If you look at a field effect set of results, test the following:
set up a model with a differential pair and vary the spacing between
them. At the same time, vary the height above the plane.
What you will see is that the closer the differential traces are to each
other, the less influence the underlying plane has. So to achieve what
you are looking for, route the traces as close as popssible, set the
underlying plane relatively far away, the the calculated impedance will
be pretty close to your design.
Two warnings:
Keep other traces (including the "reference") (relatively) far away from
the differential pair (so there is no unnecessary coupling to them).
A model like this will be more susceptible to offsets and reflections
(whether caused by the components or unequal routing lengths) without
the help of an underlying plane.
Doug Brooks
Gary Giust wrote:
When laying out microstrip 50 Ohm single-ended or 100 Ohm differential-- ********************************************
traces, it is OK to use the same layer as the reference for targeting
the trace impedance (if the laminate is very thick between adjacent
signal layers in a 2-layer stack up, e.g. 55 mils)? The signal traces
are routed next to ground on the same layer. I've only seen online PCB
calculators with a plane below the signal trace(s). Are there any online
calculators to compute the impedance when the ground reference is on the
same layer?
------------------------------------------------------------------
