Hi Neoflash, If the purpose is to preserve the appearance of the X11/X22/X12/X21 parameters compared to your previous 2-port case, I would recommend Z-parameters, not Y. However, the fit quality for Y or Z-parameters is generally worse than for properly normalized S-parameters simply because the former are un-bounded. The core of any fitting method is an SVD-like procedure that minimizes the norm of the difference between the fitted and target dependences. Hence, this procedure considers the pieces with high absolute values as "more important". But, this does not agree with our needs. In most practical cases we shouldn't care if the impedance at certain point is 1e5 or 1e6 Ohm because this will not affect the solution (i.e. power consumption is very low). Same way, with Y-parameters, we don't care if the conductance between certain nodes is 1e6 or 1e7 because there will be no perceptible voltage drop and again, in the generally non-singular structure, no considerable power consumption. S-parameters are more suitable in this sense because they show best the range where the impedance is comparable to the normalizing value. But there comes the problem of properly choosing these normalizing values. Although this could be not exactly your case, but with improperly chosen normalized port impedance the S-parameter curves may either be barely deviate from zero, or have 'flat' regions where their modules reach 1 - the maximum value - hence their real/imaginary parts be close to real(s)/imag(s) =3D=3D1 or -1. Vladimir Msg: #1 in digest Date: Thu, 20 Mar 2008 06:05:23 -0700 (PDT) From: Neo <neoflash2008@xxxxxxxxx> Subject: [SI-LIST] Re: FW: Curve fitting of center-tapped inductor Vladimir: =20 Thanks for your reply.=20 =20 If the issue comes from the improper normalized impedance, can I fit Y parameters instead of S parameters to fix this problem? =20 Also, what's the meaning of "deviate only slightly or stick flat to the lines S=3D3D1 or S=3D3D-1."? What are these two lines? Could you = explain a little more? =20 Regards, Neoflash "Dmitriev-Zdorov, Vladimir" <vladimir_dmitriev-zdorov@xxxxxxxxxx> wrote: Hi Neoflash, The cause of your problem can also be an improper choice of the normalizing impedance when creating S-parameters. The "average" impedance of the two ports you measured first is perhaps very different from the port you added later. An interesting thing about S-parameters is that even if you don't change anything about ports 1 and 2 and only add the port 3, the components of the S-matrix that corresponds to your ports 1 and 2 (such as s11, s22 and s12) may change drastically, compared to your case 1. For example, in case 1 the parameters 1 and 2 will be as if nothing is connected to the port 3 (current is identically zero). In case 2, with e.g. normalizing impedance 50 Ohm (or whatever you used for port 3) the S11/S22/S12 will appear exactly as if you terminate the port 3 with 50 Ohm resistor and still measure only the ports 1 and 2. Of course, the result might be quite different, and S12 is what probably contains most of you mutual inductance. The rule of thumb is that the S-parameter dependences - real and imaginary parts - when you observe them with some sort of viewer must use the entire range (between -1 to +1) but should not deviate only slightly or stick flat to the lines S=3D3D1 or S=3D3D-1. If you see the = =3D latter, the normalizing impedance must be changed for certain ports before you start fitting. It is possible to re-normalize S-parameters without re-creating them, provided that your frequency points are taken dense enough to cover all important resonances. If you could send it, I can investigate and transform your touchstone data. Vladimir >Msg: #1 in digest >Date: Tue, 18 Mar 2008 07:37:23 -0700 (PDT) >From: H B=20 >Subject: [SI-LIST] Curve fitting of center-tapped inductor >Hi, folks:=3D20 =3D20 >I come up with a strange thing when doing curve fitting for a differential >center-tapped inductor.=3D20 =3D20 >The original version of the inductor is a two-terminal differential >inductor. I did curve fitting to find the equivalent circuit model. And the >fitting seems good.=3D20 =3D20 >The next step I added a center-tap and got a 3-port s-parameter from field >solver. The problem is when I'm doing the curve fitting again. the mutual >inductance in the circuit model was forced to zero by the fitting tool. The >fitting seems OK but not as good as two-port one. The inductance and Q is >ok but the s-parameter related with the third port are not well fitted.=3D20 =3D20 >Any one ever experience same thing before? Any suggestions?=3D20 =3D20 >Regards,=3D20 >Neoflash=3D20 =3D20 > I've attached the link where the equivalent circuit model could be seen. > http://www.designers-guide.org/Forum/YaBB.pl?num=3D3D1205764700 ------------------------------------------------------------------ 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:=20 //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 =20 --------------------------------- Looking for last minute shopping deals? Find them fast with Yahoo! Search. ------------------------------ ------------------------------------------------------------------ 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 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