It is not the design. CST hasn't properly calculated the fields. You may need to change the energy limit, or choose the AR-filter. Or ignore the fast ripples - they shouldn't be there, or at least most of them shouldn't be, and probably all of them. You could filter with a boxcar filter the length of one period of ripple, or something similar.

The simulation is not properly converged. You have an extremely high Q cavity and any time-domain simulation is going to take a long time to dissipate all the incident energy. As indicated by Malcolm, you need to either increase simulation time or use the AR filter. Another possibility would be to simply use the frequency domain FEM solver, which is more suited for these highly resonant high Q structures.

Thanks Mr Malcolm White and Mr Andre Kuehne for helping me. I still need more detailed information, how to work on the suggestions you provide. what menu to use in CST. in My Project Design, I use resonant cavity and wave guide for 2.45 Ghz Microwave frequency. I tried to use the Microwave Oven cavity size approach which is widely used for domestic purposes. Depend on theoritical approachment i see no Problem for Wave Guide and Cavity Size. Please more advise from you.

The oven is quite big in wavelengths and has a high Q, so will store a lot of energy.

The time domain method sends in a pulse that bounces around. It will have to bounce around for at least Q bounces to get a decent answer. If Q is big then by then the computer accuracy on that many calculations may be causing problems. You haven't got there yet, you just haven't been running it for long enough so there are lots of bounces to go before the pulse dies away and the ripples smooth out. Computing errors may grow before that happens. The AR option predicts how the ripples will smooth out so avoids that problem. The frequency domain solver solves it in an entirely different way and has no problem with high Q.

I think that CST help will tell you that sort of thing if you bother to read it.

Thanks Mr Malcolm White , i try to learn some of your sugestion and that is very helpfull for me. As follow i show you the result of my configuration in AR-Filter, is this good result?

No. Not sure why, but the powers are not right. Perhaps you should try the frequency domain solver as Andre Kuehne suggested.

Yes this definitely looks like a case for the FEM solver. If you want to stick to the time domain, you need to keep an eye on the convergence plot (1DResults/Energy). The standard setting for timedomain are -40 dB convergence (which might be sufficient), but it only simulates for 20 pulsewidths. You should significantly increase this, especially since you have a really wideband excitation and hence a short pulse. You can find it in Solver Setup --> Specials --> Maximum solver duration. Increase this to 200 pulses and rerun your simulation. You shouldn't see any strong regular oscillations in the Power plot and watch out that Power Accepted + Power Outgoing all Ports

Thank you Mr Malcolm White and Mr Andre Kuehne are still helping me. In this Project I prefer use Time-Domain Solver than Frequency Domain Solver, the reason is i can set power of excitation signal source in Time-Domain easly. Also Thanks for some learn, so from your advise i found my work problem is caused by the STEADY STATE energy criterion has not been satisfied & not right in setting AR Filter .

Depend on your suggestion and some help document in CST, I do some effort to solve STEADY STATE problem :

1. By increasing Pulses to 300, but the steady state have not reached.

2. By increasing TIME duration, and

3. and i tried use AR Filter, the steady state can be reached but it spent too much time simulation. i need some information how to determine AR Filter rightly.

As Follow i attach signal form that i Use in my work, and please again your advise how to determine exactly and get better result, :

a. Number of Pulse

b. Time Duration

c. First time step, Skip time steps, Max. frequency, Max. order of filter & Relative window length, in AR-FILTER setting.

d. in reaching steady state, can we use one of NUMBER OF PULSES or TIME Duration?

I wish you can help me again. Thank You

This is the best result

I don't quite understand why you cannot use the frequency domain solver - you can easily scale the results to whatever incident power you desire (12.5W in your case, it seems). Incident power during the initial excitation does not matter if you're only looking at linear media. I would really advise you to take a closer look at the frequency domain solver again, I am almost 100% certain it will yield the results you need much more quickly.

I don't have much experience with the AR filter, the CST manual should be of more help to you. I'm also not sure how it affects power calculations.

It doesn't matter whether you use number of pulses or an absolute time, number of pulses simly takes your pulse length and multiplies it by the stated number and that's your total simulation time.

Hello,

For help, few files are attached.

Thanks,

Using the frequency domain solver over a broad band means solving at lots of frequencies, so can take as long as time domain, or longer, but it may be worth running the frequency domain solver in narrow bands around the resonances you have found to check if they are right.

I don't use the frequency domain solver, but expect Andre Kuehne is right in his comments.

The broadband sweep functionality of the CST frequency domain solver is quite powerful. It will perform a rational function interpolation using the discrete samples and doesn't have to rely on equidistantly calculating too many frequency samples. It will more or less automatically home in on the resonances and calculate more samples in these regions. From what I understood, this is a microwave oven so the broadband response is not really of interest and you're only interested in the 2.4 GHz response, is that right, Wayan Trimawiasa?

I think the oven is not designed to be resonant at 2.45 GHz. That would mean a large energy build up until a large amount of power went into leakage (not a good idea) or making the walls hot (also not a good idea). Empty microwave ovens tend to have a large reflection back to the magnetron, which is what you have predicted. It is not particularly good for the magnetron, which is why you are not supposed to run them empty.

Thank You very much Mrs. Smrity Dwivedi , its very helpful.

I interested by Mr. Andre Kuehne advise to use frequency domain. The goal In my work is to Simulate EM-Thermal Coupling Dielectric Material. I want to see correlation between its Power Accepted Vs Thermal in dielectric material in 2,45 GHz frequency. I try Design Oven in CST closly as Domestic Oven 2,45 Ghz, and oven Dimension is depend on other research before.

Mr Malcolm White , The Resonant frequency in this Oven is not My Goal. I only want to find The Most Accurate Power Accepted by the Material in this simulation. I Hope can make some variation Power accepted by the material or MW Excitation, then see its influence in thermal effect. in Time Domain I can make some variation Power easily by changing Amplitude in Waveguide Port as Microwave excitation Source.

However Also I interested By our discussion to try frequency Domain. May you can tell me how to change Amplitude or power excitation in frequency domain? I have tried read CST help but difficult to find it.

Wayan Trimawiasa, you can always scale to your desired power in postprocessing (e.g. Design Studio - Schematic). Just simulate in frequency domain using the default power and then create an AC task in Design Studio where you set the power you wish. Have a look at the attached CST project for more details.

Thank you very much Mr. Andre Kuehne . I should learn your CST project. I am very happy to still be able to discuss with you