EMI filtering attenuates unwanted high-frequency signals on power and signal lines while passing DC and the wanted low-frequency signals, so that equipment neither emits nor picks up conducted interference.
This introduction explains how interference couples through cables, where EMI filters should be placed, how insertion loss is defined and measured, and how filter configuration, source and load impedances and capacitor dielectrics determine real performance, for design engineers and component engineers selecting panel-mount, feedthrough and surface-mount filters.
The article is based on an introduction to EMI filtering by Knowles Precision Devices [1], extended with the insertion-loss equations, filter configurations and selection criteria that determine how a catalog filter performs in a real circuit. The scope is passive low-pass EMI filters for power and signal lines; RF band-pass filtering and mains-specific safety filters are mentioned only where they affect the selection.
Key Takeaways
- EMI filters are low-pass filters: they pass DC and low-frequency signals and divert or block high-frequency interference. Filters handle conducted interference, while shielding handles radiated interference; effective protection usually needs both.
- Insertion loss is the ratio, in decibels, of the load voltage without and with the filter. Catalog curves are measured in a 50 Ω system according to MIL-STD-220 or CISPR 17, and the attenuation in the application can be considerably higher or lower.
- An ideal single-element filter rolls off at 20 dB per decade, a two-element L-C filter at 40 dB per decade and a three-element π or T filter at 60 dB per decade.
- Capacitors work best against high source and load impedances and inductors against low impedances; the inductive element of an L-C filter faces the low-impedance side.
- Parasitic inductance limits high-frequency attenuation: above its self-resonant frequency a capacitor behaves inductively, which is why three-terminal feedthrough and panel-mount filters outperform standard two-terminal MLCCs.
- Class 2 ceramic capacitance falls with DC bias, temperature and ageing, and ferrite inductors lose inductance under load current; both reduce insertion loss in operation.
Why EMI Filtering Is Needed
The growing amount of electronic equipment increases the likelihood of interference between devices, while lower supply voltages and signal levels make circuits easier to disturb. Protection against electromagnetic interference (EMI) is therefore a basic design requirement rather than an add-on.
In automotive and medical applications, for example, false alarms or malfunctions caused by external interference are not acceptable. This has pushed EMI performance requirements down to the module and component level, so that filters and filtered components are specified and tested as part of the design rather than added after a failed system test.
Legislation requires both low emissions and adequate immunity. The EU EMC Directive 2014/30/EU requires that the electromagnetic disturbance generated by equipment does not exceed the level above which other equipment cannot operate as intended, and that equipment has a level of immunity that allows it to operate without unacceptable degradation [2]. In the United States, 47 CFR Part 15 sets limits for unintentional radiators, including conducted limits from 150 kHz to 30 MHz on the AC mains [3]. Product standards such as CISPR 32 for multimedia equipment and CISPR 25 for vehicle components, whose conducted voltage method covers 150 kHz to 108 MHz, define the actual limits and test set-ups [4].
Screening the enclosure or the cables is sufficient in many cases. Where it is not, low-pass filtering of the lines that enter or leave the equipment provides the additional protection. The following sections explain how interference couples into and out of equipment and the terminology used to specify effective filtering.
How Interference Couples Through Cabling
Cables are usually the weakest point of a shielded design. They connect the inside of the enclosure with the outside world and act as both receiving and transmitting antennas.
In an ideal situation, a Faraday cage would protect a piece of equipment by enclosing it completely in a conductive box. Most equipment, however, needs input and output connections for power or signal lines, and the cables that provide these connections can pick up interference and radiate it [1]. Figure 1 shows four coupling paths:
- interference enters the equipment directly through the cabling (conducted interference, path 1),
- radiated interference travels directly to the affected equipment (path 2),
- interference leaves the source through a cable and is then radiated from the cable to the affected equipment (path 3),
- interference radiated by the source is picked up by a cable entering the affected equipment (path 4).
Paths 3 and 4 show that conducted and radiated interference convert into each other along a cable. A filter at the point where the cable enters the enclosure interrupts paths 1, 3 and 4 at the boundary, while the enclosure itself blocks path 2.
Differential-Mode and Common-Mode Interference
Conducted interference appears in two forms. Differential-mode interference flows out on one conductor and returns on the other, in the same loop as the wanted signal or supply current. Common-mode interference flows in the same direction on all conductors and returns through ground, the chassis or stray capacitance. Shunt capacitors from line to line and series inductors in each line attenuate differential-mode noise, while capacitors from each line to ground and common-mode chokes attenuate common-mode noise. The safe combination of common mode chokes, Y capacitors and varistors in mains and data-line filters is described in EMC design fundamentals for varistors and common mode chokes.
Feedthrough and panel-mount filters in a grounded wall attenuate both modes on each line, because each line has its own capacitor to the enclosure. Separating the two modes matters mainly for board-level filters on cable pairs and for mains filters, where the capacitor values to protective earth are limited by safety requirements.
Filter Location
Where the filter sits relative to the shielding determines how much of its insertion loss is realized. The same filter can perform very differently in a bulkhead and on a PCB.
Panel-Mount Feedthrough Filters
To prevent interference from entering or leaving a piece of equipment, feedthrough EMI filters can be mounted in the wall of a shielded case, as shown in Figure 2. All incoming and outgoing cables then pass through the filters. Power and wanted signals pass through unaffected, whereas higher-frequency interference is removed. The screened case protects against radiated interference and the feedthrough filters protect against conducted interference; combined, they maintain the integrity of the shield [5].
For effective operation, the filter input and output must be screened from each other, which means there should ideally be no apertures in the panel around the filter [5]. Filtered connectors apply the same principle to multi-pin interfaces. The construction of the ceramic elements inside these filters is described in ceramic feedthrough filters explained.
Surface-Mount Filters
When there is no suitable bulkhead for mounting filters, surface-mount filters on the PCB are an effective option (Figure 3). In general, their insertion loss is reduced at higher frequencies unless additional screening measures are taken, because radiated interference can couple around the filter from input to output. Short tracks and connections, good grounding and placement close to the point where the line enters the board improve insertion loss performance [5].
Three-terminal surface-mount filters, shown in Figure 4, route the signal through the component and connect the capacitor to ground through separate, short ground electrodes. This feedthrough structure reduces the equivalent series inductance (ESL) to about 20 pH to 30 pH, compared with about 200 pH to 300 pH for standard two-terminal MLCCs and 80 pH to 100 pH for reverse-geometry capacitors, which extends useful attenuation to higher frequencies [6]. The ground connection to the PCB plane must be equally short, otherwise the via and trace inductance dominates.
Insertion Loss: Definition and Measurement
Insertion loss is the main performance figure of an EMI filter. Its definition is simple, but the test conditions behind a catalog curve must be understood before the value can be transferred to an application.
Insertion loss at a given frequency is the ratio of the voltages appearing across the line immediately beyond the point of insertion, before and after the filter is inserted. With a constant source, the load voltage is measured without the filter, VB, and with the filter, VC [7]:
The second form in Eq. (1) describes the equivalent constant-output method: E1 is the generator output voltage required with the component in the circuit and E2 the generator output voltage without it, for the same output voltage [1]. A network or spectrum analyzer usually keeps its source level constant and records the output-to-input ratio directly in decibels. An insertion loss of 20 dB means that the interference voltage at the load is reduced to one tenth, 40 dB to one hundredth.
Catalog insertion loss is normally measured in a 50 Ω system according to MIL-STD-220, whose current revision C specifies the measurement of passive low-pass EMI filters as a function of frequency, including the influence of temperature and DC bias [8], [7]. CISPR 17 specifies comparable methods for passive EMC filtering devices on power and signal lines from 9 kHz to several gigahertz, with procedures for unbiased and biased conditions [9]. Datasheets therefore often state both no-load and full-load insertion loss. The practical meaning of these curves is discussed in more detail in insertion loss and performance in EMI filtering.
Cut-off Frequency of a Capacitor Filter
The cut-off frequency is the frequency at which a filter starts to become effective, generally taken as the 3 dB point of the insertion loss curve. The simplest case, a shunt capacitor C between a source resistance RS and a load resistance RL, shows how strongly it depends on the circuit. The capacitor impedance is ZC = 1/(2πfC) [7], and the voltage divider gives Eq. (2):
The insertion loss reaches 3 dB at the cut-off frequency fc in Eq. (3), where the capacitor impedance equals the parallel combination of source and load resistance:
Above fc, the insertion loss of the ideal capacitor rises by 20 dB per decade. Below fc, the attenuation is small but not zero, so a wanted signal close to the cut-off frequency is already attenuated slightly; the margin between the highest wanted frequency and fc must account for this. A higher capacitance lowers the cut-off frequency, and vice versa.
Worked Example: Effect of Source and Load Impedance
A 1 nF feedthrough capacitor measured in a 50 Ω system (RS = RL = 50 Ω, parallel value 25 Ω) has fc = 6.4 MHz according to Eq. (3). Eq. (2) gives an ideal insertion loss of 5.4 dB at 10 MHz, 23.9 dB at 100 MHz and 43.9 dB at 1 GHz.
The same capacitor on a low-impedance supply line with RS = RL = 1 Ω (parallel value 0.5 Ω) has a cut-off frequency of 318 MHz and only 0.4 dB of insertion loss at 100 MHz. The published 50 Ω curve therefore overstates the performance by more than 20 dB in this application. This is why capacitors suit high-impedance circuits and inductors suit low-impedance circuits, and why real source and load impedances can increase or decrease the insertion loss relative to the catalog value [5], [7].
| Frequency | IL, 50 Ω / 50 Ω | IL, 1 Ω / 1 Ω |
|---|---|---|
| 10 MHz | 5.4 dB | 0.0 dB |
| 100 MHz | 23.9 dB | 0.4 dB |
| 1 GHz | 43.9 dB | 10.4 dB |
Self-Resonance and Parasitic Inductance
A real capacitor has an equivalent series resistance RS and an equivalent series inductance L in addition to its capacitance. The frequency at which the capacitive and inductive reactances are equal is the self-resonant frequency (SRF), Eq. (4); above it, the insertion loss decreases with increasing frequency [7]:
For the 1 nF capacitor in the example, a total loop inductance of 1 nH, including the component, vias and tracks, places the SRF at about 160 MHz, so the ideal values in Table 1 at 1 GHz are not reached with a two-terminal capacitor on a PCB. Feedthrough and three-terminal constructions reduce the inductance in the ground path, which is the main reason for their superior high-frequency insertion loss [6].
Filter Configurations and Impedance Matching
Adding elements makes the insertion loss curve steeper, but each configuration works best for a specific combination of source and load impedances. The choice of configuration is therefore made primarily on these impedances and secondarily on the attenuation required at given frequencies.
An ideal single-element filter, either a shunt capacitor or a series inductor, provides 20 dB per decade. Each additional ideal element adds a further 20 dB per decade, so an L-C filter reaches 40 dB per decade and a three-element π or T filter 60 dB per decade [5], [7]. In practice, the insertion loss curves do not exactly follow these predictions, and the datasheet curves should be used for realistic figures [5]. The general behavior of these networks is explained in passive low-pass filter fundamentals.
| Configuration | Elements | Ideal slope | Preferred source and load impedance |
|---|---|---|---|
| C | One feedthrough capacitor to ground | 20 dB/decade | High source and high load impedance |
| L-C | Series inductive element and shunt capacitor | 40 dB/decade | Low on one side, high on the other; the inductor faces the low impedance |
| π (C-L-C) | Two capacitors with an inductive element between them | 60 dB/decade | High source and high load impedance |
| T (L-C-L) | Two series inductive elements with one capacitor between them | 60 dB/decade | Low source and low load impedance |
| Multi-element (e.g. L-C-L-C-L) | More than three elements | Steeper than 60 dB/decade | Selected for very high attenuation close to the pass band |
The rule behind Table 2 is that capacitors should be placed closest to high impedances and inductors closest to low impedances [7]. A capacitor facing a low impedance, or an inductor facing a high impedance, adds little attenuation, as the worked example above shows. The inductive elements in feedthrough and surface-mount filters are often ferrite elements, whose behavior is described in RF inductors and filters.
An undamped L-C or π filter can show a resonance peak in its insertion loss curve when the actual source and load impedances differ from 50 Ω. Near this frequency the filter attenuates less than expected and may even amplify interference. Low-impedance power lines, such as the input of a switching converter, therefore require a damped filter design.
Capacitor Dielectric and Operating Conditions
Insertion loss is not a fixed property of a filter. It depends on the electrical configuration, the source and load impedances, the load current, the ceramic dielectric, the earthing impedance and the integrity of the shielding [5]. The component-related factors are covered here.
Ceramic Dielectric Class
The capacitance of a ceramic capacitor can change with temperature, applied voltage and age, and a lower capacitance gives a lower insertion loss [5]. C0G/NP0 (Class 1) dielectrics change by no more than 0 ±30 ppm/°C from −55 °C to +125 °C and show no ageing, but their relatively low dielectric constant limits the capacitance. X7R (Class 2) dielectrics allow far higher capacitance in the same size, with up to ±15 % change over the same temperature range without voltage applied and ageing below 2 % per time decade [5].
Class 2 capacitors also lose capacitance under DC bias, and this effect adds to the temperature change and ageing. If the effective capacitance falls to half its nominal value, the high-frequency insertion loss of a capacitor filter falls by 6 dB and its cut-off frequency doubles, according to Eq. (2) and Eq. (3). The voltage dependence is explained in DC bias characteristics of ceramic capacitors. Where the voltage coefficient is critical, dielectrics with a specified voltage characteristic, such as BX or BZ, are an alternative to standard X7R.
Load Current and Ferrite Saturation
For filters that include ferrite inductors, the insertion loss under load current can be lower than with no load, because the ferrite material saturates with current. The reduction depends on the current and the ferrite material; in extreme cases the ferrite becomes ineffective and the insertion loss approaches that of a C filter [5]. For power lines, the full-load insertion loss curve, not the no-load curve, is the relevant figure.
Working Voltage and Safety
The working voltage is the continuous operating voltage, which may apply across the entire operating temperature range [5]. Transients must be considered in addition to the steady-state voltage. Filters on AC mains, where a capacitor failure can create a fire or shock hazard, require safety-certified X and Y capacitors rather than general-purpose feedthrough parts; their classes and test requirements are covered in safety capacitors in power electronics.
X2Y Filters
X2Y components are integrated passive devices with very low self-inductance, used for filtering and decoupling. They contain two matched Y capacitors, A and B, in one ceramic body with a common ground electrode [10].
In the filtering configuration shown in Figure 5, the component is connected across two signal lines. Differential-mode noise is filtered to ground by the two Y capacitors, while common-mode noise is cancelled within the device; the line-to-line capacitance C2 equals half of the line-to-ground capacitance C1 [10]. The rated voltage applies from A or B to ground, and the rating from A to B is twice this value.
In the decoupling configuration in Figure 6, capacitors A and B are connected in parallel, which doubles the apparent capacitance while keeping the inductance very low. Short, multiple and opposing current paths reduce the device inductance, and mutual coupling between opposing-polarity vias reduces the mounted inductance further; a poor layout with long connections and wide via spacing loses much of this advantage [10]. Applications and limits of the technology are described in what an X2Y bypass capacitor is good for.
Common EMI Filtering Terms
The following terms are used throughout EMI filter datasheets and specifications. The definitions have been aligned with the explanations in the sections above.
| Term | Definition |
|---|---|
| Conducted interference | Interference transmitted along a conductor or cable. It is reduced by series impedance, shunt elements to ground or a combination of both. A feedthrough filter mounted in the wall of a shielded compartment removes conducted interference while maintaining the screening integrity, and it reduces both emissions and susceptibility. |
| Radiated interference | Interference transmitted through free space. Protection is provided by shielding, but unfiltered lines can act as antennas that radiate interference out of, or into, a shielded enclosure. |
| Conducted emissions | Interference generated by a piece of equipment and transmitted along its cables, measured against the conducted limits of the applicable standard. |
| Radiated emissions | Interference radiated by a piece of equipment and its cables, measured with an antenna against the radiated limits of the applicable standard. |
| Susceptibility (immunity) | The extent to which equipment is vulnerable to interference from other equipment or the environment. Immunity is the ability to operate without unacceptable degradation in the presence of such interference. |
| Differential mode | Interference current that flows out on one conductor and returns on another, in the same loop as the wanted signal. |
| Common mode | Interference current that flows in the same direction on all conductors and returns through ground, the chassis or stray capacitance. |
| Cut-off frequency (3 dB point) | The frequency at which the insertion loss reaches 3 dB. It decreases with higher capacitance or inductance and depends on the source and load impedances, Eq. (3). |
| Insertion loss | The ratio, in decibels, of the load voltage without and with the filter at a given frequency, Eq. (1). Catalog values are measured in a 50 Ω system. |
| Self-resonant frequency | The frequency at which the capacitive and inductive reactances of a component are equal, Eq. (4). Above it, a capacitor behaves inductively. |
| EMC | Electromagnetic compatibility: the ability of equipment to function in its electromagnetic environment without adversely affecting, or being affected by, other equipment. |
| EMI | Electromagnetic interference: natural and man-made electrical disturbances from DC to the gigahertz range and beyond, from sources such as radar transmitters, motors, digital clocks, lightning and electrostatic discharge. |
| ESD | Electrostatic discharge: a fast, high-voltage transient that can damage or upset electronic circuits. Immunity is tested according to IEC 61000-4-2 and its product-specific derivatives. |
| Low-pass filter | A filter that passes DC and low-frequency signals while attenuating high-frequency interference. |
| Panel-mount filter | A filter that passes a line through the wall of a shielded enclosure, with its input and output screened from each other by the panel. |
| Surface-mount filter | A filter for PCB mounting. It offers better filtering than a standard MLCC and saves board space compared with discrete elements, but its high-frequency performance is lower than that of panel-mount types unless additional shielding is used. |
| Working voltage | The continuous operating voltage, which may apply across the entire operating temperature range. |
| X2Y filter | An integrated passive component with two matched Y capacitors and very low self-inductance for balanced filtering and decoupling. |
ESD can damage components through excessive voltage spikes, and its immunity requirements are defined in IEC 61000-4-2 [11]. Ceramic filter capacitors also absorb ESD energy to some extent, but dedicated suppressors are often added on exposed interfaces.
EMI Filter Selection Checklist
The checklist summarizes the parameters to compare in datasheets and the questions to answer before a filter is released for production.
- Interference type: conducted or radiated, differential mode or common mode, and the frequency range in which the limit is exceeded.
- Mounting: panel-mount feedthrough or filtered connector where a shielded wall exists; three-terminal or other low-inductance surface-mount filters at the board edge otherwise.
- Configuration: C, L-C, π or T according to the source and load impedances, Table 2, with the inductive element toward the low impedance.
- Insertion loss: compare curves at the frequencies that matter, check whether they are no-load or full-load values, and correct for the real source and load impedances.
- Capacitance and dielectric: C0G/NP0 for stable values and signal lines, X7R or voltage-characterized dielectrics for higher capacitance; use the effective capacitance at DC bias, temperature and end of life.
- Ratings: working voltage, rated current, temperature range and, for mains connections, X and Y safety approvals.
- Signal integrity: the highest wanted frequency or data rate must lie well below the cut-off frequency, with the filter capacitance included in the timing budget.
- Installation: short ground connection, no apertures around panel-mount filters, filter placed where the line enters the enclosure or board.
- Catalog versus custom: standard parts cover most voltages and capacitances; custom configurations, filtered connectors and hermetic parts are needed for unusual impedances, space, military or medical requirements.
- Verification: confirm conducted and radiated emissions and immunity on the finished equipment according to the applicable product standard.
Conclusion
EMI filtering complements shielding: the enclosure blocks radiated interference, and low-pass filters at the points where cables enter or leave remove conducted interference. Panel-mount feedthrough filters in a continuous wall give the highest performance, while surface-mount filters need short, low-inductance ground connections and placement at the board edge to approach it.
Catalog insertion loss is a 50 Ω reference value. The attenuation achieved in the application depends on the filter configuration relative to the real source and load impedances, on the effective capacitance of the ceramic dielectric under bias, temperature and ageing, on ferrite saturation under load current and on parasitic inductance above the self-resonant frequency. Selecting the configuration by impedance, using full-load curves and verifying the finished equipment against the applicable standard turns a catalog filter into a reliable EMC solution.
FAQ
An EMI filter is a low-pass filter that passes DC and wanted low-frequency signals while attenuating unwanted high-frequency interference on power and signal lines. It consists of shunt capacitors to ground, series inductive elements or both, arranged as C, L-C, π or T configurations. It reduces conducted emissions leaving equipment and improves immunity to conducted interference entering it, and it is usually combined with shielding.
Insertion loss is the ratio, expressed in decibels, of the voltage across the load without the filter to the voltage with the filter at a given frequency. A value of 20 dB means the interference voltage is reduced to one tenth. Catalog curves are measured in a 50 Ω system according to MIL-STD-220 or CISPR 17, often under no-load and full-load conditions.
Datasheet insertion loss is measured with 50 Ω source and load impedances. Real circuits can have impedances of a fraction of an ohm or several kilohms, which raises or lowers the attenuation. Capacitance loss with DC bias, temperature and ageing, ferrite saturation under load current, ground inductance and gaps in the shielding reduce the insertion loss further.
Panel-mount feedthrough filters sit in the wall of a shielded enclosure, so the panel screens the filter input from its output and the highest attenuation is achieved. Surface-mount filters are placed on the PCB and are easier to assemble, but radiated coupling around the filter and ground-path inductance reduce their performance at higher frequencies unless additional shielding is used.
The choice depends mainly on the source and load impedances. A C filter suits high impedance on both sides, a π filter also suits high impedances with steeper attenuation, and a T filter suits low impedance on both sides. An L-C filter suits a low impedance on one side and a high impedance on the other, with the inductive element facing the low impedance.
The cut-off frequency is the frequency at which the insertion loss reaches 3 dB. For a capacitor filter it is reached where the capacitor impedance equals the parallel combination of source and load resistance, so it falls with higher capacitance and rises with lower circuit impedance. Wanted signals should lie well below it, because attenuation begins gradually below the cut-off.
Every capacitor resonates with its series inductance, and above this self-resonant frequency it behaves like an inductor and its attenuation falls. Feedthrough and three-terminal capacitors route the current through the component and connect the capacitor to ground over short paths, which reduces the effective inductance by roughly an order of magnitude compared with a standard two-terminal MLCC.
C0G or NP0 dielectrics provide stable capacitance with almost no change over temperature, voltage or time, but only low capacitance values. X7R dielectrics give much higher capacitance in the same size but lose capacitance with temperature, DC bias and ageing, which reduces insertion loss. Designs should use the effective capacitance under operating conditions when calculating filter performance.
How to Select an EMI Filter
Step-by-step procedure for selecting a passive low-pass EMI filter for a power or signal line, from identifying the interference to verifying the installed filter.
- Identify the interference
Determine whether the problem is conducted or radiated, differential mode or common mode, and in which frequency range the emission or immunity limit of the applicable standard is exceeded.
- Decide on the filter location
Use panel-mount feedthrough filters or filtered connectors where a shielded wall exists, and low-inductance surface-mount filters placed where the line enters the board otherwise.
- Estimate source and load impedances
Estimate the impedance of the circuit on each side of the filter in the critical frequency range, because it determines both the configuration and the achievable attenuation.
- Select the filter configuration
Choose a C, L-C, π or T configuration so that capacitors face high impedances and inductive elements face low impedances, with enough elements for the required slope.
- Check insertion loss under operating conditions
Compare full-load insertion loss curves at the critical frequencies, correct for the real impedances, and use the effective capacitance at DC bias, temperature and end of life.
- Verify ratings and signal integrity
Check working voltage, current, temperature range and safety approvals, and confirm that the cut-off frequency lies well above the highest wanted frequency or data rate.
- Install and verify
Mount the filter with a short ground connection and no gaps in the shielding, then confirm emissions and immunity of the finished equipment according to the applicable standard.
Further Reading
- Feedthrough Capacitors Technology and Applications: construction and use of feedthrough capacitors in filtering.
- EMC Filters Design, from Component to Design: from individual filter components to a complete EMC filter design.
- AC Line Filters and Insertion Loss Explained: insertion loss applied to mains filters with common-mode and differential-mode stages.
- Input Filters for DC/DC Converters: a damped L-C filter on a low-impedance power line, including stability.
- Ferrite Beads Behaviour Explained and its Modelling: impedance and saturation of the ferrite elements used in many filters.
- ESD Protection by MLCC Ceramic Capacitors: how ceramic capacitors contribute to ESD robustness.
- Electromagnetic Housing Shielding and Its Effectiveness: the shielding side of the filter-plus-enclosure concept.
- Knowles Introduces Hermetic, Panel-Mount EMI Filters: an example of current panel-mount filter construction for harsh environments.
References
- P. Matthews, Introduction to EMI Filtering, Knowles Precision Devices blog, 2019.
- Directive 2014/30/EU on the harmonisation of the laws of the Member States relating to electromagnetic compatibility (recast), European Parliament and Council, 2014.
- 47 CFR Part 15, Radio Frequency Devices, U.S. Federal Communications Commission, Electronic Code of Federal Regulations.
- Basic design considerations and layout recommendations for EMC robustness, Infineon Technologies, application note.
- The Need for EMI Filters – EMI filter catalogue, technical introduction, Knowles Precision Devices (Syfer).
- How to use the YFF series 3-terminal feed-through filters, TDK, solution guide.
- Multilayer Ceramic EMI Filters – tutorial, Syfer Technology.
- MIL-STD-220C, Test Method Standard: Method of Insertion Loss Measurement, U.S. Department of Defense, 2009.
- CISPR 17:2011, Methods of measurement of the suppression characteristics of passive EMC filtering devices, IEC, edition 2.0, 2011.
- X2Y Filter and Decoupling Capacitors – product flyer, Johanson Dielectrics.
- IEC 61000-4-2:2008, Electromagnetic compatibility – Part 4-2: Electrostatic discharge immunity test, IEC, edition 2.0, 2008.
- Knowles Precision Devices – EPCI premium supplier page, Passive Components Blog.
Article revised and technically reviewed in November 2026: corrected the definitions of conducted interference, conducted emissions and cut-off frequency, added differential-mode and common-mode interference, insertion-loss and cut-off equations with a worked example, self-resonance, filter configurations versus source and load impedance, dielectric and load-current effects, a selection checklist, FAQ and How-to, converted the insertion-loss equation image to math, removed duplicated images and added referenced sources.
