[SI-LIST] Re: Assessing insertion loss profiles
- From: Scott McMorrow <scott@xxxxxxxxxxxxx>
- To: Yuriy Shlepnev <shlepnev@xxxxxxxxxxxxx>
- Date: Sun, 14 Jul 2013 15:45:47 -0400
to add on to the good information that Yuriy has presented.
Rarely do I find a differential link that can be easily analyzed in the
differential-only domain. Packages, vias, and connectors often exhibit
strong common mode components that are difficult to ignore. As soon as I
allow for skew in links, due to transmitter skew, trace mismatch skew, or
laminate weave skew, I have found that the crosstalk induced can often
dominate the link analysis.
Generally, the Scd component of crosstalk is significantly higher than Sdd.
Also, the localization of crosstalk for Scd is much lower than that for
Sdd. (Yuriy and I have written about "localization" in quite a few of our
papers.) As soon as crosstalk becomes "non-localiized" we are in for a
world of hurt. This often is a problem in packages, via fields, and
around, or in, connectors. In these cases, a large enough group of
differential links must be sucked up into the analysis in order to fully
characterize the loss and crosstalk impact.
For some packages that I have seen, this means that an entire quarter of
the package needs to be included in EM extraction and analysis, since
"non-localization" means that crosstalk can literally "skip" across the
package and PCB via fields. Since a quarter of the package needs to be
swept into the analysis, then those same traces need to be extracted from
the PCB, and the connector via fields, in order to have a complete analysis
of all noise sources. As a result, to fully understand all link
impairments, it is necessary to analyze many, many links simultaneously, in
order to have complete confidence in the final results. This results in
very large s-parameter files.
On Sun, Jul 14, 2013 at 12:51 PM, Yuriy Shlepnev <shlepnev@xxxxxxxxxxxxx>wrote:
> Istvan,
>
> Just a couple of thoughts on a link analysis in case if a signal is not
> perfectly differential or link is not balanced.
> The case of un-coupled or loosely coupled pairs is trivial (though if
> requires the localization). Both single-ended and mixed-mode S-parameters
> can be used to estimate the imbalance and its impact on the link
> performance.
> In the case of tightly coupled pairs, I think only the mixed-mode
> S-parameters may be useful to estimate the signal degradation and link
> quality especially in cases if the link is not balanced and signal is not
> perfectly differential. Differential block of the mixed-mode S-parameters
> naturally describes degradation of the useful signal. If generated signal
> is
> a mixture of differential and common mode, then the common mode
> transmission
> and reflection naturally describe the problem and allows properly solve it
> (terminate common mode for instance). If link is not balanced (weave effect
> or imbalanced discontinuities), then transformation from the differential
> to
> common mode block of the mixed-mode S-parameters is useful to pinpoint and
> address the problems. Something similar to app notes #2009_01 and 2009_02
> at
> http://www.simberian.com/AppNotes.php. Description of the bends in tightly
> coupled microstrip link with the single-ended S-parameters cannot be used
> to
> do such analysis. Though, as you can see, the mixed-mode S-parameters
> provide useful information and actually define the link optimization goals.
> Instead of length compensation, the goal becomes minimization of both
> differential reflection and transformation to common mode.
>
> Best regards,
> Yuriy
>
> Yuriy Shlepnev, Ph.D.
> President, Simberian Inc.
> 3030 S Torrey Pines Dr. Las Vegas, NV 89146, USA
> Office +1-702-876-2882; Fax +1-702-482-7903
> Cell +1-206-409-2368
> Skype: Shlepnev
>
> www.simberian.com
> Simbeor - Accurate, Fast, Easy and Affordable Electromagnetic Signal
> Integrity Software
> 2010 and 2011 DesignVision Award Winner
>
> -----Original Message-----
> From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
> On
> Behalf Of Istvan Nagy
> Sent: Saturday, July 13, 2013 10:56 AM
> To: si-list@xxxxxxxxxxxxx
> Subject: [SI-LIST] Assessing insertion loss profiles
>
> Assessing insertion loss profiles
>
> Hi,
>
> Sometimes I have to look at insertion loss profiles (before running a full
> channel simulation with IBIS AMI models) to assess the quality of a channel
> on a PCB.
> Someone says we have to look at the mixed mode Sdd21 for this purpose, but
> I
> think we should look at the S21 single-ended curve. The PCB-EM simulations
> are done with single ended ports.
> The Sdd21 looks a lot better than the S21, resonances look smaller.
> I think Sdd21 would only be indicative if the channel and the return path
> was perfect without discontinuities.
> I have doubts about the signals being perfectly differential at both TX and
> RX buffers. (for example 10GBASE-KR from an FPGA) So, what is the correct
> approach?
>
> Regards,
> Istvan Nagy
>
> ------------------------------------------------------------------
> 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
>
>
> ------------------------------------------------------------------
> 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
>
>
>
--
Scott McMorrow
Teraspeed Consulting Group LLC
16 Stormy Brook Rd
Falmouth, ME 04105
(401) 284-1827 Business
http://www.teraspeed.com
Teraspeed® is the registered service mark of
Teraspeed Consulting Group LLC
------------------------------------------------------------------
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:
