through the parasitic capacitor between the primary and secondary coils [12]. To improve the topology of network, the network was designed with one radio frequency transformer with 2:1 by Mardiguian [13]. Moreover, the 0° and 180° power dividers have been used to constitute the noise separation network [14].

Based on the above models and mechanism, a lot of categories have been carried out to reduce the CE noises [15][16][17][18][19]. The EMI noises from DC-DC Buck conversion has been suppressed by employed MOSFET, decoupling capacitors and optimal design for PCB [15]. A frequency modulated (FM) source of conducted emission has an adverse effect on a DC power system and spread spectrum modulation is proposed to reduce EMI noises [16]. The Power Integrity problem for high speed systems is discussed in context of selection and placement of decoupling capacitors. The optimal capacitors and their locations on the board are found using the presented methodology, which can be used for similar power delivery networks in high speed systems [17]. A modified LLCL-filter topology is proposed to provide enough attenuation on the conducted EMI noise as well as to reduce the DC side leakage current [18]. A new method to reduce CM EMI at the DC input of variable-speed motor drives is analyzed. Unlike conventional passive or active filtering techniques that rely on impedance mismatch or active noise cancellation, the method uses a passive circuit with matched impedance to cancel the inverter CM current [19]. However, the above noise reduction methods can't obtain the EMI source and solve the EMI problem fundamentally in the economy and practical scale.

In view of above analysis and on basis of the acquired achievements about EMC of electronic equipments, the CE models and mechanisms for SMPS have been analyzed in the paper. And the CE measurement uncertainty was studies based on the voltage division factor, isolation factor and impedance under SMPS side of artificial mains network. Then, three methods have been proposed to reduce the CE, such as the capacitor shunt between the source and drain electrodes of MOSFET, CM choke and crosstalk choke, and the capacitors matrix. The experiment results show that the CE noises of the power concentrator and colonoscopy can be suppressed effectively and efficiently by using the present approaches, while can pass the EN 55022 and improve the safety and EMI performance.

Li Lin ? , Qiu Dongmei ? , Yan Wei ? & Gao Xiang ? witch mode power supply (SMPS) has been widely applied in the electrical and electronics devices, which can realize the AC-DC, DC-DC conversion and generate larger amount of electromagnetic interference (EMI), especially conducted emission (CE) from 9kHz and 30MHz [1][2][3][4][5]. The CE standards have been established by the most countries and areas, such as EN 55022 [6].

In recent years, the CE models and mechanisms for live, neutral and ground lines system have been proposed to analyze the noises paths, which contributes to noise suppression [7][8][9][10][11]. The CE models consist of common mode (CM) and differential mode (DM) noises. The CM noise path is from live/neutral line to ground line, and the DM path is from live line to neutral line. Four kinds of noise separation networks have been proposed to determine the CE mechanism by extracting CM and DM noises by Paul, See, Mardiguian and Guo [2,[12][13][14]. Paul network is composed of two radio frequency transformers with 1:1 and artificial switch, through which the high frequency (HF) noises are generated and coupled [2]. The core of See network is also two radio frequency transformers with 2:1, but the HF noises are generated

SMPS can support direct current (DC) power to the load through four diodes and metal oxide semiconductor field effect transistor (MOSFET). Voltage dependent resistance (VDR) was employed to resist and suppress the external surges and interferences. Three capacitors and common mode (CM) choke were used to reduce the high frequency (HF) EMI noises. However, a large amount of EMI noises were generated by SMPS via four diodes and FET, which will go to power supply through power line and influence other electrical and electronics devices.

CM and differential mode (DM) models were established to analyze the noise mechanism and suppress the above EMI noises based on three line system including live, neutral and ground lines, as shown in Fig. 1.

Fig. 1: CM and DM noise mechanism based on three line system

Due to the difference between the topology of SMPS (two line system without ground line) and traditional three line system, the conducted emission mechanism will be studied.

According to EN 55022 and FCC Part 15, conducted emission can be detected by EMI receiver and artificial mains network (AMN), where quasi peak detector and average detector should be fixed on the EMI receiver.

To analyze the uncertainty of CE measurement, the topology structure and HF parasitic parameters of AMN will be investigated. According to CISPR 16, the topology structure is shown in Fig. 2, where

The HF noise from 9kHz to 30MHz in power supply can be reduced by capacitor C 1 and inductance L 1 , and the 50Hz current is also ordinary and functional. The CE from 9kHz to 30MHz in SMPS can be extracted through capacitor C 2 . Resistance R 1 is used to bypass flow and resistance R 2 is designed to measure CE. Considering the HF parasitic parameters of the above capacitors, inductances and resistances, as shown in Fig. 2, L esc1 and L esc2 are the parasitic inductance of C 1 and C 2 , respectively. C esl1 is the parasitic capacitor of L 1 . L es and R es are the parasitic inductance and resistance of the interface adapter. In the paper, R es is neglected due to the adapter design and manufacturing technique.

To analyze the characteristic of AMN, the voltage division factor, isolation factor and impedance under SMPS side were defined as 20 log 20 log 20 log 20 log

Where, VIF L and VIF N denote the live and neutral lines voltage division factor of AMN, IRR L and IRR N represent the live and neutral lines isolation factor of AMN, U L and U N express the total CE of live and neutral lines, U LG and U NG express the HF EMI noise between live/ neutral and ground lines.

To analyze the voltage division factor of AMN, the above parasitic parameters were considered lonely as shown in Tab.1 and the results were shown in Fig. 3.

To analyze the isolation factor of AMN, the above parasitic parameters were considered lonely as shown in Tab.I and the results were shown in Fig. 4, where the L es can be neglected due to the measurement circuit.

To determine the impedance under SMPS side of AMN, L es was considered as 10nH and 100nH, respectively, and the other parasitic parameters can be neglected due to the measurement circuit, as shown inFig.5. As shown in Fig. 5, with the L es and measurement frequency increasing, the impedance under SMPS side was increasing greatly.

Therefore, the proper AMN should be designed and employed to determine the CE and decrease the uncertainty of CE measurement through capacitor and inductance matching.

IV.

Due to the difference between the two line SMPS system and three line system, the CE characteristic mechanism model should be established. Considered that SMPS have live and neutral lines but no ground line, the CM and DM noise transmission loop can't be formed and realized.

As shown in Fig. 2, the two different results can be obtained based on EN 55022, as follows:

Where, I L and I N denote the total CE current through live and neutral lines. Based on the formula (2), the unbalanced noise current can be defined as Where, I CM signifies the balanced noise current and can be considered as CM noise current.

Based on Fig. 2, the total live line noise have two bypass loops, such as live to ground noise I LG and live to neutral noise I LN . Similarly, the total neutral line noise also have two bypass loops, such as neutral to ground noise I NG and neutral to live noise I NL . Supposed that Based on the formula (3) and ( 9), the CM noise can be defined as

The equivalent transmission circuit of CM noise was shown in Fig. 6(a).

Where, U CM and Z CM represent the equivalent CM noise source and its impedance.

Based on the formula (3) and ( 8), the DM noise can be defined as

The equivalent transmission circuit of DM noise was shown in Fig. 6(b).

Where, U DM and Z DM represent the equivalent DM noise source and its impedance.

Moreover, Z CM and Z DM can be determined by employing the insertion method, dual current probe method, single current probe current, scattering parameter method and the proposed dual resistance calibration method. Due to the two lines system, the parasitic capacitor to ground can be considered to analyze the equivalent bypass circuits of CM and DM noises, as shown in Fig. 7.

The DM noise path was from live line to neutral line, the same as the three line system, as shown in Fig. 1. However, the CM noise path was from live/neutral line to ground line, then to the SMPS through the parasitic capacitor C PG .

Therefore, CE for SMPS still have CM and DM noises, but not the only DM noise, where noise suppression categories should be designed not only for DM noise but also for CM noise.

Moreover, the DM noise is much larger than CM noise in general because the parasitic capacitor is little, even ignored.

The switch frequency can be controlled through the grid electrode of MOSFET and the HF noise will generate and couple to the source and drain electrodes due to the MOSFET. The frequency of noises are based on the switch velocity, basically from 9kHz to 30MHz, which is the CE source.

The capacitor shunt between the source and drain electrodes can be used to reduce the above HF noises. Moreover, the value of the capacitor should be smaller than 0.1?F not only for the noise mitigation but also for the safety regulations & design, as shown in Fig. 8, where, C 4 denotes the capacitor shunt between the source and drain electrodes of MOSFET.

The CM choke can be employed to suppress the CM noises, but can't solve the environmental electromagnetic field coupling. As shown in Fig. 9(a), the live and neutral lines are in the same direction in the CM choke, which can well reduce the CM noise due to the electromagnetic field offset. Meanwhile, the CM choke can be considered as two electric dipoles which can receive the environmental electromagnetic field greatly and couple to the power lines. And the amplitude of the coupling noise is based on the length of dipoles and the noise frequency.

The crosstalk choke whose live and neutral lines are in the different orientation, can be used to solve the above problem. With different of CM choke, the crosstalk choke can be considered as the magnetic dipole, which can also obtain the environmental electromagnetic field. And the magnitude of the coupling noise resolves the area of dipole and the noise frequency. As shown in Fig. 9(b), the area of the dipole is the area between the live and neutral lines, which is very little. As shown in Fig. 12, the L C1 represents the CM choke and crosstalk choke. The single capacitor with the fixed value can suppress the certain spectrum noise due to HF parasitic parameter of the capacitor, such as the parasitic inductance of the pins. The capacitors matrix can be employed to solve the problem, which consists of 100pF, 1nF, 10nF, 0.01?F and 0.1?F. The different frequency noise will be reduced by the capacitors with different values, as shown in Tab.2. As shown in Fig. 8, the C M1 and C M2 denote the two capacitors matrix.

To verify the proposed methods, the SMPSs of the power concentrator and colonoscopy are studied in the paper. In the experiment, R&S EMI receiver ESL3 and R&S artificial mains network (AMN) ENV216 are used to determine the CE generated by the above devices.

The original CE result of the power concentrator is shown in Fig. 10(a),(b). According to EN 55022 Class B, the noise can't pass the standard from 150kHz to 10MHz. The average noises are [email protected] and [email protected]. And it exceeds 20dB?V from 3MHz to 5MHz, even critical. Based on the proposed method, the mitigation approaches are designed as follows:

1. EMI filter for SMPS of the power concentrator is designed, as shown in Fig. 10(c), where L 1 denotes 10mH CM choke, both L 2 and L 3 represent 1mH inductance, C 1 , C 2 and C 3 signify 0.01?F, 10nF and 1nF, respectively. 2. 10nF capacitor is shunt between the power supply of the carrier module to HF noises. Due to the above methods, the CE of the power concentrator can pass EN 55022 Class B, where the average noises are [email protected] and [email protected]. And the safety margin of the power concentrator can reach 10dB?V, as shown in Fig. 10(d). The experiment results show that the CE noises of the power concentrator and colonoscopy can be reduced very well by employing the present approaches.

Global Journal of Researches in Engineering ( ) Volume XVII Issue II Version I