I've had the pleasure of working with various IPC technical committees, both design- and performance-based, for the past two years now. Within that short stretch of time, I've been able to isolate several buzz words that have proven to be such harbingers of heated discussions that one would think they had come with lifetime guarantees attached to them. One set of buzz words that has caught the interest of anyone with an interest in high-speed transmission lines is "impedance models."
If you've had the opportunity to sit in on an IPC working session for any of the high-speed, high-frequency documents, then you've probably noticed that work being done these days is on impedance modeling for striplines, dual striplines and embedded microstrips. Currently, these formulas are used in four different IPC standards in one fashion or another. A new goal is to find them a single home in the future successor to the IPC-D-317A, Design Guidelines for Electronic Packaging Utilizing High-Speed Techniques. That document will keep the same name, but will have a new number: IPC-2251.
You may recall a previous article concerning the accuracy of these IPC equations in the February 2000 issue of Printed Circuit Design. Doug Brooks, owner of UltraCAD Design, indicated that the IPC equation for embedded microstrip is an "approximation at best" and that the results are "highly sensitive to the values of the variables." Doug's article was discussed during the IPC Printed Circuits Expo 2000 meetings last April.
Instead of forming a rebuttal to Doug's argument, the committee concluded that Doug was absolutely correct. The IPC equation for embedded microstrip, as well as for stripline and dual stripline, is accurate only within a certain range of dielectrics. Users familiar with the equations in the document have been advising others to use them as guidelines and then ask the board fabricator if more accurate results are needed. I was impressed by the level of cooperation among the committee attendees, despite the nature of the discussion, and quickly chided myself for initially thinking of installing metal detectors in the doorway.
So what does IPC intend to do in light of this revelation? First off, we've been able to obtain a new set of equations for the three conductor geometries mentioned above. Rick Hartley of Applied Innovation provided IPC with a set of equations he obtained from a board house. These equations have drawn praise from others within the industry for having improved accuracy over the existing IPC-D-317A equations. Armed with these new equations, the IPC committee members were determined to test both old and new against boards of known impedance. Evaluations were then processed through the early spring.
The evaluations' results were unveiled at IPC Printed Circuits Expo 2001 during the D-20 High Speed and High Frequency Committee Forum. It was revealed that the three models submitted by Hartley formed a model that was much tighter and closer to actual test results. For example, measured vs. modeled statistics for dual striplines showed that among nearly 1,300 samples, the Hartley equations accounted for an average difference of 1.78 ohms, while the IPC-D-317 model had an average difference of 7.12 ohms. It was noted that there were no changes in equipment or processes across the life of the measurements or readings taken. See Figure 1 through Figure 3 for the evaluation summaries.
The end result will be a new IPC design document that contains enhanced impedance models covering a broader range of dielectrics and trace widths, and therefore a broader range of accuracy. Will these be very high-end equations with many terms that net extreme accuracy?
No. The dielectric ranges and trace widths for which the current models are accurate will be shown, and it will be noted that if an impedance calculation is sought for traces or dielectrics outside that range, a more robust set of equations or a software tool that yields a more accurate answer should be used.
Hartley has summed up the goal of this project succinctly. First off, the inclusion of the models in the standard will illustrate the value in the determination of a transmission line impedance. And it will provide an avenue for people who do not have access to high-end software or field solvers to determine impedance of transmission lines over the range of dielectrics and trace width commonly found in present-day digital designs.
IPC would like to thank Steve Zimmer of Sanmina Corp. for contributing to the evaluations of the impedance model proposals. Zimmer, a quality control engineer at the Owego, NY, facility, also provided IPC with the evaluation presentation given at IPC Printed Circuits Expo 2001. Our thanks to Steve for the enormous time and energy put into the evaluation process.
The new set of impedance models under proposal can be obtained through copies of the IPC-2251 draft, Design Guide for Electronic Packaging Utilizing High-Speed Techniques. An electronic PDF version of this draft is accessible through www.ipc.org/html/techdocs.htm.
John Perry is a technical project manager at IPC. He can be reached at 847-790-5318, or e-mail firstname.lastname@example.org.
© 2001 CMP Media LLC.
7/1/01, Issue # 1807, page 38.
Reproduction and distribution of material appearing in PCD and on the www.pcdmag.com Web site is forbidden without written permission from the editor.
Contact information for Reprints.
NOTE: The articles presented here contain only the text originally published in Printed Circuit Design magazine. Any accompanying graphics and illustrations have not been recreated here. You may view the article in its entirety in each printed issue of PCD.