Crosstalk at Right Angles

HIGH-SPEED DIGITAL DESIGN - online newsletter -
Vol. 2 Issue 6

*---------------------(QUESTION)---------------------*

CROSSTALK AT RIGHT ANGLES

John Krawczak writes:

I own a copy of your most useful book and I have a crosstalk question: In a multilayer PC board the rule-of-thumb is that adjacent layers should be routed at right angles. But that is not quantitative. Suppose I am worried about the crosstalk from the right angle intersections and don't want to take the step of adding more ground planes. Is there any insight or reference literature that might give a good estimate of the crosstalk for that situation?

*----------------(DR. JOHNSON REPLIES)---------------*

Thanks for your interest in High-Speed Digital Design.

When two traces meet at right angles the magnetic field coupling (mutual inductance) is completely nulled out. That's because the magnetic field lines from the aggressor trace run perpendicular to the aggressor trace, and therefore parallel to the victim trace. Magnetic field lines parallel to a trace induce no voltages. All that is left is a small amount of capacitive coupling, in the region where the two traces approach each other. I'll now make some crude assumptions to answer your questions. First, we know that the capacitance per inch of a typical 50-ohm stripline trace (measured trace to ground) is going to be in the range of 3- pf per inch.

Now, as a second assumption, we will assume that when one trace crosses another trace I would expect less than that measured on a PER INCH basis across the area of the overlap. If the traces were 0.01 inch wide, I would then expect less than 0.03 pF of coupling at each intersection. At a risetime of 500 ps (corresponding to a frequency of 1 GHz) the impedance of 0.03pf is equal to 1/(2*pi*f*C) = 1/(2*3.14*(1GHz)*(0.03pf)) = 5,305 ohms.

Working into a fifty-ohm circuit, a 5,000-ohm parasitic coupling would induce crosstalk of 1 percent. This is a very crude estimate, it's probably less than this value because (1) most of the electric fields are still terminating on the nearby reference planes, not the victim trace, and (2) the impedance argument, and the effective frequency argument, both tend to overestimate the coupling.

Set up a Spice simulation with a 0.03 pF coupling capacitance to get a better idea of what will really happen.

Best Regards,
Dr. Howard Johnson

To Subscribe to this HSDD Newsletter, send an email to hsdd-request@freelists.org with 'subscribe' in the subject field.

All Publications by Dr. Howard Johnson except as noted.
Signal Integrity Training Classes taught exclusively by Dr. Howard Johnson - for full schedule, see www.sigcon.com
© 1993-2007 Signal Consulting, Inc. All rights reserved. • site map