Randall; The routing of 1GHz traces has many obstacles. Following are a few (not an exhaustive list)
1) dielectric "constant" value variance 2) cross talk 3) source termination 4) source impedance 4) end-line termination 5) common mode generation 6) component inability to handle 1GHz signals 7) mismatched impedances 8) vias 9) turns and corners of traces 10) PCB material 11) connectors and headers, etc. etc. One advice is to keep the trace on one plane only. No vias, and use rounded turns. For additional signal integrity (less emissions is a benefactor of good S.I.) you will have to use differential transmission lines, i.e. two traces for each signal. Those have to be kept precisely at the same distance apart and driven from differential drivers. The impedance of the transmission line is not so important as long it is held constant. (Unless you are driving external devices) Remember that it is difficult to get high impedances (over 50 Ohm) on a typical multilayer PCB. There are signal integrity programs that do just what you want but they are not cheap! visit the University of Missouri at Rolla web-site (something like UMR.emclab.edu or close to that) You will find a treasure of information there. Hans Mellberg EMC Consultant ______________________________ Reply Separator _________________________________ Subject: PWB Layout for high frequency traces... Author: Non-HP-owner-emc-pstc ([email protected]) at hp-boise,uugw2 List-Post: [email protected] Date: 02/03/97 03:06 PM Hello all, I am looking for a method to determine the minimum distance which I can route 1 GHz traces from other devices (including metal pin row headers) without compromising the EMC performance (characteristic impedance) of the traces, while preventing any common-mode coupling to the other devices. Any formulas, suggestions, or ideas are appreciated. Also, if anyone knows of any EMC layout modelling software which addresses this issue, please let me know. Thank you, Randall T. Flinders Emulex Corporation [email protected]
