EMI Shielding Theory & Gasket Design Guide
SECTION CONTENTS PAGE
Theory of shielding and gasketing 192
Conductive elastomer gasket design 196
Gasket junction design 196
Corrosion 198
Selection of seal cross section 202
General tolerances 204
Gasket mounting choices 205
Fastener requirements 206
Designing a solid-O conductive elastomer gasket-in-a-groove 209
Mesh EMI gasketing selection guide 214
Glossary of terms 218
Part number cross reference index 220
US Headquarters TEL +(1) 781-935-4850 FAX +(1) 781-933-4318 • www.chomerics.com Europe TEL +(44) 1628 404000 FAX +(44) 1628 404090 Asia Pacific TEL +(852) 2 428 8008 FAX +(852) 2 423 8253 South America TEL +(55) 11 3917 1099 FAX +(55) 11 3917 0817 191 EMI Shielding Theory
Theory of Shielding wave impedance is greatly different transmitted across the boundary and Gasketing from the intrinsic impedance of the and supports a current in the metal Fundamental Concepts discontinuity, most of the energy will as illustrated in Figure 2. The be reflected, and very little will be amount of current flow at any depth A knowledge of the fundamental transmitted across the boundary. in the shield, and the rate of decay concepts of EMI shielding will aid Most metals have an intrinsic is governed by the conductivity of the designer in selecting the gasket impedance of only milliohms. For the metal and its permeability. The inherently best suited to a specific low impedance fields (H dominant), residual current appearing on the design. less energy is reflected, and more opposite face is the one responsible All electromagnetic waves consist is absorbed, because the metal for generating the field which exists of two essential components, a is more closely matched to the on the other side. magnetic field, and an electric field. impedance of the field. This is why These two fields are perpendicular it is so difficult to shield against Ei to each other, and the direction of magnetic fields. On the other hand, wave propagation is at right angles the wave impedance of electric Et to the plane containing these two fields is high, so most of the energy Jo components. The relative magnitude is reflected for this case. between the magnetic (H) field and Jt Consider the theoretical case the electric (E) field depends upon of an incident wave normal to how far away the wave is from its the surface of a metallic structure Figure 2 Variation of Current Density source, and on the nature of the as illustrated in Figure 1. If the with Thickness for Electrically Thick generating source itself. The ratio conductivity of the metal wall is Walls of E to H is called the wave infinite, an electric field equal and Our conclusion from Figures 2 impedance, Z . w opposite to that of the incident and 3 is that thickness plays an If the source contains a large electric field components of the important role in shielding. When current flow compared to its potential, wave is generated in the shield. skin depth is considered, however, such as may be generated by a This satisfies the boundary condition it turns out that thickness is only loop, a transformer, or power lines, that the total tangential electric field critical at low frequencies. At high it is called a current, magnetic, or must vanish at the boundary. Under frequencies, even metal foils are low impedance source. The latter these ideal conditions, shielding effective shields. definition is derived from the fact should be perfect because the two The current density for thin shields that the ratio of E to H has a small fields exactly cancel one another. is shown in Figure 3. The current value. Conversely, if the source The fact that the magnetic fields are density in thick shields is the same operates at high voltage, and only in phase means that the current flow as for thin shields. A secondary a small amount of current flows, the in the shield is doubled. reflection occurs at the far side of source impedance is said to be the shield for all thicknesses. The high, and the wave is commonly x only difference with thin shields is referred to as an electric field. At Perfectly Ei that a large part of the re-reflected very large distances from the Conductive Plane z=0 wave may appear on the front source, the ratio of E to H is equal surface. This wave can add to or for either wave regardless of its H E i subtract from the primary reflected origination. When this occurs, the Hr wave depending upon the phase wave is said to be a plane wave, Er relationship between them. For this and the wave impedance is equal H reason, a correction factor appears to 377 ohms, which is the intrinsic z y in the shielding calculations to impedance of free space. Beyond account for reflections from the this point all waves essentially lose Figure 1 Standard Wave Pattern of a far surface of a thin shield. their curvature, and the surface Perfect Conductor Illuminated by a A gap or slot in a shield will allow containing the two components Normally Incident, + X Polarized Plane electromagnetic fields to radiate becomes a plane instead of a Wave through the shield, unless the section of a sphere in the case current continuity can be preserved of a point source of radiation. Shielding effectiveness of metallic across the gaps. The function of an The importance of wave enclosures is not infinite, because EMI gasket is to preserve continuity impedance can be illustrated by the conductivity of all metals is finite. of current flow in the shield. If the considering what happens when an They can, however, approach very gasket is made of a material electromagnetic wave encounters a large values. Because metallic identical to the walls of the shielded discontinuity. If the magnitude of the shields have less than infinite conductivity, part of the field is
US Headquarters TEL +(1) 781-935-4850 FAX +(1) 781-933-4318 • www.chomerics.com Europe TEL +(44) 1628 404000 FAX +(44) 1628 404090 Asia Pacific TEL +(852) 2 428 8008 FAX +(852) 2 423 8253 192 South America TEL +(55) 11 3917 1099 FAX +(55) 11 3917 0817 gap exists in the seam, the flow of The reflection term is largely E i current will be diverted to those dependent upon the relative points or areas which are in contact. mismatch between the incoming Et A change in the direction of the flow wave and the surface impedance of Jo Jt of current alters the current distribution the shield. Reflection terms for all in the shield as well as in the gasket. wave types have been worked out Current on 3 Rereflection A high resistance joint does not by others. The equations for the front wall due from rear wall to reflection behave much differently than open three principal fields are given by from rear wall seams. It simply alters the distribution the expressions: Figure 3 Variation of Current Density of current somewhat. A current G RE = 353.6 + 10 log10 (2) distribution for a typical seam is 3 2 with Thickness for Electrically Thin Wall f µr1 shown in Figure 4. Lines of constant 0.462 µ fG enclosure, the current distribution in current flow spaced at larger intervals RH= 20 log10 + 0.136r1 +0.354 (3) ( r √Gf √ µ ) the gasket will also be the same indicate less flow of current. 1 assuming it could perfectly fill the G x 106 It is important in gasket design RP= 108.2 + 10 log10 (4) µ slot. (This is not possible due to to make the electrical properties of f mechanical considerations.) the gasket as similar to the shield where The flow of current through a as possible, maintain a high degree RE, RH, and RP are the reflection terms for the shield including a gasket interface is of electrical conductivity at the electric, magnetic, and plane wave fields illustrated in Figure 4. Electromagnetic interface, and avoid air, or high expressed in dB. leakage through the seam can occur resistance gaps. G is the relative conductivity referred to in two ways. First, the energy can copper, leak through the material directly. Shielding and The gasket material shown in Gasket Equations1 f is the frequency in Hz, Figure 4 is assumed to have lower The previous section was devoted µ is the relative permeability referred to conductivity than the material in the to a physical understanding of the free space, shield. The rate of current decay, fundamental concepts of shielding r1 is the distance from the source to the therefore, is also less in the gasket. and gasketing. This section is devoted shield in inches. It is apparent that more current will to mathematical expressions useful for general design purposes. It is The absorption term A is the same for all three waves and is Lines of helpful to understand the criteria constant current for selecting the parameters of a given by the expression:
σg < σm shielded enclosure. (5) In the previous section, it was A = 3.338 x 10–3 x t √µfG shown that electromagnetic waves Metal Shield where incident upon a discontinuity will be A is the absorption or penetration loss Ei partially reflected, and partly trans- expressed in dB, and t is the thickness Et mitted across the boundary and into of the shield in mils. the material. The effectiveness of the shield is the sum total of these two The factor B can be mathematically effects, plus a correction factor to positive or negative (in practice it is Gasket account for reflections from the back always negative), and becomes surfaces of the shield. The overall insignificant when A>6 dB. It is expression for shielding effectiveness usually only important when metals is written as: are thin, and at low frequencies (i.e., below approximately 20 kHz). Figure 4 Lines of Constant Current S.E. = R + A + B (1) B (in dB) = 20 log (6) Flow Through a Gasketed Seam where 10 (K – 1)2 S.E. is the shielding effectiveness2 expressed in dB, 1 – 10 –A/10 e –j.227A appear on the far side of the shield. ( (K + 1)2 ))( ( ) This increased flow causes a larger R is the reflection factor expressed in dB, where leakage field to appear on the far A is the absorption term expressed in dB, and A = absorption losses (dB) side of the shield. Second, leakage B is the correction factor due to reflections from µ 2 1/ 2 can occur at the interface between K = Z S /Z H = 1.3( /fr G) the far boundary expressed in dB. the gasket and the shield. If an air Z S = shield impedance
Z H = impedance of the incident References magnetic field 1. Much of the analysis discussed in this section was performed by Robert B. Cowdell, as published in “Nomograms Simplify Calculations of Magnetic Shielding Effectiveness” EDN, page 44, September 1, 1972. 2. Shielding Effectiveness is used in lieu of absorption because part of the shielding effect is caused by reflection from the shield, and as such is not an absorption type loss. 3. Vasaka, G.J., Theory, Design and Engineering Evaluation of Radio-Frequency Shielded Rooms, U.S. Naval Development Center, Johnsville, Pa., Report NADC-EL-54129, dated 13 August, 1956.
US Headquarters TEL +(1) 781-935-4850 FAX +(1) 781-933-4318 • www.chomerics.com Europe TEL +(44) 1628 404000 FAX +(44) 1628 404090 Asia Pacific TEL +(852) 2 428 8008 FAX +(852) 2 423 8253 193 South America TEL +(55) 11 3917 1099 FAX +(55) 11 3917 0817 EMI Shielding Theory continued
The preceding equation was current to flow in the 300 See text details and correction for thin sheets Copper solved in two parts. A digital computer shield in a vertical Shielding effectiveness = absorption + reflection loss Iron Copper was programmed to solve for B with direction. A gasket 1 2 Absorption loss per mil σ = 1 250 thickness µ = 1 a preselected value of A, while K placed transverse to 3 4 Reflection loss Ð Electric fields Iron –4 3 3 σ = .17 varied between 10 and 10 . The the flow of current is 5 6 Reflection loss Ð Plane waves µ = 200
results are plotted in Figure 9. less effective than 200 7 8 Reflection loss Ð Magnetic fields The nomograph shown in Figure one placed parallel 4 8 was designed to solve for K in to the flow of current. 150 equation (6). Note that when ZH A circularly 5 LOSS (dB) becomes much smaller than ZS polarized wave (K>1), large positive values of B may contains equal 100 6 result. These produce very large and vertical and unrealistic computed values of S.E., horizontal compo- 50 7 and imply a low frequency limitation nents, so gaskets 2 8 on the B equation. In practical cases, must be equally 1 absorption losses (A) must be cal- effective in both 0 culated before B can be obtained.1 directions. Where 100Hz 1kHz 10kHz 100kHz 1MHz 10MHz 100MHz 1GHz 10GHz FREQUENCY A plot of reflection and absorption polarization is loss for copper and steel is shown in unknown, gasketed Figure 5 Shielding Effectiveness of Metal Barriers Figure 5. This illustration gives a junctions must be designed Some care must be exercised good physical representation of the and tested for the worse condition; in using these charts for behavior of the component parts of that is, where the flow of current is ferrous materials because µ an electromagnetic wave. It also parallel to the gasket seam. varies with magnetizing force. illustrates why it is so much more Nomographs difficult to shield magnetic fields Magnetic Field Reflection – than electric fields or plane waves. The nomographs presented in Figure 7: To determine magnetic Note: In Figure 5, copper offers more Figures 6 through 9 will aid the field reflection loss RH: shielding effectiveness than steel in designer in determining absorption a. Locate a point on the G/µ all cases except for absorption loss. and magnetic field reflection losses scale for one of the metals This is due to the high permeability directly1. These nomographs are listed. If the metal is not listed, of iron. These shielding numbers are based on the equations described compute G/µ and locate a theoretical, hence they are very high in the previous section. point on the numerical scale. (and unrealistic) practical values. Absorption Loss – Figure 6: b. Locate the distance between If magnetic shielding is required, Given a desired amount of absorption the energy source and the particularly at frequencies below loss at a known frequency, determine shield on the r scale. 14 kHz, it is customary to neglect all the required thickness for a known c. Place a straight-edge between terms in equation (1) except the metal: r and G/µ and locate a point absorption term A. Measurements of a. Locate the frequency on on the unmarked X scale numerous shielded enclosures bears the f scale and the desired (Example: r =10 inches for this out. Conversely, if only electric absorption loss on the A scale. hot rolled steel). field, or plane wave protection is required, reflection is the important Place a straight-edge across d. Place a straight-edge between factor to consider in the design. these points and locate a point the point on the X scale and The effects of junction geometry, on the unmarked X scale the desired frequency on the contact resistance, applied force (Example: A = 10 dB, f scale. and other factors which affect f =100 kHz). e. Read the reflection loss from
gasket performance are discussed b. Pivot the straight-edge about the RH scale. (For f = 10 kHz, in the design section which follows. the point on the unmarked X RH = 13 dB). scale to various metals noted Polarization Effects f. By sweeping the f scale while on the G x µ scale. A line holding the point on the X Currents induced in a shield flow connecting the G x µ scale essentially in the same direction as scale, RH versus frequency and the point on the unmarked can be obtained. (For the electric field component of the scale will give the required f = 1 kHz, RH = 3.5 dB). inducing wave. For example, if the thickness on the t scale. (Note that thickness is not a factor electric component of a wave is (Example: for copper t = 9.5 mils, in calculating reflection losses.) vertical, it is known as a vertically cold rolled steel t = 2.1 mils). polarized wave, and it will cause a
References 1. Robert B. Cowdell, “Nomograms Simplify Calculations of Magnetic Shielding Effectiveness” EDN, page 44, September 1, 1972.
US Headquarters TEL +(1) 781-935-4850 FAX +(1) 781-933-4318 • www.chomerics.com Europe TEL +(44) 1628 404000 FAX +(44) 1628 404090 Asia Pacific TEL +(852) 2 428 8008 FAX +(852) 2 423 8253 194 South America TEL +(55) 11 3917 1099 FAX +(55) 11 3917 0817 1 1 Figure 6 Absorption Loss Nomograph Figure 7 Magnetic Field Reflection Loss Nomograph, RH
A = 5.0 dB 0 A = 6.0 dB
A = 3.0 dB A = 4.0 dB
-5 A = 2.0 dB
A = 1.5 dB 1 kHz -10 A = 1.0 dB A = .8 dB B in dB
A = .6 dB -15
A = .4 dB |K| = 1.3[µ/fr2G]1/2 -20
A = .2 dB |K| = 2.2 x 10 -2 -25 10 -4 10 -3 10 -2 10 -1 1 |K| Figure 9 Solving for Secondary Reflection Loss (B)1
Figure 8 Magnetic Field Secondary Reflection Loss Factor Nomograph1
Magnetic Field Secondary Reflec- Find B at 1 kHz. c. At its intersection with the K tion Losses K Figures 8 and 9: a. Draw a line between copper scale, read K = 2.2 x 10–2. To determine the magnetic field on the G/µ scale and r = 2 d. Proceed to Figure 9. secondary reflection loss factor K inches on the “source to shield e. On Figure 9, locate K = 2.2 x to solve for B: distance scale.” Locate a point 10–2 on the horizontal scale. on the X scale. Given: r = 2 inches for 0.0162 in. f. Move vertically to intersect the thick copper and A = 1.3 dB. b. Draw a line from the point on A = 1.3 curve (interpolate), the X scale to 1 kHz on the and then horizontally to find f scale. B = –8.5 dB.
US Headquarters TEL +(1) 781-935-4850 FAX +(1) 781-933-4318 • www.chomerics.com Europe TEL +(44) 1628 404000 FAX +(44) 1628 404090 Asia Pacific TEL +(852) 2 428 8008 FAX +(852) 2 423 8253 195 South America TEL +(55) 11 3917 1099 FAX +(55) 11 3917 0817 Conductive Elastomer Gasket Design*
Gasket Junction Design At this stage of the design every become widely accepted. Zinc is The ideal gasketing surface is effort should be given to choosing a primarily used with steel. Consult the rigid and recessed to completely flange that will be as stiff as possible applicable specifications before house the gasket. Moreover, it consistent with the construction used selecting a finish. A good guide to should be as conductive as and within the other design finishing EMI shielded flanges for possible. Metal surfaces mating constraints. aerospace applications has been published by SAE Committee AE-4 with the gasket ideally should be 1. Flange Materials non-corrosive. Where reaction with (Electromagnetic Compatibility) Flanges are generally made of the the environment is inevitable, under the designation ARP 1481. A same material as the basic enclosure the reaction products should be discussion of corrosion control for reasons of economy, weldability, electrically conductive or easily follows later in this guide. strength and resistance to corrosion. penetrable by mechanical abrasion. Wherever possible, the flanges 2. Advantages of Grooved Designs It is here that many gasket designs should be made of materials with the All rubber materials are subject to fail. The designer could not, or did highest possible conductivity. It is “Compression Set,” especially if not treat the mating surface with the advisable to add caution notes on over compressed. Because flange same care as that given to the drawings not to paint the flange surfaces cannot be held uniformly selection of the gasketing material. mating surfaces. If paint is to be flat when the bolts are tightened By definition, a gasket is necessary applied to outside surfaces, be sure (unless the flanges are infinitely only where an imperfect surface that the contact surfaces are well stiff), gaskets tend to overcompress exists. If the junction were perfect, masked before paint is applied, and in the areas of the bolts. Proper which includes either a solidly then cleaned after the masking tape groove design is required to avoid welded closure, or one with mating is removed. If the assembled units this problem of over compression. A surfaces infinitely stiff, perfectly flat, are subject to painting or repainting groove also provides metal-to-metal or with infinite conductivity across in the field, add a cautionary note in contact between the flange members, the junction, no gasket would be a conspicuous location adjacent to thereby reducing contact resistance necessary. The more imperfect the the seal that the seal areas are to be across the junction. mating surfaces, the more critical is masked before painting. A single groove will suffice for most the function of the gasket. Perfect Ordinarily, the higher the conduc- designs. Adding a second groove surfaces are expensive. The final tivity of a material, the more readily it parallel to the first adds approximately solution is generally a compromise oxidizes – except for noble metals 6 dB to the overall performance of between economics and performance, such as gold and silver. Gold is a single-groove design. Adding but it should not be at the expense impervious to oxidation, and silver, more grooves beyond the second of neglecting the design of the although it oxidizes, forms oxides does not increase the gasketing flange surfaces. that are soft and relatively conductive. effectiveness significantly. The important property that Most oxides, however, are hard. makes a conductive elastomer 3. Flange Design Considerations Some of the oxide layers remain thin gasket a good EMI/EMP seal is its while others build up to substantial Most designers fight a space ability to provide good electrical thickness in relatively short time. limitation, particularly in the vicinity conductivity across the gasket- These oxides form insulating, or of the gasketed seam. Complex flange interface. Generally, the better semi-conducting films at the fasteners are often required to make the conformability and conductivity, boundary between the gasket and the junctions more compact. the higher the shielding effectiveness the flanges. This effect can be The ideal flange includes a of the gasket. In practice, it has overcome to a degree by using groove for limiting the deflection of a been found that surface conductivity materials that do not oxidize readily, gasket. The screw or bolt fasteners of both the gasket and the mating or by coating the surface with a are mounted outboard of the gasket surfaces is the single most important conductive material that is less to eliminate the need for providing property that makes the gasketed subject to oxidation. Nickel plating is gaskets under the fasteners. A seam effective; i.e., the resistance generally recommended for machined flange and its recom- between the flange and gasket aluminum parts, although tin has mended groove dimensions are should be as low as possible. shown in Figure 10. The gasket may
* Complete solid-O gasket design information starts on page 209.
US Headquarters TEL +(1) 781-935-4850 FAX +(1) 781-933-4318 • www.chomerics.com Europe TEL +(44) 1628 404000 FAX +(44) 1628 404090 Asia Pacific TEL +(852) 2 428 8008 FAX +(852) 2 423 8253 196 South America TEL +(55) 11 3917 1099 FAX +(55) 11 3917 0817 methods that produce longitudinal The thickness of a flange is H (circumferential) scratches or chatter governed by the stiffness required to marks. Conversely, a surface that is prevent excessive bowing between too smooth will cause the gasket to fastener points. Fewer, but larger