Simulation of a capacitor (MLCC) with voltage and Frequencies

[Melopia33]

Hello,

I have just installed OPEN EMS :-) thank you very much for this great soft , I would like to simulate a capacitor (MLCC) under voltage and frequencies with voltage an frequencies to have impedance, s parameters, strength field , Voltage visualization ... with OPEN EMS. In the tutorials we can mainly see antenna example but not capacitor. Here's an example of geometry created with OPEN EMS Could you help me please to implement the simulation =>witch excitation and options... should I use ?
Example MLCC @ 500 V and @ 1Ghz

thank you

Xav

[Melopia33]

Thank you very much your answer If I want to try solution number *1* *ideal lumped capacitance *how can I manage the simulation with OPEN EMS? Is it possible to set a voltage with a unique frequency and see the visualisation of the strength field and voltage inside the geometry? Which excitation mode should I use to input the voltage on the two metal terminations? Sorry for all the questions, I am a chemical Engineer, I know how to measure capacitors with frequencies but I have no experience in simulation tools... best Regards Xav Le dim. 8 déc. 2024 à 16:26, biergaizi ***@***.***> a écrit :

…

Unfortunately, to solve this problem we need to define a lumped components with custom S-parameters. But it's *not supported by openEMS*. openEMS can only define a single lumped capacitance, not a frequency-varying one. Since openEMS is a time-domain simulation, implementing S-parameter means we probably need to add some convolution-based algorithms. @0xCoto <https://github.com/0xCoto> said his team had some ideas about its implementation, but I'm not sure if they actually did it. Without S-parameter support, you have two choices. 1. Simulating an ideal lumped capacitance, and say it's good enough. 2. Simulating a distributed capacitance by modeling the dielectric inside the capacitor, by modeling the dielectric using the Lorentz material model (you adjust the dielectric's model parameters until it behavior looks like a real capacitor). This is theoretically possible, but likely not practical in real simulations for two problems. First, parameter fitting is non-trivial. Nobody has tried it before, it's an open question (although solvable). Second, a fine mesh is needed around the capacitor dielectric for accuracy, increasing simulation time. — Reply to this email directly, view it on GitHub <#266 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BNRIN55CMTU4QAVYED62KR32ERQKZAVCNFSM6AAAAABTHCMTRKVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCNBZHE4TOMI> . You are receiving this because you authored the thread.Message ID: ***@***.*** com>

[biergaizi]

Sorry, I misunderstood your original question. It's why I've deleted it.

I now see that, here, you're in fact simulating a distributed capacitor (with dielectrics, plates, terminals, etc). You're not simulating a lumped capacitor (which is a blackbox without any internal details).

The FDTD simulation workflow is basically the following: FDTD can be understood as virtual VNA or TDT/TDR analyzer. You sent a Gaussian impulse into the Device-Under-Test, and you check the reflection that comes out of it. Then its frequency-domain S-paremeter is calculated automatically (using a Fourier transform) at many different frequencies. Finally you analyze its S-parameter (and optionally the raw E&M field dumps) to understand its frequency response - note that it's up to yourself to interpret the data, openEMS is only a field solver.

Which excitation mode should I use to input the voltage on the two metal terminations?

So conceptually, it should be almost exactly the same as an antenna simulation, you add a thin or zero-thickness Lumped Port that starts from one capacitor terminal, and stops at another capacitor terminal. Use a Gaussian excitation signal, bandwidth 500 - 1 GHz. Run the simulation.

Finally you get the capacitor's wideband S11.

1. If the dielectric you've added is non-linear, using Drude/Lorentz model provided in openEMS (AddLorentzMaterial() with the correct physical parameters), the S-parameter should show frequency-dependent capacitance (after interpretation).
2. If the dielectric you've added is linear (AddMaterial() with a single value of permittivity). S11 may still go up and down show resonances at many different frequencies - when the size of the capacitor is comparable to the signal wavelength, you should see transmission-line behaviors such as quarter-wave transformers.

Optionally you can enable "field dump", which periodically writes raw E&M field data on disk. This can be visualized by the general-purpose data analysis tool ParaView.

Thus, I suggest you to try this option first.

However, this is only the theory... In practice many subtle details can go wrong, and the results often can be unphysical without any obvious cause. If it happens, the best hope is posting the sample code here and see if someone can help you (In fact, I'm writing a tutorial on capacitor modeling, I find even the simplest example, the parallel-plate capacitor/waveguide, is not that straightforward due to these problems)...

Is it possible to set a voltage with a unique frequency and see the visualisation of the strength field and voltage inside the geometry?

Yes, instead of a Gaussian excitation, you can also use a Sinusoidal excitation (SetSinusExcite()). The FDTD simulation will keep running until steady-state is detected. This mode is rarely used since most FDTD simulations are wideband simulations, but it would be useful if you want to take a look at the steady-state field strength inside the capacitor.

The Parallel Plate Waveguide example is the simplest demo that uses this feature:

Tutorial (slightly outdated): https://wiki.openems.de/index.php/Tutorial:_Parallel_Plate_Waveguide.html
Code: https://github.com/thliebig/openEMS/blob/f84be9f817e8cd5fc6faec1dd078c0fb101ca013/matlab/Tutorials/Parallel_Plate_Waveguide.m#L20

[Melopia33]

Thank you very much I will try in this way

[Melopia33]

Hello,

Is it to possible to check please my code for the simple Following model (I try to folow your precious advice) ?
The first goal is to see the E field inside the capacitor with volltage
Later I will try S paramaters also with your advice

Please find the main steps

Goemetric model:

X is the direction of the lenght of the electrode
Y is the direction of the width of the electrode
unit = 1e-6

% init and define FDTD parameter F=10 Mhz

FDTD = InitFDTD(100,0,'OverSampling',50);
FDTD = SetSinusExcite(FDTD,10e6);
BC = {'PEC' 'PEC' 'PMC' 'PMC' 'MUR' 'MUR'}
FDTD = SetBoundaryCond(FDTD,BC);

For the mesh I choose a simple mesh (must be improved after)

mesh.x = SmoothMeshLines([0 layerLength], 1)
mesh.y = SmoothMeshLines([0 layerWidth], 1)
mesh.z = SmoothMeshLines([-layerThickness zPos+layerThickness], 1)
CSX = DefineRectGrid(CSX, 1e-6, mesh);

% define the excitation The E=10 V/m in Z direction (unit =1e-6) =>the exitation is in the plane yz (termination), I am not sure of that..
CSX = AddExcitation(CSX,'excitation',0,[0 0 10]);
CSX = AddBox(CSX,'excitation',0,[0 0 0],[0, layerWidth,zPos+layerThickness]);

% define a time domain e-field dump box in the xz plane
CSX = AddDump(CSX,'Et','DumpMode',0);
CSX = AddBox(CSX,'Et',0,[0,layerWidth/2,0],[layerLength,layerWidth/2,zPos+met_layerThickness]);

Best Regards

Xav