[SI-LIST] Re: Stripline reference
- From: Bradley Brim <bradb@xxxxxxxxxxx>
- To: "wucjason@xxxxxxxxx" <wucjason@xxxxxxxxx>
- Date: Sat, 28 Sep 2013 12:36:05 -0700
Hello Jason,
Feng and Scott have great points you should consider when setting up your
simulation. Your original question about port definition seems motivated by
application of dual-referenced (power *and* ground) vs single-referenced
(ground *or* power). The two cases are not different and the same consideration
should be applied for each.
You must ask yourself if at the location of your port it is a reasonable
approximation to place an artificial ideal short circuit among all the points
you apply as local reference. It doesn't matter if your points are in the same
or different nets; you are creating an ideal short among them when you apply
them as a simulation port reference terminal. In your system, would it be a
reasonable approximation to enforce an artificial ideal short circuit between
power and ground at your port locations; as it operates in composite, not just
the piece you are now simulating? If so, then it is okay to do that in your
S-parameter model extraction.
For a more complete modeling you should define two ports for most
dual-reference cases: (1) signal referenced to ground, (2) power referenced to
ground.
Even for a single-reference signal you may change the way return current flows
in your system if you short together top/bottom references to form a simulation
port reference terminal. Scott's suggestion to include the transition into the
dual-referenced portion of your circuit is excellent advice. However far you
remove yourself from your transition you will still need to consider how your
port definition approximates actual return current in the PDN. For example,
shift all the way back to your on-die IO buffer and you still have a concern
with having three terminals: signal, ground and power. Unless your only
available IO buffer model assumes ideal power it would still be a good idea to
have both signal and power ports, each referencing ground.
Best regards,
-Brad
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Scott McMorrow
Sent: Saturday, September 28, 2013 7:06 AM
To: wufengthu@xxxxxxxxx
Cc: wucjason@xxxxxxxxx; si-list@xxxxxxxxxxxxx
Subject: [SI-LIST] Re: Stripline reference
Feng is correct. Where you place your ports determines your experimental
design and whether your results are close to reality or a poor
approximation. To truly model a stripline trace between a power and a
ground plane, you need to model the via transition into that structure.
Since power and ground are not directly connected you then need to realize
that the structure is "non-local", that the EM waves propagate in many
directions, which are defined by the original via transition onto the two
planes. Once you realize the structure is non-local, you have to determine
how large a containing structure is necessary to encompass all of the
propagating EM waves, or at least enough of them to show low error.
To determine the extent of the containing structure you have to find where
all the inductive loops are closed. This occurs partially at the bypass
capacitors, but it also occurs when the power plane is surrounded in the +z
and -z directions by ground planes. Ground stitch vias form the enclosing
structure to contain waves propagating between the power plane and ground
plane.
If the board is not designed well, the containing structure extent can
become quite large, often extending across the entire board, for the case
where the power plane is not covered on both sides by ground planes in the
board stack. But, if there is good containment by ground planes, and
sufficient stitch vias, the extent for simulation purposes can be quite
reasonable.
best regards,
Scott
On Sat, Sep 28, 2013 at 2:47 AM, Feng Wu <wufengthu@xxxxxxxxx> wrote:
> Hi Jason
> Before you start this long time simulation, would you please clarify what
> the purpose of extracting S parameter is ?
> For your case, there are a lot of phenomenon we can observe and study. for
> each phenomenon, we need to consider different method/tool.
> for port setting, it also depends on your purpose. Some phenomenon will be
> excluded or included based on your port location.
> Usually I try to find a point to add the port, before trace enter such a
> tricky stucture.
> As Gert point out, for your case, the trouble comes from the boundary
> design and how to include the passive components to capture the accurate
> return path and cavity resonance, besides port setting.
>
> It is very hard or time consuming to capture all the phenomenon. You need
> to trade them off, based on the purpose.
>
> --
> B.R.
> Feng Wu (吴枫)
>
> -------------
> Signal Integrity Engineer
> Cisco Shanghai
> -------------
> 2013/9/27 Jason wuc <wucjason@xxxxxxxxx>
>
> > Hello Experts,
> > I need your help.
> >
> > I am extracting s-parameter of some strupline traces. These traces
> connect
> > two flip chips on board and are routed between power and ground plane
> where
> > power is upper plane and ground is lower plane I am in doubt that
> whether
> > I should take ground/power as reference or both planes as reference while
> > assigning ports.
> >
> > Please reply.
> >
> > Jason
------------------------------------------------------------------
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:
//www.freelists.org/webpage/si-list
For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field
List forum is accessible at:
http://tech.groups.yahoo.com/group/si-list
List archives are viewable at:
//www.freelists.org/archives/si-list
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts:
