electromagnetic simulators market

I would like to have a better Idea on the following topics:

1) How big is the global market for electromagnetic simulator and what are the growth prospects in the next 5 or 10 years?
2) What are the most important application areas that are not yet served in a satisfatory manner (because of some limitation) by the present commercial offer so that these areas could rapresent a good business opportunity fo a new entrant.
3) Related to point 2: are you satisfied about the speed and accuracy of the present electromagnetic simulators when used in your application?
How often (if ever) do you feel that an important design/optimization can not be made because of the speed limitation of your EM tools ?


PS: By looking around on the internet I have already seen that Ansoft (whose business is almost entirely in the EM simulator market) has a yearly revenue greater then 50 M$ so that the global marker should be greater then that. However I am more interested in the data of individual application areas then in the global figure.

in my opinion there is a huge market for EM simulators. With the increasing performance of new CPUs, I beleive more and more people will be willing to use these simulators. In the past, it was much more reasonable to just go into the lab and use a try-error approach. But with faster CPUs EM simulators are gaining more ground.

Hi,

I guess Irfan is right.....the market for EM simulators would grow as more and more devices be it active or passive components are incorporated into smaller packages. The need for EM study for these devices would be challenging and also comprises of various mutlidisciples like electrical engineering, mechanical and thermodynamics topics.

If I were to market a EM simulator. I would try to make it more user friendly and easy to use. Because I feel that in future, feasability of software to various applications be it an electrical problem, mechanical, thermodynamics or physics would be a major criterion in choosing it over the rest. Also by feasability I mean that a mechanical engineer with a little knowledge of electromagnetics should be able to solve a complex electrical problem, which I guess is the limiting factor now.

Most of the EM software today, lets the user decide what type of setup is valid for the given problem, which sometimes would lead to errorneous approximations. So an intelligent software with inbuilt database for different models, in which the user can modify the setup according to given specifications or user defined structure provision with an inbuilt design checker...... could I guess increase the sales of the software.

Also with the advent of fast CPU's, any complex matrices can be solved in matter of seconds or say minutes. So I dont see it as a limitation. The only limitation, as I described earlier is its effectiveness to solve different kinds of problem and easy user interface.

Thanks for your replies

I agree with maverick09 in that it would be nice to have an EM simulator with a built in "Expert System" that can help in the design of a new component and also check the integity of the component structure. The problem I see is that there are so many different applications which are characterized by different techologies, different requirements and different design methodes, that it would be difficoult to create this general purpose Expert System. A more feasible approach to address this need could be that the EM simulator gives a (high level) programming language which could be exploited by some open communities of RF/Microwave/Antennas/... designers or by other commercial developers to create libraries of modules dedicated to the design of special components.

I do not completely agree with the (quite common) statement that the todays CPUs are so fast that the simulation time is no more a central point. Of course it depends on the application and, even if in many cases the statement could be true, it is not alway true. In fact in my personal experience (I am involved in the design of passive components for space antennas) I have seen several cases where the CPU speed is not enough to allow a practical design/optimization of complex structures. My experience is limited to a small segment of the global market of EM simulators but I am quite convinced that a better speed and the capability of dealing with larger structures with be a premium also in some other areas. However I do not know which is the actual relevance of these needs in the global market of EM simulators.

Hi lagrange

I think that the market for electromagnetic simulator goes towards
The IC's market there's the big money and market. they begun with
signal integrity for the moment. the increasing in frequency will lead the IC's companies to use em software. for making efficient and better performance IC's. The IC behave like little oven becouse of radiation and dielectric loss
With the EM software you can tune it to decrees the loss and get better IC's

PL

Hi,

I have read this book "microwave circuit modeling using electromagnetic field simulation"- by daniel g. Swanson Jr., Wolfgang J.R. Hoefer. Its great book and discusses whats missing in modern CAD software especially in Chapter 2.

one especially distracting thing about the EM simulators is they have a lot of bugs. During the development process very carefull bug testing is required. I found many bugs in certain EM simulators and reported these issues. But the developers should be more carefull before releasing the product to the market. I hate to see several bug fixes during a month. These bugs cost me a lot of time.

Hi irfan1

That's the way of life during making we do mistakes and then we fix
them. It's better to create something and make mistake because we are learning
to do batter things from mistake. look at Microsoft they release windows and fixing the code, after delivering it to the market.
In some way we are participating in building the software by the finding the bugs
and asking questions.

PL

I found Irfan1's problems with bugs to be so true during the 8 years I did microwave design. That's why when I started Sonnet we put big time emphasis on making the software as bug free as possible and in making the documentation as readable as possible. I think most users with experience in a variety of tools will tend to agree.

In fact, it is actually critically important that we stay as bug free as possible and keep the documentation very readable. We have over 70,000 registered copies (about 35,000 are unique, quite a few users register multiple copies) of SonnetLite out there. If there were any significant problems at all, our support staff would be completely overwhelmed. And you would see all kinds of postings on this forum asking for help with bugs and strange error messages. Instead we provide what we believe to be the best level of support in the entire industry.

I can't say anything right now, but if you are going to the MTT Symposium, stop by our booth and I will personally show you part of exactly what I think the future of high frequency EM will include. I will also be discussing the future of EDA, espeically in regard to interoperability at a panel session on Wednesday (PWA). There are some seriously amazing trends happening right now, and that is very very good news for all EDA users.

in my opinion, it may be a good idea to distinguish between research orianted and commercial purposes orianted versions of the same simulators. It is difficult to distinguish the needs of both parties. But still there are some very deep differences.

hi Rautio

I would appreciate if you can post on this forum something related to the EM CAD interoperability and (more generally) to the new trends you have spoken about. I am developing a new 3D EM simulator. This is something I am doing at home just for fun but, if possible, I would like to make of it a professional tool which can be easely integrated in the design flow of the modern RF/microwave industry. Therefore I am very interested in the topics disussed in the MTTS panel which I can not attend at. In my professional experience (as an electrical designer of antennas components) I have used some general purpose 3d EM simulators (principally CST MWS which recently has been selected as the preferred simulator of our antenna department) so I have seen some pro and contras of these tools and I made a personal opinion of what can be done to make a better EM simulator. Nevetheless my experience is limited to a small segment of the possible applications and I would like to have a broader view of the actual market needs and trends.


Best regards

Hi lagrange -- I view interoperability to be the defining issue of this decade in high frequency EDA. I will have a few slides at the panel session. I will post them here at some point after I present them at the symposium. -- Jim Rautio

I am presenting these slides in a couple hours here at the MTT show, panel session PWA. Feel free to comment.

hi Rautio

Thanks for your presentation. Looking at it and at the actual interoperability among EM software tools I see that the common approach is based on several two side agreements each one of them solves the problem for a couple of tools and exploites a proprietary interface like EMsocket of AWR or any other. In my opinion it would be better to rely on a single open standard (that is an API which is defined and maintained by an open consortium) . Do you think it could be a viable alternative or is there something I can't see that make it unfeasible?

Hi,

I would like to add my modest opinion. So far I have used (random order on the list) HP HFSS, Ansoft HFSS, IE3D, Sonnet, Momentum, CST, Ansoft Ensemble, AWR EmSight.

There were a lot of discussion which one is the best, most user friednly etc. I will not add my opinion on this topic, none of them matches perfectly my expectations, but they all serve the purpose and give reasonably reliable results, so I was happy enough with the final score in all situations.

I would like to point out some, in my opinion, important points which I encounter in my experience while using these simulators.

1. One very important issue is the drawing of the geometry regardless wether it is planar or volume simulator. All geometry editors are different! Most software demonstrations and workshops acctualy show you how to draw the structure. To some extent this is logical, the simulator has to know what one wants to analize. However, drawing is not the art to be mastered in EM simulator, so if one has to learn 4 or 6 or 8 ways of drawing the same structure it seems to me as a considerable waste of time. There are standard formats like SAT or DXF to exchange geometries, but could the editors themself be standardized?

2. As a result of 1., basic training does not include any of the real basics of the EM analysis. E.g. which port definition is most suitable or how to set the meshing parameters are some of the most frequently posted questions on this forum. As a result of this situation, there is a huge range of surprising results coming from the hands of inexperienced users. Even the best suited simulator for a certain type of problem will give wrong results if not set properly.

3. There is another related point to add here. I don't believe, at least for some time, the level of expert system for EM simulator to be reached so that any user can get the reliable result out. User must be qualified to do the job, software is just the tool. Give me a paint brush and I will make a colorful stain of a canvas, regardles if the brush belonged to Michelagelo himself! The same is with EM simulators. They can not be user friendly enough to cover ignorance!

4. Interoperability is a great idea. I saw a demonstration of AWR-Sonnet integration and will be very pleased to have this feature on my fingertips. I agree that this is a feature that will become a mast in the future and looking forward to see it implemented in all the of the missing links.

flyhigh

I think that the time domian computational electromagnetic software is have the large market.

At last! The MTT Exhibit is over, all vendors were very very busy. We set all time records (for number of leads) each day, by a wide margin...but it cost me very sore feet! I am relaxing with a nice workshop here Friday morning, and I can at last sit down.

Hi Lagrange -- The AWR com interface is not controlled by an open consortium, but it is open and anyone may interface to it. I understand that they now have 6 EM simulators interfacing to it, I'm sure more are coming. Interfacing to Agilent ADS is a bit of work (using AEL), but Agilent has been very helpful in our current update of the ADS-Sonnet interface, even though Momemtum, to one degree or another, has some competitive aspects to Sonnet. (Agilent has told us that they want the ADS-Sonnet interface to be the #1 best of all the Sonnet interfaces, we are working hard to do so!). Eagleware wrote their interface to Sonnet. The person who did it was very skilled, but I think it must have also been easy to do (things get easy when software is written properly), they did it very quicly and it works great. We spent about one year making a really nice interface to Cadence, we consider it the best interface to 3rd party EM software in the industry.

As for an interface standard for the whole industry, that would be really nice. However, we should take instruction from past failures. Specifically, IFF, a format developed under DARPA sponsorship failed. In my view it failed becasue it was a file, not a direct program-to-program interface. The other problem, much more serious, is there are so many features in so many softwares, what does software A do when it encounters a feature used only in Software B (like a specific non-linear model, or a specific kind of de-embedding, or a different kind of port...)

Hi Flyhigh -- As for Sonnet, all of our interfaces have at least one mode where you take the geometry from the main software you are using. No need to re-draw, and no need to use Sonnet drawing unless you want to. Our Cadence interface even has a mode where Sonnet remains completely hidden. If you want to try it, AWR (and I think Eagleware too) both interface to SonnetLite (free at sonnetsoftware.com, the software has no timeout).

As for ports, at least in Sonnet, we have only one kind of port. It is a gap port. You can put it on the box edge, in the interior, and make it so vias to ground (for one side of the port) are automatically added. If there are multiple ports very close together on the interior, they are not yet exactly de-embeded. We have solved the mathematics for this problem and the next major release will have the updated code. All substrate edge ports are exactly de-embeded (to within the accuracy of the underlying EM analysis), so there is no neede for any kind of wave ports.

The workshop I am attending is starting in a few minutes. Will probably not check the board again until sometime Sunday.

Hi:

I think flyhigh has some very good points. Tools are just tools. Depending upon how good you can use them, you may yield different results. As tool developers, we will try to make the tools easier to use and less errors prone. Our latest Automatic GDSII to IE3D Flow (AGIF) is one of the efforts to simply the process and make it an expert system for the users. Basically, we are aiming to let users to create some templates so that they can use one template for a class of structure without doing the same things again and again. In our AGIF initial releases, we have demonstrated how a user can go from GDSII layers directly into full 3D models and simulaitons with one single click of the button. The full 3D models are quite complicated. They can contain via holes, thick metals, finite ground plane with or without thickness, wire bonds, localized ports, differential ports, extension ports. All the visitors to our booth were amazed by what the AGIF can do and they consider it as a magic.

We have the IE3D linked to AWR's Microwave Office 2004. We are working to link IE3D to Cadence's Virtuoso and Allegro. The AGIF is the inital effort. Eventually, similar procedures will be available in Cadence's environments, as well as other formats such as DXF and GERBER. Also, we will try to integrate IE3D and AGIF into other platforms.

I also think that the remarks made by flyhigh are very interesting and I am particularly interested in what stated in point 1 about the (unnecessary ?) multiplication of efforts due to different geometrical modelers. I think that this is true on the user side (as pointed out by flyhigh) and also on the developer side.

As I have told I am in the process of developing a new 3d EM simulator and in doing that I have decided to split it in two separate codes. The first one is the actual simulator which works on a simple file describing a meshed geometry. The second one is the geometrical modeler needed to generate the mesh
and to specify the variuos parameters (material properties, ports...). The simulator is where all the innovation (it expoites a totally new method) and most of my efforts are concentrated. On the other hand the simulator can not work without a geometrical modeler and at the moment I am using an academic code which is adequate to generate the test meshes but is far from the quality of a commercial product.

So, to enter into the commercial phase (if I will even be able to do it), I need to develope a better geo modeler and I think this is a complex task (even if using a commercial 3d kernel). On the other hand this task has already been accomplished by many other developers and my feeling is that there should
be a better option then repeating the same effort again.