> > Why? Via doesn't have much more inductance than a piece of trace, > > does it? > > At high enough frequencies, traces are waveguides, not just > conductors. Vias have nontrivial geometry relative to the signal.
Indeed. Folks doing > 1GHz design (to name a rough cut-off point) spend a lot of time worrying about the waveguiding properties of their traces. Vias represent impedance discontinuities. Modeling vias is a big industry for signal integrity engineers. Here's a paper about a commerical product used to model a via (amongst other things): http://www.cadence.com/community/allegro/resources/resources_pcbsi/mgh/tp_CDN_Bayside_SingleViaPaper.pdf > That reminds me of another optimization I'd like to implement in the > future. Instead of corners, treat all traces like elastic bands, so > that they end up as sweeping curves (lines+arcs) with no corners at > all (except for 3-trace nodes). I want this myself just because it > makes the board look pretty, but I wonder if gentle curves provide > better waveguide performance than corners? This is also a subject of inquiry amongst SI engineers. At one time, it was believed that you needed semi-circular sections to make a bend for high-speed signals. Nowadays, however, the consensus is that it is enough to just miter the corners of a 90degree bend, or (better) use two 45 degree bends to transition between a 90 degree bend for all but the very highest frequencies. IN the past, microwave engineers took great pains to create semi-circular traces for microwave tracks, but that apparently is not needed. For those interested in SI, you can sign up to read a good mailing list devoted to the subject here: http://www.freelists.org/archives/si-list/ There are a number of real experts reading and posting to that list. Stuart
