scarp@sonnetusa.com wrote:
I tried out your problem on the Sonnet em Suite simulation tools.
<Sales pitch filter on>
We use an FFT Method of Moments high-frequency electromagnetic analysis=
technique for 3D planar circuits, and structures like this map into our=
software quite well (with or without vias).
<Sales pitch filter off>
I isolated on a single signal trace (S) and it's nearby ground traces
(G) with a cross-section configuration like this:
G S G
_____ _____ _____
-----------------------------------------------
///////////////////////////////////////////////
/////////////kapton dielectric/////////////////
-----------------------------------------------
------
G
Then, assuming a 5-mil dielectric between them (and no large nearby
ground planes) with the kapton dielectric constant of 3.5 I ran a few
simulations at 1.5 GHz (a steady-state stimulus is necessary since we
do the analysis in the frequency domain with a single-tone signal).
I kept the pitch between G and S at 0.8mm, and found a pretty good 50
ohm
system to exist with the following dimensions:
S Trace Width G Trace Width Zo
------------- ------------- --
12 mil (0.3mm) 16 mil (0.4mm) 50 ohms
Other combinations that get close, when you keep S and G the
same width (you can extrapolate to the "perfect value, but
there's probably 2-3% error in the simulation results--I didn't
use a real tight mesh):
S Trace Width G Trace Width Zo
------------- ------------- --
12 mil 12 mil 52 ohms
13 mil 13 mil 49 ohms
14 mil 14 mil 47 ohms
Most of the interaction is between the signal trace and the ground
trace located across from it (through the dielectric--the one located
broadside to it), so changing the combination of these two provides
the greatest change in characteristic impedance (as would shifting
the dielectric constant or the thickness of the kapton).
However, I wonder if your adhesive has the same dielectric constant
as the kapton? If not, it will have an effect on the impedance of
the system as well.
By the way, with the computed effective dielectric constant (~2.7)
for this configuration, the wavelength is ~180mm at 1 GHz, and
~18mm at 10 GHz. Your FPC (2 inches) is about 50.8mm, which may be
"electrically significant" for some harmonics of your high rise time
signals?
Hope this helps.
--Shawn
Sonnet Software, Inc.
Roland F. Portman wrote:
>
> I am designing a two layer Flexible Printed Circuit as shown below.
> A ground signal is to the left and right of each signal on the same
> layer,
> and a ground is beneath (above) the signal on the bottom (top) layer.=
> There are 40 signals on the top and bottom layers.
=