Yuriy, Thanks for making this point about proper pin/port connection; this is indeed a source of many problems. Circuit simulators, including SPICE and e.g. ADMS (ELDO) are flexible enough and allow not only assignments for ports, but also individual assignments for "pins" (pos and neg terminals of every port), leaving the user a possibility for misuse. Same with T-line. Of course, to ensure proper use, all local references could be connected to a common ground, but they also could be connected together to another node, not necessarily global ground. However in some cases this requirement may become too restrictive. There is a theory of linear multiports, where distinction is made between arbitrary N-ports and so called "2N ports" or better to say 2N terminal models. Complete N port model typically is an N+1 terminal model, with N independent port voltages (differences between terminal potentials) are independent. This model (N ports and N+1 terminals) allows any type of external connections without the risk of violating the condition of 'regularity'. (I may be mistaken about the exact term, but this is a literal translation). With "2N terminal" models, we typically have N ports and 2N terminals. Here, we are very restricted in a way we can connect this model to external circuits. Most of practical models we deal with belong to this second type. Single T-line is a 2x2 terminal model with 2 ports. Although 4 terminal model may create 3 ports maximum, this is not a legitimate way for T-line because it was created as 2 port only (or in other words, the model does not reflect the relationships other than between the currents and voltages or A/B for these two ports, hence, we cannot take from this model more info than was put into it during measurement). For example, if we connect a resistor between side A, positive and side B positive terminals, this would result non-physical solution since then the model is actually used as 3 port, and by itself cannot enforce "regularity" since it was not designed this way. Unfortunately, there is no way up to now to specify legitimate connectivity for a given S-parameter model inside the Touchstone file. We see the number of ports but, we do not have an info on how these ports should be created by selecting the pairs of external nodes. Vladimir From: "Yuriy Shlepnev" <shlepnev@xxxxxxxxxxxxx> Subject: [SI-LIST] Re: Concept of Voltage Date: Tue, 19 Jul 2011 09:20:08 -0700 Shawn, You have described a perfect example of what should never be done with multiports described with S-parameters (or equivalent t-line models). It is typical model misuse (garbage in - garbage out). Formally, each single port in a multiport has 2 terminals (or nodes) - signal and local reference. There is no information on the potentials of each terminal - only information on the difference of potentials or voltage. Moreover, as I pointed earlier, the voltage may be defined not as the difference of potentials (or line integral of electric field), but as a projection of the electric field on the electric field of a wave-guiding eigen-mode. Current flowing out of the reference terminal must be exactly opposite to the current flowing into signal terminal. Again, current may be defined either locally or as a projection of magnetic field on the eigen-mode field. The projection definition must be used even for MTL with quasi-TEM modes at high frequencies. The only possible way to use a port is to connect it to another port that is defined identically. Lumped components can be connected to local or lumped ports defined in electromagnetic analysis. Wave-ports can be connected only with the identically defined wave ports. An exception is quasi-TEM transmission lines at lower frequencies. Such ports can be connected with the lumped or local ports as long as the cross section stays much smaller than the wavelength (due to similarity of the voltage definition for MTL and lumped element). In cases when distance between strips becomes comparable with the wave-length, connection of a quasi-TEM t-line port with a lumped port becomes ambiguous and erroneous. Considering use of multiport models in a SPICE circuit simulator, the only possible and legitimate use of the local reference terminal is to connect it to a common reference node (global ground). The signal terminal must be connected to a signal terminal of the identically defined port. In case of lumped port, the lumped element must be connected between the signal terminal and the global reference node. If a lumped element contains more than two nodes, there must be a separate lumped port (possibly with common reference) constructed to connect each signal node of the lumped element. There are no restrictions in the circuit analysis as long as one follows the port construction and connection rules and build or collect a sufficient number of multiport models for each interconnect component. Not only quasi-TEM, but optical or substrate integrated waveguides for instance can be analyzed with a SPICE solver without any problems. Best regards, Yuriy Yuriy Shlepnev www.simberian.com ------------------------------------------------------------------ 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