I personally think that the simulation should be accurate as long as the following issues are clearly considered. It does not matter what tools you use to simulate the package (assuming the solvers are accurate). 1) The cascaded network's ports should have same/similar modes. For instance, you have two subsystems optimized as two wave-port networks, the first one with single mode, and 2nd one with more than one dominant mode. Cascading those two networks might lead to unexpected system response. Most hybrid solvers now can take into account multiple modes, but you have to give it the right ones first in order to see correct results. 2) Even the ports have similar modes. When they are cascaded, the interface should not generate much unexpected mode. For example, two systems may both have 50-ohm coaxial outputs, if their diameters are too different from each other, it could generate considerable amount of high order modes, which may couple back to your dominant mode resulting high return loss etc. This is typically true at high frequencies. 3) Impedance vs. modes. When we talking about impedance, normally we are talking about it for a certain mode. Even the two systems have same input-output impedance, it does not mean they will always follow the return loss = (z1-z0)/(z1+z0) at system level. Choosing the right way of calculating impedance (e.g. v/i, v**2/p, p/i**2) and properly choose the mode (e.g. avoid cascading diff type of ports like microstrip - stripline) helps when doing simulation. 4) The impact on system due to cross-talk or mutual coupling between the subsystems should be negligibly small. Otherwise, your hybrid solver may give you a prediction that is way-off from reality. Of course, it is a unnecessary worry if the solver is completely full wave. Please correct me if something is missing here. Thanks. Regards, Junwei Dong Microwave Engineering Corporation(MEC) -----Original Message----- From: si-list-bounce@xxxxxxxxxxxxx [mailto:si-list-bounce@xxxxxxxxxxxxx] On Behalf Of Brad Brim Sent: Thursday, April 14, 2011 11:13 AM To: 'Inmyung'; si-list@xxxxxxxxxxxxx Subject: [SI-LIST] Re: PEC in 3D field solver? hello Delta, Based on your text it is impossible to know what you are doing with your 3D EM solver so it is not possible to answer your general question. However, it is possible to explain the 3D EM analogy to pin grouping in a hybrid solver. First, pin grouping is not required in a hybrid solver nor in a 3D EM solver.. Next, pin grouping really only makes sense for port definitions, so I will focus the response in this manner. In hybrid solvers as well as 3D EM solvers, ports are 2-terminal devices. In a hybrid solver your port is a logical connection, not physical object. You can select multiple pins as each the (-) or (+) terminal of a port. You are simply providing an electrical short circuit amongst those multiply-selected pins for each terminal. In a 3D EM solver your port is a physical 2D object. As most 3D EM solvers are formulated this object will represent a surface through which current flows between the (-) and (+) sides of the port. (or for so-called wave ports, less common to SI/PI simulations, a face through which energy flows into the circuit) In a 3D solver, if you wish to group pins, then you must draw a physical object to short them together. Take for example the die-side connection of a BGA package. Assume you choose a VDD pin in the core power area to define as a port and you wish to use a group of local VSS pins as a return path. In a hybrid solver you simply link this single VDD pin to the (+) terminal of the port and you link each of the multiple local VSS pins to the (-) terminal of the same port. You have electrically short circuited these the local VSS pins to use as the single (-) terminal of the port. Current will flow as it desires into each of these VSS pins. If one of them has a higher impedance path than the others, then it will carry less return current for the port. For a 3D solver you must physically connect the local VSS pins you wish to use as a return path for the single VDD pin. What many people do is draw a 2D PEC object in the XY plane at the top of the solder bumps. The shape is not critical but it must touch all the VSS solder bumps. The port is then defined between this 2D PDC object and the single VDD bump. You can make a hole in this 2D object to define the port as an annular ring in the plane of the 2D object or you can shorten the VDD solder bump slightly and define a vertical port between the bottom of the 2D object above and the top of the shortened VDD solder ball below. If a VDD port bothers you, then simply replace "VDD" in the text above with "signal". Exactly the same concept applies. Based on this example, the analogy between pin grouping in a hybrid solver and a 3D EM solver should be more clear. In a hybrid solver multiple pins may be electrically linked to a single pin to provide a virtual short circuit amongst those pins to define a single circuit electrical node. However, in a 3D EM solver there must be made an explicit physical connection in the 3D geometry prior to port definition. A common concern that you did not express is: Your driver models may well be multi-pin circuits with signal pins, VSS and VDD/VCC pins. The drivers may be single or dual referenced. How do you hook-up these multi-pin, multiple power domain driver models to your simulation ports? The summary answer is that you must define multiple ports to which to connect your multi-terminal driver circuit. Sorry, but a single port will not be adequate to represent dual referenced drivers. best regards, -Brad Brim Sigrity > -----Original Message----- > From: si-list-bounce@xxxxxxxxxxxxx > [mailto:si-list-bounce@xxxxxxxxxxxxx] On Behalf Of Inmyung > Sent: Thursday, April 14, 2011 12:45 AM > To: si-list@xxxxxxxxxxxxx > Subject: [SI-LIST] PEC in 3D field solver? > > Hi everyone, > > Some engineer try to analyze the package by using 3D full > wave field solver such as HFSS. > But as you know, the power and ground nets was disconnected generally. > So they put the PEC metal to short the disconnected ground nets. > Is it a right method? > Is there any other methodology to simulate package? > > > How about Hybrid tools? > They do not use PEC but pin grouping. > I guess pin grouping means equi-potential so that is similar > with PEC right? > > Thanks, > Delta ------------------------------------------------------------------ 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