Learning Forum

[yhoshino]

I use the HFSS to calculate the transmittance of a dielectric when polarized light is incident.nI confirmed the S-parameters of TE and TM modes with floquet port, but the calculation accuracy differs greatly between TE and TM modes. When the dB value is converted to magnitude with python, TE matches the manual calculation with an accuracy of 10 ^ -6, while TM mode matches only about 10 ^ -3.nTo solve this problem, I tried the following methods.n?Expand the vacuum part above and below the dielectric to about 10 times the incident wavelength.n?Set the frequency in Analysis / Setup to 500 GHz and change the value of Maximum Number of Passes from 6 of default to 10 .n?Changed the value of Maximum Delta S from 0.02 of Default to 1e-6.nBut the accuracy did not improve.nIs it possible to obtain the calculation result for TM wave with the same accuracy as TE mode by further improving the above value? Also, is there a solution other than the above?n

[AndyJP]

you don't need python to convert db to magnitude.n1E-3 is not bad.nNumber of passes is not a precision, but a number of attempts to build a mesh producing stable results (good convergence) . Set the desired convergence Delta S and the maximal reasonable number of passes. Limit number of steps only when you do a quick approximate search for poles and zerosnalways check the log consolen

[yhoshino]

Thank you for answering.nDoes it mean that a reasonable number of passes can be determined from Maximum Delta S?n

[yhoshino]

you don't need python to convert db to magnitude.1E-3 is not bad.Number of passes is not a precision, but a number of attempts to build a mesh producing stable results (good convergence) . Set the desired convergence Delta S and the maximal reasonable number of passes. Limit number of steps only when you do a quick approximate search for poles and zerosalways check the log console/forum/discussion/comment/100012#Comment_100012

Thank you for answering. I have some questions.n・Does it mean that a reasonable number of passes can be determined from Maximum Delta S?n・Is it impossible to match the calculation of TM mode with an accuracy of 10^-6?n

[AndyJP]

>Does it mean that a reasonable number of passes can be determined from Maximum Delta S?.O.nIt is defined by your model and materials. When you are familiar with your structure, you can guess a reasonable number of passes, but there is no theory for that. It is mesher-specific, and the mesher is a pure art.n>Is it impossible to match the calculation of TM mode with an accuracy of 10^-6?nWhy no? Let it keep improving the mesh, until the accuracy is reached.n

[yhoshino]

>Does it mean that a reasonable number of passes can be determined from Maximum Delta S?NO.It is defined by your model and materials. When you are familiar with your structure, you can guess a reasonable number of passes, but there is no theory for that. It is mesher-specific, and the mesher is a pure art.>Is it impossible to match the calculation of TM mode with an accuracy of 10^-6?Why no? Let it keep improving the mesh, until the accuracy is reached./forum/discussion/comment/100208#Comment_100208

I understood your explain, thank you.nSorry, I have one more thing to ask.nIs it possible to check the latest results while sweeping the Number of Passes?.

[AndyJP]

click update report in report context menu or update field plot during simulation to see the last pass results.nIf you set solution:AdaptivePass in report config, which means EVERY pass, it will plot every pass in the same figure.nWhen the simulation in complete, HFSS should update every figure to the latest pass. But sometimes it forgets. So don't forget doing figure update yourself before reading the results.n

[yhoshino]

click update report in report context menu or update field plot during simulation to see the last pass results.If you set solution:AdaptivePass in report config, which means EVERY pass, it will plot every pass in the same figure.When the simulation in complete, HFSS should update every figure to the latest pass. But sometimes it forgets. So don't forget doing figure update yourself before reading the results./forum/discussion/comment/100479#Comment_100479

I set solution to AdaptivePass and I succeed to update and display the Pass - S-parameter plot.nI want to display the transmittance value in TM mode on the graph, but it seems that the value is different from what I expected. The value when Number of passes was set to 6 another time and the value at 6 this time are very different. nThe attached photo is the actual figure. The forecast is that the transmittance will be around -0.09dB and I cannot understand why the value changes so much.nCould you give me any other advise on what is this value displayed now and where to change the setting to see the transmittance ?.nn

[AndyJP]

Mesh. Creating a mesh that smoothly follows UNKNOWN field gradients is a pure artistic process, worth the money you pay for HFSS.nAs with art, usually it is not perfectly repeated on every new run.n