EMC Design Principles and EMI/EMC Questions for PCB Engineers

Electromagnetic compatibility (EMC) is a system-level requirement that affects circuit architecture, component choice, PCB layout, and product qualification. A product must perform its intended function without emitting excessive interference and without being upset by interference from other equipment. This part reviews five common EMI/EMC design questions and the basic principles that govern component selection, filtering, shielding, grounding, and PCB-level design. The aim is to give designers, hardware engineers, and PCB engineers a practical framework for making EMC decisions early, when they are still inexpensive to change.

Highlights:

• EMC must be designed into the circuit, software, PCB, shielding, filtering, and grounding from the start; it cannot be tested in at the end.
• At high frequency, component parasitics such as lead inductance, parasitic capacitance, and saturation behavior often determine filter performance as much as nominal component values do.
• PCB designers need signal integrity, power integrity, EMC, manufacturing, and safety knowledge; insulation spacing requirements can be checked against standards such as GB4943 or UL60950.

EMI/EMC Design: 85 Classic Questions

The questions below address the purpose of EMC design, the main design entry points, the use of decibel notation, the knowledge required for EMC work, and the broader skill set expected of a PCB designer.

Q1: Why should a product be designed for electromagnetic compatibility?

A1: To meet product functional requirements, reduce debugging time, make the product satisfy electromagnetic compatibility standards, and prevent the product from causing electromagnetic interference to other equipment in the system.

Q2: From which aspects can electromagnetic compatibility design be carried out?

A2: Circuit design, including device selection; software design; PCB design; shielding structure; filtering of signal lines and power lines; and grounding method design.

Q3: Why is the decibel (dB) unit always used in the EMC field?

A3: Because the amplitude and frequency ranges to be described are both wide, logarithmic coordinates are easier to use in graphs. The dB is the unit used when values are represented logarithmically.

Q4: I have limited knowledge of EMC, but data transmission rates in circuit design are increasing. I have run into PCB EMC problems and want to study the subject seriously, not simply follow what others are learning. I believe EMC will become increasingly important in circuit design, but I do not know where to start. What basic knowledge and courses are needed to become proficient in EMC, and what learning path is recommended? I know no subject is easy to master and I do not expect to become an expert quickly; I only want advice that helps me avoid unnecessary detours.

A4: Start with EMC standards such as EN55022 (GB9254) and EN55024, and understand simple test principles. Then learn how to use EMI components, including capacitors, ferrite beads, differential-mode inductors, and common-mode inductors. At the PCB level, understand layout, stack-up, and high-speed routing effects on EMC, along with relevant design rules. Also develop a method for analyzing and solving EMC problems when they occur. These are basic skills that a hardware engineer must master.

Q5: I am new to PCB design. What knowledge should I focus on to become a good PCB designer? Where can I find safety-related knowledge for PCB design?

A5: For PCB design, you should master: (1) relevant PCB design software, such as PowerPCB or Cadence; (2) the specific architecture of the product being designed, together with schematic and circuit knowledge, including digital and analog fundamentals; (3) PCB fabrication flow, process, and manufacturability and maintainability requirements; (4) high-speed signal integrity, electromagnetic compatibility (EMI and EMS), and SI/PI simulation design; (5) RF knowledge if the work involves RF; and (6) safety regulations for PCB design, mainly GB4943 or UL60950. General insulation spacing requirements can be obtained by looking up tables.

These questions reflect the practical scope of EMC work: standards, component behavior, PCB layout, and troubleshooting. They also show that EMC is not a separate activity added after functional design. It is part of the same engineering process that determines whether a product works reliably in its electromagnetic environment.

Basic Principles of EMC Design

Electronic circuit designers often consider only product function and fail to evaluate function together with electromagnetic compatibility. As a result, the product may perform its intended function while also generating substantial functional interference and other interference, and it may fail to meet susceptibility requirements. EMC design of electronic circuits should consider the following aspects.

In component selection, the extent to which basic circuit components satisfy electromagnetic requirements often determines the extent to which the functional unit and the final equipment meet EMC requirements. The main criteria for selecting suitable electromagnetic components include out-of-band characteristics and circuit assembly technology. EMC performance is often determined by the response of components far from the fundamental frequency. In many cases, circuit assembly also determines out-of-band response, for example through lead length, and the degree of coupling between different circuit components.

The specific rules are:

1. At high frequency, prefer feed-through capacitors or standoff capacitors with low lead inductance over leaded capacitors for filtering.
2. When leaded capacitors must be used, consider the effect of lead inductance on filtering efficiency.
3. Aluminum electrolytic capacitors may undergo temporary dielectric breakdown for a few microseconds. In circuits with large ripple or transient voltage, use solid capacitors.
4. Use resistors with low parasitic inductance and low parasitic capacitance. Chip resistors can be used in the ultra-high-frequency range.
5. Large inductors have large parasitic capacitance. To improve insertion loss in the low-frequency portion, do not use a single-stage filter; use a multi-stage filter composed of several small inductors.
6. When using magnetic-core inductors, pay attention to saturation characteristics. In particular, high-level pulses can reduce the inductance of the magnetic-core inductor and the insertion loss in the filter circuit.
7. Use shielded relays wherever possible and ground the shield case.
8. Select input transformers with effective shielding and isolation.
9. Power transformers used for sensitive circuits should have electrostatic shielding, and both the shield case and the transformer case should be grounded.
10. Internal interconnect signal lines in equipment must use shielded cable to prevent interference coupling between them.
11. To ensure that each shield is connected to its own pin, select connectors with enough pins.

Together, these rules reduce parasitic coupling, preserve filter performance, and keep interference currents away from sensitive circuits. They are most effective when applied during component selection and PCB layout, before the design is fixed and before EMC testing becomes the only remaining verification step.

EMC, Signal Integrity, and Power Integrity Design: Practical Engineering Q&A

6. Problems with Inductive Sensors Driven by Square-Wave Pulses

Q6: What problems should be considered when an inductive sensor is driven by a square-wave pulse?

A6: During signal testing, work in a shielded environment whenever possible. If that is not practical, at least shield the sensor and the front-end stage. Use a differential probe where possible, or at minimum keep the probe ground lead as short as possible; this reduces measurement error. The circuit's actual operating frequency is not very high, so ringing can be reduced through layout. For better noise performance, consider common-mode rejection. Insert a common-mode choke when necessary, and pay attention to switching-power-supply noise in the operating environment; avoid power-supply coupling. If the sensor permits, use current-amplification mode, which improves speed and reduces noise. Place the analog switch after the preamplifier. This adds one preamplifier channel, but it improves performance substantially and reduces debugging difficulty. If the waveform is critical, add frequency compensation. If the application is only digital detection, lower the operating frequency. In general, use the lowest frequency that works and block DC whenever possible. Pay attention to the anti-aliasing filter before the ADC and to software filtering to improve data stability.

7. GPS Electromagnetic Interference in Handheld Navigation Products

Q7: In a handheld and in-car GPS terminal with MP4, MP3, FM radio, and GPS navigation functions, a built-in GPS antenna is close to the MCU, SDRAM, crystal, and other components. EMI/EMC coupling severely degrades satellite reception, to the point that normal positioning is almost impossible. What can be done to solve this EMI/EMC problem?

A7: Add an ESD filter at the antenna or related interface. Such a filter provides both ESD protection and electromagnetic-interference suppression. Mobile-phone customers with GPS functions use this approach. Several suppliers offer suitable parts, including TEK (Raychem), Jiabang, Korea's ICT, and many others.

8. Differential Pairs: Definitions, Frequency Response, and Conversion

Q8: Almost all important signal lines on the board are designed as differential pairs to improve immunity. Three points remain unclear: (1) Are differential signals defined only for analog signals, only for digital signals, or for both? (2) In a real schematic, how should the frequency response of a network on a differential pair, such as a filter, be analyzed? Is the method the same as for a general two-port network? (3) How are the differential signals carried on a pair converted into a general single-ended signal? What do the waveforms on the pair look like, and how are they related?

A8: (1) Differential signaling simply uses two signal lines to transmit one signal, and the decision is based on the voltage difference between the lines. It can be either an analog signal or a digital signal. Real signals are analog; a digital signal is only the sampled result of an analog signal after quantization by a threshold level. Differential signaling can therefore be defined for both digital and analog signals. (2) The frequency-response question is a good one. The actual differential port is a four-port network, and it has two analysis modes: differential mode and common mode. When analyzing the frequency response, add a common-mode sweep source with the same polarity to both lines and a differential-mode sweep source with opposite polarity. At the corresponding port, set a common-mode voltage test point Vcm = (V1 + V2)/2 and a differential-mode voltage test point Vdm = V1 - V2. Many articles on differential impedance calculation and principles are available and can be studied in detail. (3) A differential signal usually enters a differential driver circuit and is amplified to produce a differential signal. The simplest example is the differential common-emitter mirror amplifier circuit, which is covered in general analog-circuit textbooks. The figure shows a SPICE circuit and output waveforms for a differential amplifier device. Generally, the two outputs need to be completely inverted and have a sufficient voltage difference greater than the differential-mode threshold. Of course, the signal inevitably contains a common-mode component, so a very important specification of a differential amplifier is the common-mode rejection ratio, Kcmr = Adm/Acm.

9. Power-Supply Filtering for a DC Permanent-Magnet Motor Speed-Control Board

Q9: A speed-control circuit was designed for a DC permanent-magnet motor. At the power input, 0.33 uF + a Sharp TV inductor + 0.33 uF was not satisfactory. Four inductors were then connected in series at the PCB power input, but the design still exceeds the limit by 12 dB between 30 and 50 MHz. How should this be handled?

A9: Generally speaking, an LC or pi filter is more effective than a single capacitor filter or a single inductor filter. It is not clear what "not satisfactory" means in this case. Is it radiated emissions that exceed the limit? In which frequency band? I would guess that in the DC permanent-magnet motor supply loop, the feedback noise amplitude is large and the frequency is relatively low. A larger inductance is needed for filtering, and multi-stage capacitor filtering should be used at the same time; this should give better results.

10. EMC Considerations for a 0 to 150 MHz, 80 dB Broadband Amplifier

Q10: A broadband amplifier covering 0 to 150 MHz with gain not less than 80 dB is being considered. What EMC issues should be addressed?

A10: Answer 1: In broadband-amplifier design, low-noise performance deserves particular attention. For example, the power supply must be sufficiently stable. Answer 2: (1) Pay attention to input and output impedance matching, such as common-base input and emitter-follower output. (2) Address decoupling at each stage, including high-frequency and low-frequency ripple. (3) Use deep negative feedback and prevent self-oscillation and loop self-oscillation. (4) Take care with the band-pass filter design. Answer 3: This is difficult to answer without seeing the actual design. All recommendations remain the usual ones: pay attention to the three elements of EMC, to conducted and radiated paths, and to power distribution and ground bounce. 150 MHz is the analog signal bandwidth. How fast is the digital signal rise time? If the corner frequency is also below 150 MHz, conducted coupling and power-plane radiation should be the main considerations. First do a good job of power distribution, partitioning, and decoupling. 80 dB is a high gain. Isolate and protect the preamplifier small signal and its reference supply and ground, and minimize the power-supply impedance in that section.

11. EMC Design for Low-Power DC Permanent-Magnet Motors

Q11: What EMC methods and considerations apply to low-power DC permanent-magnet motor design? A 90 W DC permanent-magnet motor (110 to 120 V, 2000 rpm) has consistently failed EMC. After production, the 16-slot design was changed to 24 slots and shaft insulation was added, but the motor still did not meet the limit. A 125 W motor is now being designed. How should this be handled?

A11: EMC problems in DC permanent-magnet motor design mainly come from the back EMF generated during rotation and from arcing during commutation. For detailed analysis, RMxpert can be used to design and optimize the motor parameters, and Maxwell 2D can be used to simulate the actual EMI radiation.

12. Impedance Boundaries and Thin-Film Resistance in Simulation

Q12: Can an impedance boundary be used? Or a layered RLC impedance? Or a Designer circuit co-simulated with HFSS?

A12: Lumped resistance can be represented by an RLC boundary. If it is a thin-film resistance, it can be represented by surface impedance or by impedance editing.

13. ESD Failure of a 32 kHz Crystal in an Enclosure with Metal Decoration

Q13: An ESD test is being performed on a product with a ring of metal decorative parts on the enclosure. Contact discharge at 4 kV leaves the 32 kHz crystal working, but air discharge at 8 kV stops oscillation. How should this be handled?

A13: With metal present, air discharge and contact discharge produce similar effects. Try spraying insulating varnish on the metal bracket.

14. Simulating PCB Electromagnetic Radiation Instead of Near-Field Probing

Q14: Measuring PCB electromagnetic radiation is cumbersome. A spectrum analyzer and a homemade near-field probe are used. Apart from accuracy, points with large voltages are frightening because the spectrum analyzer may be damaged. Can simulation solve this problem?

A14: First, EMI testing includes near-field-probe testing and far-field radiation testing. No simulation tool can replace actual testing. Second, Ansoft's PCB board noise and radiation simulation tools SIwave and HFSS can simulate the near-field and far-field radiation of a board and of a system, as well as EMI radiation in a limited shielding environment. The validity of the simulation depends on how well the EMI problem in the design is considered and on the corresponding software settings. Examples include whether the board radiation is differential-mode or common-mode, and whether the radiation source is a current source or a voltage source. In our practice and experience, the vast majority of EMI problems can be solved through simulation analysis, and the correlation with actual measurements is very good.

15. Simulation Frequency Range, HyperLynx Capabilities, and Tool Interfaces

Q15: I heard that Ansoft EMC tools generally simulate at frequencies above 1 GHz. The highest clock on our board is only 133 MHz, from the main chip to the SDRAM; most other frequencies are in the kHz range. We mainly use HyperLynx for SI/PI design because it is simple to operate, but the whole-board EMC still exceeds the limit and affects image quality. In addition, do your tools have an interface to Mentor PADS?

A15: Ansoft tools can simulate signals from DC to tens of GHz and above. Relative to other tools, their lossy transmission-line models are more accurate above 1 GHz. As far as I know, HyperLynx mainly performs SI and crosstalk simulation, plus a small amount of EMI radiation analysis for a single signal line; at present it does not have a PI analysis function. Many factors affect board EMC. Solving signal integrity and crosstalk is only one aspect. Power-plane noise, decoupling strategy, shielding method, and current distribution path all affect EMC performance. These can be examined in Ansoft SIwave through simulation. In addition, Ansoft tools do have an interface to Mentor PADS.

16. Split Ground Planes vs. Ground-Plane Partitioning

Q16: When should the bottom layer be split to reduce interference, and when should ground-plane partitioning be used to reduce interference?

A16: I have not heard the term "splitting the bottom layer" used in this context. What does it mean? Could you give an example? Ground-plane partitioning is mainly used to improve the isolation between an interference source and a victim, such as isolation between analog and digital sections. Of course, partitioning also introduces signal-integrity problems such as crossing a split. Ansoft SIwave can conveniently check the isolation between any two points. There are other ways to improve isolation as well: layering, decoupling, and single-point connection are all options, and the effectiveness of a specific application can be simulated with software.

17. What Does a Capacitor Bridging Two Power Zones Actually Pass?

Q17: A capacitor is connected across two different power-copper zones to serve as a high-frequency signal return path. It is well known that a capacitor blocks DC and passes AC, and that the higher the frequency, the more freely current flows. My confusion is that the supply levels reaching the PCB have mostly been filtered to remove AC. What does the capacitor pass in this case? Is it an "AC signal"?

A17: Answer 1: This question is somewhat subtle, and I have not seen a completely convincing explanation. For AC, the ideal situation is that power and ground are "short-circuited," but in reality the impedance between them cannot truly be 0 ohms. The capacitance you mention should not be too large, so that the arrangement reflects the principle of "single-point grounding at low frequency and multi-point grounding at high frequency." This is roughly the value of the capacitor's presence. It is common to encounter a situation where two units, each with its own power supply, are connected and produce unexplained interference; a ceramic capacitor connected across the two supplies makes the interference disappear. Answer 2: This capacitor is used for voltage stabilization and EMI suppression, and it passes an AC signal. It is true that "the supply levels reaching the PCB have mostly been filtered to remove AC," but do not forget that digital circuits themselves generate AC signals and disturb the power supply. When many switching transistors act at the same time, the fluctuation imposed on the power supply is very large. In practice, however, this capacitor mainly plays an auxiliary role and is used to improve system performance. If the rest of the design is good, it can be omitted entirely. Answer 3: AC means changing. For a so-called DC level such as a power supply, wiring impedance means that when the load changes, the demand on the supply changes, either larger or smaller. In this case, the series wiring impedance produces a larger or smaller voltage drop. As a result, the DC supply carries an AC signal. The frequency of this signal is related to the frequency of the load change. The role of the capacitor is to store a certain amount of charge energy locally so that the energy needed for the change can be obtained directly from the capacitor. Approximately, the capacitor, which can now be regarded as a power source, and the load form an AC return loop. That is roughly the role of the capacitor.

18. Radiated-Emission Failure at 50 to 60 MHz in a Mobile Phone

Q18: A new mobile phone from the company failed one radiated-emission item during 3C certification. The frequency is 50 to 60 MHz, and the limit is exceeded by 5 dB. The charger is suspected. Several capacitors were added and nothing else was changed; the capacitors were 1 uF and 100 uF. What is a good solution if the charger cannot be changed and only the phone circuit can be modified? Would adding capacitors at the charger input on the phone board solve the problem?

A18: Answer 1: Make the large capacitor larger and the small capacitor smaller, and add a ferrite bead in series. However, the battery is not very likely to be the cause. Answer 2: Try shorting the case of the variable-frequency inductor to ground and shielding it.

19. Avoiding High-Frequency Interference in PCB Design

Q19: How can high-frequency interference be avoided in PCB design?

A19: The basic approach is to minimize interference from the electromagnetic fields of high-frequency signals, which is commonly called crosstalk. Increase the spacing between high-speed signals and analog signals, or add ground guard traces or shunt traces beside the analog signals. Also pay attention to noise from the digital ground interfering with the analog ground.

20. Resolving the Conflict Between High-Speed Routing and EMI

Q20: How can the conflict between high-speed routing and EMI be resolved in PCB design?

A20: Resistors, capacitors, or ferrite beads added for EMI must not cause the signal's electrical characteristics to violate specifications. Therefore, the best first step is to use routing and PCB stack-up techniques to solve or reduce EMI, such as routing high-speed signals on inner layers. Only as a final step should resistors, capacitors, or ferrite beads be used, so that damage to the signal is minimized.

21. Ground Connections Between PCBs in a Multi-Board System

Q21: When several PCBs form a system, how should the grounds between boards be connected?

A21: When signals or power transfer between boards, for example when board A sends power or a signal to board B, an equal current must return from the ground layer to board A. This is Kirchhoff's current law. The ground current will find the lowest-impedance path back. Therefore, at every interface where power or signals are connected, the number of ground pins assigned to the ground layer must not be too small, so as to reduce impedance and lower ground-layer noise. In addition, analyze the entire current loop, especially the high-current portion, and adjust the ground layer or ground wiring to control the current path. For example, create a low-impedance point at a certain location so that most of the current flows there, reducing the effect on other more sensitive signals.