Conformity Magazine

See this article in our Digital Edition
Download a pdf of this article (right click to save the pdf)

Many of today’s complex electronics depend at least in part on shielding to provide the EMC performance mandated by regulatory and quality considerations. Shields act as barriers to electric and magnetic fields, keeping unwanted signals in (emissions containment) or out (immunity).

Most engineers take it as an almost unshakable axiom of engineering faith that a conductive surrounding will provide adequate shielding protection. While shields can be very effective, designers will get the most out of them when some key issues are kept in mind.

Shielding In A Nutshell

There are a handful of factors which interact to decide whether a shield will perform as intended:
Conductivity (and to a lesser extent, permeability) of the shield
Continuity and connectivity of the different pieces of the shield
Thickness (important for low-frequency magnetic field applications, but not at RF frequencies)
Treatment of I/O and power leads which enter and exit the shield enclosure

In this article we’ll use these categories to organize our discussion of the major issues affecting shielding performance. Conductivity, connectivity, and shield thickness interact to determine a shield’s effectiveness. The shielding or filtering of I/O leads determines whether the anticipated shield performance will be maintained or compromised.

First, let’s not overlook the most important point: topology. If a shield doesn’t completely enclose the product it is protecting, it won’t work very well, if at all. A bunch of conductive pieces that are not designed to form a more or less continuous enclosing surface are not going to provide reliable, wide-frequency shielding.

How Shields Work
Shields work by two primary methods. The first is reflection, which occurs as an electromagnetic disturbance is incident on the shield, while the second is absorption as the fields attempt to traverse the shield. At RF (say MHz frequencies and above), both are extremely effective. The ultimate shielding effectiveness of an unflawed metal shield of reasonable thickness (i. e., a seamless box of sheet metal thick enough to supporting its own weight) is in the vicinity of 100 dB to 200 dB, which is more than adequate for commercial applications. Even thin conductive layers, such as those provided by conductive paint or electroless plating, are capable of very high shielding effectiveness at RF frequencies, although their magnetic shielding effectiveness generally falls off at low frequencies (only an issue in special situations).

Lets examine at the basic mechanisms underlying reflection and transmission. When electric and magnetic fields impinge on a conductor, there will be some reflection and some transmission at the boundary. Reflection occurs at a boundary where there is a difference in the conductivity, permittivity, and/or permeability of two neighboring materials. The greater this difference is, the greater the reflection. For plane waves, and metallic levels of conductivity, reflection will provide attenuation of the transmitted signal of 100 dB at any frequency of interest, although it does decrease with frequency.

Absorption occurs in conductors as fields attempt to travel through them. Sinusoidal electric and magnetic fields are attenuated exponentially with distance in a conductor. This is commonly called the “skin effect” phenomenon. Alternating fields travel primarily at the surface of conductors, in a skin layer. The is a characteristic skin depth in conductors given by:

Skin depth =

meters,

where:
ω = angular frequency (or 2πf)
μ = permeability
σ = conductivity

Figure 1 shows the values for some common materials as a function of frequency. Fields are attenuated by a factor of (1/e), or 8.78 dB for each skin depth into the material. We’ll discuss the effect of skin depth in more detail below, under the heading of “Thickness.” The attenuation provided by the mechanism of absorption increases with frequency. Both reflection and absorption are at work for most applications.

Figure 1

Conductivity
The first factor that affects the shield is the conductivity of its surface. This directly reflects the attenuation due to reflection, and also affects the amount due to absorption. There are many possibilities for the shield material, but for all of them, the conductivity will generally be great enough to provide a high degree of reflection.

Practical shields always involve metal in some form, but the shielding material may be “thick” (i.e., multiple skin depths at the frequencies of interest) or thin. The metal itself may vary substantially in form and conductivity. The metal may be pure or plated; it may be a solid thick enough to be structural or it may be a coating or deposition on a non-conductive plastic, or it may come in the form of a paint.

Conductive paints are usually applied so they are several thousandths of an inch thick, or about .01 cm. The usual figure of merit one looks at with conductive paints is the “surface resistivity” (or if you like, its reciprocal, the surface conductivity), which has the units of “ohms per square.” This is the resistance that will be seen from one edge to the other of a “square” section of painted surface, which, it happens, will be the same regardless of the size of the square (Figure 2) shows why this is so.

Figure 2

Practical experience has shown that the rough dividing line between paints that are effective as shields and those which are not as reliable occurs in the vicinity of 1 to at most 2 ohms per square. Lower values work better; higher ones are problematic. Since paints are “thin” coatings, the mechanism by which they shield is primarily reflection. The greater the difference between the wave impedance in air (377 ohms in the far field) and the value in the conductor, the greater the shielding provided by the mechanism of reflection.

Typically, nickel paints are in the 0.75 to 2 ohms per square range; copper paints (either alone, or alloyed with other metals such as nickel or small percentages of silver) are about 4 or 5 times more conductive.

Interestingly, it turns out that these paints, conductive as they are, are approximately two orders of magnitude less conductive than a pure metal layer of the same thickness as the applied paint. This is at first glance a bit surprising, because the volume percentage of metal in these paints is quite large (of the order of 50%). Apparently, the conductive particles contact each other frequently enough to give a decently low resistance, but particle-to-particle contact is far less intimate than in a solid, because of the insulating effect of the organic compounds which hold everything together once the paint dries. For this reason, it is important that conductive paints be well stirred during the application process, to ensure that the particle to organic chemical ratio remains at the designed level.

There are several alternatives to paint when a conductive coating is required. Among them are:

electroless plating
zinc arc spray
vacuum deposition
laminates

Electroless plating is a technique whereby a thin layer (of the order of 100 microns) of pure metal is electrically plated onto the plastic. Typical metals used are nickel, or copper and nickel in combination. Electroless plating results in a coating of very high conductivity.

Zinc arc spray involves spraying a hot metal alloy, primarily composed of zinc, onto the surface to be shielded. The metal is fed into a flame or arc, melted, and blown onto the plastic. The resulting coating is of very high conductivity, but it is somewhat rough and may affect dimensional tolerances.

In vacuum deposition, a heated metal is deposited on the plastic in a vacuum chamber. Vacuum deposition isn’t as commonly used for shielding as the other techniques. It can provide adequate performance, but only if the deposition is thick enough to provide a film that is thick enough and durable enough to make reliable contact with. Some vacuum depositions are not thick enough to do this, particularly if aluminum is used as the deposited material.

Table 1 contains a summary of the major coating methods and their main characteristics.

Coating	Conductivity	Comments
Nickel paint	~ 1 ohm/sq	At upper level of conductivity for effective shielding.
Copper, Copper alloy paint	~1/4 – 1/2 ohm/sq	Capable of very good performance if properly applied and connected.
Electroless Coating	very high	Very high conductivity, excellent performance.
Zinc arc spray	very high	Roughens surface; must be carefully applied. Not a noble metal. Very conductive.
Vacuum deposition	varies widely	May not provide reliable contact if deposition too thin.

Table 1: Characteristics of Common Surface Coatings

Continuity/Connectivity
Problems arise when the conductivity provided by the main shield materials is not put to proper use to create a continuous overall shield. Continuity is our second key factor. It can be compromised in many ways. These can be grouped into “mechanical design” and “unintentional insulating” factors. The mechanical issues are primarily topological—dealing with the shape or fit of the various pieces which comprise the shielding enclosure.

There may be deliberate gaps and slots in the shielding for ventilation which are “electrically large” – i.e., significant fractions of a wavelength at some of the radiated frequencies. Sometimes these gaps are due to an inadequate number of fasteners between panels, or between panels and an underlying frame or cabinet.
The pieces and panels that compose the shield surface may not mate mechanically. This may be through design error (e.g. two panels which fold over but do not touch—as in an “overbite”), or because the components bow slightly under mechanical stress when assembled
Gaskets, if used, may not be properly dimensioned. Every gasket has a working dimension range and pressure. Too little, and the designed gap won’t be filled mechanically or electrically. Too much, and the gasket will be too deformed to work on subsequent installations.

It is well known that a slot or gap can be excited by fields and radiate in a manner similar to that of a wire of the same length. (See references at end of article). The longest dimension of such a gap governs the frequency above which shielding will be compromised. Deliberate large holes and slots are becoming less common as designers become more aware of the need to shield. Unintentional gaps, due to poor tolerances or incompatible finishes are still common. Remember also that screwing two surfaces together does not guarantee contact along their entire length. Contact will occur at the points of attachment and at most at the two highest points in between.

In what I’ve called the “insulative” group of problems, we are dealing with unanticipated effects which interfere with the electrical contact of two parts that are mechanically connected. Unintentional insulation can break up contact just as surely if it were deliberate. Look for the following:

Improper masking of paint can lead to insulation of surfaces that were meant to be in contact.
Non-conductive anti-corrosion treatments may insulate the pieces from each other or from intervening gaskets
Galvanic incompatibilities between dissimilar metals may introduce insulating surface corrosion. It is not uncommon for this to occur between gaskets and the portions of the shield they are meant to connect. When this happens, shielding effectiveness will degrade over time.

Protective metal treatments are potentially problematic. The oxygen in our atmosphere is a very active chemical; commonly used metals want to oxidize. Aluminum forms a thick tenacious surface oxide after long exposure to moisture; steel simply rusts. Protective treatments keep this from happening, but by and large also form an insulating layer. The fact is, just about everything, even conductive finishes like nickel plating, form some sort of an oxide. The question is whether the protective layer is thin enough or brittle enough to scratch or part when light pressure mechanical contact is made. So, a coating may be effectively conductive or insulating depending on how thick the treatment is. For example, “yellow chromate” is conductive if lightly applied, when it will have a gold-tinted iridescent appearance, but will insulate the surface under modest contact if the parts are left in the tank until they are a deep uniform golden yellow. “Clear anodizing” of aluminum can be quite difficult to penetrate and leave the part difficult to make contact to.

How can you tell if there is a problem? Coated or treated materials need to be checked for conductivity under light pressure, especially if the contact will be made with low pressure gaskets, such as those made of woven cloth over foam. The contact pressure one gets from gouging a pair of pointed test probes into the metal is very high, and may give misleading assurance of the contact that will be made through the coating in the assembled product. Check the likelihood of contact your coating will provide under the conditions of use with broad area pads made of soft conductive gasket material to simulate the actual installation.

Another way to create an insulating layer that will lose contact is by galvanic action. Whenever two dissimilar metals contact each other in the presence of an electrolyte, a short circuited micro-battery is created. Since some moisture is always available, dissimilar metals won’t stay connected for very long. The galvanic scale positions materials in terms of the relative potential they will exhibit. Materials that are far apart on the scale will corrode; those that are closer together are “galvanically compatible.” It is unfortunate, but incompatible materials are commonly used, and practically unavoidable.

The gasketing materials used are almost always at the noble end of the scale, so they are incompatible with aluminum or zinc plated steel. The details of the gasket geometry can make the situation worse by varying the size of the contact area. Low area contacts—point or line—are going to be more susceptible to corrosion than those made over a broader rectangular area.

Here’s a historical example from your author’s experience. Years ago, imported personal computers sometimes used zinc finished sheet metal chassis connected to a cover of similar material with beryllium copper spring fingers. These materials are far apart on the galvanic scale. After a fairly short time—days to weeks—the shielding performance deteriorated noticeably, and higher radiated emissions would be seen. Upon disassembly, a fine dark line of corrosion could be seen at the contact between the materials. The zinc, being less noble than the spring finger material, would corrode. In addition, the contact area was minimal, consisting of a line where the fingers curved against the case. When the case was flexed, or if it were disassembled and the surface cleaned, the shielding effectiveness would return to its original level.

These problems have not entirely left us today, although they are usually less severe. A common combination of shielding/gasketing seen today involves aluminum panels and conductive cloth over foam gasketing. This provides a wide area contact, which is helpful, but the contact doesn’t provide a hermetic seal, and the conductive part of the gasket is made from cloth bearing fine threads of a galvanically incompatible metal such as silver, nickel, or copper. Sometimes products which have been stored in a humid environment need to be cleaned to return them to optimum shielding performance.

Type	Comments
Gasket—Cloth over foam	Conductive cloth over open cell foam. Very compressible. Wide variety of form factors (rectangular, P-shape, hinged, with/without adhesive, etc.). Related varieties include plated rubber foam. Wide contact area slows onset of galvanic problems.
Gasket—Loaded rubber	Includes very high-performance types; military heritage—combines hermetic seal with wide contact area; limited material compressibility (unless hollow extrusion); wide variety of extruded or cut shapes possible. Can be expensive, sometimes intolerant of rubbing contact.
Gasket—mesh	Wire and wire over foam types. Not currently popular in commercial applications.
Spring fingers – strip	Relatively noble metal spring (e.g., beryllium copper, sometimes w/nickel or tin plating), can lead to galvanic issues. High conductivity. Generally good tolerance to wiping contact on insertion along finger orientation. Relatively small area of contact—line or multiple points. Wide variety of shapes. Highly compressible.
Spring coil	Continuous flat coil set in groove. Spring finger variation.

Table 2: Properties of Common Types of RF Gaskets and Fingers

Thickness
For materials that are multiple skin depths in thickness, there is sharp attenuation of any electric and magnetic fields that are transmitted through the surface boundary. Every skin depth traversed means the field drops by a factor of 1/e, or 8.7 dB. Looking again at Figure 1, we can see that any structural shield will be many skin depths thick in the RF range. At low frequencies, from the 50/60 Hz mains frequency through the multi-kHz frequencies used by CRT monitors and switching supplies, this may not be the case. And, thin, or somewhat less conductive coatings (i.e., conductive paints) may be electrically thin (less than a skin depth or two thick) even well into the MHz radio frequency range.

“Thin” materials can still work by reflection, but when does thickness matter? Reflection works well enough by itself for far-field electromagnetic waves and for waves that are in the moderate near field. However, for magnetic sources (loop-like sources very much in the near field, where the E/H ratio is small), reflection is not a factor, and absorption—attenuation on transmission through the depth of the shield material—is the only mechanism available. Here, the amount of absorption is determined by how many skin depths thick the shield material is. Looking again at the formula for skin depth, if the frequency of concern is fixed by the electronics to be shielded, the only way we can get more attenuation is by making the material a greater number of skin depths thick. This means either a thicker material, or a thinner skin depth.

Making the material physically thicker may not be practical, so we are left with trying to change the skin depth. If frequency is fixed, the only variables left are conductivity and permeability. Conductivity can only be made so high—copper and silver are as good as it gets. Fortunately, low frequency permeability can be quite high in magnetic materials.

There are a number materials that offer a high permeability, including steel, permalloy, and specialized compounds such as “mu-metal” (an alloy of 77 % Nickel, 15% iron, with additional copper and molybdenum). These can provide very effective low frequency magnetic shielding without being very thick. Note, however, that the permeability of these materials drops with frequency. At RF frequencies, the relative permeability of these materials drops to near one, so skin depth and absorption performance becomes governed primarily by the materials electrical conductivity.

Entries And Exits
The last piece in the puzzle concerns the treatment of leads that must enter and leave the shielded enclosure. Sadly, the performance of a well-formed enclosing shield can be completely compromised by inadequate attention to the leads that penetrate it—signal I/O, control, and power connections. It is easy for common mode energy to exit or enter the enclosure if these leads aren’t shielded or filtered.

The idea of shielding equipment connections makes intuitive sense if we consider the shield to be a tubular extension of the enclosure. For shielded cabling to work, then, it must be connected to that enclosure. The key point is that it must be well connected to the right place with a connection that is low-impedance at all frequencies of interest. This usually means a 360 degree conductive connection directly to the enclosing chassis/shield. Drain wires won’t do—too much inductance. Similarly, tying shields to signal ground, or routing them to the chassis via pins and PCB etch won’t work, either. This is the wrong place, and doesn’t form an extension of the enclosure. Cable construction can also be a factor—sometimes the high-impedance connection occurs underneath the cable molding which transitions from the cable shield to the connector.

If you aren’t going to contain the energy on the wires inside of a shield, you have to get rid of it. This means low pass filtering (R, C, and occasionally L). RC filtering will band-limit differential as well as common mode signals, which may impact the decision as to whether to filter or shield. If filtering is done, the filters must be referenced to the case shield at the point of exit.

Remember, however, that it is usually common mode energy that we are trying to suppress. With proper attention to layout, magnetic (i.e., transformer) coupling and provide common mode containment.

Summing Up
Shielding works if you remember the basic principles. Create a continuous enclosure and keep an eye out for mechanical or electrical breaks in its continuity. When power or I/O signals leave the box, think about how you are going to treat the exiting leads. If you pay attention to the details involved, your shields will serve you well. n