It’s hard to imagine that a shielded enclosure could provide 3,000-dB RF attenuation, but based on the equations that define such values, it’s not a problem. Shielding effectiveness is a function of frequency, and the higher the frequency, the greater the material attenuation. It just keeps rising.
At 100 MHz, a 125-mil, all-welded aluminum box will provide greater than 3,000-dB attenuation. To achieve maximum attenuation, the welding rod should be made from the same material as the enclosure, and the box should be fabricated with no holes, sharp corners, or edges. It’s possible to get 3,000 dB from any reasonably conductive material, but the thickness will have to change.
We’ll never know if we succeeded because test-equipment limitations restrict the measurement to about 150 dB. Still, that is plenty. In fact, most commercial specifications generally only call out 60 dB; the military typically specifies 80 dB to 100 dB.
Building a shielded enclosure seems easy enough. After all, it’s just a metal box. Sometimes it’s very, very small like those in cell phones. Sometimes it’s very, very large such as the shielded buildings usually owned by multinational organizations.
Shielding often is called a mechanical solution to an electrical problem. That’s not really the case because it’s actually an electrical solution to an electrical problem. However, the devil is in the mechanical details—particularly the holes.
RF leakage through holes is the number-one problem that must be overcome to achieve a high quality shield. To sum it up in a few words, the shielding effectiveness of a hole is not very good.
In fact, depending on the resonant frequency characteristics of the hole, its presence may increase the radiation levels in certain directions way in excess of the levels produced by the circuit alone. Plus, the more holes in the enclosure, the worse it gets. Often we find that it would have been better not to use a shield and spend additional effort hardening the circuit.
Equation 1, based on the radiation efficiency of a slot antenna, provides a simple model for calculating the worst-case shielding effectiveness of a single hole or slot aperture in a thin conducting sheet
where: SE(dB) = slot attenuation
? = incident wavelength
L = slot length, which is ? ?/2
At or above the cutoff frequency for the hole when L = ?/2, the SE = 0. It’s not zero at all frequencies because of the antenna resonances. Murphy’s Law prevails. The material may be adequate, but the holes limit the shielding. Figure 1 shows a graph of Equation 1 for a 2-inch slot.
Some error results from using Equation 1, but it’s adequate for design because the actual shielding effectiveness is greater than the calculation indicates.
There is a problem relying on Equation 1 because one hole is never enough. Plus, holes of equal area can have significantly different shapes. A slot is a good example. Slots have long lengths with small widths and can be quite long without having a large opening area.
For an array of equal-sized holes, Equation 1 simply can be corrected for the number of holes. That results in
where: k = 20 for slot, 40 for round hole
The multiple hole correction (MHC) accounts for the RF illuminated hole area.
For holes illuminated in phase, the transmitted electric field from each hole recombines in phase on the other side of the enclosure wall as if it came though a hole with this equivalent area
If all the holes are the same size, which is highly unlikely, then the total area is
where: n = the number of holes
Since the first term in Equation 2 provides the shielding effectiveness for the area for one hole, the correction in dB for the additional holes is
Same-size holes that are not illuminated in phase recombine out of phase on the other side of the enclosure wall with this equivalent area
Because we are only looking for the correction for multiple holes, the correction in dB is
Unfortunately, there is no guarantee that the holes will always be illuminated out of phase. As a result, the first hole correction term normally is used to model the effects of multiple holes. As an example, two holes would be -6 dB, 10 holes would be -20 dB, and so on. You might incorrectly conclude that a large aperture array with many holes would have no attenuation at all.
Fortunately, in the near field, the number of holes that can contribute to the reduction in shielding is limited. This is determined by the separation between the holes. As the spacing between the holes increases, coherent recombination of the RF energy from the widely spaced holes becomes impossible, and the holes act independently.
As an approximation, the array attenuation is limited by the holes contained in a circular area (A = ?r2) with a radius of ?/2, Even then, this relationship fails because it makes a counter-intuitive statement: If the sizes of the openings are made smaller so more are enclosed by area (A), the shielding will be degraded. This is simply not the case. A comprehensive analysis of this deficiency was provided independently in 1982 by Bereuter and Chang1 and again in 2000 by Belokour, et al.2
Aperture shielding can be improved (as ?/2 approaches the value L) by increasing the thickness of the material equal to or greater than L. At this point, the opening begins to act as a waveguide being operated below its cutoff frequency.
For frequencies less than Fco/3, waveguide attenuation is independent of frequency and given by
where: A(dB) = attenuation
t = thickness
L = length or diameter of the opening
This equation is for one waveguide.
The constant, 30, is an approximation. The actual value differs because of hole shape and ranges from about 28 to 32.
For a t/L ratio of 4, typical of high-performance honeycomb, the calculated attenuation of one cell is 120 dB. In the frequency range between Fco/3 and Fco, the attenuation decreases, reaching 0 dB at Fco. Even so, the waveguide has significantly greater attenuation in this frequency range than the same size hole in a thin sheet. This characteristic is fundamental to the superiority of honeycomb.
Figure 2 shows the attenuation in dB/inch for a single waveguide as a function of frequency, hole size, and thickness. It is possible to run fiber-optic cable or other nonconductors through waveguides without degrading their performance. But a waveguide with an added center conductor becomes a short section of low-impedance coaxial cable.
Seams
A lot of time and money is expended trying to overcome shielded enclosure compromises that occur from two primary sources: cables and connectors and seams. The major compromises from cables and connectors are illustrated in Figure 3. In general, these can be overcome by using shielded cables and filtering the leads.
The more difficult problem is hardening the larger apertures, things like LCD screens, keyboard bezels, ventilation panels, honeycomb material, filter and access panels, and doors. These items, especially access panels and their seams, create the largest apertures.
Seams tend to be the worst-case apertures. As a result, using smaller enclosures limits the maximum size of any seams. Smaller enclosures also have fewer penetrations, and these also are smaller.
Minimum enclosure size is determined by circuit size and should completely enclose the offending or susceptible circuits and nothing more. Cell phone implementation of PCB shielded boxes is a good example of how this is done.
Even though a solid metal access panel could provide thousands of dB attenuation, if it is not adequately bonded around its perimeter, it may as well not be there. Any panel that is not adequately bonded to the enclosure behaves as a lossy antenna structure located in a hole surrounded by metal.
Single-point grounding of the panel will reduce the antenna efficiency and may even solve some radiation problems at low frequencies but will not eliminate leakage from the rest of the seam. For example, if a 120-dB honeycomb panel is improperly installed over a small 6-inch muffin fan, the perimeter leakage can limit the enclosure attenuation at 100 MHz to about 20 dB.
The best installation methods are welding, brazing, soldering, or riveting—in that order. But these methods do not allow easy access for maintenance and repair.
Using closely spaced threaded fasteners or clamps helps but does not facilitate field removal. As a result, RF gaskets generally are used on larger panels to reduce the number of fasteners and still maintain contact across the seams.
There are three basic seam designs: the isolated seam or butt joint, the compression seam, and the shear seam. These are illustrated in Figure 4.
The isolated seam is popular in weight-sensitive applications because there are no overlapping surfaces that increase material content. Plus, the shielding material does not need to handle gasket compression forces and can be very thin. Foil, in fact. This fixed configuration is used frequently in lightweight spacecraft and satellite applications. There even is one low-cost hard-drive manufacturer using this approach.
Compression seams are the most popular, not because they’re so good but usually because there were no initial plans for EMC protection for the enclosure. It’s just a happening that occurs at the last minute when everything is failing. Using this technique, it’s possible to convert a low-cost unshielded box with an overlapping seam into a shielded enclosure by adding an RF gasket and more fasteners into the design.
The shear seam is the only dynamic configuration and much different than the other two. It is constructed in several different configurations, such as pan-edge, knife-edge, modified knife-edge, or longitudinal. These designs align the mechanical forces parallel to the panel surface and do not require clamps or fasteners to provide RF gasket compression force. Solid metal gasket materials normally are used for this application.
Because of the difficulty retrofitting to a shear seam, this design configuration should be considered from the beginning of the project. Accordingly, the shear configuration is lower in cost than the compression configuration because the structural materials can be much lighter, there is a savings in fasteners, and there are no labor costs to install them.
RF Gaskets
Based upon geometry and materials, there probably are several thousand gasket varieties that can be loosely categorized as solid metal gaskets (flat spiral coil spring, linear spring fingers, and rings), wire mesh/oriented wire (BeCu hollow core tubing, Monel, and SnCuFe), conductive elastomers, and conductive fabric-covered foam. Each of these materials has distinct advantages and disadvantages.
Total design cost is the major factor in the RF gasket choice, not the cost of the gaskets themselves. Regardless of the gasket type, the most important factor to be considered when choosing a gasket is its RF impedance
where: R = resistance
j X = inductive or capacitive reactance
because this determines its shielding effectiveness. Long term, we also must consider a material’s compatibility, corrosion control, compression force, compressibility, compression range, compression set, and environmental sealing.
When selecting a gasket to be used for DC or low-frequency grounding/bonding, low-contact resistance and high conductivity across the seam are the most important parameters. At the higher RF frequencies, having low gasket impedance becomes the most important parameter.
This can be accomplished by having very high capacitance or very low inductive reactance. Because most gaskets are conductive, minimizing inductive reactance is the goal. This parameter is directly related to self-inductance that, for a rectangular conductor, is related to its geometry as follows:
- Increasing the thickness by a factor of 10x decreases inductance approximately 10%.
- Increasing the width by a factor of 10x decreases inductance by approximately 30%.
- Decreasing the length by a factor of 10x reduces inductance by approximately 93%.
Consequently, the shortest conductor with the largest width/length ratio will provide the lowest inductance.
Determining which gasket to use from the many thousands of choices is a daunting task. It’s not possible to select an RF gasket based on visual inspection, and sales personnel and literature can’t possibly tell everything.
The real choice should be made by installing what appears to be a good solution in the prototype and then sending everything to the EMC lab to find out if it meets the requirements. If it does—great. If it doesn’t, then do it all over again.
Fortunately, competition within the industry has resulted in RF gasket products being tested so that the various manufacturers can claim bragging rights about how much better their product is than the competition. Even then, arguments abound regarding which test method is better and why.
There are two principal test methods used to measure RF gaskets. One is based on measuring the transfer impedance of the gasket material. The other measures the transmission loss of a radiated signal from the outside to the inside of a shielded enclosure.
Although there are several variants for both approaches, the two most widely used in the RF gasket industry are MIL-DTL-83528 (radiated) and SAE ARP 1705 (transfer impedance). Most RF gasket manufacturers test gaskets using both methods.
Application of Shielding Measurements
Radiated
A radiated procedure using the MIL-STD-285 shielded enclosure measurement methods has been used for RF gasket evaluation since the early 1960s. Originally, it consisted of removing the cable entrance panel from an electromagnetically tight shielded enclosure, replacing the OEM RF gasket material with the RF gaskets to be evaluated, and then performing a MIL-STD-285 test on the enclosure to evaluate the shielding using the test RF gasket material. Different shielded enclosure manufacturers used different-sized cable entrance panels that generally were rectangular, not square. This resulted in major differences between vertical and horizontal polarizations.
In an attempt to standardize the MIL-STD-285 procedure for testing conductive elastomers, Chomerics created a new standard called MIL-G-83528, which now is known as MIL-DTL-83528. This procedure uses a standardized 24″ x 24″ hole in the enclosure side about midway between the floor and the ceiling. That hole is constructed with a standardized seam and covered by a metal plate that has a shielding effectiveness equal to or greater than the unmodified enclosure.
The panel is held in place using stand-ardized bolt spacing. A MIL-STD-285 shielding effectiveness test is done by placing antennas on either side of the open hole. The metal plate and the test RF gaskets then are installed, and the shielding effectiveness test is repeated. The difference is presumed to be the shielding effectiveness of the RF gasket that is assumed to be a poorer shield than the metal plate. This error in the procedure stems from the assumption that the attenuation is limited by the RF gasket and the measurement represents the gasket.
To simplify the procedure, the test ignores the problems of reflections from the conducting wall surfaces, enclosure resonance, and antenna loading. The minimizing process is time-consuming and not repeatable. In addition, the cover plate acts as a slot antenna which has significant directivity whose main lobe typically is not in the direction of the measurement antenna, plus the opening size provides its own shielding at the lower frequencies. These problems introduce errors that may be in excess of 20 dB. Refer to IEEE 1302-1998 for a discussion of the various errors.
Regardless of the many problems, the radiated procedure has a number of supporters because it is easy to visualize that it measures the shielding effectiveness of an enclosure using the test RF gaskets. It does not measure the RF gasket material. There is a big difference. On the other hand, it is a standard, and so long as every manufacturer does the test the same way, the data can be compared.
Transfer Impedance
The SAE ARP 1705 transfer impedance test measures the electrical characteristics of the RF gasket material. Shielding effectiveness is directly related to the material’s surface transfer impedance (Zst). This is the voltage-to-current ratio for the tangential electric field developed on the interior of the shield to the driving surface current density on the exterior shield surface created by external RF excitation sources.
Figure 5 illustrates the relationship between the radiated and transfer impedance methods. In the radiated case (A), an antenna-coupled EF induces current into the RF gasket material. For transfer impedance measurements (B), a signal generator is used. The Zst is a complex quantity that combines the material’s DC surface resistance (Rs) and its surface inductive reactance (Xsl).
The intrinsic impedance and the surface transfer impedance become equal for material thicknesses greater than approximately three skin depths. Accordingly, the measurement of the surface transfer impedance can be used to determine variations in the material characteristics that most directly influence the shielding effectiveness of that material.
The SAE ARP 1705 transfer impedance measurement is performed using a coaxial fixture that allows placement of the sample between the inner and outer conductors of the coaxial fixture. Over the test frequency range, RF current is driven through the test sample.
The resistive and inductive losses that make up the combination of transfer impedance plus any series contact impedances cause an RF voltage to be developed across the current conduction path. The ratio of the measured voltage to the current that created it defines the transfer impedance. This test is graphically illustrated in Figure 6.
Since the conductor path length influences the transfer impedance and different test samples can be of different lengths, the data is normalized to 1 meter. Transfer impedance measurements provide data that is directly related to the characteristics of the shielding materials that affect their capability to act as a shield.
In addition, the measurement can be performed with easily obtainable test equipment and has excellent repeatability with variations of less than ±2 dB. Radiated measurements are very uncertain and can vary as much as 20 dB.
A Few Words of Advice
Shielding works quite well once its limitations are understood. It is frequently used as a stand-alone EMC solution. It is the only noninvasive suppression technique available and can be used to reduce both radiated emission (RE) and susceptibility/immunity (RS/I).
Since shielding is not inserted into the circuit, it solves EMC problems without affecting a circuit’s high-speed operation. It only is necessary to completely surround the problem circuit with a solid homogenous metal enclosure. Therein lies the problem.
Solid and homogenous make it difficult to design and build. It costs lots of money to build a high-performance shielded enclosure that provides greater than 100-dB to 120-dB attenuation, possibly as much as 10x what it would cost to build a 60-dB enclosure.
For that reason, it is prudent to harden the electronics as much as possible before designing the shielding. When shielding is necessary, take advantage of any inherent shielding and use multiple small enclosures. This approach generally is lower in cost because it reduces the requirement for filters and RF gasket materials. An enclosure not only will reduce RF coupling of the circuit with the external RF environment, it also will reduce or eliminate crosstalk among circuits.
References
1. Bereuter, W.A. and Chang, D.C., “Shielding Effectiveness of Metallic Honeycombs,” IEEE Transactions on EMC, Vol. EMC-24, No.1, February 1982.
2. Belokour, I., LoVetri, J., and Kashyap, S., “Shielding Effectiveness Estimation of Enclosures With Apertures,” 2000 IEEE EMC Symposium, Record 0-7803-5677-2/00.
About the Author
Ron Brewer currently is a senior EMC/RF engineering analyst with Analex at the NASA Kennedy Space Center. The NARTE-certified EMC/ESD engineer has worked full-time in the EMC field for more than 30 years. Mr. Brewer was named Distinguished Lecturer by the IEEE EMC Society and has taught more than 385 EMC technical short-courses in 29 countries and published numerous papers on EMC/ESD and shielding design. He completed undergraduate and graduate work in engineering science and physics at the University of Michigan. e-mail: [email protected]
February 2008