Discussion Forum

Earth ground? An electrical ground? Once you decide/understand what you mean by "ground" in terms of its impact on the particular physics you are studying, then you will be better able to understand how (or if!) to define it in your problem. Might you simply mean a (or any) perfect electric conductor (PEC)? Well, that's already the default boundary condition on any external computational boundary, in the RF module. Now, if you mean Earth ground, that's different, since you'll need to properly represent some subset of the Earth (that is, assuming it is somehow interacting with your device of interest) in your problem. This could range from something as simple as a PEC to a 3D model of a suitable volume of material (e.g., soil, concrete, water, etc.) with an appropriate finite resistivity. Lying between those extremes of simplicity and complexity, an "impedance boundary condition" may be appropriate. Regardless, if you are willing to explain more about your problem, readers here will likely be able to provide you with much more focused and useful advice. Note that if your "microwave resonators" are ordinary standalone components, not involved in a separate circuit model or interacting with the Earth, etc, then you probably don't need to define any "ground" anywhere. Rather, there exist simply surfaces and domains of various materials (metals, plastic, air, etc.) and you will need to specify boundary conditions corresponding to external boundaries, which will variously be PECs, ports (of various types), or (in some cases) other boundary conditions, especially if you are taking advantage of a symmetry plane. You might want to look in the Application Library at the Comsol-supplied example models of some simple microwave components.

Hey Robert, thank you for your answer, I'll specify my reply as you asked.

I would like to know specifically how ground is defined in the "Electromagnetic waves, frequency domain" physics module. My simulations have only the physics domains and boundaries present in the "cpw_bandbass_filter.mph" example and are identical when it comes to the physics with my own models. To my understanding in the "cpw_bandbass_filter.mph" there is no explicit definition of ground (by explicit I mean a ground physics domain) such as in the "AC/DC physics module" (electric currents (ec) in AC/DC for example). As such it is quite confusing how ground is defined in this specific physics module (emw).

My question could be summed as "How is ground defined in these examples I have linked?".

For example in a past discussion forum I asked about ports and I was adviced to put it in between ground and the conductor. My question is, how does the lumped port in the "cpw_bandbass_filter.mph" know which one is ground and which one is the conductor if there is not an explicit definition? Same for numeric ports, how does it know what is the ground and what is the conductor in the "cpw.resonator.mph" model?

Regards, Nico

If I am diagnosing your confusion correctly, you seem to misunderstand the concept of "ground," as well as some EM theory. In low-frequency electronics, one often considers a "circuit ground," or perhaps a "chassis ground," to be a surface of either zero scalar potential or held at some fixed reference potential, or equivalently, an infinite reservoir of charge. But in full-wave RF formalism, the concept of the potential is necessarily generalized (there is both a scalar and vector potential to consider). There is then often no value in even attempting to define a "ground." That said, circuit designers will often (not always) tie a PC Board backplane to a traditional electrical grounding wire which then may (but often may not) ever connect somewhere to an Earth ground or the "ground" wire on some power supply involved in the system. And yes, people talk about microstrip lines over a "ground plane" all the time. But that's just an electrically-conducting plane. There isn't anything really special about it, other than that. Now, if it makes you happier to label a "ground," then simply declare some surface to be one (write it down, stick a label on it, tell all your friends, etc...) that your largest metal surface (or any other metal surface you prefer in your problem) is hereby your offical "ground" and... be satisfied with that. But what the computational model requires is for you to set the boundary conditions (and any other conditions) that constrain the partial differential equations (PDEs) that govern the RF, to yield an unambiguous solution to those PDEs. If this is still confusing to you, then I encourage you to consult one of many available college-level textbooks on classical electromagnetism and electromagnetic waves.

As always Robert is right. Ground is just a name for a certain electric potential value. Particularly in RF it is not of much value. E.g. symmetric antennas like dipoles and yagi-uda do not need the ground concept. In other RF systems we often use the term ground plane, but there is nothing special about it and no specific BC is needed for correct modeling. The ground plane is just a conducting surface.

In the model file cpw_bandpass_filter, the emw node is setup with the two lumped ports being sources and Perfect Electric Conductor 2 as being a return path for the PCB. There is no specific ground node in emw so ground is an ambiguous term in this case. There is also PEC1, which are the airbox sides. While one might say that the box sides are ground as well, this is not 100% accurate. In a real physical system one may consider the outer airbox to be at ground potential and this may be true at low frequencies. It may be better to talk about return paths rather than "real grounds" at higher frequencies. Coaxial cable is very similar in concept as the return path is the shield but it may be attached to mechanical ground in an overall system.