Thanks Jason. I recall Bruce Archambeault of IBM referencing FSV in a DesignCon presentation 4 or 5 years ago. As I understood it at the time, FSV seemed like a very useful method to replicate the way engineers visually analyze data (like what I describe as the eyeball method, but more repeatable/consistent). Are you aware if anyone has made an implementation of this method generally available? It would be cool to perform a comparison between what folks use... Thanks again, -Brett -----Original Message----- From: Jason.R.Miller@xxxxxxx [mailto:Jason.R.Miller@xxxxxxx] Sent: Monday, January 25, 2010 10:08 AM To: Grossman, Brett Cc: si-list@xxxxxxxxxxxxx Subject: Re: [SI-LIST] model to measurement correlation - (taken from Re: 2D vs 3D EM based signal integrity simulators) Feature Selective Validation (FSV) is a good technique for quantifying the data comparison between simulated and measured s-parameters: https://www.dora.dmu.ac.uk/xmlui/handle/2086/279 Jason Miller Grossman, Brett wrote: > For my own curiosity... > I've seen the term 'correlation' used a lot when considering whether > s-parameter simulation results (a.k.a. model results) agree with > measurements. I am curious as to what people on this list generally consider > as correlation. > > To achieve a measure of 'correlation', what I've personally observed being > applied much of the time is a technique we call "The Eyeball Method." Use of > this technique is often characterized by the user making some statement like > "you can 'see*' that the measured and simulated insertion loss curves are > right on top of each other." I'd say in the majority of papers I review this > is the method employed. > > [*hence the reason we call it the eyeball method] > > Probably the next most common method I've seen is to take a scalar difference > between a measured and simulated response, and allow the residual to be the > measure of agreement. This is probably the second most used method in papers > I review, and it is a very distant second (IMHO). > > We presented a method based on EVM a couple years ago which maintains the > vector nature of the data to a degree. It was meant to compare a simulation > to a distribution of measurements, but has also been applied to one > measurement vs. one simulation comparisons. > > There are other methods we've described in past papers, and I believe that > all of these methods have their place. I don't think I could describe them > as well as I can if I hadn't used them all myself, so I don't feel as if I am > picking on anyone by asking a question. > > What I am curious about from this list is: > > > 1. What method do you use to quantitatively describe the agreement between > simulated and measured s-parameters? > > Any comments? > > Thanks, > -Brett > Brett Grossman > Sr. Staff SI Engineer > Signal Integrity Pathfinding - Sort Test Technology Development > [cid:image001.gif@01CA9DA3.2E08F710]<http://www.linkedin.com/in/brettgrossman> > > > > ------------------------------------------------------------------ > 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 technical documents are available at: > http://www.si-list.net > > 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 technical documents are available at: http://www.si-list.net 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