This is a true story. When I first joined IBM as an EMC engineer, my new manager handed me a document titled ‘EMC Design Process at IBM’ and asked me to comment. I quickly read the short document that basically said that the EMC engineer would provide the design engineers a list of EMC rules, which would be largely ignored.
The product would be built and when tested in the EMC laboratory it would fail. The EMC engineer would then spend anywhere from a week to a month to try various band-aid fixes (not the term used in the document, of course) before making recommended changes back to the design team. The changes (or some of the changes) would be implemented, and a new version of the product built. Testing would be repeated, and this process might need 2-3 iterations before completed and the product was ready to ship.
I handed the document back to my brand new manager, and asked if he had fired this EMC engineer yet? He was shocked, and told me this was one of the more senior EMC engineers! I told him that if I told my boss I expected to fail every time I should be fired. Of course, this was not the design process that was desired, but rather the one that had evolved.
Using software tools helped IBM turn this process around completely. So now, instead of failing the first time, every time, products usually pass the first time in the EMC chamber! Using these tools (along with education of the design engineers) made all the difference!
There is a variety of tools available and they operate at different levels. This article will discuss using these tools and point out the benefits and where they can be used most effectively in the design process. Many people had told me they have no time to learn how to use new tools. I equate this to a story a friend of mine told me years ago. A woodsman is tasked to clear five acres of forest in a very short time. He begins with his double-bladed axe and is working hard when someone tries to show him a new invention, called a chain saw. The woodsman replies that he has no time to learn new tools! He is busy with a short deadline!
Before the Design Begins – Simulation
There are many EMC rules, and some of them are in direct conflict with each other! These rules need to be evaluated to see which ones will work for your particular product family.
For example, EMC rules for large main frame computers may or may not apply to a small hand held device where large metal shields and finger stock can not be used. Furthermore, some published EMC design rules do not follow physics! All rules need to be examined to make sure that they make sense, and are appropriate for your product types.
One of the best ways to validate rules is to use full wave simulation software. There are a variety of vendors offering a variety of different software simulation tools. These tools use a variety of different simulation techniques, each having areas where they excel and areas where they are not the best tool for the job. A tool box approach is strongly recommended so the user will have a variety of tools at their disposal and can optimize their particular simulation for the type of problem at hand. Figure 1 shows a number of different possible problems that might be simulated (courtesey of CST) and represent only a small number of examples possible for simulation tools. However, each of these problems are very different, and so a different simulation technique would be best for some of the problems. There is no one size fits all in the world of simulation techniques.
For example, the heatsink in Figure 1A would usually require an open boundary condition and would likely be easiest to simulate using Finite-Different Time Domain (FDTD), Finite Integration Technique (FIT), or the Method of Moments (MoM). PCB problems often require dielectric materials to be included as well as open boundaries, so FDTD or FIT might be best suited. If the PCB problem includes many discrete components (such as capacitors, equivalent inductances, etc.) then the Partial Element Equivalent Circuit (PEEC) would probably be the most efficient way to perform the simulation. Internal shielded air vents (Figure 1C) could easily be solved with FDTD, FIT, or the Finite Element Method (FEM). Problems such as the coax cable in Figure 1D might be optimized using FEM, since it is a problem with metal boundaries (open boundaries not needed) and the non-rectangular shape of the grid can be well suited to curved surfaces.
Figure 1A: Heatsink example
Figure 1B: PCB example
Figure 1C: Internal shielded air vent example
Figure 1D: Coax cable example
Very seldom is a single simulation run to determine pass/fail of a system. Usually there are a family of simulations, each with something slightly different, to help define the grey area between the absolutes. A classic example would be to determine how many posts are required to connect the heatsink in Figure 1A to the ground-reference plane in order to reduce the emissions over a certain frequency range. In this example, a number of simulations would be performed, each with a different number of grounding posts, to observe the frequency range where the emissions are reduced. Using these multiple simulations, a set of design guidelines can be created that are optimized for the specific type of product that is to be designed.
A word of caution should be mentioned here. Simulation tools are very powerful and useful. They can help fill in the grey areas, and also help understanding of the engineers and non-engineers who often must be convinced to implement a certain design rule even though it might add a little cost, weight, etc. to the product. However, all simulations should be validated. The software vendors spend a lot of time to insure their tools give an accurate answer to whatever question was asked. However, the user is often the primary source of error. A good rule of thumb about validating simulations: if you have never made a mistake in your life, you might be safe to ignore the recommendations for validation!
One of the primary potential sources of error are the types of source used in the simulation. Wave ports are an easy source of error. If the boundaries of the wave port are too close to a microstrip (for example), then the fields will interact with the perfect electrical conductor boundary and incorrect wave modes are established (see Figure 2).
Figure 2: Incorrect and correct electric fields in wave port
Validation can take many forms. Probably the most common, and often the most difficult is to use measurements to validate the simulation. After all, measurements are a great emotional comfort! However, there are a lot of measurement artifacts that may or may not be included in the simulation. Antenna patterns, ground plane reflections, and equipment input impedance loading for direct measurements can all make it difficult to compare measurements and models unless all these effects are included in the simulation.
Another popular way to validate simulations is to use a completely different simulation technique. For example, using FDTD and MoM for the same problem will use very different
physics for the simualtion. Of course, this means the simulation must be run twice, but if both simulations give the same results, then the user must have understood the problem well enough to create models for the different simulation techniques, and the results are probably good.
Before the Design Begins – Training
Training is an important part of the preparation for the coming design project. While there are a number of training seminars available, it is important to make sure that it is not just a listing of EMC design rules collected over the years, but rather training that explains how the physics work, why the rule is important, and how to determine if the rule is appropriate for this product/project or not. This does not mean that a lot of heavy math is required! We can leave the math for the universities and those who love to solve equations. Understanding the physics means that the students should learn the fundamentals of how current flows, the true nature of ground vs. return current path, how shielding really works, and especially a good understanding of inductance concepts. Remember, once the seminar is completed, the student/engineer must rely on the knowledge gained during the seminar to be able to know when a rule must be enforced, when the rule can be bent a little (and how far) or when a rule does not make sense for the product under design.
During the Design – Rule Checking Software
Once the design has begun, there is seldom time to do multiple simulations, etc. The design rules that were vetted prior must be used since time is usually short. When designing many layer high speed printed circuit boards (PBCs), it can be impossible for an engineer to double check all the proper design rules were followed. There are software tools available that can read the PCB CAD design file, quickly check against a wide variety of EMC and Signal Integrity (SI) design rules, and highlight the areas where design rule violations occur.
These software EMC/SI design rule checking tools usually include a variety of rules that are more complex than the simple manufacturability Design Rule checkers (DRCs) that are included in PCB layout CAD tools. For example, a typical EMC design rule is that high speed traces must not cross a split in the nearby reference plane. However, depending on the data rate, rise time, etc. for the signal on that trace, a stitching capacitor might be used to allow the return current to cross from one plane to the other if the capacitor is located within a certain specified distance from the crossing point. Complex rules, such as this one, are too complex for the DRC in the CAD tools.
One of the major advantages of these EMC/SI rule checking tools is that they will highlight the area where the violation occurs, turning on only the PCB layers involved, and often even drawing a box around the violation to draw the engineer’s eyes to the right location quickly. Figure 3 shows an example screen shot of a violation of the trace crossing a split reference plane rule (courtesey of CST).
Figure 3: Example of EMC Rule checking tool Violation Viewing
Typical rules for printed circuit boards cover a wide range of potential violation, including distance from decoupling capacitor (and IC) pads to vias, decoupling capacitor density, traces close to the edge of a PCB, distance from signal via to return current via, and many more. Users can tailor the limits for the various rules depending on the product specific requirements, data rates etc.
Rule checking software tools are usually very fast, doing an entire high speed PCB in minutes or at most, tens of minutes. This is in contrast to most full wave simulations which typically take hours or even days to complete. The full wave simulation gives a complete solution to Maxwell’s equations, vs. a relatively simple geometry checking against a rule. Therefore the rule checking tools can be incorporated into the typical product design process easily and quickly. The visual aid of the violation viewing allows the engineer to quickly evaluate which violations are important and to make the necessary changes before building the hardware and possibly failing during EMC testing.
After the Design is Completed
Once the product has been successfully designed, built, and passed the EMC testing, feedback into the EMC rules can help the next product development as well as help reinforce the importance of the tools used before and during the design process. Of course, if the product happens to fail during initial EMC testing, once the offending portion of the product is determined, the feed back into the EMC rule checking tool will tighten the appropriate rule limits as necessary.
Summary
A variety of software tools are available to design engineers that can help increase the probability of passing EMC requirements the first time. Full wave tools are most useful to help understand the shades of grey for various design approaches, and are less useful to predict the pass/fail performance directly (due to the excessive amount of details required and excessive simulation run times for such complex models).
Rule checking software tools are very fast, accurate and helpful to identify potential design issues for high speed complex PCBs. The engineer still must make a decision about the relative importance of the violation and whether or not it must be corrected. The visual feedback and focusing on a violation allows engineers to make quick and informed decisions.
The bottom line is that none of these tools replace the need for the engineer to have a fundamental understanding of the physics of high frequency electromagnetics. These
are simply tools to help the engineer, not replace the engineer! Imagine taking your auto to a repairman who knows nothing about engines, but has a full set of mechanics tools. Equally absurd!
In this time of short design cycles, product cost pressures, and increasing RF noise from wireless devices etc., no one can afford to not use these tools to their fullest potentials. Don’t be like the woodsman and ignore things that will help you be successful! 
 |
Dr. Bruce Archambeault is an IBM Distinguished Engineer at IBM in Research Triangle Park, NC and an IEEE Fellow. He received his B.S.E.E degree from the University of New Hampshire in 1977 and his M.S.E.E degree from Northeastern University in 1981. He received his Ph. D. from the University of New Hampshire in 1997. His doctoral research was in the area of computational electromagnetics applied to real-world EMC problems. He is the author of the book “PCB Design for Real-World EMI Control” and the lead author of the book titled “EMI/EMC Computational Modeling Handbook”. |
