calculating pec

I want to simulate some thin PEC structures, i.e slanted PEC plate with 0 thickness. anybody know how to realize it by FDTD? I just want to know the basic idea of calculating such thin structures. please...

regards,
tenderne

anybody, any hints or suggestion.... please...


regards,
tenderne

there are some free FDTD codes in ths forum u can use them

hi, akedar,

do you mean the codes for "thin" structure calculation? i have searched this forum, but cannot find any codes related. would you please show me where it is, just a link...? thanks a lot in advance.

regards,
tenderne

PEC structures are easily simulated in FDTD by forcing the electric fields inside of them to zero. I implement my codes this way:

1. Update fields as normal. Pretend here there are no PEC's.
2. Force fields inside of PEC's to be zero.

That's it! The last remaining trick, and most important, is assigning where to place the PEC structures on the Yee grid. You will get much better results if you place these such that tangential field components lie on the boundaries.

Hope this helps!

-Tip

hi, rrumpf,

thank you very much for your reply. but how to calculate very thin PEC board (<0.1mm). 'thin' here means that maybe PEC only account for 5% or 10% of a cell, and it maybe computing demanding if i use very dense mesh.

would you please give me some suggestion regarding such applications? thanks a lot in advance.

regards,
tenderne

Hmmm...

I suppose the most obvious approach is to use a non-uniform or conformal grid, but there is probably an easier solution to this. I will look through Taflove's book on FDTD tonight and see what I find.

In the mean time, can you just model it as a thicker PEC? If it is a PEC, then no fields can penetrate it regardless of thickness. For this reason, I suspect thickness will have little effect in your model.

-Tip

thanks a lot, rrumpf,

but i don't thin it is proper to model 'thin' pec as thicker one, suppose i have a 10mm cubic cell, and pec is only a 1mm thick slice, seems not so reasonable to model the whole 10mm cubic cell as pec block, is it? kindly let me know your idea. thanks in advance.

regards,
tenderne

Depending on what you are modeling, it can be. For example, if you are modeling a microstrip transmission line or a patch antenna, you can get very accurate results using this approximation where the PEC's are much thicker than they would be physically.

What are you modeling?

-Tip

thanks a lot, rrumpf,

i calculate some simple antenna structures with hard source excitation, as i am still a rookie in fdtd simulation.

take the structure in the attached figure as a example, in cell_2, only a very thin slice is pec, i think it is not right to model the whole cell as pec, is it? please let me know your opinion. thanks in advance.

regards,
tenderne

After paging through Taflove's book, essentially all of the solutions are "conformal" FDTD approaches. I recall a message where somebody left a link to some free CFDTD software with that capability.

I have never coded CFDTD myself, but I have run simulations of strange transmission lines and microstrip patch antennas. I always just set one layer of grid cells to be PEC. By doing this, the thickness is actually half of one grid cell because I only include the tangential fields in the PEC. I have always gotten pretty accurate results. Right now I have a simulation running of the original left-handed material with double split rings printed onto a printed wiring board. Thickness of the metal traces was very thin I am sure, but I just set one layer of cells (actually half) to be PEC and I am getting accurate results.

I recommend trying it and seeing what happens. It will get you started anyway.

You will also need to use a "soft" source and I think the total-field/scattered-field technique works pretty well. At some point I uploaded my dissertation to a discussion on 3D UPML. I devote Chapter 4 to FDTD and I describe the TF/SF technique in section 4.2.8. I could also provide some papers on this as well. For me, the biggest trick was realizing that all of the fields are offset on the Yee grid and you have to take this into account in terms of phases and delays between the different field components.

-Tip