Why many projects are still using numerical methods like FDTD or MoM?
the simulation is not enough?
Solving a problem means that we solve the Maxwell's equations. As you know from electromagnetics we are only able to solve some certain very easy problems by hand. In practice we want to solve more complicated problems, and this requires computational numerical techniques. Numerical methods in EM is a very large and important field called Computational ElectroMagnetics (CEM). The most common methods are the Finite-Difference Time Domain method (FDTD), the Finita Element method (FEM) and the Method of Moments (MoM). For example FDTD finds a solution by approximating the Maxwell's equations with finite differences. An important thing in CEM is to use the correct method for your application since different methods have different strenghts and weaknesses. For example, MoM is suitable for small wires and plane structures like dipole and patch antennas. On the other hand MoM is very inefficient for problems involing complicated structures and materials. There is no method which is perfect for all problems. All of these methods typically results in solving matrices. These matrices can be very large and we then want to use the method which gives the smallest and the most sparse matrix. You can read more about these methods in for example Wikipedia.
Thank you for your complete answer.But my main quastion is when commercial simulators such as CST MWS and HFSS can solve our problem what is the necesity of numerical method.because these simulator use same method for example CST uses FIM(Finite integration method) and so on...
Added after 4 hours 49 minutes:
Running a simulation in any software means that we solve the problem numerically using some method, for example HFSS uses FEM, CST uses FIM. Then we have solved our problem, using some method supported by the software. It is good to understand the underlying method when you use some commerical method, because then you know how to build an efficient model for your application and how to set different parameters in the software for best computational efficiency and accuracy.
hey.... i thought that CST uses FDTD ?
CST MWS actually has several solvers available. The standard is a Finite Integral Technique (FIT) solver, which is basically FDTD with integration instead of differentiation. It also has a direct MoM, iterative MLFMM, and eigenmode solvers.
There are a variety of reasons that people might choose to use their own implementation of some numerical technique, but I think two reasons stand out.
The primary reason is cost. Since most commercial software licenses cost over S50,000US for one seat, they are not an option for many small companies and universities. Universities can get a substantial discount, but it's still several thousand dollars. While individual users could probably download a less-than-legal copy, corporations and universities are held to much higher standards, and would be in serious trouble if they were caught.
The second reason is flexibility. As has already been pointed out, some methods solve specific problems more easily than others. If a user has a very specific problem, he can optimize the technique of choice for that problem to make it more efficient than a general purpose solver used in commercial software. Many people simply prefer to know exactly what's happening in the code.
There are other reasons, but I think those are the most common.
CST is based on Finite integration technique.
I have made a study on the top EDA software on the market today.
You can read it at my website www.mariomansour.com
Hi Mansour,
It's a great website. I guess you already try all of these simulators. I interested in Microstripes. Can you help me more about it? I need more tutorial related with Microstripes. I have any questions to you.
1. How about with the accuracy?
2. Is it can use to simulate reflector antenna?
Thank you.
Hi mansour
Thank you for your reply and help.I want to know when we can use for example CST or HFSS what is the necessity of writing the Numerical method such as MoM or Fdtd?
Hi,
Nothing personal meant here but in my opinion this doesn't seem to be a balanced view of the simulators - I really don't agree with quite a few of the strengths and weaknesses of some tools. For example, there are only two weaknesses associated with one of the packages but from my experience there are a lot more. Similarly, you have come to some conclusions that I would never have reached with some of the other tools. My message would be to take such comparisons with a small pinch of salt - one of the simulators can solve a wide range of problems and its interface is one of the best around - the opposite was claimed here despite the fact that some others really do have relatively primitive GUI's. GUI is also a matter of opinion - they may all look the same but you should always ask the question, when something seems to be arduous, "could I do this with the other simulators"? The devil's in the detail.
My point is actually that the best way to compare these tools is to test them yourself. Most comparisons, despite the good intentions, are always a matter of opinion. Secondly, who is the author of the comparison and which is his/her favourite package and how good is the author's knowledge of the other tools? Everybody has a simulator/package of choice and is more or less an expert in one of them and more so than in others.
Another point made earlier in the thread : cheap software is not equal to saving money. Okay, you've either got the money or not but if you are working on the preprocessing for a full day (and not even for a CAD import!) and a few hours on the post-processing then how is the software cheap? What about parameterisation - Something that I, many years ago, inefficiently spent a lot of time on? Engineering cost of a project is generally dominated by the engineer time per project, then software price, then hardware price etc. What about bugs and service packs, reliability, support? It is not the strongest argument but it's better to go for a popular tool which many companies use and count on for their work.
aw
In an interview for multinational company, the director said to my friend. "Its job of a student to write his/her own code for simulating a problem".
Hi, thanks for your review on the EM simulators. By the way, do you know of any simulators that support multicore simulation by default without any add on license or upgrade? thanks.
I use microstripes to simulate my microstrip antenna design. It can run in dualcore microprocessor (AMD Athlon 64 X2 Dual Core Processor 3800+). Is it what you mean?
Good news to users, Bad news to other EM Simulator softwares
CST announced today the acquisition of Flomerics EM business including MicroStripes and FLO/EMC
Best Regards,
Mario
Wireless Solutions
You could consider:
GEMS, a FDTD tool, 3 month demo available
http://www.2comu.com