Power Supply Meets EMC

Feb 5, 2019
In this whitepaper we give a short introduction to basic EMC knowledge and which EMC considerations are key in designing power supplies. The shown EMC tests can trace the most common failure causes.

1. Introduction

The main purpose of this document is to give a short introduction to the basic knowledge of Electro-Magnetic Compatibility EMC requirements and connect this knowledge with the topic of power supply. The first part is concerned with the principles as well as the definition of the topic of EMC. Additionally, the main test procedures that are far more than radiated and conducted emission concerning EMC are listed and explained. Therefore, this part should provide an awareness of the vast variety of EMC test procedures.

The second part explains more about coupling mechanisms, which are usually included in every EMC presentation. The information in this section can be applied to all EMC issues.

When talking about power supplies and EMC, it is inevitable to mention protective devices that are necessary against transient disturbances. Therefore, the third part presents these protective devices with special consideration when placing them. The fourth part is concerned with the EMC-focused designs of three different DC/DC converters. By explaining the single measures, a connection to the second part can be made.

The last part of this paper covers information gathered from experience about how to find the noise source as well as how to start with the design keeping EMC in mind. Therefore, it contains not only the information on how to handle a failed design but also what should be done before starting the design in order to archive a first time pass.

2. Principles

Every electrical device emits and receives electromagnetic disturbances. Therefore, usually, every device is both source and sink as depicted in Figure 1. However, due to the performance of these devices, some of them are more likely to be a significant source, and some are prone to be a sink. A pacemaker, for example, is more likely to be a jamming sink than a source.

There are two ways to emit or receive disturbances. The first one is through free space; no medium is necessary. Mobile phones use this phenomenon to operate. They emit encoded electromagnetic waves and also receive and decode electromagnetic waves to send and receive information. However, electromagnetic waves are not the only ones that take this path; electric and magnetic fields can also spread and couple into a device.

The second way a disturbance can take is over a wire, a metal sheet, or basically anything that is conductive. Just as with radiated emissions, several types of disturbances may occur.


Fig. 1: Principles
Besides the frequency-dependent disturbances, we also test (e.g., according to EN 55022) transient disturbances such as surge, burst, and Electro-Static Discharge ESD that can be very critical.

3. EMC Directive

The definition of EMC according to the current European EMC directive 2014/30/EU requires that a device must be able to function satisfactorily in its electromagnetic environment without introducing intolerable electromagnetic disturbances to other equipment in its environment or being adversely affected by the electromagnetic interference EMI that they generate. The directive refers to both radiated emissions and susceptibility (EM fields) and conducted emissions and susceptibility (interference along the cables). Additionally, this directive requires that emission tests are conducted at the mode that probably emits the most disturbances, and immunity tests must be conducted at a mode where the device is most sensitive to external disturbances.

4. Typical EMC Tests

When considering EMC tests or EMC issues, radiated or conducted emissions often come to mind; however, there are several additional topics to be considered. Nevertheless, experience has shown that conducted and radiated emission tests are mainly the most critical tests the engineers must face.

Conducted emission

The measurement of the disturbing voltage in the lower frequency range between 9kHz and 30MHz on the lines going to or leaving the device is called conducted emission. The purpose of this measurement is to avoid interferences between several devices connected over cables or other common conductive paths. If the emission of the tested device remains below the limit, it probably will not cause any disturbances to other equipment.

Radiated emission

The measurement of the disturbing electrical field above 30MHz is called radiated emission. Similar to the conducted emission measurement, also this measurement is usually set up according to the Comité international spécial des perturbations radioélectriques (CISPR) standard number 16. Based on the particular CISPR standard, the limits are applied. – for example, CISPR 11 covers industrial, scientific and medical equipment, while CISPR14 covers household appliances and electric tools.

This measurement determines whether or not the emitted electromagnetic waves of the tested device will probably disturb any other nearby electrical equipment.

ESD

Another type of critical test covers transients, for example, ESD. For the ESD test, a pulse with a specific shape is applied to conductive parts of the device, conductive metal parts nearby with a specific distance and set-up, or non-conductive but touchable parts of the device. These pulses usually cause efields and high currents over a very short period of time in the systems that might lead to an interrupt, a shutdown, a reset, or any other unwanted behaviour. Therefore, the device should be designed to be robust against these electromagnetic phenomena. Depending on the typical environment in which the device is used, different test levels are valid if not mentioned otherwise in a particular standard.

The ESD pulse is quite high (up to +/–15kV) and very fast (ns range). Due to the very fast pulse applied directly and indirectly to the device, this test is one of the most critical EMC tests.

Surge

The surge test applies a pulse with a specific shape over galvanic coupling onto the lines of the tested device. The purpose is to test the robustness of the device against indirect lightning strikes. Unless otherwise stated, the test levels must be applied according to the environment in which the device is used. The pulse of the surge test is not very high (up to +/–4kV) and very slow (µs range), which leads to a high amount of dissipated energy. This high energy is the critical part of this test, which easily leads to a damaged device.

Burst

The burst test has a property of the surge and ESD tests. The pulse itself is very fast (ns range) like the ESD pulse but is much lower (about maximum of +/–4kV) like the surge pulse. However, this pulse is repeated over a fixed period at a specific interval.

The burst test simulates the brush sparking of a motor. Even though the pulse is fast with a rise time in ns, since the coupling is capacitive, the device does not have to drain the current directly; this is the least harmful of the transient tests.

Harmonics

The harmonics test is one of the most important tests that we have to cope with as a power supply distributor. This measurement indicates if there is a harmonic distortion superimposed onto the fundamental frequency of 50Hz.

Flicker

A device consuming very high energy for short periods can cause voltage dips that may cause problems to other devices connected to the same main power line. If this happens too often or too intense, or the influence may be too high, the device will not pass the flicker test. Therefore, this test may be critical for devices switching high energy in short intervals. The value for the pass or fail criteria was determined empirically and is dependent on the magnitude of the drop and the frequency of the event. The voltage drop caused by the device is allowed to occur many times per second if the drop is small, and if the magnitude of the drop is large, it is only allowed to happen at a very low frequency.

E-field immunity

The e-field immunity test is complementary to the radiated emission test. Here, the device is tested if it is immune against external electrical fields that can be produced by other electrical equipment. The test is typically performed in a frequency range from 80MHz to several GHz, dependent on the applicable standard. The applied field is an amplitude modulated signal with 80% modulation at 1kHz.

HF-induced disturbances

Here, high-frequency disturbances are induced to the lines connected to the device. The purpose of this test is to analyse the robustness of the device against external disturbances caused by other equipment connected to the same electrical system as the device. The HF-induced disturbance test is often considered an important complement to the conducted emission tests.

Voltage dips, variations, and interruptions

This test determines how an electrical device behaves when the main power supply voltage changes. There are three different parts of this test, each with different types of voltage variations.

The response criteria that specify how the device should behave in response to voltage fluctuations must be selected. For example, depending on the application of the device, it may be allowed that the device shuts off in response to a voltage fluctuation. Or it could be required that the device remains operational despite voltage fluctuations.

Voltage dips are applied to check the influence of an abrupt voltage drop on the device.

Magnetic power field

The magnetic power field test simulates a homogenous magnetic field with the frequency of the main power supply of 50Hz/60Hz. When placed in a magnetic field typically produced by a Helmholtz coil, the device has to work properly. Devices like e.g. cathode tube monitors were highly affected by this test.

5. Types of coupling

The following figure shows a model where Rk, Ck, and Zk represent parasitic elements responsible for coupling in a circuit. Galvanic coupling is depicted by the resistor Rk. Parasitic capacitive coupling is represented by the parasitic capacitance Ck. Parasitic inductive coupling is represented by the parasitic elements Zk.

The only coupling mechanism not pictured in Figure 2 is the airborne coupling since it cannot be sketched through parasitic elements.


Fig. 2: Coupling Mechanism

Galvanic coupling

Figure 3 shows a circuit branch with a supplementary noise signal represented by a resistive value. The noisy current in the circuit creates a noisy voltage on the parasitic resistive trace, which affects the voltage on the resistive load. In this case, an ideal source can become a noisy source due to the addition of noise on a common branch.



Measures to reduce coupling interference


Fig. 3: Galvanic Coupling
This type of coupling can be reduced by decreasing the impedance and therefore the common resistive path of the two circuits since the effect on the voltage on the resistive load is directly proportional to the impedance of the common trace.

If the location and source of the noise are known, the noise current can be reduced. This can be accomplished by reducing the noise directly at the source, by using a filter so that the noise is reduced on its path to the jamming sink, or by reducing the frequency of the noise. The effect of the impedance of the common trace is directly proportional to the frequency of the noise since impedance is a frequency­dependent value. The frequency in a system should be as fast as necessary and as slow as possible

To avoid a common trace, a star-point topology can be used for several circuits in order to separate the noisy part from the load circuit. The ground systems are connected at one single point.

To avoid uncontrollable coupling, the return path as well as the forward path should be considered when doing the layout. As the design usually consists of several forward and return paths, every single path should be prioritized and categorized as aggressive, sensitive, or indifferent traces.

The most aggressive loop has the highest di/dt, which means it should be as small as possible and have the highest priority among the aggressive traces.

Capacitive coupling

The parasitic element Ck in Figure 4 is present in circuitry between adjacent wires in parallel. Over this capacitance, which is usually very low HF signals may couple to other components and become problematic.



Measures


Fig. 4: Capacitive Coupling
Several measures may be made in order to reduce capacitive coupling:

  1. Short traces are one countermeasure to deal with capacitive coupling since the length of the trace is directly proportional to the coupling of the conductive parts. Traces in parallel over long distances are more prone to capacitive coupling that may become problematic for (for example) an analogue measurement line in parallel to a cock line.
  2. Additional capacitors create symmetry for highly sensitive signals so that the noise is coupled back early on before the noise can cause trouble.
  3. The capacitive coupling can be reduced by shielding with conductive material.
  4. The capacitive coupling can also be decreased by reducing the frequency.
  5. In case where cables are run over large distances, twisted pairs (possibly shielded) are very helpful in order to avoid unwanted coupling.

Inductive coupling

Live wires can be inductively coupled through magnetic fields. These fields can couple into nearby metal parts (which might be sensitive traces) and induce the current.



Measures


Fig. 5: Inductive Coupling
Several measures may be to consider in order to reduce inductive coupling:

  1. As with capacitive coupling, traces for different signals should not be designed in parallel since live wires in parallel induce a current in one another.
  2. Large loops in noisy circuitry generate large magnetic fields, so the most critical loops with the highest di/dt should be designed as small as possible. Additionally, highly sensitive loops should also be small since circuits with larger loops are more prone to disturbances.
  3. In order to design small loops, the designer must always consider the return path.
  4. Traces are less sensitive to frequencies when the loops (and therefore the impedances) are small.
  5. As the frequency is reduced, so is the inductive coupling. A design should use a frequency, which is as slow as possible and as fast as necessary.
  6. There is also a way to shield magnetic fields with non-conductive materials such as permalloy and µ­ metal, which can be found in RFID and NFC applications.
  7. Ferrite plates may be used on controllers to shield magnetic fields.

Airborne coupling

Radiated emission measurement is a form of airborne coupling. If the measured emission is very high, then there is a high level of coupling between the Equipment-Under-Test (EUT) and the antenna. Since the antenna should not be altered, the coupling can only be modified by shielding the device or reducing the noise level caused by the device. There are two different measurement types for radiated emission. Typically, radiated emissions are measured in µV/m, and normal metal shielding is sufficient; however, when measuring according to CISPR 11 Group 2 devices, the H field is measured in µA/m additionally.

Depending on the size of the device under test and the length of the cables, the emission usually is quite high λ/4 of the wavelength. In most cases, the cables work as an antenna while the source is on the board. The emission starts at about λ/10 of the wavelength.

Either shielded cables or filtering before the cabling can reduce the radiated emission. Regardless, every outgoing and incoming signal should be filtered.

Radiated emission also decreases at lower frequencies, which in turn lowers the airborne coupling of the device and the antenna.

6. Protective Elements

Protective circuitry is one of the most important subsystems for AC/DC converters. These elements must protect the device against surge, burst, and ESD.
There are several possibilities to protect the device:

  1. Spark air gaps, which can dissipate most of the additional energy
  2. Gas discharge tubes
  3. Varistors
  4. Series resistor
  5. MLV (Multilayer Varistor)
  6. TVS
  7. Filter
For protecting ICs with much smaller amplitudes, a suppressor-diode or Zener-diode is in most cases the right solution.

All these elements are important to consider since each device has its voltage and energy range where some or a combination of these elements should be used. For power supplies, normally, two or even three elements are used since the energy pulses can be very high.

An element that can convert higher energy also allows a significantly high voltage to pass; for example, the spark air group can remove the highest energy but will not able to reduce the voltage to a safe level. The gas discharge tube can remove less energy but will reduce the voltage lower than the spark air gap. The same holds for varistors, diodes, and so on.

This leads to the optimal placement shown in Figure 6. The current firstly passes ...

Want to read the whole Whitepaper?

Download / Login
Applications