In addition, even if everything is perfectly matched, cancellation only works if the fields are identical. This is only possible when all aggressors are infinitely far away. In reality an aggressor running parallel to the pair will be significantly closer to one signal in the pair than to the other, so the crosstalk induced will be worse in the nearest signal and cancellation will not take place. Either keep a wide spacing between the pair and any possible aggressors, or better, put a GND plane under the pair and a GND track/fill between the aggressors and the differential pair, making sure it is well stitched to the plane below. Do this on both sides of the pair and this then makes it a differential coplanar waveguide. Better still, add another GND plane above the diff pair and make sure all GNDs are stitched through. Regards Jonathan -----Original Message----- From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On Behalf Of Zabinski, Patrick Sent: 16 March 2012 12:21 To: krishna@xxxxxxxxxxxxxxxxxxx; si-list@xxxxxxxxxxxxx Subject: [SI-LIST] Re: Differential Traces and GND planes > Differential traces perfectly cancel out the noise induced on the trace > provided the length of the two lines are same and are closely tied up > all along with the controlled impedance maintained throughout the > routing. This is true assuming ideal conditions. However, these conditions never exist. For example, even if the true and complement signals are identical in physical length, weave effect and other proximity-based features will instill different velocities between the traces. As a result, the true signal will arrive sooner (or later) than the complement signal, which subsequently creates common-mode signals. Also, it is incredibly challenging to ensure the true and complement signals are identical in physical length. Bends/corners, vias, and breakouts from packages are difficult to manage in board design to exact levels. Plus, the physical length of the interconnect within the package is often unknown to the engineer, thus it cannot be exactly compensated for. Beyond the interconnect, transmitters and receivers typically are non-ideal as well. That is, they inherently inject common-mode signal that cannot be overlooked. In brief, differential signals and routing help along the lines you mention, but they are not perfect. Instead, engineers must continue to consider the common mode portion that is inherent - whether intentional or not. Based on this, I encourage the use of solid planes for differential traces. Having splits in planes will create reflection, discontinuity, and EMI/EMC issues. The magnitude of these issues might be smaller than when using traditional single-ended signals, but the issues will exist and need to be carefully considered. $0.02, Pat Zabinski ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List forum is accessible at: http://tech.groups.yahoo.com/group/si-list List archives are viewable at: //www.freelists.org/archives/si-list Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu ------------------------------------------------------------------ To unsubscribe from si-list: si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field or to administer your membership from a web page, go to: //www.freelists.org/webpage/si-list For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List forum is accessible at: http://tech.groups.yahoo.com/group/si-list List archives are viewable at: //www.freelists.org/archives/si-list Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu