| T O P I C R E V I E W |
| tomed |
Posted - Apr 02 2015 : 17:38:54
Dear Enrico,
Please would you like to tell me how to correctly define the ground plan meshing by seg1/seg2?
Is there any rules like for nhinc/rh and nwinc/rw linked to the skin depth?
Thank you
Appreciate your help and continuous support.
Best regards,
Tomed |
| 9 L A T E S T R E P L I E S (Newest First) |
| Enrico |
Posted - May 11 2015 : 18:24:44
quote:
However, for the code above, i've got something strange.
In fact, i've compared fasthenry2 output file Zc.mat.
For "together and broken" examples Zc.mat deals with admittance, and on the file got from this code only impedance.
I don't understand the missmatch, please can you help?
Actually this is not due at all to the current visualization, but to the '-x' option that specifies to calculate the currents only related to 'trace1'. This is an optimization option, since the problem is symmetrical, but does not apply in your case, as you have one trace only.
So why removing the second conductor yields an admittance matrix instead of an impedance matrix? Because actually FastHenry solves the equation Is = Y'*Vs where Is is the vector of port source currents, Vs is the vector of port source voltages, and Y' is the ammittance matrix of the system. Now when Y' is found, it is inverted to provide to the user the impedance matrix Z'. This can be easily done since Y' is small (the dimension depends only on the number of ports).
Now when you remove one or more port, to speed up the solution, you are left with a rectangular matrix Y', that cannot be inverted. So FastHenry2 will provide you directly Y' in this case. Up to you to consider if symmetry plays a role, and if you can re-build the whole matrix, provided you are interested in it.
quote:
Also, on the matlab plots, what are the X-label, Y-label?
Guess you mean what is represented on the X axis and Y axis. The reference system is relative to the ground plane. Using the MatLab commands specified in the User's manual, you have on the X axis the current phasors for the filaments along Seg1, and on the Y axis the current phasors for the filaments along Seg2. Now these are phasors, and are complex-valued. This is why the user's manual suggest to use the low freq and hi freq cases, showing the real part in the former and the imaginary part in the latter, if your purpose is current visualization.
Best Regards,
Enrico
|
| tomed |
Posted - May 07 2015 : 16:08:01
|
| tomed |
Posted - May 04 2015 : 18:28:05
Dear Enrico,
Please this is the code i've used to try current distribution visualization:
*exemple simple d'une piste au dessus d'un plan de masse
.units mm
.default sigma=5.8e4
g1 x1=0 y1=0 z1=0 x2=100 y2=0 z2=0 x3=100 y3=100 z3=0
+thick=1.2 nhinc=11
+seg1=20 seg2=100
+nin (50,0,0)
+nout (50,100,0)
N1 x=50 y=0 z=5
N2 x=50 y=100 z=5
E1 N1 N2 w=0.254 h=0.0254 nhinc=5 nwinc=11 rh=2 rw=2
.external N2 nout trace1
.equiv nin N1
.freq fmin=1e-1 fmax=1e19 ndec=0.05
.end
On the fasthenry examples(together and broken), using matlab commands i've got the figure 15, 16, 18 and 19 from the user's guide.
However, for the code above, i've got something strange.
In fact, i've compared fasthenry2 output file Zc.mat.
For "together and broken" examples Zc.mat deals with admittance, and on the file got from this code only impedance.
I don't understand the missmatch, please can you help?
Also, on the matlab plots, what are the X-label, Y-label?
Thank you
Appreciate your time & continuous support
Best regards,
Tomed |
| tomed |
Posted - May 01 2015 : 12:32:56
Dear Enrico,
Thank you for your help.
I'll review the user's guide and may be recome back to you if needed.
Best regards,
Tomed |
| Enrico |
Posted - Apr 30 2015 : 18:36:05
The user's guide can be simply opened from FastHenry2's main page of the embedded help, see at the bottom of the page.
Best Regards,
Enrico
|
| tomed |
Posted - Apr 30 2015 : 17:52:50
Dear Enrico,
Thank you for your help.
I'll used matlab.
I guess there is some missing pages on the user's guide i've downloaded.
Where can i download the user's guide?
Best regards,
Tomed |
| Enrico |
Posted - Apr 28 2015 : 23:33:16
This is explained in some detail in chapter 4 of the FastHenry's user's manual. If you intend to use the files as they are, you would need MatLab.
You may use also other softwares for visualizing the current distributions, but in this case I'm afraid that in most of the cases you need either to elaborate the file, or to write your own import routine.
I would recommend to use SciLab, a free and powerful alternative to MatLab, where you can use the command 'loadmatfile' to load a matrix in MatLab .mat format.
Best Regards,
Enrico
|
| tomed |
Posted - Apr 24 2015 : 17:55:08
Dear Enrico,
Thank you for all your support.
I've got some issue with current distribution visualization.
By defining -d grids, -k text, -x portname, i've got a lot of .mat files(Grid1_x, Jimag1_x, Jmag1_x, Jreal1_x, and b).
Would you like to help me understanding the purpose of each file?
Also please, how to plot the current distribution using those files?
Thank you
Appreciate your time & continuous support
Best regards,
Tomed
|
| Enrico |
Posted - Apr 06 2015 : 11:06:50
Again, it is difficult to give general rules. As a rule of thumb, you should not need a discretization finer than the other features you have in your geometry; this however is valid for "high" frequencies, when the current tends to crowd under the metallic structures you have over the plane (e.g. wires). You have anyway an option to visualize the currents in the gnd plane, exporting them in Matlab-compatible format, please check the user's guide for details on hiw to do it.
So you can:
1) refine step by step the discretization for the same set of frequencies and see if there any significative change in the results (expecially at the highest frequency), and in case there us, refine again, since you have not reached convergency (in terms of geometrical refinement). Increasing the discretization is easy as you just need to increase the number of segments. Take care about nhinc and nwinc as these may lead to a huge overall number if filaments. Better very low nhinc, nwinc (compatible with skin depth) but higher number of segments.
2) check the current distribution (visually) and refine if too coarse
Best Regards,
Enrico
|
|
|
