Simulation Software

Walter Chipley of Octagon Systems, Inc., writes:

I am curious if you have had any experience in working with EMC analysis software tools such as Incases, Hyperlynx, Quad, etc. I am in the process of trying to select an EMC analysis tool for use in pre-prototype EMC screening of single board computer products. My engineering group is using the VeriBest PCB tool suite for design, and we have a complete dedicated EMC lab to enable us to screen our board designs once in prototype form. I have briefly looked at Incases EMC Workbench product and also Hyperlynx. If you can provide any input relative to your direct experience with these or other EMC analysis tools it would be greatly appreciated.

Thanks for your interest in High-Speed Digital Design.

In direct answer to your question, no, I have not had much hands-on experience with software-based EMC analysis tools.

Here's why

EMC modeling tools are grouped into two categories, (1) three-dimensional modeling tools that predict the radiation from a single component lead frame, a single connector, or the like, and (2) less-than-three dimensional tools that predict radiation from an entire PCB.

OK, now it's 2011 and I've got to revise this part of the article. You can now get tools from Sigrity, Ansoft, CST microwave, and others that will perform 3-D S-parameter extraction on whole IC packages consiting of hundreds or thousands of little parts. That is highly useful for power-system analysis and crosstalk estimation, however, it still doesn't really help much with predicting EMI.

Three-dimensional tools

It is my understanding that the true three-dimensional tools will generate an accurate answer, provided that you set up ALL the relevant physical parameters. As there are many, many physical parameters that may affect the problem, it often turns out that one or more crucial terms are omitted from the simulation. To address this problem, we must invent ways to verify our simulations (like, for example, building a physical model and measuring the result). If you are manufacturing a large number of variations of one rather simple, basic package then the simulation approach may pay off. Once set up, it is useful for examining small perturbations. If you are making only a few variants, the setup time is not worth the effort. If your problem is complex (more than a few hundred leads or connections), then you can kiss the 3-D simulator approaches goodbye. They won't converge before you have to ship the product.

PCB analysis tools

It is my understanding that the PCB analysis tools predict radiation from looking at the radiating loop area of each individual trace. They assume that the power/ground return system is perfect. They in many cases assume that the bypass capacitors are perfect. They may also assume that chips are mounted flat against the board, with no ground bounce, and with no radiating area underneath the chip bodies. These assumptions are needed to make the simulation approach computationally feasible.

Having spent many long years tracking down and fixing EMI problems, it is my opinion that imperfections in the power/ground return system, bypass capacitors, chip packaging, connectors, cables and the like are the most usual sources of EMI difficulties. Tools that ignore these aspects are useless.

Conclusion

I don't mean to sway you from checking into these tools. What with the pace of progress in the field of software engineering, you never know what will happen, but, I do mean to inform you as to what are the primary issues at hand, and what are the important questions to ask before you get yourself mired in a project that may not pay off.

Best regards,
Dr. Howard Johnson