Re: Zeland Fidelity : post-processing issue (.fdfield file)
Does anybody know how is organized the file .fdfield ?
three first columns = position
(for personnal post-treatment, it is better with a uniform mesh)
fourth and fifth columns ? real and imaginary parts ?
Is there any logic concerning the field components (Ex Ey Ez Hx,...)
Linear or dB values ?
Thank you
Added after 3 hours 21 minutes:
Everybody is on holiday
Well I will give my own partial answer to my question !
This forum is done for sharing its knowledge, isn't it ?
In the .fdfield file, I?v found the following line: FdVolume 0 0 74 0 40 0 71 1 1 1
74 40 and 71 depends on my grid choice.
74 for X, 40 for Y and 71 for Z
It does not indicate the position, only the 'number' of the vertex
Then you have to build a dependence between this number of the vertex and the real position (care to non-uniform meshing !)
Then if you search a same localization (for example, the last position: 74 40 71), you?ll find six times this information on the .fdfield end to end.
?corresponding to the six fields components.
But, what is the order of those components? I don?t know yet?
All those components does not begin at the same localization
Component n°1 : 1 0 0
Component n°2 : 0 1 0
Component n°3 : 0 0 1
Component n°4 : 0 1 1
Component n°5 : 1 0 1
Component n°6 : 1 1 0
Is there a relation between the FDTD method which shifts the grid for Efield and Hfield?
Hi:
The .fdfield contains the frequency domain near field information. Whe a user defines SAR calculation (or frequency domain near field calculation), it will create this file. A user has a chose to ouptut E field only in the simulation setip. The .fdfield is supposed to be used by the GUI for post processing the f-domain near field and SAR only. Its data may not be useful to users. The near field is not the absolute value. It is related to some input power which may not be in the file. That is why the near field data here may not be useful to you. In the post processing, the GUI will allow you to decide the excitation and display the final data with the specified excitation to you. Regards,
Those data may be useful for me?
For my specific application I?d like to get, for each voxels (1x1x1 mm^3), the simulated values of each component of E and H fields on a media facing the antenna?
Fidelity is pehraps not the appropriate soft for me ?
Nevertheless, as mentioned before in my post, I think that this .fdfield file contains this information, isn?t it?
[If related to an input power of 1W ?]
When only E near-field components are plotted, does the following organization of the file is correct?
?PosX? ?PosY? PosZ? ?Re_Ex? ?Im_Ex? ___ with (Nx)x(Ny+1)x(Nz+1) lines
?PosX? ?PosY? PosZ? ?Re_Ey? ?Im_Ey? ___ with (Nx+1)x(Ny)x(Nz+1) lines
?PosX? ?PosY? PosZ? ?Re_Ez? ?Im_Ez? ___ with (Nx+1)x(Ny+1)x(Nz) lines
With Nx, Ny and Nz, the number of vertexes chosen for the grid?
I?m not a member of the zeland staff, I just try to interpret this file and serious elements let me think that it would be possible to get my needed information thanks to this file?
----------------------------
If what my interpretation is wrong (which is quite possible :- ), how can we get, with Fidelity, the 6 near-field components corresponding to each simulated position ?
Does Fidelity allows us such a result ?
I'm looking for extracting Ex, Ey and Ez fields in the frequency domain (f-domain near field).
For a given frequency, I need to extract those components on a given volume facing the antenna.
Does Zeland Fidelity allows this post-treatment ?
Thank's for your contribution.
Hi, Littlindian: I am sorry not to be able to respond to you earlier because I had been on vocation. Theoretically, you can get the t-domain near field and do a DFT for the f-domain near field. However, it wil involve too much data. For the current situation, we normally suggest users to save about 10-50 frames of t-domain near field. If you want to do DFT on the near field, you may need to save many frames. However, each frame may take a lot of hard disk space. If you want to save thousands of frames, it may blow up your hardrive for the near field in a not-very big volume. This can be a problem.
I trust you if you say that this action may be quite-hard (long & bulky) to implement...
If you propose this action, this is because the fdfield file is not abble to give me the needed informations...
... despite the fact that a visualization of all E-fields components is availabe when a SAR calculation is run ?
I 'm still not totally convinced and you still not write : 'no it's not possible'
Let me say that, whatever would be your answer, I'm happy to take advantage of your knowledge on Fideltity.
Regards
Hi, Littlindian:
I checked the code. The .fdfield is for the frequency domain field. There are two possiblities:
1. If you require FIDELTIY to solve the SAR, E-field and H-field, each line in the file store total 19 items.
i j k SAR P_ABS ReEx ImEx ReEy ImEy ReEz ImEz ReHx ImHz ReHy ImHy ReHz ImHz
The i j and k are the 3D grid indices.
The SAR is the SAR calculated in the cell and the P_ABS is the power in the cell. The Ex, Ey, Ez, Hx, Hy and Hz are the f-domain fields are the grid point.
One question you may have is "what is the input power?". The input power is defined in the .poy? file.
2. If you do not require it to save the H-field, each line stores 11 items (no Hx, Hy and Hz).
Hope this help. Thanks!
Yes ! I'm relieved.
My interpretation of the fdfield was correct
Well, It's a good new because I had already write matlab "code" for extracting those components and made some post-treatments.
The button "helped me" has never been so meaningfull for me.
Hope this will hope other users of Fidelity ... and edaboard
See you next time.
Regards.
I did not see this file that way !!!
In this file, I have not 19 (or 11 colums) but a series of 3 sections separated by
"fdplane", "fdvolume" and "fdmedia"
which all contains 5 data : " i j k unknown1 unknown2 "
Are you really talking about the fdfield0 from Zeland Fidelity ????