Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

HFSS number of passes

Status
Not open for further replies.

Karen1992

Newbie level 6
Joined
Jul 18, 2016
Messages
12
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
80
Hello,
Who can explain what is maximum delta S and maximum number of passes in detail?
 

Welcome Karen1992,

These topics have to do with convergence (or 'correctness') of the solution, which in HFSS for Driven simulations is evaluated on the basis of the scattering parameters observed at each port.

HFSS uses adaptive meshing, which is to say that it sets up an initial mesh, solves the fields, and then re-meshes based on where the fields have a high concentration and/or gradient. Each re-meshing step is called an "adaptive pass". Importantly, at each step, the scattering parameters are evaluated at each port, and compared to the previous step. The difference between the two is called "delta S".

So, in order to make sure a simulation is correct, HFSS does adaptive passes until the delta S falls below a set threshold. HFSS also gives you a large degree of control over this process: you can set delta S to be whatever you want, and you can also tell it to do a maximum number of passes, such that the simulation will stop whether or not the scattering parameters have converged.

Hope this helps!
 
Can you please clarify me further about the adaptive meshing.
My issue is that I have to sweep from low frequency to high frequency range to plot my dispersion curve data.
So should I define my mesh with respect to highest occurring frequency or define an optimal starting mesh and HFSS will take care automatically for higher solution frequency.
As far as I know that mesh is directly related to frequency so higher frequency means more refined and more dens mesh
 

Usually, you want to mesh at the higher frequency, with the assumption is it will be as valid at low frequencies.

This isn't always correct though: what's more important than meshing for the correct frequency is meshing for the correct mode. Make sure that at your highest frequency, the fields correspond to the desired mode. This will be the optimal, automatically generated mesh.

Even more optimal than this is a manually specified mesh, since you can set it to be exactly what you want.

You may need multiple solutions at different frequencies if there are a large number of dispersive modes in the frequency range of interest.
 
Thanks for your explanation.
Now my question is what you mean when you say mesh for the correct mode?
I am working for the Wake Field Analysis in Multi Bunch Normal Conduction Accelerating Cavities. Fundamental frequency for my structure is 2.998 GHz. Now for wake filed analysis I have to find all possible higher order modes and plot their corresponding dispersion curves.
Lets say I start with 10 Higher order modes highest among them is around 7GHz.
So how can I mesh in this case for correct mode?
Your help in this regard will be highly appreciated as I am getting very strange modes very close in frequency range.
 

Attachments

  • modes_PEC.png
    modes_PEC.png
    17 KB · Views: 341

In the case of multiple modes, you just need to ensure that the mesh is adequate for all of them. For a large number of modes, I would suspect this would be approaching a uniform density mesh.
 

So it means concept of meshing is same that is meshing will depend on the frequency and we have to define mesh keeping in the view the max frequency.
During simulation HFSS use adaptive meshing depending upon the definition of delta s defined in solution setup.
This adaptive meshing is not sufficient to take care for meshing at higher frequency. I mean we define an optimal mesh at 2.998 GHz and then let HFSS take care for meshing at higher frequencies and accuracy.
Please let me know if my assumption is correct.
Your help will be deeply appreciated
 

Dear PlanarMetamaterials thank you for replying.I have another problem

For the fixed value of the Maximum Number of Passes 6 and frequency when we change the value of the Maximum DeltaS from 0.02 to 10^-4,10^-5 the value of S12 changes from 29.1028 to 28.7552 , 28.1552.
Then, for fixed of the Maximum DeltaS 10^-4 , if we change Maximum Number of Passes from
6 7 8 9 10 15 20 30
S12 =28.1552 , 27.9465, 27.8402, 27.6423, 27.4979, 27.0001, 26.7901, 26.6412
Could you please answer me how we can choose Maximum Number of Passes in our calculation?
 

There is always message in HFSS indicating if the solution has converged or maximum number of passes condition is reached , So define the maximum number of passes in the way that your solution converges.
You can see very clearly that error in S11 is large for smaller number of passes but as you increase the number of passes then this error is reduced, so it depends on how accurate solution criterion you have defined. If you want very accurate solution then you have to define more number of passes to allow it to converge. Goal is to reach to convergence.
 

Could you please answer me how we can choose Maximum Number of Passes in our calculation?

You want the highest number of passes possible. I'm pretty sure it is impossible for properly converging computations to become more inaccurate with a higher density mesh.
 

I'm pretty sure it is impossible for properly converging computations to become more inaccurate with a higher density mesh.
I agree that in case of properly converging computations increasing the number of passes can not increase the error and that comes from proper definition of model and the most importantly proper definition of mesh.
Also nothing comes free as increasing the number of passes will require more computational resources. So what is the most crucial is proper implementation of model and then the core proper definition of mesh and then if you can use symmetry plans you can save lot of computational resources and can get very accurate solution with smaller number of passes
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top