[SI-LIST] Re: PCB Gnd discontinuity on simulation software

  • From: ruscino <Ettore.Ruscino@xxxxxxxxxx>
  • To: si-list@xxxxxxxxxxxxx, Wolfgang.Maichen@xxxxxxxxxxxx
  • Date: Thu, 08 Mar 2012 12:19:26 +0100

Hi Wolfgang et all, first of all many tanks for your help to my problem.

My circuit is flexible and has this stack-up:
1) kapton
2) top layer (routing layer)
3) kapton
4) copper (gnd layer)
5) kapton.

So the circuit is composed of two parts:
-) a bus, where lvds nets are routed on gnd plane
    (embedded microstrip topology);
-) the wings where the lvds nets go to the connectors.
Unfortunately the wings must be of only one routing layer (top) so
the lvds net topology is no more microstrip. I try to visualize this
with the draw:
       Wing1   Wing2  Wing3 ...
           ||          ||          ||
           ||          ||          ||
           ||          ||          ||
=======          ||          ||
===============        ||
======================         LVDS bus

Sorry if it is unclear.....
The lvds bus is 40 cm length, the wings are 3 cm each.
The lvds signal frequency can vary from 40 Mhz to 160 Mhz
with rise/fall time of about 1-2ns (for the 160Mhz signals).
I certainly will try to simulate this topology with Sonnet Lite but
I ask you if it is possible to simulate it with linesim hyperlynx to
estimate reflection. To know the impedance I  built
two stack-up on two files, one is with ground plane (lvds bus)
and one is without ground plane (kapton - top layer- kapton).
These information could be pass to LineSim where I have:
LVDS driver -> embedded microstrip -> Simple Transmission line-> receiver.
Is it correct?


Il 08/03/2012 08:35, Wolfgang.Maichen@xxxxxxxxxxxx ha scritto:
> Ettore,
> actually a full 3D field solver may be a bit overkill; any decent 2.5D 
> (planar) field solver can give pretty accurate answers for split ground 
> planes, and makes model creation easier for an occasional user. As a rule of 
> thumb, routing single ended lines over a split in the reference plane sets 
> you up for a lot of trouble unless both the frequencies involved are very low 
> and the signal rise times are very long. Things are much better for diff 
> pairs which can tolerate breaks much better as long as the two lines of the 
> pair are not spaced too far from each other (1-3 line widths is usually fine 
> with typical PCB dimensions). But only a field solver can give a good answer 
> for a specific case. One place to start could be the free version of Sonnet, 
> called Sonnet Lite - can actually handle such situations and I've used it 
> successfully myself to simulate split grounds. (I'm not affiliated with 
> Sonnet in any way, I'm just a user of Sonnet Lite). One thing to watch out 
> when simulating
>   _single ended_ lines with such a solver are boundary conditions - e.g. 
> Sonnet encloses the model in a virtual metal box, so if you are not careful 
> this box can short your ref planes together, resulting in unrealistically 
> good performance of such a split.
> Wolfgang

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:

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field

List forum  is accessible at:

List archives are viewable at:     
Old (prior to June 6, 2001) list archives are viewable at:

Other related posts: