Building A Faraday Cage To Contain EMI, Part 2
Part 1 of this article addressed the various ways in which EMI impacts products, and covered how to build a Faraday cage to contain it. This part of the article will focus on getting power into a product without letting EMI out; building a Faraday cage for rack-mounted products, determining whether or not logic ground should be connected to a Faraday cage while also providing some additional ways to build one.
How to Get Power into a Product Without Letting EMI Out
As noted in Part 1 of this article, a Faraday cage seals in signals and heat unless methods are provided for them to enter and exit. This creates a possible exit for EMI. In addition, power may be delivered to the circuits inside the Faraday cage while inadvertently allowing EMI to escape on the power leads. This is prevented by placing some form of low pass filter in series with the power lines. As cited in Part 1, for conducted EMI, the frequency band of interest is from 150 KHz to 30 MHz. This frequency band is low enough that ordinary discrete components will function properly.
One form of low pass filter is built into the AC connector module as power enters the product. For those products that are powered from DC, such as the 48 Volt DC supplies that are common to Telco product installations, the problem is a little more complex.
When a product is comprised of a backplane and a series of plug-in cards that are supplied raw 48 VDC, the most successful strategy for containing conducted EMI is to place low pass filters on each module as the raw DC enters the PCB from the backplane. This way, the potential EMI never gets a chance to leave the module by this path. There are several different component suppliers who have modules designed for this purpose. It’s also possible to construct these filters from discrete inductors and capacitors.
For low-cost products that use wall plug mounted power supplies, the low pass filter must be installed on the main PCB as the power enters it. There are also commercially available modules that are designed for this purpose.
Building a Faraday Cage for Rack-Mounted Products Containing Plug-in Cards and Backplanes
To build a Faraday cage around a card cage, backplane and a group of plug-in modules, it’s necessary to find a way to surround the cards, card guides, and backplane with a conductive enclosure that has six sides. At the same time, it’s necessary to allow for cooling air to be forced through the cards usually from the bottom to the top through the slots that are formed by the card guides.
The least expensive way to do this is to form two sides of the Faraday cage from the two solid sides of the card cage. The back is formed by the ground planes in the backplane. These are bonded to the flanges of the card cage with strips of plated copper on the backplane. This is shown in Figure 1.
Figure 1. Backplane with Copper Bonding Strips
Using this approach, it is not necessary to have a special plane on the backplane (referred to as “chassis ground”). There are some EMI gurus who insist that the edges of the PCB be connected with grounding strips along the edges of the card guides. This is not necessary for passing EMI and it may actually make EMI worse by providing more than one connection between logic ground and the Faraday cage. The reason this is undesirable is detailed in Part 1 of this article. It’s also not necessary to plate the edges of the backplane to contain EMI.
In the aforementioned six-sided Faraday cage, the top and bottom of the card cage have slot-like openings between the card guides which are usually large enough to allow EMI to escape. As cited in Part 1, a honeycomb will block these paths while still allowing cooling air to flow through. This honeycomb needs to be tightly bonded to the top and bottom of the card cage.
Thus far, we have addressed five of the six sides of the Faraday cage but still need to deal with the front of the cage. This cage front is filled with plug-in modules. It’s necessary that the faceplates of these modules act as a single surface to block EMI from escaping while still allowing modules to be put in and taken out. It is possible to put EMI gasket material on the sides, top and bottom of each module that will bond one faceplate to another which thus forms a good EMI seal. Two examples of this are shown in Figure 3.
Figure 3. EMI Gasket Material on Plug-in Module Faceplates
The seal on the left is compressible foam covered with a very fine woven metal mesh. The seal on the right is a series of spring fingers that contact the neighboring card. This completes the six sides of the Faraday cage. Signals and power that enter and exit must be handled as described in the above sections.
In the product containing the backplane in Figure 1, the power delivery system and cooling fans were housed outside the Faraday cage and connections were made between them using various types of filtering. Using these techniques, this product—which was a terabit router consuming 7 kilowatts of power and filling one-half of a rack—passed the specified EMI tests on the first try.
Where Should Logic Ground Be Connected to the Faraday Cage
In part 1 of this article, I noted that it was not necessary to make a DC connection between the logic ground of the Faraday cage in order to contain EMI. In the case of those products that have RS232 interfaces, this is not allowed if the product has a green wire ground connection. If logic ground is not connected to the Faraday cage, all of the circuitry inside the product will float at a voltage potential that is different from the Faraday cage. The voltage at which the circuits float may be of such a frequency that it could cause EMI to appear on exiting unshielded wires such as the UTPs used to connect Ethernet circuits.
In order to prevent these problems, it is essential to connect the logic ground to the Faraday cage. The question becomes where and how often. In other articles here, it’s been pointed out that voltage gradients exist across the structure that is referred to as logic ground. It’s also been cited that if you connect logic ground to the Faraday cage in more than one place, you run the risk of impressing the voltage gradient on the Faraday cage, turning it into an antenna. That’s why we state that no more than one connection between the logic ground and the Faraday cage is appropriate. This appears as though this is many more connections than we recommend.
Why does this work? Keeping in mind that we want the entire ground plane of the backplane to form one side of the tightly sealed Faraday cage, the backplane must be bonded as shown in Figure 1. To prevent currents from flowing in as a result of this action, it is necessary to ensure that there is no voltage gradient in the ground plane of the backplane. In Figure 1, the backplanes don’t carry any power, so there is no current flow and no voltage gradient. This enables us to make the connections noted above.
The question becomes ‘where is the power distributed in the backplane, if it is not in the ground planes?’. In this case, power is 48 Volts DC, which is distributed in the unused spaces in the signal layers. As a result, the planes of the backplane serve only as partners for the transmission lines.
In the case of “pizza box” style products, where there is a row of RJ-45 connecters along the front edge, it is possible to connect all of the housings of these connectors to the Faraday cage along the front of the box. There is no EMI problem as long as the ground plane of the PCB is not cut up in any way. Note: There are EMI gurus in the industry who advise cutting the ground plane under the output transformers to control EMI. In fact, this approach always creates an EMI problem, and as a result should never be done.
The Other Ways to Build a Faraday Cage
Most commercial products consist of a single PCB. This includes devices such as (some) cell phones, small routers and hubs, and printers. These products are characterized as having the need to contain costs while ensuring performance and conformance with EMI standards.
With cell phones, which usually have plastic cases, a common solution is to vacuum deposit a thin metal film on the inside of the case and bond it to the logic ground plane of the PCB. This is a very low cost, effective way to build a Faraday cage. Millions of other products, such as routers, switches, and hubs, are brought to market with the same cost constraints as cell phones. These products are commonly housed in a stamped metal case that serves as the Faraday cage. In some instances the product housings are molded plastic with the same vacuum-deposited metal lining noted above.
Low cost products such as inkjet printers and video game consoles cannot be housed in Faraday cages due to their construction. These types of products do operate at high enough frequencies that they can fail EMI tests. Since Faraday cages can’t be used, this means EMI must be contained in some other way.
A common way to do this is with spread spectrum clocking (SSC). In spread spectrum clocking, the clock period is altered from cycle to cycle so that the radiated noise is spread out such that the amount of energy at a given frequency is reduced. This is done by modulating the clock with a noise source, and it works quite well.
EMI can cause a product to not operate properly or perhaps fail altogether. Understanding the origin of EMI, how it impacts a product’s operation and the different ways a Faraday cage that can be built to contain it will ensure that a product will work from the first cycle of operation through the entire product’s lifetime.
Would you like to find out more about how Altium can help you with your next PCB design? Talk to an expert at Altium or continue reading about advanced routing and verification of ground and power supply PCB layout traces with Altium Designer®.
Ritchey, Lee W. and Zasio, John J., “Right The First Time, A Practical Handbook on High-Speed PCB and System Design, Volume2.”