|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
November 6, 1997 Tools and techniques stifle EM emissions JIM LIPMAN, TECHNICAL EDITOR Faster clocks, deep-submicron chips, denser boards, and stricter government EMC regulations are forcing you to tighten your pc-board designs. Using the right EDA tools and good high-speed-board engineering techniques helps you design EM-compliant systems. If you design high-speed pc boards, one specification you need to consider is how the board and the system in which you place it handle electromagnetic (EM) radiation. Two EM-radiation problems that board circuitry and traces can have are excessive EM emission and susceptibility to EM emission. These problems are so prevalent that many national governments set strict limits on the allowable radiation magnitude for an electronic system. (The European EMC Directive also sets EM- susceptibility limits.) The traditional way to check if a pc board is EMC-compliant is to measure EM emissions and EM susceptibility on a prototype. However, shrinking design cycles and end-product price pressure have made waiting for a physical prototype a nonviable way to design many products, particularly those for the consumer market. You need to check for EMC compliance during component placement and routing on a board, not after you reach prototype status. Fortunately, you have a variety of EDA tools available to check EMC-design rules; to simulate pc-board, cable, and enclosure EM emissions; and even to offer advice on how to correct EMC problems. Board equals radio transmitter High-speed digital chips and on-board connections between these chips are major sources of EM emission from pc boards. Radiated emissions result from EM waves from both electric (E) and magnetic (H) field vectors. The E and H fields are perpendicular to each other and to the propagation direction of the EM wave. High-speed clocks, fast signal-edge rates, long signal lines, and poor grounding all contribute to high board-radiation levels and to unacceptable EM susceptibility. Potential signal-integrity problems, such as ringing, undershoot, and overshoot, can also contribute to high radiation emission (Reference 1).
Inadvertent voltage drops in a system, often on the ground plane of a pc board, result in common-mode radiation currents (Figure 1b). When you connect cables to the board, common-mode currents also flow through the cables. Common-mode currents travel in the same direction on both signal and ground conductors, resulting in a monopole antenna radiating energy proportional to the current, the length of current path, and the radiating frequency. Because a monopole antenna is a high-impedance source, the E field dominates EM emissions. Energy from both differential- and common-mode radiation sources de-creases inversely as the distance from the source increases. Whereas differential-mode radiation results from intentional signal-to-receiver-to-ground loops, board-based systems unintentionally generate common-mode radiation, making it more difficult to predict and control. Designing a ground system with a minimum of discontinuities reduces common-mode radiation. Minimizing ground-path length on both the board and the connecting cables also reduces this radiation source. For both differential- and common-mode radiation, high-frequency and high-current signals are big radiation sources. These signals include clocks, strobes, and read/write lines. Repetitive signals, such as clocks, are also potentially troublesome, because periodic signals concentrate their energy in discrete harmonics. Asynchronous signals spread their energy over a broader and lower frequency range and do not generate relatively high-energy levels at any one frequency. When analyzing pc boards for EM radiation, you should first concentrate on the components, board traces, and cables carrying high-current signals, synchronous signals, or both. At first glance, it appears that modern pc boards, which have a number of enhanced features, would produce lower levels of EM radiation. Higher trace density, small component packages, and surface-mounted components result in smaller loops, which radiate less power. In addition, multilayer boards have lower impedance ground planes and additional trace shielding, both of which also reduce emissions. However, higher clock rates and faster logic have probably more than made up for these reductions, putting even more emphasis on your use of EMI-analysis and advisory tools to meet EMC requirements. The "cut-and-try" methods you use on simpler systems just aren't adequate for leading-edge pc-board designs. Analyzing EM radiation When designing a board to meet EMC specifications, you should be concerned with how much EM radiation your design generates and in what frequency region the design produces this energy. Many countries, including the United States and those in Western Europe, have stringent EM-radiation requirements for electronic systems (see box "US and European EMC standards"). You also need to know how susceptible your design is to self-generated EM radiation and to EM emissions from other sources. You have many EDA tools to choose from for estimating or measuring generated EM radiation. However, no good tools exist that can directly tell you how EM-susceptible your board is. Similar to what you should do for designing boards with speed and power constraints, you should design high-performance pc boards with EMC constraints as well. This procedure entails simulating EM radiation during component placement and routing. You can use advisory or simulation EMC-design tools. You use advisory tools (Table 1) to tell if your board complies with design rules for meeting EMC specifications. Advisory tools are "expert" systems that let you know, during component placement and routing, if you have some topological or other problem that would likely result in excessive EM emission. Some EMC advisors may include EM-emission simulation, whereas other tools are just design-rule checkers based on built-in rules or a combination of a built-in-rule set and user-defined design rules. The advantage of an advisory tool with user-definable EMC rules is that you can add your own company- specific design rules, eliminating some checks you would normally do during a design review. An advisory tool is no substitute for a good EMC-simulation tool; the two are complementary. You use an advisory tool in the early stages of board placement and routing, when you can make quick "what-if" EMC analyses and modify your pc-board design on the fly. However, adherence to EMC-design rules does not eliminate the need for careful EMC modeling and simulation after you finish board placement-and-routing design. No EMC- advisor design-rule set is complete or can guarantee that you'll meet EMC specifications.
An example of an EMC advisory tool with no simulation is Zuken-Redac's EMC Advisor. Designed as an option for the company's pc-board placement-and-routing-tool packages, EMC Advisor has a set of predefined design rules against which it checks a placed-and-routed design. These design rules, covering board traces, component placement, termination quality, and other design parameters, check EMC compliance based on your inputted design-specific constraints. The tool graphically highlights problems and displays histograms showing the seriousness of those problems. In addition, EMC Advisor also offers suggestions on how to fix problems to meet EMC specifications. Another knowledge-based EMC advisory tool with no simulation is Cadence's EMControl, which works with Cadence's Allegro pc-board-design system. EMControl has built-in EMC-design rules, design-rule checking of pc-board-component placement and routing, graphical problem display, and advice on how to correct problems. Similar to how you work with Quiet, you can add your own company- or design-specific design rules. EDAnavigator, Xynetix's virtual-prototyping tool, takes a modular approach to working with design advisors. The basic EDAnavigator software lets you add electrical, thermal, physical, and manufacturing constraints to your pc-board placement and routing. From schematic input, you use the tool to place components and route a board, guided by physical and electrical design rules. EDAnavigator's "advisor backplane" lets you connect three design advisors to the tool: first order, which offers quick analysis and order-of-magnitude analysis; second order, which offers more detailed analysis and medium runtime; and third order, which offers very detailed analysis and slow runtime. The first-order EMI advisor calculates only rough component EM-emission noise based on simple parameters, such as the clock frequency, the signal rise time, and the number of package I/O pins. You can have a third-order EDAnavigator advisor by buying an interface to Viewlogic's Quiet. Xynetix also has optional second- and third-order EMC-advisor interfaces available for Incases' EMC Workbench. EM-radiation-simulation tools When choosing an EDA tool for modeling and simulating EM radiation, keep in mind how well the tool models your design and how long the tool takes to run a simulation. Modeling a board, onboard components, cables, and connectors is complex. To simulate radiated emissions, a simulator must first read in the geometry of the system being simulated, either from a solid modeling program or from a geometry file, such as GDSII or Gerber. Then you define the materials and their electrical properties that the system uses. The simulator uses finite or integral methods, or a combination of both, to solve Maxwell's equations and produce electric- and magnetic-field patterns representing the EM emissions (Reference 3). Computer-memory requirements depend on the solution method a simulator uses. Finite-difference and transmission-line-matrix methods need memory proportional to the volume of the space they are modeling. The finite-element-method memory usage is proportional to the square of the space it models, with some reduction possible if the simulator uses sparse-matrix techniques. Integral methods, on the other hand, work with a reduced simulation space and usually need less memory for a problem. No matter which solution method a simulator uses, memory requirements and runtimes increase so rapidly with problem complexity that most system EM simulations need some type of problem simplification to run in a reasonable time. You can simplify the system description that the simulator uses by eliminating structures that are not critical for EMC design. However, how you eliminate those structures without invalidating simulation results may be difficult and is best handled by a designer with experience designing for EMC specifications.
Another point to consider is that EM-emission-simulation tools are mode-dependent; they model and simulate designs for an operational mode. The mode, and hence EM emission and susceptibility levels of pc boards and their chips, changes as system operations change. Because you run simulations at fixed points in time and accurate simulations are time-consuming, the only way to guarantee EMC compliance is to measure EM emission on a board prototype spanning system operation over many modes. Mike Radhanauth, product-marketing manager, and Rick Morgan, manager of military, aerospace, and commercial hardware, of Spectrum Signal Processing (Burnaby, BC, Canada) note that EMI requirements vary widely among designs. Spectrum is a DSP-systems provider, offering VXI boards and other high-performance-board products. Spectrum's products must comply with US-government EMC standards, such as MIL STD 461 and FCC Class A and Class B. Compliance requires those products to meet both conductive (through wiring) and radiative (through air) radiation and susceptibility specifications. Morgan controls conductive emissions with good board-design practices. The techniques he uses, including minimizing transients on the power supply and proper I/O-connector shielding, also help reduce radiated emissions. Minimum board traces; board self-shielding using inner and, where feasible, outer ground planes; and board shrouding using a surrounding metal shield also reduce EM emission. Because of the high cost, the long runtimes, and the difficulty of modeling boards of even moderate complexity, Morgan feels that current EDA tools are insufficient to economically simulate his pc-board designs. His solution: Design well and thoroughly test a physical prototype. Peter Wicher, product-marketing manager of mobile graphics for S3 (Santa Clara, CA), has the same opinion about chips designed with EMC in mind that many other chip designers have. Wicher designs graphics-controller chips for notebook PCs. High-end displays require very fast data rates. For example, a 1280×1024 display refreshed at 75 Hz needs a graphics data rate of 135 MHz. Wicher's concerns about EMI are not just for self-contained notebook systems, but also for notebooks connected to a desk display that are near other electronic equipment potentially susceptible to EMI. Wicher also forgoes EM-analysis tools for good engineering techniques. Rather than performing EMI-chip analysis, he analyzes the major sources of EMI, such as the digital connection from a graphics controller to a flat-panel display, and employs design methods to minimize the EM emission. For example, by employing a high-speed serial-data link in place of a parallel link between the controller and display, Wicher can reduce EMI by a factor of two. Of course, engineering techniques such as serial-data links require special chip circuitry, such as low-voltage differential-signal interfaces. How you employ EMC-design tools to complement your high-speed pc-board and board-based-system design skills depends on your design's complexity and the degree of EMC compliance to which you are designing. Just remember a few points. First, the tools for EM-emission analysis vary greatly in cost and capability from vendor to vendor. Next, the earlier in a design you use an EMC-analysis or advisory tool, the easier it is to correct any EMI problems the tool may flag. It is cheaper and less time-consuming to find problems during board layout than it is with a physical prototype. Finally, tools are imperfect EMC-compliance predictors. There still is no substitute for accurate EM-emission testing of a final-product prototype.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| EDN Access | Feedback | Table of Contents | |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| Copyright © 1997 EDN Magazine, EDN Access. EDN is a registered trademark of Reed Properties Inc, used under license. EDN is published by Cahners Publishing Company, a unit of Reed Elsevier Inc. | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||