RF-Electric Field with respect to an applied voltage

Please login with a confirmed email address before reporting spam

Hello Dear Readers,

This is my first forum entry -- please be patient with my unexperienced discussion skills.

I am working on Silicon-Organic Hybrid-Modulators and its modulation efficiency. For the electrode structure, I am using CPW.

My problem I am trying to troubleshoot since many days is that in the RF-Module, I am not able to set a voltage on the (center) conductor of the CPW and thus be able to alter the electric field in the gap between the conductors in dependence of the applied voltage. I am trying to run a Mode Analysis Study with RF Module (emw) and Electrostatics (ES)/ Electric Currents (EC) interfaces of the AC/DC Module. In the ES or EC interface, I am applying Electric Potentials and Grounds on the respective conductors. But still, the Mode Analysis Study computes the Electric Field in the gap independently of the applied voltage in the ES/EC inteface -- the Electric Field in the gap is the same at 1V at the center conductor as at >>1V...

May I ask you for your help, how to overcome this problem? Is there any method to apply a voltage on a conductor (of a CPW) and compute the RF-Electric Field (in the gap between the conductors) and the mode profile with respect to the applied voltage?

I sincerely hope for your help. Thank you very much!

Best Regards Adil Han Dogan


8 Replies Last Post Jan 23, 2021, 12:20 PM EST
Robert Koslover Antennas, Waveguides, Electromagnetics

Please login with a confirmed email address before reporting spam

Posted: 1 month ago Jan 20, 2021, 7:04 PM EST
Updated: 1 month ago Jan 20, 2021, 7:06 PM EST

A subtlety arises here because electric potential is not a strictly-defined quantity, when working with RF. That's why you don't see it listed among the boundary condition settings in the RF module. (Caveat: the best ways of dealing with this have evolved over the years as Comsol keeps adding new features, my approaches may not be fully up-to-date). For an electrically-small region (in your case, the gap between conductors), you can relate the integrated tangential component of the electric field along a short path between the two conductors. There are a number of ways to manage this. I suggest you add a line segment (not a wire, just a geometric line segment) between the conductors, oriented so that you already know it will be tangential to the electric field there. Make sure it has at least a few segments of mesh along it. Identify (later) your "voltage" as the integral of the electric field along that line segment. You can do this in advance or in post-processing. If you do it in advance, you can define (Component 1 --> Definitions) an "Edge Probe," with Probe Type: Integral, and for the expression, choose the appropriate component of electric field. For the selection, choose only the subject line segment. You might also want to change the default name and/or label of the probe, to "Vin" (for example). Strictly speaking, you still won't be specifying the voltage. Rather, you will be computing/displaying it, as one of the results later on, and can use it in post-processing computations too. Now, if you want a different voltage from what you get, and if your problem is linear, just scale everything proportionately! (If your problem isn't linear, you may need to run it a few times, with different-strength excitations, to get it to fit the input "voltage" that you really want.) Also, in regard to that/those excitations, there are many ways to accomplish it/them. When possible, use ports. In your case, you might even consider using lumped ports (which can simplify things sometimes, since they behave much like circuits), although this can result in a less-correct field distribution than if you are using a solution from a mode analysis study. I hope that all helps and makes some sense to you. Good luck.

A subtlety arises here because electric potential is not a strictly-defined quantity, when working with RF. That's why you don't see it listed among the boundary condition settings in the RF module. (Caveat: the best ways of dealing with this have evolved over the years as Comsol keeps adding new features, my approaches may not be fully up-to-date). For an electrically-small region (in your case, the gap between conductors), you can relate the integrated tangential component of the electric field along a short path between the two conductors. There are a number of ways to manage this. I suggest you add a line segment (not a wire, just a geometric line segment) between the conductors, oriented so that you already *know* it will be tangential to the electric field there. Make sure it has at least a few segments of mesh along it. Identify (later) your "voltage" as the integral of the electric field along that line segment. You can do this in advance or in post-processing. If you do it in advance, you can define (Component 1 --> Definitions) an "Edge Probe," with Probe Type: Integral, and for the *expression*, choose the appropriate component of electric field. For the *selection*, choose only the subject line segment. You might also want to change the default name and/or label of the probe, to "Vin" (for example). Strictly speaking, you still won't be *specifying* the voltage. Rather, you will be computing/displaying it, as one of the results later on, and can use it in post-processing computations too. Now, if you want a different voltage from what you get, and if your problem is *linear*, just scale everything proportionately! (If your problem isn't linear, you may need to run it a few times, with different-strength excitations, to get it to fit the input "voltage" that you really want.) Also, in regard to that/those excitations, there are many ways to accomplish it/them. When possible, use *ports*. In your case, you might even consider using *lumped* ports (which can simplify things sometimes, since they behave much like circuits), although this can result in a less-correct field distribution than if you are using a solution from a mode analysis study. I hope that all helps and makes some sense to you. Good luck.

Please login with a confirmed email address before reporting spam

Posted: 1 month ago Jan 21, 2021, 1:05 PM EST

Thank you very much for your insightful explanation!

Indeed, measuring the voltage between the conductors by integrating the electric field along a geometric line segment and adjusting the excitation afterwards helps me to apply the desired voltage on the conductors in RF Module!

Regarding the excitation: In case of 3D, the RF Module (emw) offers Port (Boundary) Conditions to vary the excitation of the model by specifying the Port input power in the Port Properties Settings. Thus, I can set up the voltage I am looking for!

In case of (cross-sectional) 2D though, I can't find any method to excite the model, since Ports and Lumped Ports can only be defined as 1D Edges. And since I am dealing with non-ideal conductors and its losses, Lumped Ports can't be attached between the conductors, because of the non-idealistic conductive boundaries (e.g. no PECs at the conductor surfaces, …).

Are there any sophisticated methods of exciting a (cross-sectional) 2D of a CPW in RF Module such that I can alter the voltage on the non-ideal conductors by varying the excitation?

Thank you very much again!

Best Regards Adil Han Dogan

Thank you very much for your insightful explanation! Indeed, measuring the voltage between the conductors by integrating the electric field along a geometric line segment and adjusting the excitation afterwards helps me to apply the desired voltage on the conductors in RF Module! Regarding the excitation: In case of 3D, the RF Module (*emw*) offers Port (Boundary) Conditions to vary the excitation of the model by specifying the *Port input power* in the *Port Properties Settings*. Thus, I can set up the voltage I am looking for! In case of (**cross-sectional**) 2D though, I can't find any method to excite the model, since *Ports* and *Lumped Ports* can only be defined as 1D Edges. And since I am dealing with non-ideal conductors and its losses, *Lumped Ports* can't be attached between the conductors, because of the non-idealistic conductive boundaries (e.g. no PECs at the conductor surfaces, …). Are there any sophisticated methods of exciting a (**cross-sectional**) 2D of a CPW in RF Module such that I can alter the voltage on the non-ideal conductors by varying the excitation? Thank you very much again! Best Regards Adil Han Dogan

Robert Koslover Antennas, Waveguides, Electromagnetics

Please login with a confirmed email address before reporting spam

Posted: 1 month ago Jan 21, 2021, 11:38 PM EST
Updated: 1 month ago Jan 21, 2021, 11:55 PM EST

You're welcome.
In regard to going from the cross-section to the 3D model, there are a number of approaches; I only know/use a subset. First of all, the requirement that the boundaries be PEC conditions, when you use lumped ports, is more of a mild annoyance than a genuine limitation, because there's at least one workaround: Set up just a very short length of the structure to be excited as PEC, with the rest of it set to the non-ideal conducting materials of your choice. I.e., your transmission line (TL) can have two smoothly connected pieces, with identical cross-sections (one can be an extrusion of the other, for example). Set PEC conditions only on the short piece of TL where you want to place your lumped source, but assign the actual material properties everywhere else. This won't impact your calculations (of things like RF losses) much, since it is only PEC for a very short distance. There are other ways to force the exciting fields to be what you want them to be, but I think I'll stop there. If you want more specific advice, I suggest you post your model to the forum. And you may get a better answer from some of the other folks who follow along here. Good luck.

Added: Oh, in regard to that 1D excitation question... Well, you could always just excite it along a line segment, but that isn't very realistic. Better to solve for the 2D field distribution (electrostatics will do this well enough, normally) as a first/separate step, then use that computed electric field as a user-specified field for the boundary excitation in a second computational step. It takes some moderate care to set this up properly. You are essentially doing a multiphysics task: (1) 2D electrostatics, followed by (2) 3D RF. The fields from the first step will have names like es.Ex, es.Ey, es.Ez. In the second step, you'll be entering those variable names as the x, y, and z components of your user-specified electric field. Actually, you'll only need the tangential components (so for a z-directed TL, that would be es.Ex and es.Ey). You have to set up each step of problem carefully and connect them/execute these two steps in order. This is vastly easier to do nowadays than it used to be in older versions of this code! Take a shot at it and let us all know if it all works out. :)

Finally, you might want to read about numeric ports, which could potentially also be useful to you here. (I can imagine that some folks here might recommend you try that before what I said above. And they might even be right. Or not.) :)

You're welcome. In regard to going from the cross-section to the 3D model, there are a number of approaches; I only know/use a subset. First of all, the requirement that the boundaries be PEC conditions, when you use lumped ports, is more of a mild annoyance than a genuine limitation, because there's at least one workaround: Set up just a very short length of the structure to be excited as PEC, with the rest of it set to the non-ideal conducting materials of your choice. I.e., your transmission line (TL) can have two smoothly connected pieces, with identical cross-sections (one can be an extrusion of the other, for example). Set PEC conditions only on the short piece of TL where you want to place your lumped source, but assign the actual material properties everywhere else. This won't impact your calculations (of things like RF losses) much, since it is only PEC for a very short distance. There are other ways to force the exciting fields to be what you want them to be, but I think I'll stop there. If you want more specific advice, I suggest you post your model to the forum. And you may get a better answer from some of the other folks who follow along here. Good luck. Added: Oh, in regard to that 1D excitation question... Well, you could always just excite it along a line segment, but that isn't very realistic. Better to solve for the 2D field distribution (electrostatics will do this well enough, normally) as a first/separate step, then use that computed electric field as a user-specified field for the boundary excitation in a second computational step. It takes some moderate care to set this up properly. You are essentially doing a multiphysics task: (1) 2D electrostatics, followed by (2) 3D RF. The fields from the first step will have names like es.Ex, es.Ey, es.Ez. In the second step, you'll be entering those variable names as the x, y, and z components of your user-specified electric field. Actually, you'll only need the tangential components (so for a z-directed TL, that would be es.Ex and es.Ey). You have to set up each step of problem carefully and connect them/execute these two steps in order. This is vastly easier to do nowadays than it used to be in older versions of this code! Take a shot at it and let us all know if it all works out. :) Finally, you might want to read about *numeric* ports, which could potentially also be useful to you here. (I can imagine that some folks here might recommend you try that before what I said above. And they *might* even be right. Or not.) :)