CHAPTER 4 MEASUREMENT OF NOISE SOURCE IMPEDANCE

69 CHAPTER 4 MEASUREMENT OF NOISE SOURCE IMPEDANCE 4.1 INTRODUCTION EMI filter performance depends on the noise source impedance of the circuit and the noise load impedance at the test site. The noise source impedance is due to the circuit parameters and parasitic elements in the power converter and its environment. Its magnitude and phase vary with frequency. If the parameters of the EMI filter are not selected properly, the EMI filter amplifies the noise at certain frequencies. The knowledge of the noise source impedance and the noise load impedance of a SPC is essential in its power line EMI filter design. The filter design principle maximizes impedance mismatch so that the noise energy delivered to the load is minimized. The load impedance is the LISN and the source impedance varies among the power converters. 4.2 METHODS OF MEASURING NOISE SOURCE IMPEDANCE Due to the complexity of CM and DM noise coupling mechanism, derivation of complete theoretical models is very difficult. Hence, the best way to determine their characteristics is through measurement. Three ways to get this measurement are as follows: (i) (ii) Resonance method Two current probes approach (iii) Attenuation method

70 4.3 RESONANCE METHOD The resonance method is usually employed to calculate the noise source impedance of a SPC by terminating the power input of the SPC with a reactive part that is indifferent to the noise source reactance. The resonance of Conducted EMI is determined by assuming that the SPC is a Norton circuit with reactive impedance. The basic setup of the resonance method is shown in Figure 4.1. Added Inductor I L Current Probe I SC C P Converter R P Figure 4.1 Basic Setup of Resonance Method First, the Norton current source is obtained by short circuit at the load side. Second, the resistive and capacitive impedance of SPC is achieved by the load side inductor s resonate. Moreover, the quality factor (Q) source is found at the single frequency. The derivation of CM and DM input impedance of SPC by resonance method is obtained from Equations (4.1) to (4.3). IL Q = (4.1) I sc where, I L is the current when adding the inductor at the load side. I sc is the short circuit current.

71 The input resistance of SPC, R P = Q L (4.2) The input capacitance, Q C P = (4.3) R Figure 4.2 shows the CM setup of the input impedance measurement. AC Mains Ground LISN 27nF L' B 3 F 5 5 3 F Current Probe 0.5 F Converter 27nF L' A Figure 4.2 CM Input Impedance Measurement Setup Figure 4.3 shows the DM setup of the input impedance measurement. The experimental set up of the resonance method of noise measurement is shown in Figure 4.4. AC Mains Ground LISN 27nF L B 3 F 3 Current Probe Converter 27nF L A Figure 4.3 DM Input Impedance Measurement Setup

72 E.U.T. LISN AC Main Components used to Measure Input Impedance Current Probe Spectrum Analyzer Figure 4.4 Experimental Setup of Noise Source Impedance Measurement Using Resonance Method The experimental result of parallel resistance R p for CM is shown in Figure 4.5 and the experimental result of parallel capacitance C p for CM is shown in Figure 4.6. The average of CM parallel resistance R pcm is 46 k and CM parallel capacitance C pcm is 0.71nF. Common mode Rp versus frequency 250 200 Resistance(K ) 150 100 50 0 0-50 2 4 6 8 10 Frequency(MHz) Figure 4.5 Experimental Results of CM Parallel Resistor R pcm

73 Common mode Cp versus Frequency Capacitance(nF) 8 6 4 2 0 0 2 4 6 8 10 Frequency(MHz) Figure 4.6 Experimental Results of CM Parallel Capacitor C pcm The experimental result of parallel resistance R p for DM is shown in Figure 4.7 and the experimental result of parallel capacitance C p for DM is shown in Figure 4.8. The average value of DM parallel resistance R pdm and DM parallel capacitance C pdm are 320 k and 0.1nF, respectively. Differential mode Rp versus Frequency Resistance(M ) 3 2.5 2 1.5 1 0.5 0-0.5 0 2 4 6 8 10 Frequency(MHz) Figure 4.7 Experimental Results of DM Parallel Resistor R pdm

74 Differential mode Cp versus Frequency Capacitance(nF) 0.5 0.4 0.3 0.2 0.1 0 0 2 4 6 8 10 Frequency(MHz) Figure 4.8 Experimental Results of DM Parallel Capacitor C pdm The R p and C p of the DM input impedance is high due to the diode reverse recovery voltage and low source impedance component. Its inductance (L) value ranges from 0.5 H to 1 H and the resistance (R) value is 0.5. In this method, choosing the values of reactive components and tuning for resonance is very tedious and cumbersome. If the measured frequency is high, the parasitic effects of non-ideal reactive components become major and the resonance method based circuit topology becomes unsuitable. In case of DM noise source impedance measurement, it is very difficult to achieve parallel inductance capacitance (LC) resonant as the impedance nature of DM noise source is low.

75 4.4 TWO CURRENT PROBES APPROACH The advantage of two current probes approach is its premeasurement calibration procedure for the Radio Frequency (RF) combination circuit. Hence, the possible error contributed by the RF coupling circuit is eliminated to achieve good accuracy for measurement of noise source impedance of the SPC. In this method, noise source impedance is measured without interrupting normal operation of the SPC. Using appropriate calibration, two probes approach develops a lumped circuit model with resistive and reactive components to characterize the noise source impedance with acceptable accuracy. From this lumped circuit model of the noise source impedance, the most efficient filter design and the selection of proper component values becomes simple. With one probe as an injecting current probe and the other probe as a receiving current probe, CM and DM noise source impedances are determined in EMI regulated frequency range of any SPC within the acceptance level. With careful calibration of the measurement setup, good accuracy is achieved for measuring both CM and DM noise source impedances by using two 1µF capacitors (one for line to ground and another for neutral to ground) jointly with the injecting and receiving current probes from the RF coupling circuit. The perception of measuring the magnitude of unknown impedance by two current probes method is shown in Figure 4.9. This approach involves an injecting probe, a receiving probe, spectrum analyzer and signal generator. Z X at b-b 1 represents the unknown noise source impedance to be measured.

76 Spectrum Analyze r Receiving Probe I W a b Injecting Probe V W Z x Unknown Impedance Signal Generator Coupling a 1 b Capacitor 1 Figure 4.9 Two Current Probes Approach The two current probes and coupling capacitor from a RF coupling circuit are used for the measurement of magnitude of Z X. The coupling capacitor has an Equivalent Series Resistance (ESR) of R C and ESI of L C. The injecting current probe and the signal generator are used to induce a continuous wave signal V W, into the coupling circuit. The magnitude of the resultant current I W is measured in the wire of the coupling circuit through the receiving current probe and the spectrum analyzer. The frequency range of measurement has been made at 150 KHz to 30 MHz. By adjusting the signal generator to the correct level of signal output, the spectrum analyzer detects the induced signal from injecting current probes through the receiving current probes. If Z X is the unknown noise source impedance of a SPC under actual power ON condition, the receiving probe also pick up Conducted RF noise generated by the SPC, other than the signal

77 induced by the injecting probe. In some frequency ranges, the induced signal is masked by the noise generated in the SPC. Therefore, network analyzer which is based on the sweep frequency techniques is not suitable in this measurement. With the signal generator and spectrum analyzer, it is quite easy to identify the induced signal from the RF noise by adjusting the signal generator s output level by shifting its output frequency slightly. The measurements are made at sufficient selected frequencies that cover the required frequency range. Using the spectrum analyzer, only the magnitude of noise source impedance of the SPC is obtained. However, by observing the trend of the noise source impedance s magnitude with frequencies, a reasonably accurate noise source impedance model in terms of resistor, inductor and capacitor is derived. 4.4.1 CM Noise Source Impedance Measurement CM noise source impedance measurement setup for an unfiltered SPC is shown in Figure 4.10. To ensure input impedance Z in of the RF coupling is repeatable, the two capacitors are mounted on a Printed Circuit Board (PCB) and two fixed positions have been marked on the PCB for receiving probes. The final wire connections from PCB to the points of measurement have been made as short as possible to minimize the parasitic effect due to wire positioning. The advantage of fixing the coupling circuit for both CM and DM noise measurement setup is that it is previously calibrated to obtain Z in. So it is used for both CM and DM setup to get the noise source impedance measurement faster. Z s is the actual noise source impedance to be measured. Z L is the impedance of a choke which is used to provide the RF isolation. The actual measured noise impedance of the RF coupling circuit is Z T, which is the

78 parallel connection of Z L and Z s. They provide the sufficient isolation, Z L >> Z S, then Z T Z S. Z T L L CM Z L Z S L N N LISN L CM 1 F 1 F Nonfiltered Half bridge AC-DC Converter E E Signal Generator Spectrum Analyzer Figure 4.10 CM Noise Source Measurement Setup For the CM setup, an inductor of 100 µh is inserted between LISN and SPC to provide DM RF isolation. The value Z in of the coupling circuit is obtained by a resistance of 1.12 ; an inductance of 240 nh and a capacitance of 2.2µF which are connected in series. The coupling circuit measures the unknown impedance by monitoring the change in measured magnitude of Z in from 150 KHz to 30MHz. 4.4.2 DM Noise Source Impedance Measurement Two DM chokes are added between LISN and SPC for DM setup. Initially, SPC is removed from AC mains and now the coupling circuit ensures Z T. For full frequency range Z T << Z L, the DM chokes provide good RF

79 isolation for entire frequency range. The two current probes setup allows measurement of the noise source impedance of the SPC under normal power ON condition; it has flexibility to determine the noise source impedance of SPC for different conditions. Z T L DM Z L Z S L L 1 F 1 F N LISN N Nonfiltered Half bridge AC-DC Converter E L DM E Signal Generator Spectrum Analyzer Figure 4.11 DM Noise Source Measurement Setup DM noise source impedance measurement setup for an unfiltered SPC is shown in Figure 4.11. The DM noise source impedance is usually capacitive in nature; it resonates with the inductive element of the RF coupling circuit at some frequency below 30 MHz, which imposes the maximum frequency limit for the DM noise source impedance measurement. The resonant frequency is pushed up further by reducing the inductive element of the RF coupling capacitor with the lowest ESI to make the RF coupling circuit more compact. SPC with high power rating, special attention must be paid to ensure that CM and DM chokes are not saturated while providing the necessary RF

80 isolation. This is usually be resolved by putting two chokes of lower inductances and higher ratings in series. The two 1µF capacitors are connected together with injecting and receiving current probes from the RF coupling circuit. For DM set up, two 350µH DM chokes are added between LISN and SPC to provide DM RF isolation. The current rating of the preferred DM choke is 3A. 4.4.3 Hardware Setup The hardware set up of two current probes approach noise measurement is shown in Figure 4.12. Experimental results are presented for an isolated AC-DC half-bridge converter with a nominal input voltage of 220V/50Hz, and an output voltage of 12 V DC at a full load current of 5 A. LISN L 22mH E.U.T. N 22mH 1 F 1 F E Injecting Probe Receiving Probe Signal Generator Spectrum Analyzer Figure 4.12 Experimental Setup of Noise Source Impedance Measurement using Two probes Method

81 Noise source impedance is not constant throughout the measurement. It varies with frequency. Even after the design of the filter, noise source impedance varies with frequency and the complexity of the CM and DM noise coupling mechanism makes it difficult to derive the complete theoretical models. 4.4.3.1 LISN LISN is mainly used for noise voltage measurements and to isolate noise, then provide constant impedance Conducted emissions, without affecting the normal power flow required by the EUT. For power line frequency, the LISN presents a low impedance path for power flows from source to load and provides high impedance path for load to ground. The 50 impedance to ground is essential for the input impedance of the spectrum analyzer or EMI meter which is used to measure the Conducted noise. The LISN provides direct connection between the input terminals of an EMI receiver and the 50 connector provided on the LISN. The LISN is sometimes called a Line Stabilization Network (LSN) or Power Line Impedance Stabilization Network (PLISN). An artificial main supply in a coupling unit is used to measure the Conducted emissions from EUT power leads instead of main power supply to these leads. The LISN is a buffer network which connects the power leads of the EUT to the power mains by following methods. (i) (ii) (iii) Passing AC or DC power to the EUT. Preventing the EUT s noise getting back into the power bus. Blocking the power mains RF from coupling into the test sample.

82 The schematic diagram of the LISN is shown in Figure 4.13. The series inductance provides RF isolation between the power leads from the power converter main bus and from the EUT. It is obvious at 50 Hz and provides shortest coupling between the power source and EUT. Noise generating on the power bus is moved to ground through the coupling capacitor Cp, and any RF noise on the EUT s power leads are coupled to the 50 connector jack through this capacitor. L AC line source N 50µH L 1 50µH L 2 A B C 1 C 2 0.1µF 0.1µF Equipment Under Test R 1 50 R2 50 Measurement Points Figure 4.13 Line Impedance Stabilization Network 4.4.3.2 Specification of LISN used Product code : EP660-6 Rated Voltage : 250 or 300 V AC at 50Hz Rated Current : 1Amp Leakage Current : 2 0.3 ma High Voltage test for 1 Minute : 2.0 KV AC

83 4.4.3.3 Choke For common mode choke : 16 mh For differential mode choke : 350 µh The choke is the inductance used for LISN and it is connected in line and neutral terminals. Spectrum analyzer is used for Radiated and Conducted emission applications with EMC and associated communities. Modern spectrum analyzers are normally characterized by relatively high noise figures, unturned front ends and built-in electro-optical display unit with variable persistence amplitude calibration of intercepted signals which is achieved through narrow-bands. Due to wide-open front ends, spectrum analyzers offer minimal dynamic range to impulsive signals. The benefit of the spectrum analyzer is functional displays and flexibility. The theory of the two current probes approach has been described and its validity has also been experimentally verified. The measured CM and DM noise levels in db are shown in Table 4.1. The two probe setup permits SPC noise level measurement during normal power ON condition. The noise level is measured for the frequency range upto 10 MHz. The noise level in CM noise is more than the DM noise.

84 Table 4.1 Measured CM and DM Noise Level using Two Probe Method Common Mode Noise level in db Frequency in MHz Noise level in db 2 26 4 24 6 22 8 21 10 20 Differential Mode Noise level in db Frequency in MHz Noise level in db 2 24 4 22 6 21 8 20 10 18 4.5 ATTENUATION METHOD To overcome the problems faced by the previous methods, attenuation method is used to measure the noise source impedance of SPC as proposed in this thesis. The information obtained through the proposed method enables the prediction of EMI filter performance and the design of a suitable filter for an SPC. The impedance of the inserted component must be much larger than the noise source impedance. If these conditions are not fulfilled, accuracy deteriorates. However by proper selection of insertion element, accuracy is maintained.

85 In the basic theory of attenuation method, the noise source is modeled as voltage source V s in series with noise source impedance Z s which is shown in Figure 4.14. If a piece of filter element (Z series or Z shunt ) is inserted between Z s and R load, the noise voltage across R load will change. This change is measured by attenuation Equation (4.4) which is defined as the ratio of voltage across R load before and after the filter element is inserted. V A 20 log V noise noise without filter with filter (4.4) LISN Converter R load Z S + - V S Figure 4.14 Conducted Emission Model Attenuation value is a complex number, but usually magnitude alone is measured. Now attenuation and CM filter impedance is a known quantity, from which CM noise source impedance Z scm is calculated. In order to obtain noise source impedance Z s accurately, any one of the following methods is used, depending on the relative magnitude of either Z s versus R load. Series insertion method Shunt insertion method

86 4.5.1 CM Noise Source Impedance The major components of the CM noise source impedance are the unintentional capacitance between the switching devices, heat sink and parasitic capacitances between other devices or wires, which carry the pulsating voltage waveform and the grounded frame. The charging and discharging of insulator capacitance is the main cause for CM noise. Line Z scm Neutral Ground Half Bridge AC-DC Converter Figure 4.15 CM Noise Source Impedance Measurements The CM noise source impedance is measured between new terminal formed by shorting line and neutral with the terminal ground, when looking into the Equipment under test as shown in Figure 4.15. In practice, as the CM paths have large impedance, shorting the line and neutral terminals does not make little difference. When necessary, a cross capacitor is used to short the two terminals. 4.5.2 DM Noise Source Impedance The major components of DM noise source impedance are the diode on state resistance and the ESR, ESI of the bulk capacitor. Other factors, such as the PCB layout, component placement and wiring layout also influence the noise source impedance. The DM noise source impedance measurement between line and neutral with noise ground current is shown in Figure 4.16. In practice, the ground current is minimized by either inserting a CM choke or floating the EUT, when measuring DM noise source impedance Z sdm.

87 Z sdm Line Neutral Half Bridge AC-DC Converter Figure 4.16 DM Noise Source Impedance Measurements 4.5.3 Series Insertion Method Series insertion as shown in Figure 4.17 is used in case of Z s >> R load for better accuracy. A series component with assumption Z series >> Z s is used for measuring only CM input impedance. LISN Z series Converter R N Z S + - V S Figure 4.17 Series Insertion Method For CM noise source impedance measurement, it is assumed that: (i) CM inductor only suppresses the CM mode noise (ii) CM noise current alone transfers from the noise source to earth. The expression for attenuation is simplified using Equations (4.5) to (4.7).

88 A = R Rload VS R load + ZS Rload V + Z + Z load S series S (4.5) Z series = 1+ R load + Z S Zseries 1 + (4.6) Z S Since A is normally much greater than 1, then Z S» Z se ries A (4.7) where Z series is given, and A is obtained by attenuation measurement. In general, larger the attenuation value, the more accurate Z s will be obtained. When R load + Z s value turns out to be much greater than R load, series insertion method is used. CM noise source impedance measurement setup is shown in Figure 4.18. CM Choke CM Noise Source 50 R Load CM Current Z scm I scm 0 o Power Combiner + 50 R load CM Noise to Spectrum Analyzer Figure 4.18 CM Noise Source Impedance Measurement Test Setup

89 In order to measure the maximum and minimum value of CM noise source impedance, a test inductor of value 100 H is added at the input side of SPC. The circuit in Figure 4.18 is simplified to its equivalent circuit as shown in Figure 4.19. This equivalent circuit is used to derive the expression for noise attenuation. Z CM L CM FILTER + V noisecm R CM Z scm I scm _ N LISN EQUIVALENT CM NOISE SOURCE Figure 4.19 CM Equivalent Circuit From the CM equivalent circuit as shown in figure 4.19, attenuation is calculated using Equation (4.8). A TCM = RCM ZsCM R CM + Z ZsCM R + Z + Z scm CM scm CM I scm I scm R CM Z CM A TCM = 1 + (4.8) R CM + Z scm In Equation (4.8), Z scm is the CM noise source impedance and R CM = 25 is the LISN CM load resistance. A TCM is the CM noise attenuation measured after a test CM inductor is added. Z CM is the impedance of the test

90 CM inductor at the test frequency point. From the above relationship, Equation (4.9) is obtained. Z CM R CM + Z scm = (4.9) A TCM - 1 For A TCM >> 1, Z CM R CM + Z scm = (4.10) A TCM In Equation (4.10) A TCM and Z CM are known values, so CM noise source impedance Z scm is easily calculated. From Equation (4.10), it is clear that the value of Z CM /A TCM is a real number. The noise source impedance has a significant impact on the attenuation of the EMI filter. In order to effectively attenuate the EMI noise over the frequency range of interest, the EMI filter must be designed to match the noise source impedance. It is projected that (i) if the maximum amplitude and the minimum amplitude of the noise source impedance is determined and (ii) if the EMI filter components are selected properly, then the EMI filter is designed effectively. 4.5.4 Impact of the CM Noise Source Impedance The attenuation of an EMI filter is defined by Equation (4.11). A T Vnoise without filter (4.11) Vnoise with filter V noise without filter is measured at the LISN output when no CM inductor is added, which corresponds to the situation when no EMI filter is added.

91 V noise with filter is measured at the LISN output when a CM inductor is added. The filter attenuation A T is a function of frequency, so that measurements must be taken and the attenuation is calculated at several frequency points. The CM noise spectrum before a CM inductor is added at the input of the SPC is illustrated in Figure 4.20. Figure 4.20 CM Noise Test Results before CM inductor is added The noise voltage V noise across R loadcm, before and after the CM inductor added, is expressed by Equation (4.12) as V noise without filter and Equation (4.13) as V noise with filter respectively. V noise without filter RloadCM ZsCM = I RloadCM ZsCM scm (4.12) V noise with filter RloadCM ZsCM = I RloadCM ZsCM ZCM scm (4.13) The CM noise spectrum after a CM inductor is added at the input of the SPC is illustrated in Figure 4.21.

92 Figure 4.21 CM Noise Test Results after CM inductor is added inductor is added If Equation (4.14) is satisfied, V noise will be amplified after a CM R loadcm + Z scm + Z CM < R loadcm + Z scm (4.14) At high frequencies, the inductance dominates (Z CM >> R loadcm and Z CM >> Z scm ). The CM inductor effectively suppresses the CM noise at high frequencies. From the test results, the noise at 525 KHz is amplified, but at higher frequencies the noise is effectively suppressed. This demonstrates that the EMI filter should be designed properly to attenuate the noise for the entire frequency range. It is generally desirable to design a filter so that its noise spectrum is below the 3 db limit to account for varying environmental conditions and component tolerances. It is worth noting that more complex topologies for a CM filter is used, but if the CM filter is not designed to match the CM noise source impedance, the effects of the deteriorated impact of the EMI filter will still exist.

93 4.5.5 Determining the Maximum Value and Minimum Value of the CM Noise Source Impedance The objective is to find the maximum and minimum noise source impedances Z smax and Z smin respectively. If the ratio Z scm /A TCM is the radius r, then the unknown variable in Equation (4.10) is Z scm, which is a complex number. If Z scm = x+jy, then Equation (4.10) is expressed by Equation (4.15), where it is clear that Equation (4.15) represents a circle centered with radius r. The maximum CM noise source impedance is on the real axis on the left side of the circle by Equation (4.16). The minimum CM noise source impedance is on the real axis on the right side of the circle by Equation (4.17). 2 2 (R CM + x) + jy = r (R CM + x) + y 2 2 2 = r (R CM + x) + y = r (4.15) The maximum and minimum valve of the CM noise source impedances are approximated as follows. Z (4.16) CM ZsCM Max R CM + A TCM -1 ZCM (4.17) ZsCM Min RCM - A TCM +1 At full load, the CM noise spike of the SPC at 0.5 MHz is 54 db as in Figure 4.20. After a test CM inductor is added, if the CM noise at 0.5 MHz is 36 db as in Figure 4.21, then 54 44 20 ATCM 10 3.2

94 Using an impedance analyzer, the CM inductor s impedance is measured to be 314 at 0.5 MHz. Z scm Max and Z scm Min are calculated to be 474 and 324 using (4.16) and (4.17), respectively. The CM noise source impedance varies with frequency, therefore some points in the frequency range of interest are selected and then the maximum value and minimum value of the CM noise source impedance are calculated at these frequency points. The maximum value and minimum value of the CM noise impedance from 0.15 to 9 MHz are calculated at different frequency points using measurements of the noise voltage along with Equations (4.16) and (4.17). The resulting impedances are calculated as shown in Figure 4.22. Z scm Max 72 Z scm Min 39 900 800 Max Min CM Noise Impedance ( ) 700 600 500 400 300 200 100 0 0 2 4 6 8 10 Frequency (MHz) Figure 4.22 Maximum and Minimum Values of CM Noise Source Impedances

95 4.5.6 Shunt Insertion Method Shunt insertion method as shown in Figure 4.23 is used in case of Z s << R load for better accuracy. By using a shunt component with assumption Z shunt << Z s, it is used for measuring only DM input impedance. LISN Converter R load Z shunt Z S + - V S Figure 4.23 Shunt Insertion Method and (4.19). The expression for attenuation is simplified using Equations (4.18) A = Rload VS R load + ZS R load // ZShunt V R load // Z Shunt + ZS S (4.18) = 1+ R load // Z Zshunt S Z S 1+ (4.19) Z shu nt Equation (4.20). Since A is normally much greater than 1, Z s is approximated as Z S Z sh u n t A (4.20)

96 where Z shunt is given, and A is obtained by attenuation measurement. In general, the larger the attenuation value, the accuracy is more. When R load //Z shunt value is much less than R load, then shunt insertion method is used. To measure the DM noise source impedance, test setup in Figure 4.24 is used. 50 R Load DM Current C X Z S DM I S DM 180 o Power Combiner + 50 R load DM Noise to Spectrum Analyzer DM Noise Source Figure 4.24 DM Noise Source Impedance Measurement Test Setup C X is the DM test capacitor added at the input side of SPC. In order to measure the maximum and minimum value of DM noise source impedance, a test capacitor of value 1 F is added at the input side of SPC. To calculate the attenuation, A TDM simplified equivalent circuit given in Figure 4.24 is used. Z DM is the filter impedance which is chosen to be less than the noise source impedance. For the DM noise source impedance measurement, it is assumed that (i) X capacitor alone suppresses the DM noise. (ii) DM noise source impedance Z sdm is smaller than the LISN equivalent resistance of 100.

97 (iii) DM noise current is only conducted from the noise source to the power ground. L DM FILTER + I sdm V noisedm R DM Z DM Z sdm _ N LISN EQUIVALENT DM NOISE SOURCE Figure 4.25 DM Equivalent Circuit From the equivalent circuit as shown in Figure (4.25), attenuation expression is derived as follows: DM sdm A TDM = 1+ R DM + Z sdm Z DM R Z A TDM ZsDM 1+ (4.21) Z DM calculated. In Equation (4.21), A TDM and Z DM are known, from which Z sdm is 4.5.7 Impact of the DM Noise Source Impedance The DM noise spectrum before a DM capacitor is added to the input of the SPC is illustrated in Figure 4.26.

98 Figure 4.26 Test Result of DM Noise before an X capacitor is added If -1 < Z sdm / Z DM < 0, the attenuation is less than one, and the DM noise is amplified. This phenomenon is illustrated in Figure 4.26 and Figure 4.27, where the DM noise spectrum at 7.85 MHz is amplified when an X capacitor is added at the input of the SPC. Figure 4.27 Test Result Illustrating Amplification of DM Noise after an X capacitor is added

99 Similar to the case for CM filter design, more complex topologies for the DM filter is used. However, if the DM filter is not designed to match the DM noise source impedance, the effects of the deteriorated impact of the EMI filter will exist. The above analysis and experimental results, for both CM and DM cases, prove that the designed filter should match with the noise source impedance to effectively suppress the EMI noise. If the inductor faces the input of the SPC, the inductor s impedance should be much larger than the noise source impedance. If the capacitor faces the input of the SPC, the capacitor s impedance should be much smaller than the noise source impedance, so that the noise may be suppressed via the capacitor and not the parallel noise source impedance. 4.5.8 Determining the Maximum Value and Minimum Value of the DM Noise Source Impedance The maximum and minimum value DM noise source impedance measurement is similar to the CM noise source impedance measurement. Equations (4.22) and (4.23) are used to calculate the maximum and minimum DM noise source impedances. Z sdm Min = ZDM ATDM 1 (4.22) Z sdm Max = ZDM ATDM 1 (4.23) At full load, the DM noise spike of the SPC at 0.5 MHz is 63 db as in Figure 4.26. After a test DM capacitor is added, if the DM noise at 0.5 MHz is 44 db as in Figure 4.27, then

100 63 44 20 ATDM 10 8.9 Using an impedance analyzer, the DM capacitor s impedance is measured to be 0.318 at 0.5 MHz. Z sdm Min and Z sdm Max are calculated to be 0.128 and 0.192 using (4.24) and (4.25), respectively. ZsDM Min 0.318x3.6 = 1.145 ZsDM Max 0.318x5.8= 1.844 The maximum value and minimum value of the DM noise impedance from 0.15 to 9 MHz are calculated at different frequency points using measurements of the noise voltage along with Equations (4.22) and (4.23). The resulting impedances are calculated as shown in Figure 4.28. DM Noise Impedance ( ) 7 6 5 4 3 2 1 Max Min 0 0 2 4 6 8 10 Frequency (MHz) Figure 4.28 Maximum and Minimum Values of DM Noise Source Impedances

101 4.6 CONCLUSION In this chapter, the noise source impedances are calculated by Resonance method, two probes approach and attenuation method. From these three methods, the noise voltage level satisfies the FCC class B regulation in attenuation method. So the attenuation method is used for filter analysis. Table 4.2 gives the noise source impedance values measured at 0.5 MHz through a spectrum analyzer. Table 4.2 Comparison of Noise Source Impedance Values Filters Description Filter Impedance Experimental Results 314 CM Attenuation 3.2 Noise source Max 72 Impedance Min 39 Filter Impedance 0.318 DM Attenuation 8.9 Noise source Max 1.145 Impedance Min 1.844 The EMI filter design method is based on noise source impedance because it is an efficient and simple method to design EMI filter for SPC. Further, this method is independent of SPC topology.