Richard, Even though you are a "mechanical type," you have put your finger on a week area in EMC design. The brutal truth is that there are no good design rules for shielding. While this may be shocking, since it is far from a new topic, it is due to the complexity of the problem. If you look at Maxwell's equations, they are difficult to solve analytically for very simple geometries and impossible for realistic, real life geometries.
Many of these published "rules" were developed more than a decade ago (some three decades ago) and are based on a blend of theory and some empirical data. They all have a common flaw and work better in some configurations than in others. The common flaw is that they are based on plane wave behavior (this means that you are in the far field) with some corrections for being close in. My experience is that you may survey several sources of theory and either take the one that makes most sense to you, or take portions of several to make your own composite theory. Here are some other sources: 1. Reinaldo Perez, ed., Handbook of Electromagnetic Compatibility, Academic Press, Inc., New York, 1995, pp. 401 - 443, ISBN 0-12-550710-0. This is a chapter written by Richard B. Schulz. Schulz's original IEEE article is found in time #2 below. 2. Schulz, R.B., et. al., "Shielding Theory and Practice," IEEE Transactions on Electromagnetic Compatibility, 1968, EMC-10, pp. 168 - 175. 3. Violette, J, et. al., Electromagnetic Compatibility Handbook, Van Nostrand Reinhold Co., NY,1987, ISBN 0-442-28903-0 If you are a real diehard, you may be interested to know that the EMC Lab at University of Missouri - Rolla has been developing theory along these lines. Contact Dr. James L. Drewniak, 573-341-4969, or look up their web site at http://www.emclab.umr.edu Their approach may have practical promise. (They also have a good list of EMC books on their site.) To avoid the inaccuracies of the published theories on shielding (top two paragraphs), many people have taken to numerical modeling of specific problems. With this approach, you solve these pesky Maxwell's equations numerically, but this requires specific geometries and produces results only appropriate to specific geometry addressed. Finite Difference Time Domain (FDTD) is one of the more common types of numerical techniques that are appropriate. These also have some well-known problems, but I will not bore you or the list-server with those unless some one asks. I am new in the computer industry. Hence, I have asked myself these same questions not too long ago. My experience and the experience of many colleagues in the industry that have spoken to is this: if you follow the guidelines based on classical theory that have been published (the Schulz stuff is an example) you will usually overdesign. These methods drive you to very small vent holes and you equipment sounds like a vacuum cleaner when forced air is used for cooling. Good Luck, Jim Knighten ------------------------------------------------------------------------ --------------------------- Dr. Jim Knighten NCR 17095 Via del Campo San Diego, CA 92127 Telephone: 619-485-2537 Fax: 619-485-3788 e-mail: [email protected] ---------- From: WOODS, RICHARD [SMTP:[email protected]] Sent: Monday, July 20, 1998 5:57 AM To: 'emc-pstc' Subject: Desiging Openings for EMC Compliance I am about to give a short seminar to our Mechanical Engineers and Designers on enclosure design for EMC compliance. There is only one problem - I have no faith in the theory I have for the attenuation through openings. The following formula is from the "EMC Handbook", Vol 3, by Don White. Assuming the frequencies of interest are below the waveguide cutoff frequency, the formula is A(dB) = KL/G - 20 log N where, K = 32 for round holes or 27 for square holes L = thickness of panel G = hole diameter N = number of holes. According to this formula, one 1/4 inch hole in a 0.090 inch panel would have an attenuation of 11 dB, and ten holes would have no attenuation whatsoever. This does not match my experience in typical ITE. Does anyone have any usable "rules of thumb" for Mechicanical types? Richard Woods Sensormatic Electronics [email protected] Views expressed by the author do not necessarily represent those of Sensormatic.
