distribution:Special requirements for clocksClock repeatersMultidrop clock distributionClock jitterPower filtering for clock sources Simulation:Frequency-domain simulation methodApplicability of Spice and IBIS HOW THIS BOOK IS ORGANIZED Each chapter in this book treats a specialized topic having to do with high-speed signalpropagation. They may be studied in any order. Chapters 1 and 2 present the underlying physical theory of various transmission-lineparameters, including the skin effect, proximity effect, dielectric loss, and surfaceroughness. Chapter 3 develops a generalized frequency-response model common to all conductivemedia. Chapter 4 outlines the calculation of time-domain waveforms from frequency-domaintransfer functions. Chapters 5 through 11 discuss specific transmission media, including single-ended pcbtraces, differential media, general building wiring standards, unshielded twisted-pair wiring,150- shielded twisted-pair wiring, coaxial cables, and fiber. Chapter 12 addresses miscellaneous issues concerning clock distribution. Chapter 13 explores the limitations of Spice and IBIS simulation methods. PREREQUISITES A basic understanding of the frequency domain representation of linear systems is assumed.Readers without the benefit of formal training in analog circuit theory can use and apply theformulas and examples in this book. Readers who have completed a first-year class inintroductory linear circuit theory will comprehend the material at a deeper level. RELATION TO PRIOR BOOKS This book is a companion to the original book by Johnson and Graham, High-Speed DigitalDesign: A Handbook of Black Magic , Prentice-Hall, 1993. The two books may be usedseparately or together. They cover different material. The original book deals with a broad spectrum of high-speed phenomena. It builds asolid understanding of ringing, crosstalk, ground bounce, and power supply noise as theyexist on printed circuit boards. It emphasizes basic circuit configurations where these effectsmay be easily understood and learned. It treats supplementary subjects including chippackages, oscilloscope probe, and power systems for high-speed digital products. This High-Speed Signal Propagation book is more highly specialized, delving intoissues relevant to transmission at the upper limits of speed and distance. If you need totransmit faster and further than ever before, this book shows you how. High-Speed Digital Design and High-Speed Signal Propagation together comprise agood reference set for persons working with high-speed digital technology. Those of you familiar with my other books will recognize similarities in style.Notably, I've tried to impart, as best I can, the same sense of realism born of longexperience. ACKNOWLEDGEMENTS Literally thousands of people have taken the time to communicate with me about high-speedissues, either through email or in person at my seminars. These conversations have inspiredme to investigate and collect together the material in this book. To all of you, I owe a debt ofgratitude. The following people contributed specific comments or questions that are discussed inthe text (in alphabetical order): Sal Aguinaga, James C. Bach, Eric V. Berger, RaymondBullington, Doug Butler, Tim Canales, Bruce Carsten, Code Cubitt, Dave Cuthbert, BillDaskalakis, Martin Graham, Paul Greene, Gary Griffin, Bob Haller, John Lehew, John Lin,Raymond P. Meixner, Craig Miller, Mitch Morey, Dan Nitzan, Bhavesh Patel, Dipak Patel,Jim Rautio, Ravi, Boris Shusterman, Kevin Slattery, Bob Stroupe, Bill Stutz (twice), andFabrizio Zanella. Thanks to all of you for many hours of good correspondence. I especially thank those who volunteered for the difficult task of reviewing the text.This group of intrepid individuals spotted numerous errors and suggested many new topicsfor exploration. They deserve a large measure of credit for helping make this a more usefultext (in alphabetical order): Jacob Ben Ary at Aquanet, Greg Dermer at Easystreet, SteveEms at Lecroy, Alexandre Guterman at Nortel, Valery Kugel at Juniper, Professor WillMoore at Oxford University, Jose Moreira at Agilent, Gopa Parameswaran at Cisco, BobRoss at Mentor Graphics, Bert Simonovich at Nortel, Palani Subbiah at Cypress, and GeoffThompson at Nortel. My editors at Prentice-Hall, Bernard Goodwin, Nicholas Radhuber, and Carol J.Lallier, have contributed their professional expertise (and patience) during the long processrequired to complete this project. Without my dutiful and highly accurate assistant Jennifer Epps this book would nothave been possible. All the articles adapted for publication in this book are reprinted with permission fromEDN magazine, a publication of Reed Business Information, Electronic Design Magazine, adivision of Penton Media, Inc., or PC Design Magazine, a publication of UP Media Group,Inc., as noted in the header of each article, respectively. Bob Ross, Mentor Graphics Corp., past chair of the EIA IBIS Open Forum, wrote afine discussion about the future of IBIS modeling for Chapter 13. Bruce Archambeaultcontributed the article in Chapter 12 about reducing emissions, which I only edited. BradCole and Matt Hudale of Ansoft simulated the capacitance of many via configurations forChapter 5. To Gopa Parameswaran at Cisco, thanks for your simulations of via capacitance,although your data did not appear in the final version of the book. Steve Ems and RobertTalambiras of Lecroy piqued my interest in non-TEM modes of propagation during a visit tomy ranch in October of 2000. Roger Billings of Wideband Corporation deserves mention asthe world record-holder for fastest data conveyed across barbed-wire cabling (Chapter 2).Jeff Sonntag at Accelerant Networks was the first to focus my attention on howbackplane performance changes with temperature (Chapter 5). Thanks also to Jim Tavacoliat Accelerant for sending the cool pictures showing adaptive equalization at work inChapter 3. The discussions I have been privileged to hold with Michael King, Ed Sayre, andDoug Smith have been of enormous value to me. Thank you for your friendship.My technical mentor Dr. Martin Graham of U.C. Berkeley has contributed hisenduring support, encouragement, and technical assistance over the past twenty years, aswell as having been the first to direct my attention toward the general features oftransmission line attenuation and how it varies with frequency. Thank you, Martin. To my wonderful and understanding wife Liz, thank you for taking care of all thedetails of my life so that I could have time to write a book like this. Regardless of the assistance of others, any remaining errors are entirely mine. Information contained in this work has been obtained by Prentice Hallfrom sources believed to be reliable. However, neither Prentice Hall norits author guarantees the accuracy or completeness of any informationpublished herein and neither Prentice Hall nor its authors shall beresponsible for any errors, omission, or damages arising out of thisinformation. NOTATIONAL CONVENTIONS This book uses metric units, except for some common printed-circuit board dimensions,which are denoted in English units. Variables and general function names appear in italics.Constants, enumerators, and specific well-known functions appear in ordinary type (e.g.,f(x) = 1 + sin x). Matrix and vector-field quantities occur in boldface type. CONTACTING THE AUTHOR Should you spot something out of place in the text, or merely wish to discuss the finer pointsof transmission-line theory, I may be reached at my ranch high in the mountains near thetown of Twisp, Washington: howiej@sigcon.com. A great place to keep up with the latest developments in high-speed signaling is myweb site, www.sigcon.com. At that site I maintain a growing collection of articles abouthigh-speed digital phenomena, and information about my schedule of public seminars. Ifyou would like to read even more about signal integrity issues, sign up to receive mynewsletter. An errata page for this...