Who makes the cheapest FDTD solver

[biff44]

Have some large structures (5 foot long pipes, wr430 waveguide parts, feeds) That I need to do some emag analysis on. Have read thru many of the posts on this forum, but have a few questions:

It seems like large structures prefer an FDTD solver. But I have a waveguide feed illuminating the inside of a much bigger pipe. Can FDTD solvers handle waveguide ports, or do they need a voltage/current port as an input?

I am on a shoestring budget. Does someone make a fairly cheap FDTD solver that is relatively user friendly (ie import autocad files, or have easy to use drawing user inputs)? Am hoping for something free, or at most $2000 US.

Any suggestions?

 

[iyami]

It seems like large structures prefer an FDTD solver. But I have a waveguide feed illuminating the inside of a much bigger pipe. Can FDTD solvers handle waveguide ports, or do they need a voltage/current port as an input?

Quickwave should be able to do so. They seem to have one of the best toolkits for handling waveguides.

I am on a shoestring budget. Does someone make a fairly cheap FDTD solver that is relatively user friendly (ie import autocad files, or have easy to use drawing user inputs)? Am hoping for something free, or at most $2000 US.

They may have student/trial versions.

 

[biff44]

Interesting. It does look like it has the features I need. Any idea on the price? All of these programs take some effort to learn how to use, and I would not want to download a trial and then find out it was a small fortune to buy.

Rich

 

[iyami]

Interesting. It does look like it has the features I need. Any idea on the price? All of these programs take some effort to learn how to use, and I would not want to download a trial and then find out it was a small fortune to buy.

I don't think the regular price is within reach. I would expect something in the $20-40,000 range.
But if it is a onetime simulation the trial version might do. You might also contact them by e-mail
asking about short term licenses or academic versions etc.

I think if you look at some of the samples in the documentation it should not take more than a day
or two to get up to speed.

 

[biff44]

Yes, you are right! It does not fall under the "cheap" constraint. Any other ideas?

 

[iyami]

Yes, you are right! It does not fall under the "cheap" constraint. Any other ideas?

It is unclear to me for how long you need the simulator.
If it is a one time thing I would still try to talk to the Quickwave people.
(I also don't where you are located on the commercial/academic scale)

If you plan to use it for a long time then I only see two options

1) get the money you need from somewhere
2) write it yourself (using perhaps some of the open source codes as a starting point).

The high price tag of commercial engine is usually caused by

a) flexibility of the engine (support for many different applications)
b) optimizations (speed/memory footprint)

If neither a nor b is important to you then I think a reasonably skillful coder who
understands FDTD and your needs should be able to produce a working program
in a month. (Two if you want to use C or Fortran).

Ideally you could join forces with other people in search of a free FDTD engine.

Optimizations/extra features you can add then at your own pace.

 

[biff44]

$20,000 to $85,000 USD, plus yearly maintenance fees is a heck of a lot of money if you are an independent consultant who might need to do an EMAG analysis once every 2 months! Unfortunately, I might need the program today, and then again at the end of the year, and then in the spring....so a 1 month trial period is of little use. Plus, you really have to keep using these programs regularly in order to learn their tricks/quirks.

I am suprised that there is not a fairly good program, maybe missing the fancy bells and whistles, available somewhere in the world that sells for maybe $2000! Maybe some program written by a university for their students that is commercially available, or some older software that does the job but has not been extensively updated?

As far as writing my own...not going to happen! I am more of a system/component designer, and leave the heavy emag analysis to the big boys.

 

[katko]

I haven't used it personally, but MIT's Meep might meet your needs. All I know is that it's open-source FDTD which runs on linux and is largely scripted.

http://ab-initio.mit.edu/wiki/index.php/Meep

Take a look through it and see if it works.

 

[loucy]

If would be helpful to give more details of the geometry, frequency range and parameters desired. You might not be able to simulate it with a normal PC even if they allow you to try with the most expensive FDTD software.

 

[iyami]

I am suprised that there is not a fairly good program, maybe missing the fancy bells and whistles, available somewhere in the world that sells for maybe $2000! Maybe some program written by a university for their students that is commercially available, or some older software that does the job but has not been extensively updated?

Often for such programs you have to modify the code to set up the model. Another problem
is that adding new functionality is very difficult because of some hidden assumptions.

Actually I think what is needed is an open source EM engine (FDTD and/or MOM and/or ... ).
The problem with all the existing open source code is that they are too hard to extend.
A good design would be needed and a group of serious coders.

Be that as it may from all open source FDTD engines I have seen so far Meep left the best
impression. I do not know its capacity (memory requirements/cell, speed etc.) but among
open source engines it may be your best bet. I haven't followed what progress qucs has
made recently but it probably has not so much to offer for your problem.

Good luck.

 

[biff44]

Sorry, details:

2450 MHz +/- 10 MHz
Input WR430 rectangular waveguide (4.3 x 2.15 inch)
Waveguide runs into a wall, and there is some sort of transition thru the wall (rectangular iris, small horn, etc) feeding thru the wall.
On other side of the wall is a cylindrical vessel, 5 feet long, 2 feet circular diameter. With the cylindrical vessel lying horizontally, the bottom half of vessel is filled with lossy material of εr=3.

I want to play around with the various geometries to get uniform heating of the material in the vessel.

Added after 1 hours 25 minutes:

Meep looks interesting.

After reading about that one, I noticed several more free FTDT solvers at the bottom of this page:

https://en.wikipedia.org/wiki/Finite-difference_time-domain_method

Sounds like a plan.