dispersion diagram
I am trying to construct the dispersion diagram of Sievenpiper's structure by using an eigensolver in HFSS. Basically, I have a unit cell of his structure. I have PML, which is situated at about 5-6times the height of the structure. I define master/slave boundary on the sides.
I get reasonable agreements with his results between the light lines in his paper. However, when the wave vector is very close to Gamma of the Brillouin zone, there is a lot of loss. Basically, my imaginary frequency is very high in the solutions at these points.
Anybody uses HFSS to construct dispersion diagram? I am not sure how to use the PML boundaries here. Please help.
Thanks in advance!
You can email me if you still in need for an answer
Hi George137,
I would like to know. What is your email?
Thanks,
hertz
Hi Hertz,
Did you ever solve the dispersion diagram problem in HFSS? I have a similar problem...if I only solve for the 1st mode, convergence and the generated dispersion diagram is accurate. However if I tell HFSS to solve for 4 modes, the results are very bad, that is the eigenvalues for the 2nd, 3rd, and 4th modes do not appear correct...Any help is appreciated.
Since the 4th eigenmode (particularly at the M-symmetry point) appear at quite a high frequency. A dense mesh is usually necessary.
What should I set the convergence to and how many modes should I solve for? Also should I regenerate the mesh for each eigen-point or can I use the same mesh? Also what does converge on real-frequency only do? Shouldnt it alwasy be real?
I'm not running HFSS these days - I'm using the Eigenmode solver in CST MWS to solve these types problems now. Nonetheless, I did frequently refer to a solution of the Sievenpiper structure using HFSS to guide my own work in MWS. Attached are a presentation and MWJ article by Richard Remski - an Ansoft applications engineer - from a few years ago showing the procedure for modeling an EBG structure and generating the dispersion diagram using HFSS.
Hope this helps,
eckypting
I set the convergency to 0.0001 so that it almost never reached, therefore the simulation will stop when the number of passes are completed.
You must use the same mesh otherwise you won't get a smooth curve, but using the same mesh is only possible in version 8 and 10, not version 9 - Why the developer neglect this option in version 9 beats me.
Eigenvalues are presented as real and imaginary values. Therefore one can converge on only the real part if the option is enabled. Eignenvalues shouldn't be just real.
How can I use the same mesh in version 10?
Added after 3 hours 31 minutes:
An imaginery eigenvalue means there is loss, is that right? But, if the structure has no metal/dielectric loss and is in the slow-wave region (no radaition), then the eignevalue should be purely real (imaginery part is insignficant). So for higher order modes (fast-wave region, possible radiation loss), the eigenvalues will have both real and imaginery values even if the structure itself is perfect (no metal/dielectric losses). Therefore, one should never check the 'converge on real values only' if one wants to solve for higher order modes.
Can someone please tell me if this is correct? Thank you.
Hello,
I too am using hfss to create dispersion diagram of sievenpiper structure. How many modes should I solve for and at which beta point should I generate my mesh? I will reuse mesh for other eigenpoints. Please reply.
does anybody konw the method of simulating this dispersioin digram with CST.
The Eigen Mode Solver is required to generate a dispersion diagram in CST MWS. If you have it, then the periodic boundary is applied to the ±X & ±Y bounds of a unit cell of the structure to be analyzed, with the phase between the boundaries as parameter. The phase parameters are then stepped through the desired range of values to create the dispersion diagram β sweep. See the Ansoft app note for a good reference w.r.t. to a planar EBG structure.
Thanks. But the problem is what kind of boundary shall i set in Zmax? H=0 boudary or E=0 boundary.What is the physical explanation of this boudrary definition ?
Many thanks.
Added after 7 minutes:
Thanks. But the problem is what kind of boundary shall i set in Zmax? H=0 boudary or E=0 boundary.What is the physical explanation of this boudrary definition ?
Many thanks.
go to emtalk.com , they can help you there.
[quote="huilailiu"]Thanks. But the problem is what kind of boundary shall i set in Zmax? H=0 boudary or E=0 boundary.What is the physical explanation of this boudrary definition ?
Many thanks.
Added after 7 minutes:
If your solver supports an open boundary condition (which I believe HFSS does), then that would be preferable...and its physical interpretation is rather obvious. If you're restricted to a PMC or PEC boundary condition (whick I recall is the restriction in MWS), then the mode results that are above the light line will be influenced by whatever boundary you select. It should be noted that for determining the surface current bandgap, only modes that are below the light line are considered.
My thinking is that these modes (the ones above the light line...) are the same surface current modes that, in the case of an open boundary solution would either radiate off the structure (in the case of the TE modes that are above the first bandgap freq. range) or be bound to the light line (in the case of the TM modes that are below the first bandgap freq. range). Refer to Sievenpiper's seminal paper...
Care should be taken that sufficient distance be included between the Zmax of the solution space and the max Z dimension of the structure, to avoid and direct impedance loading of the structure.
Thanks for all post, does anyone has the methode for developping the structure's bandwidth ? thanks.
hi
does anyone simulated the sivenpiper structure for disperson diagram using CST MWS?
it would be great if someone post the procedure to obtain this diagram in cst
regards
I'm trying to simulate this diagram with cst 2008 but I don't know how to plot the results.
I think in result watch, i must add something more than the frequency mode to plot it but I don't know what.
Maybe with userdefined.
Anybody has an idea ?
That is the drawback in CST MWS.you can't do this with MWS.
Hi all,
Just came across this forum, and it definitely sounds heated up with respect to Sievenpiper's work and the dispersion diagram! I've found a dosument on ansoft's website that guides you to plot the 2D and 1D dispersion diagram (would like to share it here, except, am not too sure whether this is allowed, can somebody tell me if i can upload it here?)
Anyways, my main aim when plotting the 2D dispersion diagram is to find the eigenfrequencies of the high symmetry points (gamma, X and M). These values, along with the Bloch impedance will let me extract the inductance and capacitance values (LL, CL, LR, CR). My question is, does anyone have an idea about plotting the Bloch impedance? I've found a book on metamaterials that says that, the bloch impedance can be found by the formula ZB=2P/I*I, where P can be found from surface interation of E and H field of the port surface. This is in essence how the caracteristic impedance Zo is calculated on HFSS, using Zpi.
Does this mean I need to do the simulation in modal mode, and construct a port at the unit cell? If so, what's a good port size, since varying the port size will vary the terminal impedance. And, am i doing this correctly? Please help!
Thanks alot!