CircuiTree - July 2008 - (Page 29)

[ Intelligent Design ] By Lee W. Ritchey An Update on the Effects of Laminate Weave on Jitter and Skew in Gigabit and Higher Signals n my April 2007 column I reported on the effects of laminate weaves on the quality of gigabit per second and higher signals. There had been several reports of laminate weave causing system failures due to the lack of uniform distribution of glass in the woven reinforcements used to manufacture laminates and prepregs. The failures were jitter and skew in differential pairs. Figure 1 is a photo of 1080 glass style with a 3.5 mil wire running across it. This wire is approximately the size of PCB traces that are used to conduct signals on a high-speed PCB. As can be seen, the wire is much narrower than the pitch of the weave as well as the width of the threads and gaps between them. This results in impedance variations along the length of the trace as shown in Figure 2. The center line is 50 ohms. The solid line above it is +10 percent and the dotted line below it is -10 percent. As can be seen, the impedance varies widely along the length of the trace as the trace ﬁrst travels over a glass ﬁber and then travels between glass ﬁbers. A study by McMorrow and team intended to solve the problem. A test PCB was constructed that had traces routed at angles from horizontal to vertical in increments of 10 degrees. The hope was to ﬁnd an angle at which the effects mentioned previously were averaged out. The results of the study showed that routing traces at an angle of about 15 degrees to the warp or woof yarns produces the desired results. A patent was applied for covering routing traces at such an angle and fabricators were asked to build PCBs rotated at this angle to the weave. It is not hard to imagine what the reaction to this request was in terms of wasted materials and increased processing costs. In my April 2007 article, I showed a much simpler solution to this problem. It was to use a ﬂat weave with even distribution of glass— in this case, 3313. The impedance variation was shown to be eliminated and, along with it, the jitter and skew problems that it caused. I had also stated that there might be other weaves that could result in the same kind of improvement but that I had not had time to verify this. Since then, we have constructed PCBs from other glass styles, notably 2113 and 3070, with similar results. It is reasonable to believe that any weave that has ﬂat ﬁbers that result in a uniform distribution of glass across the entire surface as shown in Figure 3 will produce similar favorable results. Moreover, the only two commonly used weaves that do not have uniformly distributed glass are 106 and 1080 (see page 109 of my handbook Right the First Time, a Practical Handbook on High Speed PCB and System Design, Volume 2, for photos of these two weaves and other commonly used glass styles). The loss of 106 and 1080 cloth means that 2.5 and 3 mil (0.63 and 0.076 mm) cores are not available for use in highperformance PCBs. This problem has been partially addressed by Compunetics Corporation, who introduced a novel weave called Novaspeed 1080. This material has the uniform distribution of glass that is necessary for high-speed signaling with 3 mil cores. Figure 3 3313 Glass Weave With 3.5 mil Wire Conclusion Lack of uniform distribution of glass in certain cloths used in the manufacture of PCB laminates can result in severe degradation of high-speed differential signals. The two weaves that are known to cause this problem are 106 and 1080 glass styles. Fortunately, there are adequate choices of uniform weave glass styles to allow construction of PCBs with signals as fast as 10 Gb/S without resorting to unusual materials or costly manufacturing techniques. ■ References Bogatin, E., “Skewering Skew, Laminate Weave Induces Skew,” Printed Circuit Design, April 2005. Dudek, R., et al., “Advanced Glass Reinforcement Technology for Improved Signal Integrity,” Printed Circuit Design and Fab, Feb. 2008. McMorrow, S., et al., “Impact of PCB Laminate Weave on Electrical Performance,” DesignCon. Ritchey, L. W., et al., Right the First Time, a Practical Handbook on High Speed PCB and System Design, Volume 1, Speeding Edge, 2003. Ritchey, L. W., et al., Right the First Time, a Practical Handbook on High Speed PCB and System Design, Volume 2, Speeding Edge, 2007. Figure 1 1080 Glass With 3.5 mil Wire Lee W. Ritchey is currently president of Speeding Edge, a leading training and consulting company specializing in the design of high speed PCBs and systems. He has spent his 40-year career designing high-speed PCBs for supercomputers and high-performance Internet products. Email: leeritchey@earthlink.net circuitree.com • July 2008 29 Figure 2 Impedance Versus Length of 50 ohm Trace on 1080 Glass http://circuitree.com

Table of Contents Feed for the Digital Edition of CircuiTree - July 2008

CircuiTree - July 2008 Contents My Line Industry Review Tech Talk Flexible Thinking Toward a PCB Production Floor Metric for Go/No Go Testing of Lossy High-Speed Transmission Lines Intelligent Design 20-Year Retrospective Ask the Flexperts Environmentally Speaking BPA Growth Curves Considering Design Variants to Maximize Process Efficiency Market Outlook Technical Product Spotlights Classified Ads Upcoming Events Ad Index

CircuiTree - July 2008

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.