hi thanks for the comprehensive answers. i will think about these. the part with the z-to-s conversion was quiet clear: dividing very big numbers with very small numbers, with noise and finite number of digits... i read about this somewhere (i.novak: frequency domain characterisation of power...). i was looking for books about s-parameters, but most of what i found just spends 2 pages on it and costs more than 100$, they spend 95% of the content for introduction to signal integrity. maybe i will buy the "Microwave Differential Circuit Design Using Mixed Mode S-Parameters" (but it has a very bad review on amazon.com), or one of them adviced by you, Yuriy... how does the characteristic impedance of a tline coming through a touchstone file? to produce matching when connectied terminations in the system simulator. is there some relationship between Z11/Z12/Z21/Z22 parameters and Z0? (at what frequency value) istvan cct ----- Original Message ----- From: "Yuriy Shlepnev" <shlepnev@xxxxxxxxxxxxx> To: "'Istvan Nagy'" <buenos@xxxxxxxxxxx>; <si-list@xxxxxxxxxxxxx> Sent: Tuesday, January 06, 2009 6:22 PM Subject: [SI-LIST] Re: 3D-em simulation and terminations: macromodelling > Hi Istvan, > > I see that you have got two answers on the S-parameters normalization: > 1) J.X. Zheng: "It is independent of excitations and terminations." > 2) G. Havermann: "...the port selection and geometry has a huge impact on > the results." > Both statements are correct in an appropriate context that may confuse you > further. > So, let me try to explain it starting from the first principles. > > Behavioral model of interconnect or power distribution network (PDN) is a > black-box description of the system in form of a multiport. Each port of > the > multiport is defined by two terminals with a voltage V between them and a > current I going in and out (both can be defined as special or modal > variables). Multiport can be characterized by the impedance matrix Z as > V=Z*I - defined with the open-circuit terminations. Short circuit > termination produces admittance matrix Y description of a multiport - > I=Y*V. > For non-degenerate cases (always the case for non-ideal physical systems) > two descriptions are equivalent and Z=Y^-1 and vice versa. > Most of the frequency-domain electromagnetic solvers use immitance > formulation and produce directly either admittance Y or impedance Z > matrices > (whatever is more convenient). These matrices can be written into a > Touchstone file and used by a system level solver without conversion into > S-parameters. Conversion to S-parameters is usually done as a > post-processing step in that case. Note that most of the system-level > solvers use the immitance formulation too (so called modified nodal > analysis). Immitance description may be called "un-terminated" description > because of no particular termination impedance is specified. > > S-parameters are convenient for measurements and can be introduced as a > simple change of variables: > a=0.5*(V/sqrt(Zo)+sqrt(Zo)*I) - incident waves, if t-line is attached > b=0.5*(V/sqrt(Zo)-sqrt(Zo)*I) - reflected waves, if t-line is attached > b=S*a; > S=(Z/Zo-U)*(Z/Zo+U)^-1; or S=(U-Y*Zo)*(U+Y*Zo)^-1; > Where Zo is the normalization impedance (assumed to be identical for all > ports for simplicity), that is technically the termination impedance. Z is > the impedance matrix and Y is the admittance matrix. > As you can see the S-parameters describe multiport with a particular > termination of all ports. If your electromagnetic solver requires setting > up > the termination impedances before the analysis - it may use the scattering > formulation. Changing the termination impedances may change the > S-parameters > dramatically as was pointed out. But S-parameters with any termination can > be easily transformed into "un-terminated" admittance form as > Y=(1/Zo)*(U-S)*(U+S)^-1 or into impedance form or renormalized. In fact, > practically all system-level solvers transform S-parameters back to > "un-normalized" Y (there are exceptions) to perform the analysis of the > multiport with the connected models. > Thus, theoretically, all descriptions of multiports are completely > identical > for non-ideal physical systems. As a practical experiment, try to use > Touchstone files with matrices of different type and with different > normalization. The results of the system-level analysis of the terminated > interconnects should be nearly the same (at least in frequency domain). > Nearly, because of the numerical accuracy during the descriptors > conversion > - it may become an issue if the normalization impedance and diagonal > elements of the impedance descriptor are different by multiple orders in > magnitude, that may happen in case of PDN analysis for instance. A > substantial difference may be observed in the time domain due to the > reduced > order modeling issues - the normalization or termination have to be > carefully selected in that case (pointed out earlier by Vladimir > Dmitriev-Zdorov). > > Finally, note that the "real" t-line model in your system-level solver is > also a multiport described by admittance parameters of a multi-conductor > transmission line with t-line p.u.l. parameters computed either with a > formula or with a static field solver. You should expect good > correspondence > of such model with a model on the base S-parameters extracted with > electromagnetic solver at lower frequencies (up to 1-3 GHz for typical PCB > geometries). > > To learn further on multiports and S-parameters I would recommend the > following books: > B. Young, Digital signal integrity: Modeling and simulation with > interconnects and packages, 2001 > J. Choma, Electrical networks: Theory and analysis, 1985 > H.J. Carlin, A.B. Giordano, Network theory: An introduction to reciprocal > and non-reciprocal circuits, 1964 > > Best regards, > Yuriy Shlepnev > www.simberian.com > > > > -----Original Message----- > From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] > On > Behalf Of Istvan Nagy > Sent: Monday, January 05, 2009 1:34 PM > To: si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] 3D-em simulation and terminations: macromodelling > > hi > if i simulate a pcb structure (a microstrip trace) with an 3d > electromagnetics simulator to create a macromodell (s-parameter touchstone > file), then do i have to connect terination resistors in the em simulation > setup? or all the ports in all simulators are automatically terminations > too? do i have to use the same term.resistor value in system simulation > too > (when using the macromodel. example in Agilent ADS or spice) to get > accurate > results? > the problem is that in the em simulation we can not use the correct > termination, since it comes with using IBIS buffer models and those are > not > compatible with any of the EM simulators, they can be connected in a > circuit > simulator only. so do macromodelling: first EM simulation, create > macromodel, then system simulation with the other elements of the signal > path. > > I was thinking: > if the S parameter data ( S21(f) ) should depend on the actual > termination? > if we change the termination of a transmission line, then it changes the > reflections (matched, unmatched...as we studied in the school). is it the > same when we use an s-param (touchstone) macromodel instead of the real > tline? > if i dont terminate in the em simulator, then the strong reflections will > cause strong loss and not-flat response, and that loss will go into the > s-param file. then if i connect correct termination in the system level > simulation, it will not to fix those reflections/loss anymore. > am i right? > > so, how much usable is an s-parameter macromodell, and macromodelling > anyway? > > Istvan Nagy > CCT > > ------------------------------------------------------------------ > 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 > > > > ------------------------------------------------------------------ > 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 > > > ------------------------------------------------------------------ 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