Learning Forum

[mabdelr3]

Greetings all,

Is it possible to use the Eigenmode solution type to estimate the self-resonant frequency and quality factor for a planar spiral coil (example in the figure below...)?

[Praneeth]

Hi Muhammad,nPlease find responses inline to your questions - nHow do I define my excitation in this case? For example, how do I define my port in this case? You do not require any excitations for Eigenmode solver type.nDo I need an air box? This depends on your requirement. Please be specific on your need for air box.nThe eigenmode solution type should give me the same self-resonant frequency that I observed from the modal solution type, correct? May I know why you would like to use Eigenmode solution type for your model if you were able to solve it using Driven type?nIncorporating the lossy dielectric layer (in green) and a lossy Cu conductor (in orange) would produce some losses. These collective losses can be found from the Quality Factor that is given in the results of the Eigenmode solution type, correct? Yes, quality factor information will be present in the Eigenmode solution data if applicable.nI noticed that since I have a thin layer of dielectric, the mesh generated is huge... Any advice on how I can reduce the mesh without compromising the correctness of the solution? You can make use of layered impedance boundary. For detailed information on this please refer the HFSS Help document.nAll the very best.nBest Regards,nn

[AndyJP]

2. I guess, you don't need an airbox where the system does not radiate. You can make a small box around connectors, if you wish.n3.,Obviously, for finding out resonances. How would you do it for sure in driven modal with unknown matching?n4. Impedance boundary. But remember, it is 1D in the depth direction. And it is not so good for thin layers between two guiding volumes. Shell element feature, which should treat any coupling between volumes, does not work in Eigenmode; so they become independent, which is not always true.n

[mabdelr3]

... Thank you very much gentlemen for your recommendations and answers. I will implement what you suggested, especially the idea of using impedance boundary on the dielectric layer.As for modifying my 3D model to be suitable for an eigenmode simulation, I think all I need to do is remove the port (shown in the figure) and connect the two wires directly with a piece of Cu (close/short the loop).As soon as I have some results, I will attach the results here and any further questions to make this post more useful for other HFSS users (hopefully).nnThanks!!nMAn

[AndyJP]

since your port has an impedance, the correct substitute would be a resistor, not a conductive wire.n

[mabdelr3]

@Andyjp... Hello gentlemen,nnI have tried a couple of simulations based on your suggestions. I will display my results and hope to receive your feedback and suggestions.First of all, I replaced the actual 3D thin layer of coating/insulator with the layered impedance boundary as illustrated below:nIn order to specify the thickness of my insulation layer and the material type, I selected the two-sided option as shown below:nInstead of the port that I defined previously in the driven modal simulation, I defined an RLC boundary and set my R value to 50Ohm (to mimic an actual port input resistance). Here are the convergence and simulation results:nnAfter that, I wanted to set my RLC boundary R value to 0Ohm to neglect the effect of the port (since results in the driven modal simulation are renormalized with respect to the 50ohm port anyway and is not normalized in the eigenmode simulation)... Here are the results in this case: nR = 0 Ohm in RLC Boundary:nnReplaced the RLC boundary with a piece of Cu connection (shorted the port):nAs you can see, the results are comparable, however, there is some mismatch that is clear in the Q value. It is worth noting that I have simulated my system before getting rid of the thin insulation layers (took a lot of time) and performed a freq sweep that showed the first self-resonant frequency to be at around 81MHz.... So it is kind of matching what I am getting now from the eigenmode simulation.nnI wanted to check If I will get the same first resonant frequency If I simulated the modified model (with layered impedance boundaries) in a modal simulation. I used two types of excitation:nLumped port nCircuit Port (to have better accuracy at lower frequencies in the Hz range)nUnfortunately, the position of the first resonance was less that what I got previously from the unmodified 3D model and the eigenmode simulation. (See the following attached impedance vs. freq curves):nLUMPED PORT RESULTS:nCIRCUIT PORT RESULTS:nThe resonance now lies at either 57 or 64 MHz rather than the 81~82 MHz that I had previously. In my freq sweep I tried both interpolation and discrete sweep types and they both yielded the same results.nnMy main question now is that which one is giving me the accurate results that I desire? (note that I checked the shell element option in the layered impedance boundary in the modal simulations)...nnI know this is a long post, however, I hope I can get some feedback from you soon!nnI greatly appreciate you help!nRespectfully,nMuhammad n

[mabdelr3]

... I hope to get your feedback soon!Thank you,nMAn

[AndyJP]

well... heh. I wish I could find a bug here. Normally I set Delta-f around 1%, but the deviation is more than 20%. input impedance of multipole is not z11, but lt should have the same zero... btw, your feeding appeared not that I thought from the first images. You've missed the variant with a resistor, but I guess it gave lower Q, first of all.n

[mabdelr3]

I think I may have found the issue... It is the Adaptive Solutions settings. I was selecting a single solution frequency and setting it to an arbitrary value. When I had a low solution frequency, the results of the Freq Sweep had a different self-resonant frequency from the Eigenmode simulation.nI just tried a 1GHz solution frequency instead of the low one I had before (in the modal simulation) and now, I have my resonant-frequency closer to the 81MHz.nI am glad I figured out this technicality but I still want to understand what is going on in the background...nnAny insights?.Thanks,nMAn

[AndyJP]

in the eigenmode solver you should set more than one frequency. 5-6 frequencies would be fine, I think. Usually, it does not cost much.n