Since EMI may seriously hinder the design progress in the later stage and waste a lot of time and money, it is necessary to consider EMI at the beginning of design. Switching mode power supply (SMPS) is one of the most widely used circuits in modern technology. In most applications, SMPS can provide higher efficiency than linear regulator. However, there is a price for this efficiency improvement, because the switching of power metal oxide semiconductor field effect transistor (MOSFET) in SMPS will produce a lot of EMI, which will affect the reliability of the circuit. EMI mainly comes from the discontinuous input current, the fast slew rate at the switch node and the additional ringing at the switch edge caused by the parasitic inductance in the power supply loop.
Figure 1 takes buck converter topology as an example to illustrate the influence of various factors in different frequency bands. With the increasing design pressure, the EMI problem becomes more serious by increasing the switching frequency to reduce the size and cost, and by increasing the slew rate to improve the efficiency. Therefore, it is necessary to adopt EMI mitigation technology that does not affect the power supply design, but is cost-effective and easy to integrate.
Figure 1: an example of EMI source in SMPS.
What is EMI?
In the system requiring electromagnetic compatibility (EMC), the design should reduce the interference of interference source components and the sensitivity of vulnerable components. When terminal equipment manufacturers integrate components from different suppliers, the only way to ensure that the interfering components and the vulnerable circuits can not affect each other is to establish a set of common rules, in which the interference of the interfering components is limited to a certain range, so that the vulnerable circuits can reduce the impact within this range.
These rules are based on Industry General specifications such as CISPR25 for the automotive industry and cispr32 for multimedia devices. CISPR standard determines the final performance of any EMI mitigation technology, so it is very important for EMI design. Since SMPS is a typical source of electromagnetic interference, this white paper focuses on how to reduce such interference.
In addition to understanding the appropriate standards for a given application, it is also important to understand how to measure EMI, which will help you gain insight into how to reduce EMI. EMI measurement is usually divided into conducted EMI measurement and radiated EMI measurement. As the name suggests, it also explains the measurement method and generation mechanism of EMI. Although conducted emission is usually related to lower frequency (> 30MHz) and radiated emission is usually related to higher frequency (> 30MHz), the difference between the two is not so simple, because the conducted frequency range and radiated frequency range overlap.
Conducted emission measurements are designed to quantify the EMI generated from the device and returned to its power supply. For many applications, reducing these emissions is critical because many other sensitive circuits are usually connected to the same power line. In modern vehicles, the number of long wire bundles is increasing, so it is particularly important to reduce the conducted EMI of long wire bundles.
Figure 2 shows the general test setup for conducted emissions, including power supply, line impedance stabilization network (LISN), EMI receiver, power supply line, and device under test (DUT). LISN plays a key role as a low-pass filter to ensure the repeatability and comparability of EMI measurement and provide accurate impedance for DUT. Figure 2 also illustrates the subdivision of conducted emission into common mode (CM) current and differential mode (DM) current. DM current flowing between the power line and its return path is the main factor in the lower frequency range. CM current flows between each power line and ground, which is the main factor in the higher frequency range.
Figure 2: General test setup for conducted emission measurements with DM and cm loops highlighted in cyan and red, respectively.
The setup of radiated EMI measurement is similar to that of conducted EMI measurement. The main difference is that the EMI receiver of the former is not directly connected to the LISN, but connected to the nearby antenna. The radiation energy in SMPS comes from the fast transient current loop which generates magnetic field and the fast transient voltage surface which generates electric field. Because the current loop that generates the radiated magnetic field also generates DM conducted emission, and the voltage surface that generates the radiated electric field also generates cm conducted emission, many EMI mitigation technologies can reduce conducted emission and radiated emission, but they may be specific to one of them.
Usually, large passive filters are used to mitigate lower frequency transmission, which will increase the circuit board area and cost of the solution. High frequency emission faces different challenges in measurement, modeling and mitigation, which is mainly due to its parasitic nature. Common high frequency emission mitigation techniques include controlling the shimmy rate and reducing the parasitic effect. Figure 3 summarizes the mitigation technologies included in this white paper, the frequency bands for which these technologies apply, and examples of frequency ranges in the CISPR25 standard.
Figure 3: summary of EMI mitigation technologies.
Conventional methods to reduce EMI
When other systems share common physical contacts, the input voltage ripple caused by discontinuous current in SMPS may be transmitted to these systems. Without appropriate mitigation measures, excessive input or output voltage ripple may affect the operation of power supply, load or adjacent system. In the past, you could use an EMI filter based on a passive inductor capacitor (LC) to significantly reduce the input ripple, as shown in Figure 4. LC filters provide the attenuation necessary to meet EMI specifications. The cost is to increase the size and cost of the system (depending on the attenuation required), which will reduce the total power density. In addition, the large inductor used in the design of input EMI filter can not attenuate in the frequency range higher than 30MHz due to its low self resonant frequency, so it needs additional components such as ferrite beads to deal with the high frequency attenuation.
Figure 4: typical LC based passive filter for EMI reduction, and implemented attenuation.
Another traditional way to alleviate EMI is to use spread spectrum (or clock jitter) to modulate the switching frequency of SMPS, which will reduce the spectrum peak related to the basic switching frequency and its harmonics, but at the cost of increasing the background noise, as shown in Figure 5.
Figure 5: examples of SMPS spectrum with and without spread spectrum technology.
Spread spectrum is an attractive technology because it is easy to implement and you can use it in combination with other EMI reduction methods. But this technology is not a panacea, because it can only relatively reduce the existing EMI, and according to its characteristics, its performance will be reduced when the switching frequency is low. In addition, you can usually only apply spread spectrum to a single band, for the reasons explained in the next section.
In order to reduce the size of filter inductor to a greater extent, you can choose higher switching frequency for SMPS design. However, for switcher operation, sensitive bands need to be avoided. For example, in the past, the recommended switching frequency of automotive power solutions has been below am (about 400kHz). By choosing a higher switching frequency to significantly reduce the inductor size, you have to avoid the entire AM band (525khz to 1705khz) so that basic switching spurious does not occur in the more stringent automotive EMI band.
The switching frequency of Ti switch converter is higher than 1.8mhz, which can meet the requirements of EMI band. In order to reduce the switching loss and improve the switching frequency, the requirements for the rise and fall time of switching are more strict. However, the switching nodes with very short rise and fall times can maintain high energy even at high frequencies close to the 100th harmonic (as shown in Figure 6), which once again highlights the trade-off between high efficiency and low EMI.
Figure 6: EMI diagram of square wave with different rise time.
Due to the parasitic inductance in the power supply path of DC / DC converter, the high voltage swing rate can also cause high frequency switching node ringing, which further increases the ringing frequency and the emission at higher frequency. Figure 7 shows how yaw rate and switch node related ringing affect the transmission. The traditional way to limit the EMI emission caused by switching is to reduce the speed of EMI emission by deliberately adding a resistor in the gate drive path of the switching device. This causes the transition to occur more slowly, resulting in a faster roll down of the emission and a drop of 8 to 10 dB at the ring frequency. However, the slowing down of the switch edge will reduce the peak current efficiency by 2% to 3%.
Figure 7: effects of different switching node runout rates and associated ringing on high frequency emissions. Lower slew rate will affect EMI roll off in 30MHz to 200MHz frequency band, while lower ringing rate will affect EMI at 400MHz frequency band.
Innovative technology to reduce low frequency emission
Let’s take a look at some of the technologies Ti uses to build its converters and controllers, which strike a fundamental balance between efficiency, EMI, size, and cost.
Spread spectrum technology uses the principle of energy conservation to reduce the EMI peak by dispersing the energy in multiple frequencies. However, the peak energy of the sensitive circuit may not decrease; It depends on the relationship between the bandwidth of the sensitive circuit and the frequency modulation mode. When measuring EMI, spectrum analyzer belongs to sensitive circuit, and industry standard specifies resolution bandwidth (RBW). Therefore, it is very important to modulate the frequency according to the actual standard in a more effective way. The general experience is that the modulation frequency FM is approximately equal to the target RBW and the bandwidth is expanded Δ FC is about ± 5% to ± 10%。 Figure 8 illustrates these parameters in the time and frequency domains.
Figure 8: FM sum of spread spectrum parameters in time domain and frequency domain Δ fC。
In CISPR25 and other standards, FM is usually set at about 9KHz to optimize the low frequency band, which is just within the audible range. To solve this problem, you can further implement triangular modulation in a pseudo-random manner to propagate audible energy without significant impact on conducted and radiated EMI performance.
Figure 9 illustrates the modulation curve in time domain and frequency domain, which is a characteristic of the synchronous buck / boost converter tps55165-q1.
Figure 9: at the end of each modulation cycle, the audible noise is reduced by pseudo randomly modulating the triangular wave.
EMI is not limited to a single band (so it is not limited to a single RBW), but exists in multiple bands, which brings a dilemma, because spread spectrum is usually only used to improve a single band. A digital spread spectrum technology called double random spread spectrum (DRSS) brings a new solution to this problem. The basic principle of DRSS is to superimpose two modulation curves, each for different RBW.
Figure 10 shows the DRSS modulation curve in the time domain, where the triangular envelope is for the lower RBW and the superimposed pseudo-random sequence is for the higher RBW.
Figure 10: time domain modulation curve of DRSS.
Spread spectrum technology is suitable for both non isolated and isolated topologies, because their EMI sources are similar, spread spectrum can provide the same advantages.
Active EMI filter
In order to significantly improve the transmission in low frequency spectrum, lm25149-q1 buck controller adopts active EMI filtering method. The integrated active EMI filter can reduce the DM conducted emission at the input by acting as an effective low impedance shunt. Figure 11 shows how the active EMI filter of the buck controller is connected to the input line. The sensing and injection pins are connected to the input through their respective capacitors. The active components in the active EMI filter block will amplify the induced signal, and inject the appropriate reverse polarity signal through the injection capacitor to significantly reduce the overall interference on the input line. This reduces the filtering burden of the required passive components, thus reducing the size, volume and cost of these components.
Figure 11: active EMI filter showing induction and injection capacitors and components for compensation.
Figure 12 shows the EMI measurement results of a buck converter operating at 400kHz switching frequency, in which the active and passive EMI filtering methods are compared. In order to effectively meet the requirements of CISPR25 spectrum shielding, a 3.3 GHz passive EMI filter is needed μ H DM inductor and a 10 μ F DM capacitor. The active power filter method is implemented by a 1 μ The same effective attenuation can be achieved with the DM inductor of 100nF and the induction and injection capacitor of 100nF. This helps to reduce the size and volume of the passive filter to about 43% and 27% of the original value, respectively. For the high current converter, the cost can be further reduced and the efficiency can be improved by reducing the DC resistance of the inductor.
Figure 12: EMI attenuation using passive and active filters for 12V input and 5V / 5A output buck converters, and comparison of the passive inductors for filtering in the two methods.
Elimination of winding
Different from the non isolated converter, the additional transmission path across the isolation boundary is the main reason for the common mode (CM) EMI of the isolated converter. Figure 13 shows the parasitic capacitance of the isolation transformer in a standard flyback converter. CM current can flow directly from the primary side to the ground through the parasitic capacitance associated with each switch node. The CM current also flows from the primary side to the secondary side due to the parasitic capacitance between the windings, resulting in an increase in the measured CM EMI. In general, you can reduce this additional interference by using a larger cm choke in the input power path.
Figure 13: CM EMI with parasitic effect in flyback converter.
In order to help reduce the size of passive filter devices to a greater extent, the 65W active clamp flyback reference design for high power density 5V to 20V AC / DC adapter using Silicon FET adopts winding elimination and shielding method for isolated converter. As shown in Figure 14, an improved internal transformer structure inserts an additional auxiliary winding layer (shown in black) between the internal primary and secondary layers to achieve cm balance. The auxiliary cm balance layer shields the interface between the inner semi primary layer and the secondary layer, which helps to generate cm elimination voltage to eliminate cm injection from the outer semi primary layer. By balancing the parasitic capacitance from the auxiliary winding and the primary outer layer to the secondary layer, the CM current injected from the outer semi primary layer to the secondary layer can be eliminated (by injecting the inverse CM current from the elimination layer). The net effect (the CM current flowing into the secondary layer is almost zero) reduces the CM emission, so that the design can meet the requirements of EMI spectrum standard with very few cm filters.
Figure 14: using shielding and elimination windings to reduce EMI in flyback converters.
Innovative technology for reducing high frequency emission
So far, the EMI mitigation technology we have introduced can usually reduce the low frequency emission (30MHz), and correspondingly reduce the amount of passive filtering required, as well as the related size, volume and cost. Now let’s take a look at technologies designed to mitigate high frequency emissions (> 30MHz).
HotRod ™ encapsulation
One of the main methods to reduce the high frequency emission is to reduce the inductance of the power loop to a greater extent. Hotrod packages flip the silicon wafer and place it directly on the lead frame, thus greatly reducing the parasitic inductance caused by the bonding wires on the pins running the switching current. Figure 15 shows the structure and advantages of the hotrod package. In addition to improving the inductance of the power loop, the hotrod package also helps to reduce the resistance in the power path, thereby improving efficiency and reducing the size of the solution.
Figure 15: Standard QFN, with bonding wire, can be electrically connected to the bare wafer (a); Hotrod package with copper post and flip chip interconnect between lead frame and die (b).
Another advantage of using hotrod package devices is that they are easy to implement parallel input path pin arrangement (input capacitor layout of DC / DC converter). By optimizing the pin arrangement of the DC / DC converter, the layout of the input capacitor is symmetrical, and the reverse magnetic field generated by the input power loop will be in the symmetrical loop, thus reducing the emission to the nearby system to a greater extent. Parallel input paths can further reduce high frequency EMI, especially in the more stringent FM band.
Enhanced hotrod ™ QFN
Enhanced hotrod quad flat leadless (QFN) package can provide all the EMI reduction functions of hotrod package, and has switch function
The additional advantage of lower node capacitance is to reduce ringing to a greater extent. Compared with hotrod package, the input voltage (VIN) and ground (GND) pins of devices with enhanced hotrod QFN package are lower.
Integrated input bypass capacitor
As mentioned earlier, a larger input power loop results in higher emissions in the high frequency band due to higher switch node ringing. The integration of high-frequency input decoupling capacitor in the device package helps to reduce the parasitic effect of the input loop to a greater extent, thus reducing EMI.
Effective control of yaw rate
Despite the above technologies, high frequency EMI (60MHz to 250MHz) may still exceed the specified standard limits in some designs. One way to mitigate and improve margin to meet industry standards is to use a bootstrap of a resistor to switch converter
The capacitors are connected in series. The use of resistors can reduce the swing rate of the switch edge, thus reducing EMI, but also reducing efficiency.
The rapid development of electronic products has brought great pressure to the design of power converters. Complex systems need to be installed in smaller and smaller space of power converters. Each sensitive system is close to each other, so it is difficult to suppress EMI. Special care must be taken in the design of power converters to comply with the limits specified by the standard bodies to ensure that critical systems can operate safely in a noisy environment.
Editor in charge: PJ