| Author |
 Topic  |
|
|
Capponero
6 Posts |
 Posted - May 26 2015 : 15:05:51
|
Hello,
I am using FastHenry2 for simulating the coplanar waveguide. In the real case I have input and output electrodes, which are bevel, but I have no idea how to create them in FastHenry. The straight waveguide is created simply by defining two points and connect them with defined width w. So my question can be simplified as:
How can I draw a trapezoidal segment (waveguide)?
(see picture in following link - right hand side)
ht*p://imagebank.osa.org/getImage.xqy?img=cCF6ekAubGFyZ2Usb2UtMjAtMjQtMjY2OTYtZzAwMQ
Thanks a lot in advance |
|
|
Enrico
545 Posts |
Posted - May 26 2015 : 16:16:47
|
In FastHenry, all primitives are segments. They can be rotated along any axis, but stay segments. However this does not mean that you cannot approximate different shapes. Actually a segment is broken into filaments automatically by FastHenry in any case. The filament is the basic element whose current is calculated by the solver, but actually a filament is nothing else than a segment that won't be broken down.
So you can use a bundle of segments (just like a segment is a bundle of filaments) to approximate any shape; in your case, you should use a sort of ladder on the sides. Up to you to decide if your segments, or some of them, or none at all, can be split into filaments (using the 'nwinc' and 'nhinc' directives) for simulating skin & proximity effects.
Best Regards,
Enrico
|
 |
|
|
Capponero
6 Posts |
Posted - May 26 2015 : 16:37:31
|
Thanks a lot for your answer. Actually this solution came to my mind, but it seemed to be very obsolete, especially in the case, when I want to experiment with with different dimensions of the trapezoid electrodes.
I was hoping, that there is a simpler way, such as: I define 4 points (x and y coordinates) and then join them to a planar quadrilateral. And the nwinc and nhinc will be taken from the default and applied automaticaly on the created segment.
My goal is tu tune up the dimension of the electrode according to following image
segment 2 can be easily created by two points. I am searching a way how to create segment 1, where can be value of x easily changeable.
Do you have an idea how to do that?
Thanks
quote:
Originally posted by Enrico
In FastHenry, all primitives are segments. They can be rotated along any axis, but stay segments. However this does not mean that you cannot approximate different shapes. Actually a segment is broken into filaments automatically by FastHenry in any case. The filament is the basic element whose current is calculated by the solver, but actually a filament is nothing else than a segment that won't be broken down.
So you can use a bundle of segments (just like a segment is a bundle of filaments) to approximate any shape; in your case, you should use a sort of ladder on the sides. Up to you to decide if your segments, or some of them, or none at all, can be split into filaments (using the 'nwinc' and 'nhinc' directives) for simulating skin & proximity effects.
Best Regards,
Enrico
|
|
|
|
Capponero
6 Posts |
Posted - May 26 2015 : 16:42:27
|
here is the mentioned picture (please rewrite the ht*p at the begining)
ht*p://s24.postimg.org/mtjp0rmbp/Untitled.png
btw, i have no idea, why the forum code for images doesn't work here:
 |
|
|
|
Enrico
545 Posts |
Posted - May 26 2015 : 18:31:43
|
Forum images are disabled due to spam. This is the same reason why w*w and ht*p are automatically masked, so links must be manually copied & pasted, and this is implying a willing action (cannot be clicked by mistake, bringing you to illegal or infected sites).
Sorry about that, the other option was to authenticate every Forum user, and at the time we just chose the first option.
Regarding your question, well the fastest option is still to build up the bevel via approximation, but to do it via a script. You could also implement this option into FastHenry2, as the source code is fully available, so it will be fully integrated in the solver.
If this task is too time consuming for you, we also can customize FastHenry2, if this is of your interest, but as your feature is unplanned at the moment, I'm afraid we would need to charge the cost of the modification; please let me know if you could be interested in this option.
Best Regards,
Enrico
|
|
|
|
Capponero
6 Posts |
Posted - May 27 2015 : 16:43:27
|
Well, since I am PhD student working on my dissertation right now, I am not very interested to "buy" a new option of FastHenry, also because of the fact, that I am not sure that I will need it also in future. I am more interested about the first option.
My numerical program is little bit more complicated. I am working with matlab, where I calculate all the parameters. The program then changes the source code for FastHenry and runs it in command prompt. Then it reads the output of impedance and works with it further.
Would you be more specific, how exactly can I implement the script in the FastHenry2 source code in order to simulate the impedance of the above mentioned waveguide?
Thanks a lot
quote:
Originally posted by Enrico
Forum images are disabled due to spam. This is the same reason why w*w and ht*p are automatically masked, so links must be manually copied & pasted, and this is implying a willing action (cannot be clicked by mistake, bringing you to illegal or infected sites).
Sorry about that, the other option was to authenticate every Forum user, and at the time we just chose the first option.
Regarding your question, well the fastest option is still to build up the bevel via approximation, but to do it via a script. You could also implement this option into FastHenry2, as the source code is fully available, so it will be fully integrated in the solver.
If this task is too time consuming for you, we also can customize FastHenry2, if this is of your interest, but as your feature is unplanned at the moment, I'm afraid we would need to charge the cost of the modification; please let me know if you could be interested in this option.
Best Regards,
Enrico
|
|
|
|
Enrico
545 Posts |
Posted - May 28 2015 : 11:36:40
|
Well if you are a student, I fully understand your point. Actually I would say there are two kind of users:
- students / academic people in general, who don't have much money, but can provide contributions
- industry professionals, who don't have time or deep knowledge to contribute, but can spend money and are interested in fast result
You are very welcome to contribute, and this is also a benefit of the open source ecosystem.
Before going in further details, which version of FastHenry are you using? Are you working under Windows, or are you compiling your own version under *nix?
Enrico
|
|
|
|
tomed
France
28 Posts |
Posted - Jun 07 2015 : 19:21:02
|
Hello,
I'm also student and my concerns came from studying transmission lines theory.
Please for this topic, the cross section is non uniform? Isn't it?
Is there any link with transmission lines theory?
By this way, how fast henry2 calculates R and L?
How fast Henry2 takes into account this kind of section?
Thank you
Appreciate your time & support
Best regards,
Tomed |
|
|
|
Capponero
6 Posts |
Posted - Jun 09 2015 : 16:52:24
|
Hello again
Sor for longer time to answer, I was at conference for one week.
Well, I work with FastHenry2 3.32 (I am using it in simulating superconducting coplanar waveguides). I am using it under Windows 7 operating system and I am mostly running it via command prompt (ran by matlab/octave using the system command). The layout I occasionally check in FastFieldModel.
I would like to solve above mentioned problem, since it is very obsolete to do it manually (draw trapezoidal segments with changeable base)
quote:
Originally posted by Enrico
Well if you are a student, I fully understand your point. Actually I would say there are two kind of users:
- students / academic people in general, who don't have much money, but can provide contributions
- industry professionals, who don't have time or deep knowledge to contribute, but can spend money and are interested in fast result
You are very welcome to contribute, and this is also a benefit of the open source ecosystem.
Before going in further details, which version of FastHenry are you using? Are you working under Windows, or are you compiling your own version under *nix?
Enrico
|
|
|
|
Enrico
545 Posts |
Posted - Jun 09 2015 : 23:20:12
|
The best way to make the community benefit from your improvements in FastHenry2 is to work on a branch under revision control.
Before digging into where in the code you can act, are you familiar with git?
Actually we were planning to host the sources on GitHub (the FastHenry project is there, even if still empty at present). You can fork the project there, and once cleaned up, we can even pull it into the master branch for everybody's benefit.
Best Regards,
Enrico
|
|
|
|
Capponero
6 Posts |
Posted - Jun 10 2015 : 12:50:14
|
About Git I know only, that it is used in Linux Kernel, but I have never worked with it. I am physicist so I have experiences with Matlab/Octave, C++, and little bit Java.
So to clarify the topic. It is not possible to solve my problem now, because that kind of feature (create and simulate non-perpendicular segments) is not part of the FastHenry2, but it is possible to implement this feature additionally. Am I right?
quote:
Originally posted by Enrico
The best way to make the community benefit from your improvements in FastHenry2 is to work on a branch under revision control.
Before digging into where in the code you can act, are you familiar with git?
Actually we were planning to host the sources on GitHub (the FastHenry project is there, even if still empty at present). You can fork the project there, and once cleaned up, we can even pull it into the master branch for everybody's benefit.
Best Regards,
Enrico
|
|
|
|
Enrico
545 Posts |
Posted - Jun 10 2015 : 19:42:48
|
quote:
About Git I know only, that it is used in Linux Kernel, but I have never worked with it. I am physicist so I have experiences with Matlab/Octave, C++, and little bit Java.
Actually git, although born for the Linux Kernel, is independent from it and very flexible. It is not difficult to use in the basic commands. Under Windows you have the nice option to install a GUI called TortoiseGit that allows you to do most of the operations via menus and dialogs. I would recommend you a short reading at ht*p://rypress.com/tutorials/git/index, then you will tell me if this is ok for you. Otherwise of course you can still download and modify the sources, but keeping trace of the modifications etc. may be complex, while Git allows you to experiment over different 'branches', and you can always go back to the original code, or move to a different branch, so you don't need to keep multiple copies of the sources.
quote:
So to clarify the topic. It is not possible to solve my problem now, because that kind of feature (create and simulate non-perpendicular segments) is not part of the FastHenry2, but it is possible to implement this feature additionally. Am I right?
Well, not completely right. You can solve your problem even without changing the source code, but you still need to approximate the shape you are interested in with a bundle of rectangular segments.
Changing the source code will give you additional flexibility, and as this solver is fully free, it will be for everybody's benefit.
Best Regards,
Enrico
|
|
|
|
Enrico
545 Posts |
Posted - Sep 23 2015 : 17:32:59
|
I am guessing you solved your problem otherwise, but for completeness, let me inform you that under the 'Download' section you can find the full sources for FastHenry2 ported under MS Visual Studio 2013 Community Edition, that is free for Open Source project (check detailed license conditions on the MS site), supporting both 32 bits and 64 bits compilation and with zero warnings, i.e. the ideal starting condition if you need to customize the software.
The sources are now hosted on GitHub, see details under the 'Download' section.
Best Regards,
Enrico
|
|
|
| |
Topic |
|
|
|
Create a trapezoidal segment in FastHenry
