Addressing electromagnetic compatibility late in the design cycle is becoming less and less tenable as product complexity and densities increase while design cycles continue to shrink. The rules of thumb commonly used to calculate EMC are breaking down at higher frequencies and can easily be misapplied, resulting in high late-stage redesign costs and often even higher lost sales costs if the product ship date is delayed. Designers should consider the use of collaborative, conceptual analysis-based EMC simulation early in the design process to identify and fix problems at a much lower cost.
The embedding of faster and faster processors into appliances leads to more challenging EMC concerns at higher frequencies than in the past. In the Gigahertz world, enclosure resonances enhance emissions, making apertures and seams problematic, and ASIC heat sinks can exacerbate radiated emissions. In addition, regulations are evolving to ensure compliance at higher and higher frequencies. To top it off, the trend towards integrating wireless capabilities, such as Wi-Fi, Bluetooth, WiMax, ZigBee, and UWB, into virtually all appliances presents further challenges, as intentional radiators (antennas) must be designed into systems and be kept from interfering with critical functions.

Traditional approach

Normally, EMC design is considered by both the electrical hardware designers and mechanical designers in parallel, with little if any communications between the two groups. Rules of thumb are often used during the design process with the hope they will be sufficient for the device being designed. Many of these EMC rules are becoming obsolete at higher frequencies, leading to failure during testing.
After the design stage, a prototype is built and tested for EMC compliance. This all too often results in EMC problems being identified at a point when it is too late to design in EMC compliance. Often, expensive fixes on the existing design are the only options available. Design changes generally increase by an order of magnitude or more as the design moves from conceptual to detailed to validation stages. So a change that would only cost $100 at the conceptual level might exceed tens of thousands of dollars at the testing stage, not to mention the negative impact on time-to-market.

Simulation challenges

Fig. 1. With same source and input power in all three cases, emissions increase solely due to physical configuration. In the upper left, the radiated emissions are due to a single PCB in a backplane. The upper right shows the same for two PCBs in the backplane, and the lower figure for three. In all cases, only one PCB is an active source.

It has become essential to include EMC design as an integrated part of the product cycle to obtain first-pass compliance in the test chamber in order to ensure on-time delivery within budget. This can be achieved with a 3D solution of Maxwell’s equations, which provides an elegant mathematical representation of electromagnetic interactions. But EMC simulation presents specific challenges not always encountered in other areas of computational electromagnetics.
A typical EMC problem involves an enclosure that is very large relative to specific features such as slots, holes and cables, all of which are important to EMC performance. Accurate modeling requires that the large and small details be included in the model. This results in a high aspect ratio (the ratio of the largest to smallest feature), which in turn requires very fine grids to resolve the finest details. Compact model technology can allow large and small structures to be included in a simulation without prohibitive simulation times.
The sources of noise for EMC emissions are usually at the PCB or IC level. Including detailed source models in a large system level EMC simulation can be cost prohibitive using conventional modeling approaches, yet an accurate source model is necessary for accurate system-level predictions. New modeling procedures can be used to create compact source models for system-level EMC simulations that accurately capture the source characteristics without the need for geometrically complex and computationally intensive source models. In addition, creation of compact source models can be done using data available from existing tools being used for PCB design.
Another challenge is that the EMC characterization must be performed over a very wide frequency range. The time required to calculate electromagnetic fields at each sample frequency would be prohibitive. The transmission line matrix method, and other time domain full-wave methods, perform the field solution in the time domain using broadband excitation, yielding data over an entire band in a single simulation run. In TLM, space is divided into cells modeled at the intersection of orthogonal transmission lines. Voltage pulses are transmitted and scattered at each cell. Electric and magnetic fields are calculated from voltages and currents on the lines at each time step.
EMC simulation at the system level can yield quite accurate results. Fig. 1 compares the computed radiated power in dBuV/m (red) to the measured radiated power (blue) for three configurations of a PCB module plugged into a backplane. The small discrepancies in resonant peak location for the multimodule cases can be attributed to difficulties in obtaining precision alignment of the modules in the measurements. It is interesting to note that the differences in resonant peaks and amplitude of radiated power between the plots is due solely to the physical (mechanical) layout of the system, as the input power is the same in all cases.

Application range

EMC simulation is applicable for examining components and subsystems such as radiation profile versus frequency in heat sink grounding, as well as assessing different grounding techniques, the impact of heat sink shape, and so on. The shield effectiveness of varying metal thicknesses and different sizes and shapes of air vent holes can also be compared. Recent work in this area has included a study to evaluate the use of large-hole air vents to allow for air-flow while controlling EMC by placing two such panels back to back.
EMC simulation is also well-suited to EMC design and optimization at the system level to compute broadband shielding effectiveness, broadband radiated emissions, 3-D far-field radiation patterns, cylindrical near-field radiated emissions to mimic a turn-table type measurement scenario, as well as current and E and H field distributions for visualizations that help to locate EMC hot spots. Typical system level EMC applications include:

• Designing enclosures to ensure maximum shielding effectiveness.
• Assessing the EMC ramifications of component location within an enclosure.
• Computing cabling coupling both internal and external to the system.
• Examining the effects of radiation from the cables.

EMC simulation also helps identify specific mechanisms for unwanted electromagnetic transmissions through chassis and subsystems such as cavity resonances; radiation through holes, slots, seams, vents and other chassis openings, conducted emissions through cables; coupling to and from heat sinks and other components; and unintentional wave guides inherent to optical components, displays, LEDs, and other chassis mounted components.

Enclosure shielding

Fig. 2. Evaluating different thicknesses and hole shapes of panels. The geometry is shown above while the graphs show the shielding provided by the panels for different thicknesses (left) and different hole shapes (right).

A typical application of EMC simulation is to evaluate the shielding of different ventilation panels. While there are rules for designing air vent panels for EMC leakage, EMC simulation can accurately predict more exotic configurations, such as back-to-back panels with large holes, waveguide arrays, etc., while keeping thermal and cost constraints in mind. The application in Fig. 2 shows the computation of shielding for panels with round and square hole geometries and different thicknesses. The graphs show the shielding provided by the panels for different thicknesses (left) and different hole shapes (right).

Heat sink radiation

Fig. 3. Radiated power spectrum for three heat sink configurations: a.) small floating ungrounded heat sink (yellow), b.) small grounded heat sink (green), and c.) large floating heat sink (red).

The EMC simulation application shown in Fig. 3 examines the radiation from a heat sink. In this simple model, the heat sink is excited by a broadband signal source located directly underneath it, representing electromagnetic coupling to the heat sink from an IC to which it is bonded. The plot shows the radiated power spectrum for three different configurations. Clearly, the radiation level depends on the geometry and frequency. While grounding the smaller heat sink provides and improvement at lower frequencies, the radiation is increased in the middle part of the frequency range.

Cable coupling solution

Fig. 4. Multiple rack-mounted hubs. Currents are coupled via cabling between the hubs. The filter reduces the coupling about 50 percent at 600 MHz and 75 percent at 800 MHz.

The example in Fig. 4 shows the use of EMC simulation to examine system-level cable coupling. The geometry consists of three network hubs in a 19-inch rack as shown. A four-wire ribbon cable connects the PCBs in the top and bottom hubs to the middle hubs. The center hub has the only EMC source in the model. EMC simulation computed the currents coupled from the center hub to the connection at a PCB in the upper hub. The coupled current displayed two strong resonances at 600 MHz and 800 MHz. A common approach for dealing with this sort of problem is to add filtering to the affected cable, and then gauge the impact with simulation. The lower plot shows that adding a low pass filter reduces, but does not eliminate, the magnitude of the coupled current at the resonant frequencies. This is a band-aid fix because it does not address the problem at its source.

EMC simulation was used to visualize the internal physics of the cable coupling application in order to find the root cause of the problem. This type of diagnostic visualization is usually impossible to do using lab measurements. Examining the electric field distribution inside the center hub at 600 MHz made it possible to identify electric field hotspots that identified a cavity resonance that generated high field levels near the cables. By adding a metal partition to the hub, the resonant cavity mode was suppressed and the coupling was eliminated as shown in Fig. 5. This solution eliminates the source of the problem and resulted in a solution superior to the brute-force filtering approach.

Fig. 5. Adding a metal partition solved the root cause of the cable-coupling problem.

Compact sources

When using simulation for system level EMC design, it is very useful to be able to replace geometrically complex internal noise sources, such as PCB’s, with equivalent compact models that accurately represent the electrical characteristic of the noise source without requiring a high fidelity geometric model to be included within the large system level model.
In many cases, the PCB’s in a system are designed and analyzed or tested separately from the system level design and it is very desirable to use the information from these design streams in the EMC design and simulation. Modern computational electromagnetics methods provide ways of using known PCB emission characteristic in large system level models as the noise sources in computationally efficient ways. For example, if the near field radiated emissions of a PCB can be obtained from other PCB-specific simulation tools or measurements, then this data can be used in the system level EMC simulations to drastically reduce simulation time while retaining a high degree of accuracy.
In addition, a re-usable library of compact models for commonly used PCB’s and components can be maintained for use in system level EMC design. The information needed for the creation of these compact sources is simply the near field mapping of electric and magnetic fields close to the board. An example near field scan is shown in Fig. 6.

Fig. 7. System enclosure with aperture and internal PCB used to demonstrate compact source modeling.

To demonstrate the compact source modeling, a simple enclosure with an aperture containing a PCB with multiple traces and sources as well as terminations is considered as shown in Fig. 7. The detailed traces and components on the PCB normally would require a very fine resolution computational grid if modeled explicitly in the enclosure leading to large simulation times.

By using the compact model, significant computational savings are realized; reductions by several orders of magnitude are possible, meaning the difference between practical and prohibitive simulation times. As seen in Fig. 8 where the radiated emissions are shown, the accuracy level using the compact model is very good compared to the highly accurate detailed model.

Fig. 8. Radiated emissions from enclosure using a detailed (blue) and compact (red) PCB model in the enclosure. Results are shown up to 2 GHz.

Thermal design and EMC

In another example, EMC simulation was used to identify and solve a problem that arose from a thermally driven design change. The example is based on a model of a controller node, essentially a dual-processor Pentium computer, for an enterprise storage system. After this design was committed to hardware, the standard Pentium heat sinks were replaced with heat-pipes that occupied the same footprint as the heat sinks, but were taller and had fins that were oriented horizontally instead of vertically. The simulation models for the two cases are shown in Fig. 9.

A broadband simulation was performed to compute the radiated emissions of the system. Engineers were specifically interested in isolating the emissions due to a 120-MHz oscillator signal present in the system because they had measurements indicating a problem. Therefore, after computing the broadband response, an indirect excitation was used in post-processing to extract the response to the desired source signal. It is very clear from the radiated emissions shown in Fig. 10 that the radiation increases significantly (~40 dB) at the fundamental harmonic of the oscillator frequency (120 MHz). It’s remarkable that such an innocuous thermal design change has such a major and alarming impact on EMC compliance.

Fig. 11. Emissions using P4 heat sink vs. heat pipe.

Fig. 11 shows the visualization of the electric field in a plane through the center of the controller node (side view). Note that the controller node can be seen in outline in the figure and the visualization plane extends beyond the enclosure in order to see the radiated field outside of the box. The figure on the left shows the design with heat sinks and the one of the right shows the heat-pipes. It is obvious that the field levels both inside and outside the controller node with the heat-pipes are significantly higher than for the baseline case. When examining field levels for various visualization planes and locations, it quickly became apparent that the close proximity of the horizontal fins to the lid of the enclosure created a large capacitive coupling path, creating a resonance with high field levels that generated high levels of radiation.

Having identified the root cause, cost-effective solutions were explored. In this case, eliminating the capacitive coupling path by making a ground connection between the top of the heat-pipes and the enclosure lid provided an excellent, low-cost solution. This was achieved by placing a small section of EMI gasket with a conductive adhesive on the top fin of the heat pipes such that the contact with the lid compresses the gasket and forms an electrical ground connection. That solution eliminated the resonant phenomena. Fig. 12 shows the radiated emissions plot with the results for the grounded heat pipe included. The fix resulted in emissions that are virtually identical to the baseline case, improving thermal performance without having a negative impact on emissions.

Fig. 12. Grounding the heat pipe fixed the EMC problem.

Summary

As can be seen from the previous examples, using simulation early in the design process makes it possible to investigate and predict key EMC phenomena, and, therefore, optimize electronic product design in terms of EMC requirements and shielding effectiveness before building a prototype. Modern simulation tools enable designers to evaluate more designs than is practical to prototype, and optimize products from an EMC perspective to a level that wasn’t possible in the past.