Returning to this subject, I just posted some ideas on how to formally
compare S-parameters for automation of conclusions on the analysis to
measurement correlation with practical cases at LinkedIn:
https://www.linkedin.com/posts/yuriy-shlepnev-7b39136_signalabrintegrity-ele
ctromagnetics-simbeor-activity-6802703023052328960-cKxz
If it is not clickable - un-wrapped the link.
It is also available as #2021_05 (long version with code snippets) and
2021_06 (short version explaining the idea) at
https://www.simberian.com/AppNotes.php ;
Give it a try on your cases to see how it works.
As usual, I will appreciate any feedback.
Best regards,
Yuriy
Yuriy Shlepnev, Ph.D.
President, Simberian Inc.
www.simberian.com
Simbeor THz - Accurate, Productive and Cost-Effective Electromagnetic Signal
Integrity Software to Design Predictable Interconnects!
Simbeor SDK - The Industry-First Signal Integrity Analysis Automation Kit!
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Yuriy Shlepnev
Sent: Monday, April 26, 2021 3:47 PM
To: robertkapoor81@xxxxxxxxx
Cc: 'SI-LIST'
Subject: [SI-LIST] Re: Comparing s-parameter files - using single value
Robert,
Cases like one with the offset resonances spoil the approach with the formal
metrics based on the vector norms :-) I guess authors of FSV tried a lot of
things before they came up with the FSV idea.
Though, you can always try something that will work for a particular case -
just make sure that your metric satisfy the norm definition
(https://en.wikipedia.org/wiki/Norm_(mathematics)). Otherwise you end up
with false positive or false negative outcomes due to the metric failure.
Best regards,
Yuriy
Yuriy Shlepnev, Ph.D.
President, Simberian Inc.
www.simberian.com
Simbeor THz - Accurate, Productive and Cost-Effective Electromagnetic Signal
Integrity Software to Design Predictable Interconnects!
Simbeor SDK - The Industry-First Signal Integrity Analysis Automation Kit!
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On
Behalf Of Robert Kapoor
Sent: Monday, April 26, 2021 2:07 PM
To: shlepnev@xxxxxxxxxxxxx
Cc: SI-LIST
Subject: [SI-LIST] Re: Comparing s-parameter files - using single value
Thanks Yuriy and Al, the idea is to compare s-parameter files
extracted from the same domain either simulation only or measurement only.
No test card or reference trace de-embedding needed. In simulation only
case getting most info would be useful, L0,L1.. vector as you mentioned.
I'll try to see how this would be implemented in python or other script to
create a stand alone tool.
If there is a shift in resonant frequencies say in S11 combing comparison,
even though the two s-para files are extracted with same sampling points,
any suggestions on how to include resonance shift tolerance? I was thinking
an additional variable declaration with +/- tolerance limits you could
specify which would pass the phase shift and give one number as though the
two files were in phase. I know its not SI specific question, but will
appreciate if anyone has worked on similar situation.
On Mon, Apr 26, 2021 at 10:59 AM Yuriy Shlepnev <shlepnev@xxxxxxxxxxxxx>
wrote:
Hi Robert,And
I see many answers to your questions already - here are my 2 bits on
that:-)
Each element of S-matrix may be treated as a complex vector (a set of
complex numbers).
If 2 S-matrices are sampled at the same frequency points (N points for
instance) you have 2 complex vectors to compare. Simple difference of 2
vectors and use of L2 norm for instance (as Tramp suggested) or any other
existing norms for complex vectors would be one way to have a one number
metric. Use of difference of magnitudes or difference of angles for
instance may be also useful for a particular purpose. In addition to L2,
L0,
L1, or Linfinity vector norms can be also useful for a particular purpose.
Those methods are useful for optimization or for investigation of
convergence of algorithm for instance.
However, it may be useless for comparison of simulations with the
measurements (try it and you will see). That is why the feature selective
validation (mentioned by Gert) was invented - to have a formal conclusion
on
closeness of measured and simulated data (or 2 models or measurements).
the closeness metric is based on the experts evaluation. FSV washttps://www.simberian.com/Presentations/SLIDES_14_40GHzPCBInterconnectValida
standardized by IEEE.
Another approach is simple plotting of S-parameters magnitudes, phases or
delays and corresponding time domain response matrix elements for the
visual
inspection of the closeness or defects. That is what I usually use for my
"objective conclusions" - see examples for CMP-28
https://www.simberian.com/Presentations/CMP-28_Simbeor_Kit_Guide.pdf
Or for EvR-1
tion_Marin.pdf<https://www.simberian.com/Presentations/SLIDES_14_40GHzPCBInterconnectValid
columns.
Best regards,
Yuriy
Yuriy Shlepnev, Ph.D.
President, Simberian Inc.
www.simberian.com
Simbeor THz - Accurate, Productive and Cost-Effective Electromagnetic
Signal
Integrity Software to Design Predictable Interconnects!
Simbeor SDK - The Industry-First Signal Integrity Analysis Automation Kit!
-----Original Message-----
From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx]
On
Behalf Of talented.libran@xxxxxxxxx
Sent: Wednesday, April 21, 2021 4:27 PM
To: SI-LIST
Subject: [SI-LIST] Comparing s-parameter files - using single value
Hello SI Listers,
I am trying to compare two s-parameter files. If we want to produce a
single value like a floating point number to indicate how closely the two
sNp files match or don't match. Would there be a way to calculate this
number?
I was thinking maybe Excel to plot s-parameter touchstone sNp file? There
is a tab delimited option to view sections of the file in separate
But I found out plotting SDD21 and SDD11 not straight forward. There is agives
regression option in Excel under Data Analysis which could 'predict'
relationships between independent and dependent variables.
Or would I need to go with single value metrics like ERL or ICN which
out a single value which might need matlab. Let me know your approach.
Thanks
Robert
------------------------------------------------------------------
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
