EMI reduction of boost APFC based energy system

Indiana University - Purdue University Fort Wayne Opus: Research & Creativity at IPFW Engineering Faculty Presentations Department of Engineering 11-215 EMI reduction of boost APFC based energy system Istiaque Mauf Ahmad Indiana University - Purdue University Fort Wayne Ming Li Regal-Beloit A. Eroglu Indiana University - Purdue University Fort Wayne, eroglua@ipfw.edu C. Pomalaza-Ráez Indiana University - Purdue University Fort Wayne, raez@ipfw.edu Roger Becerra Regal-Beloit Follow this and additional works at: http://opus.ipfw.edu/engineer_facpres Part of the Engineering Commons Opus Citation Istiaque Mauf Ahmad, Ming Li, A. Eroglu, C. Pomalaza-Ráez, and Roger Becerra (215). EMI reduction of boost APFC based energy system. Proceedings of the 2th International Conference on Transformative Science and Engineering, Business and Social Innovation. 156-16.Presented at SDPS 215 (Society for Design and Process Science) Conference, Fort Worth, TX. http://opus.ipfw.edu/engineer_facpres/2 This Presentation is brought to you for free and open access by the Department of Engineering at Opus: Research & Creativity at IPFW. It has been accepted for inclusion in Engineering Faculty Presentations by an authorized administrator of Opus: Research & Creativity at IPFW. For more information, please contact admin@lib.ipfw.edu.

SDPS-215 Printed in the United States of America, November, 215 215 Society for Design and Process Science EMI CHARACTERIZATION OF BOOST APFC BASED ENERGY SYSTEM Istiaque Maruf Ahmad (1), Ming Li (2), Abdullah Eroglu (1), Carlos Pomalaza-Raez (1), Roger Becerra (2) (1) Indiana University- Purdue University Fort Wayne, (2) Regal Beloit Corporation {ahmaim1@ipfw.edu Ming.Li@RegalBeloit.com eroglua@ipfw.edu raez@ipfw.edu Roger.Becerra@RegalBeloit.com] ABSTRACT Electro-magnetic interference (EMI) in the boost active power factor correction (APFC) power supply circuit occurs due to common mode (CM) and differential mode (DM) noise. Performance of conducted EMI varies depending on different parasitics and EMI filter elements such as Y capacitor, CM choke and variation of the switching. Furthermore, load variation impacts the level of EMI in the circuit. This paper analyzes the EMI noise variation in boost APFC circuit due to the change of circuit when circuit parasitics, switching and load conditions are changed. The results of this work provide guidance for the EMI filter design of boost APFC power supply circuit to meet the strict electromagnetic compatibility (EMC) standard, such as Federal Communication Commission (FCC) class B. INTRODUCTION While designing EMI filter, designer has to meet EMI standards, which is always a challenge in high density and low cost power supply system design. Despite of this challenge, designer also has to take into account of the size and weight of the filter. Reduction of CM noise generation can greatly decrease the size of the filter and also reduce cost. CM noise current is mainly caused by displacement current within the inter-winding parasitic capacitances of the transformers and the parasitic capacitances between semiconductor switches and the ground (Majid, Saleem, Alam & Bertilsson, 213). Further enhancement in the filter profile can be made by increasing switching to decrease the size of the energy storage elements. This is important requirement for the high density power supply design. However, increasing switching increases the current and voltage slew rates, i.e. di/dt and dv/dt and that conversely affects the EMC performance of the power supplies. For power supply design over 75W in Europe (EPSMA, 21), AC-DC Boost power supply circuit is always been used as APFC type, because of its theoretical high power factor over 95%. APFC automatically controls the wide range of input voltage to maintain higher power factor in the AC input. This results in cost increase due to complex of implementation of the circuitry and additional filtering. Hence, to design a better filter for boost APFC power supply circuit, we have to analyze the EMI produced in the circuit. In addition, the filter performance should meet the EMC standards in order to be commercially available. In recent years, U.S. and Europe both have imposed strict EMI standards for the power supplies which require efficient and high performance design methods for power supplies. In this paper, we have investigated and presented the performance of a boost APFC power supply circuit when switching and load conditions, EMI filter elements such as X capacitor, CM choke value have been varied with switching. We also have included the performance results when switching was varied for a fixed load, also verified the performance in reverse condition. BASIC CIRCUIT MODEL The basic diagram of a Boost APFC circuit that is considered in this paper is illustrated in Fig. 1. Input voltage goes through a bridge rectifier, which gives full wave rectified output voltage. Unless line voltage gets boosted over than the voltage in holdup capacitor, no current will flow through holdup capacitor. Control circuit compares the line voltage with current at the input by measuring it, and boost the voltage. CM current always flows through both wire and return to the ground via parasitic capacitance. Both CM and DM currents can produce EM field outside the wire which leads to radiate EMI issue. If the generated EM field cannot be eliminated, then it can couple to nearby circuits. In this work, boost APFC power supply and analog control circuit have been designed. Output of the converter is fed back to the controller to control the boosted output voltage. Amplifiers are also been used in the control circuit. Fig. 1 Basic Boost APFC circuit 156

1 CONST M1. MUL2.5 CONST M2. MUL3 CM DM Diode6_5 Diode6_1 CONST 1 E4 Ls.5mH E5 L7 R2 Llisn1 Llisn2 C13.1mH.1ohm 25uH 5uH C14 C15 8uF 2uF R19 5ohm Rs.1ohm R16 5ohm C21 C22 C17 2uF 8uF Llisn4 Llisn3 C16 Llisn5 M1 R18 5ohm 2uH Llisn6 R17 5ohm 2uH M2 C18 C7 TFR1P2W1 C8 C1 C11 A M3 R13 M6 M9 Diode6_3 Diode6_4 R9 OP5_3 R8 - GAIN1 M7 R6 R7 R1 R12 Diode6_2 L1 C2 C1 M11 C12 C4 R4 C6 R11 C5 - OP5_4 M1 M12 R15 1megohm R3 OP5_2 E2 MPWM - R1 2475kOhm Mdc R2 25kOhm EC1 E1 R5 C3 OP5_1 - M5 25uH 5uH EC2 C2NC R14 C9 2kOhm GAIN M4 MUL1 E3 M8 C2NC Fig. 2 represents our experimental circuit, where input feeds to the line impedance stabilization network (LISN). LISN gives us stable line impedance, also in practice LISN is being used for measurement of the conducted emission. The general practice for measuring EMI is that noise generated by the equipment under test (EUT) is separated by LISN and fed to the spectrum analyzer. Before the signal is fed into the bridge rectifier, it s been fed to CM choke. Active devices such as IGBT have been modeled with implementation of parasitic capacitors in their model. For the measurement of the CM, DM and line noise we have made different measuring systems in the circuit. We obtained the response of the output signal with respect to the range of FCC class B which is 15 3MHz. Fig. 2 Boost APFC circuit with LISN PERFORMANCE ANALYSIS & RESULT ariation of the loading conditions As discussed in the previous section, the performance and change in CM, DM & line noise versus the variation of the load condition has been studied. Resistive load has been used for the simulation, which dissipates the energy in the form of heat. CM, DM and line noise when switching and load are respectively 4 and 5W are sown in Fig. 3, respectively. In Fig. 3, measured CM noise is much higher than DM noise under the same condition. CM noise decreases as the increases. And line noise response is similar as the CM noise in all ranges. It can be seen from Table 1 that CM noise for 5W is 32.3 dbµ at 4, whereas noise for 1KW 29.29 dbµ at 4. DM noise for 5W is -9.8 dbµ at 4, and -11.85 dbµ at 4 for 1kW. It is much lower than the CM noise for the same settings. The line noise is 31.99 dbµ at 4 and 5W load, but it has been reduced to 29.33 dbµ when is 4 and load is 5W. These are very close to the CM noise values. Fig. 3 CM, DM & line noise spectrum at 4 switching and 5W load Table 1 EMI values at different point for various loading conditions at 4 switching Load Noise at CM noise (dbµ) DM noise (dbµ) Line noise (dbµ) 5W 4 32.3-9.8 31.99 5W 8 14.57-25.46 14.58 5W 3.2 MHz -14.49-26.9-15.43 1KW 4 29.29-11.85 29.33 1KW 8 14.94-3.9 14.95 1KW 3.2 MHz -17.14-36.33-17.53 157

Fig. 4 CM, DM & line noise spectrum at 4 switching and 1KW load CM Noise analysis versus switching Frequency is another key factor for the performance of switching power supplies. The switching has been changed to different levels such as 8, 6 and 4 with the same loading conditions to observe the impact on the EMI noise. Fig. 5 is the representation of the CM noise at 4, 6 and 8 when load is 1KW. Table 2 is the collection of data at different points. At 24, noise is 41.16 dbµ when switching 4. Noise is obtained to be 45.3 dbµ when the switching is 6. Noise is found to be 49.49 dbµ at 8 when other elements unchanged. So, noise level is increased by approximately 1 dbµ with every 4 intervals in switching. DM Noise analysis versus Frequency The DM noise versus switching has been simulated without tuning the parameter of the boost APFC. Fig. 6 is the DM noise spectrum for the different switching frequencies. Change due to the variation of the switching can easily be observed from the figure. From Table 3, at 24 DM noise is 8.27 dbµ when switching 4. It increases to 14.35 dbµ for 6. When switching is increased to 8, DM noise increases to15.84 dbµ which is almost twice of the noise level at 4 switching. It becomes clear that with the increase of switching, DM noise level also increases. However, in high range the DM noise level decreases as the switching increases as tabulated in Table 3. Fig. 6 DM noise versus switching Table 3 DM noise spectrum versus switching Frequency 4 Frequency 6 Frequency 8 24 8.27 dbµ 14.35 dbµ 15.84 dbµ 48-8.46 dbµ -1.74 dbµ -14.8 dbµ 72-9.24 dbµ -16.86 dbµ -17.47 dbµ 96-18.23 dbµ -23. dbµ -26.34 dbµ Fig. 5 Frequency spectrum of CM noise versus switching Table 2 CM noise versus switching Frequency Frequency 4 6 Frequency 8 24 41.16 dbµ 45.3 dbµ 49.49 dbµ 48 23.66 dbµ 32.86 dbµ 35.72 dbµ 72 17.97 dbµ 21.83 dbµ 28.87 dbµ 96 9.55 dbµ 17.69 dbµ 22.65 dbµ Line Noise analysis versus switching Line noise spectrum also has been studied and simulated when the switching frequencies are 4, 6 and 8. It is shown in Fig. 7 that the line noise has been increased when switching increases. Table 4 gives tabulation of the data illustrated in Fig 7. In Table 4, line noise level is 17.9 dbµ at 78 when switching is 4. At same noise, line noise is 23.44 dbµ when switching is 6. When switching is 8, line noise increases to 28.88 dbµ. Comparison of Tables 2 and 4 shows that noise levels for CM and line remain almost 158

same. Hence, it can be concluded that the change in line noise follows the change in CM noise for this boost APFC circuit. values, CM noise level decreases by 12 dbµ when other components of APFC circuit were unchanged. Fig. 7 Line noise spectrum versus switching Table 4 Collection of noise at different point from Line noise spectrum versus switching 4 6 8 24 41.16 dbµ 45.31 dbµ 49.44 dbµ 48 23.7 dbµ 3.21 dbµ 35.72 dbµ 72 17.9 dbµ 23.44 dbµ 28.88 dbµ 96 9.58 dbµ 14.75 dbµ 22.65 dbµ CM noise response versus Y-cap and CM choke values Y capacitor, is generally used to filter out CM noise in circuit which is connected between line and chassis. With change of the value of Y-cap, the level of CM noise also changes. CM choke always produces high impedance to reduce the CM currents flowing from both lines. In CM choke, two coils are wounded on a single core, where DM current flows since choke produces small impedance against DM current due to the cancellation of magnetic field (DM type) in the choke. To observe the response of CM noise due to the change of Y-cap and CM choke, capacitor value and choke value have been changed to three different combinations while other parameter and switching (4 ) was remained same. Fig. 8 is the CM noise response spectrum of APFC when Y-cap and choke value are changed. Initially, the value of Y- cap has been changed to 22nF and choke value is set to.6h. Then, Y-cap is changed to 11nF and choke value is set to.3h. Finally, values have been set to 44nF for Y-cap and.12h for the choke. Noise spectrum for these three combinations is plotted in the Fig. 8. It is given in Table 5 that CM noise for different combinations of Y-cap and choke values is different. When Y-cap and choke value are decreased by half of their values, there is approximately 12 dbµ increase in CM noise level. When Y-cap and choke value are increased to double of their Fig. 8 Spectrum of CM noise for different Y-cap and choke values Table 5 CM noise at different point versus Y-cap and choke values Noise at Y-cap=22nF Choke=.6 H Y-cap=11nF Choke=.3H Y-cap=44nF Choke=.12H 8 6.11 dbµ 72.35 dbµ 48.28 dbµ 12 16 2 54.12 dbµ 66.19 dbµ 42.12 dbµ 48.89 dbµ 6.89 dbµ 36.96 dbµ 44.56 dbµ 56.53 dbµ 32.62 dbµ DM noise response versus Y-cap and CM choke values Although the change in Y-cap and CM choke does not affect the DM noise generation or the variation, it is still important to verify it. Fig. 9 is the DM noise spectrum for the variation of Y capacitor and CM choke values. Table 6 gives the DM noise at different point when Y-cap and choke values have been changed to different combination. In practice, X capacitor is usually used to reduce DM EMI which is an indication that Y capacitor does not blocks DM noise. In Table 6, it is clear that DM noise is not changed on with the variation of Y capacitor and CM choke values while all other parameters were unchanged. As a result, there is no effect on DM noise level, when the Y capacitor value and the choke value have been changed. It can be concluded that DM noise is independent of the Y capacitor and CM choke value in this case. 159

Table 7 Line noise at different point versus Y-cap and choke values Y-cap=22nF Choke=.6H Y-cap=11nF Choke=.3H Y-cap=44nF Choke=.12H 8 59.23 dbµ 71.76 dbµ 59.5 dbµ 12 16 2 53.99 dbµ 66.15 dbµ 44.14 dbµ 48.91 dbµ 6.89 dbµ 37.11 dbµ 44.43 dbµ 56.49 dbµ 32.8 dbµ Fig. 9 DM noise spectrum versus Y-cap and Choke values Table 6 DM noise at different point versus Y-cap and choke values Y-cap=22nF Choke=.6H Y-cap=11nF Choke=.3H Y-cap=44nF Choke=.12 H 8 66.14 dbµ 66.19 dbµ 66.1 dbµ 12 46.65 dbµ 46.77 dbµ 46.63 dbµ 16 31.22 dbµ 31.23 dbµ 31.35 dbµ 2 19.27 dbµ 19.89 dbµ 19.35 dbµ Line noise response versus Y-cap and CM choke values The line noise for the APFC with the variation of Y capacitor and CM choke when switching is 4 is also simulated. Fig. 1 is the line noise spectrum for different Y-cap and CM choke value when switching is 4. Table 7 gives the line noise levels at different frequencies, with the change of Y-cap and choke values. When Table 5 and 6 are compared, it is observed that the line noise follows the change in CM noise. We d like to also report that while reducing the values of the Y-cap and choke, high line noise has been observed on the system. Whereas increasing the value the value of Y cap gave reduced line noise on the system. CONCLUSION Accurate modeling of EMI simulation of boost APFC circuit that leads to an efficient design for the power supply circuit is presented. The detailed EMI simulation results using Ansys Simplorer for various conditions have been reported. It is found that the EMI performance of single-phase APFC power supply circuit is dominated by CM noise. Double switching can increase the envelope of the noise spectrum by 1 dbµ in wide range. It is also found that front-end EMI filter is critical to suppress EMI when the switching is increased. It is also observed that the CM choke and Y-capacitor play important role for CM EMI suppression. REFERENCES Majid, A., Saleem, J., Alam, F., & Bertilsson, K. (213). EMI Suppression in High Frequency Half Bridge Converter. Electronics and Electrical Engineering ElAEE. European Power Supply Manufacturers Association [EPSMA] (21), Harmonics Current Emissions Guidelines to the Standard EN 61-3-2, Power Factor Correction- Guide from EPSMA Ott, H., (29). Electromagnetic compatibility engineering. Hoboken, N.J.: John Wiley & Sons Fig. 1 Line noise spectrum versus Y-cap and choke values at different combination 16