[SI-LIST] Re: Via modeling & de-embedding
- From: "Hassan O. Ali" <hassan@xxxxxxxx>
- To: Ivan Ndip <ndip@xxxxxxxxxxxxxxxxx>
- Date: Fri, 16 Jul 2004 12:17:35 -0400 (EDT)
Ivan,
To know how long is long enough, you need first to understand what a via model
should
accomplish.
Assuming a stripline on either side of the via, a TEM wave gets into the via
and comes
out as a TEM wave. At the via itself, though, the incident signal in TEM mode
gets
converted into several other (higher-order) modes and then back to TEM mode as
it exists
the via. The types of higher order modes depend on the environment around the
via. Some
of the modes are evanescent (non-propagating) and some are propagating modes.
Some of
the higher-order propagating modes have much higher attenuation constants (i.e.
they die
off short distance away). Ultimately, one cares only about the TEM mode that
propagates
far into the striplines on either side of the via.
So, in this case, the via model should essentially capture the mode conversion
at the
via. For that reason, you need to place your reference planes at points where
only the
TEM mode exists (substantially). Unfortunately, the determination of such
points is
problem-dependent. Certainly you shouldn't just use the attenuation constant of
the
first propagating stripline higher order mode to determine the location of such
points.
That's because you also need to be away from the effects of evanescent
(non-propagating)
modes and other modes that propagate through the via in a manner not supported
by
stripline.
For each via configuration, I would suggest you start with very short line
segments
(e.g. of length equal to the pad radius) and increase the length in several
steps to
determine S21 (insertion loss) convergence for the frequency band of interest.
You
should have to fix your own convergence criterion. I would be comfortable with
0.5% S21
covergence.
Regards.
Hassan.
On Jul 16, Ivan Ndip <ndip@xxxxxxxxxxxxxxxxx> wrote:
>
> Hi Hassan,
> How "long enough" must the trace segments be? Do you have any experience
> in determining the exact length?
>
> Best regards,
> Ivan Ndip
>
> Hassan O. Ali wrote:
>
> >It seems that "de-embedding" in this thread has been used to suggest that
> >one can
create
> >an "isolated" via model that can be used in various other environments such
> >as with
> >different trace widths and thicknesses; different signal rates, etc. If that
> >is the
> >case, then I think that suggestion is not accurate enough.
> >
> >I tend to believe that via characteristics (hence its model) cannot be
> >isolated from
the
> >surrounding environments - especially for higher frequencies. The
> >trace/via-pad
> >transition itself presents a discontinuity that contributes to the
> >parasitics of the
> >via. The pads have their own parasitics. Neighboring ground/power vias also
> >can
> >influence the signal via performance. All of those along with the via barrel
parasitics
> >contribute to the overall via characteristics and have to be appropriately
> >considered
in
> >model creation.
> >
> >As to de-embedding, I think you won't get accurate results if your reference
> >(de-
> >embedding) plane is right at the transition. In other words, an accurate via
> >model
> >should include trace segments long enough to ensure that the field at the
> >segment
ends
> >are mainly of the dominant mode of field propagation. In fact most (if not
> >all) de-
> >embedding techniques make that assumption.
> >
> >Regards.
> >
> >Hassan.
> >
> >
> >On Jul 15, David Siadat <dsiadat@xxxxxxxxx> wrote:
> >
> >
> >>Hello Ivan,
> >>
> >>De-embedding the transmission line feeding the via is limited to the
> >>edge of the anti-pad region that the transmission line is referenced to.
> >>That could mean an additional 17.5 mil transmission line for 35mil diameter
> >>anti-pad. The inductance value of this transmission line can be significant
> >>compare with the inductance of the via itself.
> >>
> >>Regards,
> >>David
> >>
------------------------------------------------------------------
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 FAQ wiki page is located at:
http://si-list.org/wiki/wiki.pl?Si-List_FAQ
List technical documents are available at:
http://www.si-list.org
List archives are viewable at:
//www.freelists.org/archives/si-list
or at our remote archives:
http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
http://www.qsl.net/wb6tpu
Other related posts: