Signal- For a diff pair in a PCB, some amount of current does have to flow in the reference ground planes, because each signal line is coupled to the planes to some degree, even though they are also coupled to each other to some degree. This degree of coupling is reflected in the relative magnitude of the even and odd impedances of the diff pair. As coupling between the signals increases and/or coupling between signals and planes decreases, the even mode impedance rises relative to the odd mode impedance. Consider two extreme cases: 1. Two separate coax cables. There will be extremely little coupling (call it zero) between the two signals, as they are shielded from one another by the coax ground traces. Therefore, the even mode impedance will be very nearly (call it exactly) the same as the odd mode impedance, equal to the impedance of the cables (assuming they are the same). 2. Twisted pair cable. In this case, consider ground to be infinitely far away. Then, the even mode impedance will be infinite, but the odd mode impedance will be finite, dictated by the coupling between the signal traces. The diff pair in PCB case will be somewhere in between. One can tune this to some degree by changing the relative spacing of the signal lines to the planes and to each other (within practical limits). When stringing together a system as you describe (PCB - cable - PCB), because the even and odd impedances will typically differ in the different parts of the string, you can only match either the odd or even mode impedances at the interface. In your case, you'd almost certainly design to match the odd mode impedances, but you'd want to be aware of the even mode impedance mismatch, as this will lead to common mode reflections which may or may not impact your system performance. There are a lot of past SI postings on this issue. - Erik ================================================================== Erik Daniel, Ph.D. Voice: (507) 538-5461 Mayo Foundation Fax: (507) 284-9171 200 First Street SW E-mail: daniel.erik@xxxxxxxx Rochester, MN 55905 Web: www.mayo.edu/sppdg/ ================================================================== > -----Original Message----- > From: signal hoss [mailto:simails@xxxxxxxxx] > Sent: Tuesday, July 24, 2001 7:35 PM > To: si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] Re: diff pair questions > > > > Chris, > > I think you made the assumption on the drivers that > return current has to be on the planes. > What about designing a driver to return current on > opposite trace on pcb, so we don't need to care about > reference to planes. A similar example is eithernet > Cat-5 cable, which has no ground refernce. > > The reason I asked is the scenario: > > diff pair with planes refernce on pcb --> coaxial > cable --> other pcb > > From pcb diff pair to cable diff pair, there seems > to have mismatch even though both have same > impendence- > one has reference to planes while the other don't. > > This really puzzled me. Please advise. > > Signal Hoss > > --- Chris Rokusek <crokusek@xxxxxxxxxxxx> wrote: > > > > Signal, > > > > Try thinking about it as two single ended traces. > > Then in terms of return > > current, a large proportion of the return current > > for each diff trace > > actually returns on the planes rather than the > > opposite trace. So what you > > really have in terms of currents is a separate > > signal/return loop for each > > diff half. These two loops are equal and opposite > > but spaced very near each > > other (in fact on the planes they overlap at all but > > the highest frequencies > > causing cancellation, less noise, good stuff). > > Another way to visualize > > this is that the E/H field lines are "contained" by > > the planes. Thus > > impedance will vary as the plane separation is > > varied. > > > > Chris Rokusek > > Innoveda > > > > > __________________________________________________ > Do You Yahoo!? > Make international calls for as low as $.04/minute with > Yahoo! Messenger > http://phonecard.yahoo.com/ > ------------------------------------------------------------------ > To unsubscribe from si-list: > si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field > For help: > si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field > > List archives are viewable at: > //www.freelists.org/archives/si-list > or at our remote archives: > http://groups.yahoo.com/group/si-list/messages > 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 For help: si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field List archives are viewable at: //www.freelists.org/archives/si-list or at our remote archives: http://groups.yahoo.com/group/si-list/messages Old (prior to June 6, 2001) list archives are viewable at: http://www.qsl.net/wb6tpu