Joel, depends very much on your specific board layout (and the degrees of freedon you have inchoosing layers, dimensions etc.). Sometimes there are simple solutions that give reasonable performance. Best (easiest) case would be if your routing density is not too high, and the transition goes from a layer at or at least close to the top surface to a layer at or close to the bottom surface of your board, so the resulting stubs are very short and can be neglected in first order (i.e. most of or all of the via is part of the intended signal path). Of course if there are significant stubs you could still do backdrilling to remove them. Then you could add several vias (4 is a good number that still allows for trace escape from the central signal via) for the return current (i.e. stitching the reference planes together) around that signal via in a quasi-coaxial configuration (i.e. on the four corners of a square centered at the signal via). Next step is to clear all planes in the area between those return vias (and maybe even slightly beyond, except below the signal trace leading to the via). You can now use a simple 2D field solver or even some closed formulas to adjust the distance between signal and return via, and also the via diameter, to obtain 50 Ohm impedance. Obviously that will only give an approximation to a true 50 Ohm structure - it neglects stubs (if the signal traces are not at the surface), fringing fields at the ends of the via, the effect of the other planes (you can't cut out too far), the annular rings of the via (will add capacitance), and finally the trace section close to the via will be inductive (because there is no ground plane). But I recently consulted on such a "quick and dirty" design and at first pass achieved a bandwidth of over 20 GHz through the via, and the structure itself came out to 47 Ohms (I'm not ruling out a certain amount of luck here - don't hold me accountable if your first try doesn't work out as nicely :-) With a very fast TDT I was also able to observe all the parasitic effect mentioned. Also, keep in mind that even with the best 3D field solver and many months of effort your result will only be as good as your input data. In your model you know and control exactly every single parameter. But in practice there are tolerances everywhere - the dielectric constant may be off by a few percent, the traces and planes have etching tolerances, the via may be drilled slightly off-center, etc. etc. So even if your field solver predicts 50.00000 Ohms for a given configuration, the structure in the real board can be off by several Ohms. (and that doesn't even take into account that you may have a mistake in your model setup). So I'd also make a judgement call how much simulation effort / time / cost it is really worth, as opposed to e.g. do a combined approach of simulation + some simple test board(s) to gain confidence and experience. Once your test boards match your simulation I'd be much more confident that my simulation can accurately predict my real, complex design. (and in addition you'll get a better feeling as to how precisely your simulation model can predict different structures). Wolfgang "Joel Brown" <joel@xxxxxxxxxx> Sent by: si-list-bounce@xxxxxxxxxxxxx 01/09/2008 05:51 PM Please respond to joel@xxxxxxxxxx To "'SI LIST'" <si-list@xxxxxxxxxxxxx> cc Subject [SI-LIST] 50 Ohm Via? Is there such a thing as a design methodology for designing a PCB via with 50 ohm impedance, or does it have to be done iteratively using a 3D field solver? Are controlled impedance vias necessary, worthwhile or helpful for multi-gigabit serial links running at 1 to 5 Gbps? Thanks - Joel ------------------------------------------------------------------ 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 technical documents are available at: http://www.si-list.net 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 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 technical documents are available at: http://www.si-list.net 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