Control Simplification and Stress Reduction in A Modified. PWM Zero Voltage Switching Pole Inverter

Transcription

1 Control Simplification and Stress Reduction in A Modified Xiaoming Yuan Power Electronics and Electrometrology Laboratory Swiss Federal Institute of Technology Zurich PWM Zero Voltage Switching Pole Inverter Ivo Barbi Power Electronics Institute Federal University of Santa Catarina P. 0. Box: 5119, , FlorianopolisSC Brazil ETHZentrum / ETL, CH8092 Zurich Switzerland Phone: Fax: Phone: Fax: em.ee.ethz.ch inep.ufsc.br Abstract This paper discusses a modified PWM zero voltage switching pole inverter which allows for zero voltage switching of the main switch and zero current switching of the auxiliary switch. The proposal features simplified control and reduced auxiliary circuit stress in comparison with the AuxiliaryResonantCommutatedPoleInverter (ARCPI). Operation and designing aspects of the proposal are detailed in the paper. The proposal is verified by a 4.5kW IGBT half bridge prototype. I. INTRODUCTION While hard switching inverter has been firmly established as the state of the art, the swarm of ills such as switching losses, EMI, excessive voltage rate of change etc., are warranting the extensive attention for soft switching inverter alternatives [ 11. Among the family of the candidates, the AuxiliaryResonantCommutated PoleInverter (ARCPI) [2][3][4] has been appearing to be the most plausible circuit for high power installation. However, the ARCPI circuit suffers from such other problems as control complexity, centertap potential variation as well as auxiliary device protection etc.. For which the TruePWMPole inverter [ 101 seems offering an interesting solution while the autotransformer resetting problem arises. This paper proposes a modified PWM zero voltage switching pole inverter trying to solve this problem. necessitates additional monitoring and controlling with highly demanding resolution. Without the boost stage, the pole voltage may not be able to arrive at the rail level during the resonance in practice and zero voltage switching may be lost [5][6]. Note that freewheeling diode reverse recovery current and DC bus inductivity can in theory contribute to boost the resonance [3][9]. However, they can not be counted on as both are second order circuit parameters. T Fig. 1. Circuit diagram of the.4uxiliaryresonantcommutated Pole Inverter (ARCPI). 11. PROBLEMS OF THE ARCPI CIRCUIT The circuit diagram of the ARCPI and the relevant commutation waveforms are shown in Fig. 1 and Fig. 2 respectively. Before turningoff the main switch, the corresponding auxiliary switch must be turned on in advance which allows for the rampup of the auxiliary switch current until the valve current tboost for the main switch is reached [4]. Such ramp stage enables the pole voltage to swing to the rail level in the presence of the resonant loop losses. To maintain a constant level of Ibwst, the ramp time tramp must be controlled following the load current variation instantaneously, which :... Fig. 2. Relevant commutation waveforms of the ARCPI during a switching cycle /99/$ IEEE. 1019

2 Controlling of trnnlp must also take into account the capacitor center tap potential variation. Or, alternatively, the center tap must be kept constant by controlling. Both may require significant control complexity. The problem may appear especially when a half bridge inverter supplying a low frequency high current load, in which case both capacitors have to be chargeddischarged during the half cycle of the low frequency period [7]. Even for a single or three phase system, where the net current flowing into the center tap may add to zero within the high frequency cycle ideally [3], this problem may still be outstanding as close match of the two storage capacitors assumed in the control designing may be lost. In addition to controlling aspects, effective measures for protecting the auxiliary devices from overvoltage must be taken [8]. During the reverse recovery of the auxiliary devices, a releasing path must be established for the energy stored in the resonant inductor. Such measures always lead to extra loss and circuit complexity. III. AUTOTRANSFORMER RESETTING OF THE TRUEPWMPOLE INVERTER The TruePWMPole inverter, as shown in Fig. 3, introduces an autotransformer in the auxiliary circuit. The induced voltage on the secondary (N2) joins the DC voltage forcing the commutation resonance. By setting the transformer ratio (N2/NI) to a value less than <%, the pole voltage is able to swing to the rail level during the resonance in the presence of losses without any additional monitoring or controlling. The turnoff of the main switch and the turnon of the corresponding auxiliary switch can then take place at the same instant, which significantly simplifies the control requirement. The half bridge configuration for the auxiliary devices eliminates the need for the capacitor centertap, the associated problem in capacitor centertap potential variation is therefore avoided. In the meantime, the auxiliary devices are now directly clamped to the DC bus, overvoltage protection measures are no longer necessary. In particular, due to the shunt branch established by of the autotransformer primary (NJ, the actual current flowing in the auxiliary switch is nearly half [( 1k) times] of the resonant inductor current. In comparison, the auxiliary switch in the ARCPI flows the full resonant inductor current. Despite the potential advantages of the TruePWM Pole inverter over the ARCPI, the magnetization of the autotransformer in the TruePWMPole inverter can not be reset appropriately following each commutation, rendering it not applicable in practice. This phenomenon is explained as following. l l 1 Fig. 3. Circuit diagram of TruePWMPole inverter.... in2... freewheeling,,'... current /... L...?... 4 '. residual current '.'\....>....~ I..., 'Dsa2 4 freewheeli ne....j...;::!q c..<.., \L.,..*.,.. \;' t,, I tl t2 t3 k t5 10NDiv; 2uS/Div Fig. 4. Experimental waveforms of the currents in the autotransformer windings and the auxiliary freewheeling diode in the TruePWMPole inverter. Referring to Fig. 4 where the relevant experimental waveforms during D, to Sz commutation are shown. Due to the freewheeling current flowing through L, N2 and Dsa2 prior to the commutation (tl), the resonant inductor current does not return to zero after the commutation (t2). Instead, this residual current flows through L,, NZ and Sal, causing hard switching (turnoff) of the auxiliary switch Sal (t3) as a result. Upon turnoff of Sal, the opposite freewheeling diode Dmz starts conduction, which forces the current decreasing in the resonant inductor. In the meanwhile, a voltage is reflected on the autotransformer primary side due to the current decreasing in the secondary side, which forces Do? into conduction carrying an increasing current. The two currents add to each other and the resulted net core flux changes at a rate which maintains the voltages on both windings of the autotransformer. 1020

3 The autotransformer secondary current reverses its direction at t+ Afterwards, when the summing current in Dsa2 declines to zero (ts). both currents get equal and flow steadily through Da2, NI, N2 and L, until the next switching action. The autotransformer flux can not be reset to zero as a result. Moreover, the steady current flow in the resonant circuit during noncommutation period creates considerable loss. Obviously, the freewheeling current is resulted from the residual current and each becomes augmented by the other. The residual current is resulted primitively from the autotransformer magnetizing current. Iv. THE MODIFIED PWM ZERO VOLTAGE SWITCHING POLE INVERTER From the analysis in section 111, there are potentially two approaches solving the autotransformer resetting problem, either by breakingup the freewheeling path between the two windings of the autotransformer during the noncommutation period, or by decoupling the interaction between the freewheeling current and the residual current when the auxiliary switch is turned off. The latter approach is taken in this paper. Decoupling can be done simply by using an unidirectional auxiliary switch in the auxiliary circuit, instead of the bidirectional one, as shown in Fig. 5. Also shown in Fig. 5 is the modification of the positioning of the autotransformer primary. The new positioning allows the autotransformer secondary to carry the same current as the auxiliary switch, rather than the resonant inductor. Thus the secondary current is nearly halved [l/(l+k) times]. Further, the DC link voltage is now shared by the autotransformer primary and secondary, rather than by the primary alone. The autotransformer primary voltage rating is almost halved. The autotransformer size, as a result, will be almost halved [l/(l+k) times]. To ensure zero voltage switching, the autotransformer ratio must be set less than 1, rather than 112 in the Fig. 3 circuit. The predicted commutation waveforms of the modified circuit are shown in Fig. 6. The step diagrams for the commutations during a whole switching cycle under positive load current are shown in Fig. 7. Assuming an initial current flowing through D2, the commutations proceed in the following steps: step l(totl): Freewheeling diode D2 carries the load current, pole output is connected to the minus rail. step 2(tlt2): At instant tl, the main switch S2 is turned off while the corresponding auxiliary switch Sd is turned on. Auxiliary diode Da2 gets conduction. NI and N2 share the DC link voltage and an auxiliary voltage source of V,J(l+k) (k=n2/ni is the autotransformer ratio) appears across NI. Current in D2 starts decreasing. step 3(t2t3): At instant t2, D2 is blocked. Resonant capacitor Cr2 is charged and C,I is discharged. Pole voltage starts swinging toward the plus rail level. sfep 4(t3t4): At instant t3* resonant capacitor CrI is fully discharged. Currents in C,I and Cr2 transfer to the freewheeling diode Dl and Dl becomes conducting. Gating signal for SI is released by the zero voltage detecting circuit installed across SI. step 5(t4t5): At instant t4, the resonant inductor current arrives at the load current level and the main switch SI becomes conducting. step 6(15t7): At instant ts, the resonant inductor current reaches zero, the main switch SI carries the full load current. Auxiliary switch gating signal is withdrawn at k. Pole output is connected to the plus rail. step 7(t7t8): At instant t7, the main switch SI is turned off while the corresponding auxiliary switch Sal is turned on. Auxiliary diode DaI enters into conduction and an auxiliary voltage source of Vdc/(l+k) appears across NI. Resonant capacitor CII is charged and Cr2 is discharged. Fig. 5. The modified PWM zero voltage switching pole inverter. 3 PWM gating signal Fig. 6. Predicted relevant commutation waveforms of the modified PWM zero voltage switching pole inverter. 1021

4 step l(btl) step 2(tlt2) step 3(t2t3) step 4(t3t4) step 5(t4t5) step 6(t5t7) step 7(t7t8) step 8(t8ty) step 9(tYtlo) Fig. 7. Commutation step diagrams of the modified PWM zero voltage switching pole inverter during a switching cycle under positive load current. step 8(r8tY): At instant tg, Cr2 is fully discharged. Currents in Cr2 and CrI transfer to the fieewheeling diode D2. D2 becomes conducting. Gating signal for the main switch S2 is released by the zero voltage detecting circuit installed across S2. step 9(t9tlo): At instant ty, the resonant inductor current reaches zero. The freewheeling diode D2 carries the full load current. Gating signal for Sal is withdrawn at tlo. Pole output returns to the minus rail. In practice, due to the small voltsecond product, the magnetizing current at the end of the commutation t5 or ty is small depending on the commutation duration. This current will be reset during the extinction transience of the winding voltages. Small RC snubbers can be introduced across the series diodes DSI and Dd to suppress the possible voltage spike during the extinction processes. The use of the unidirectional auxiliary switch is penalized as it is no longer clamped directly to the rail. 1022

5 I % k=0.67 k=054 I '%ht I 'load I 'load (a) Diode to switch commutation (b) Switch to diode commutation Fig.8. Variations of commutation durations tst and t,, with load current and autotransformer ratio 'LrPl ilrp I I (a ) Diode to switch commutation (b )Switch to diode commutation Fig. 9. Variations of the resonant inductor peak currents with load current and autotransformer ratio. 0. I I k=o67,'=90 k=l, T=90 e k=o43,td5 k=o67 T+ k=l. TA5 ilmmls n " nn., ""i I W (a) diode to switch commutation (b) switch to diode commutation Fig. 10. Variations of the resonant inductor RMS currents related to load current, autotransformer ratio and switching cycle. 'load 1023

6 V. ANALYSIS AND DESIGNING Under the following assumptions, the commutation duration, the resonant inductor peak current and the resonant inductor RMS current (averaged over the switching cycle) in relation to the load current, autotransformer ratio and switching cycle are shown in Fig. 8, Fig. 9 and Fig. 10 respectively. Details of the mathematical procedure is discussed in [l 11. Resonant frequency resonant impedance z,=jm, snubbing capacitance Crl=C~=Cr, and switching cycle T. Unit current S=iZO/vdc, unit voltagev=v/vd, unit time t = two. Circuit parasitics, device switching delayes, losses etc are neglected in the analysis. From Fig. 8, for diode to switch commutation, the commutation duration increases with load current but decreases with autotransformer ratio. For switch to diode commutation, it decreases with both load current and autotransformer ratio. From Fig. 9, for diode to switch commutation, the peak resonant inductor current increases with load current but decreases with autotransformer ratio. For switch to diode commutation, it decreases with both load current and autotransformer ratio. The auxiliary switch peak current is actually l/(l+k) times of the resonant inductor peak current. Moreover, from Fig. 10, the resonant inductor RMS current resulted from diode to switch commutation increases with load current, but decreases with autotransformer ratio and switching cycle. The resonant inductor RMS current resulted from switch to diode commutation, however, decreases with load current, autotransformer ratio and switching cycle. The reverse proportion relation of the commutation duration, resonant inductor peak current and resonant inductor RMS current with the load current for switch to diode commutation highly justifies the auxiliary switch gating strategy in this paper to trigger the corresponding auxiliary switch simultaneously with the main switch not only for diode to switch commutation where such triggering is indispensable for achieving zero voltage switching, but also for switch to diode commutation where such triggering becomes dispensable above certain load current level. Such triggering strategy facilitates significant control simplification while the extra loss so accrued during switch to diode commutation is negligible. Note that the auxiliary switch RMS current equals l/(l+k) of the diode to switch commutation component for one load current direction. It equals l/(l+k) of the rr switch to diode commutation component for the other load current direction. The obtained auxiliary switch RMS current stress reduction has been deemed an important advantage of this modified PWM zero voltage switching pole inverter over the more known ARCPI circuit. For designing of the auxiliary circuit, the autotransformer ratio should be set less than 1 dependent on the actual resonant loop resistance to ensure the pole voltage swinging to the rail level during the commutation resonance. This requirement is hard and allows no bargaining. In addition, the resonant capacitance can be designed as per the device turnoff loss and the thermal condition associated. Resonant inductance can then be decided according to the accepted resonant circuit RMS stress based on Fig. 10, taking into account the system operating frequency. Rating of the auxiliary device can be made according to the auxiliary device RMS current stress, and the peak current stress as well given in Fig. 9. Gating signal width for the auxiliary switch must cover the maximum commutation duration given in Fig. 8 and remains constant over the low frequency cycle. In the meantime, the minimum pulse width in the inverter PWM pattern should not be less than this duration. The next commutation should not start before the conclusion of the previous commutation. On the other hand, autotransformer designing can be based on the knowledge of commutation (magnetization) duration and the RMS resonant inductor current stress. With peak and RMS resonant inductor current information, the resonant inductor can be designed. DC input voltage output nnwer VI. EXPERIMENTATION RESULTS A 4.5kW IGBT half bridge inverter prototype has been built. Specifications of which are given in Table 1. The resonant and autotransformer parameter parameters are: L=15uH, CrI=Cr2=0.1uF and k=0.67. An output lowpass filter is installed with Lf=l.45mH and C~12uF. SI S2 and SalSa2 used are SKh450GB123D (1200V/50A), DalDa2 and DslDs2 used are HFA30T60C (600V/30A). Besides, a simple zero voltage detecting circuit is designed to interface the SEMIJCRON driver to each main IGBT for releasing of the gating signal. The commutation waveforms of the main switch S2 are shown in Fig. 11 and Fig. 12 respectively. Vk Output V,,,, Modulati M= =4OOV voltage =IOOV onindex 0.78 P" Load I,,,, Switching f,= =4 SLW ciirrent =4SA fremtenrv 6 5kH7 1024

7 . inverter. The autotransformer size has been nearly halved due to the new autotransformer positioning. Compared with the ARCPI circuit, it allows for significant control simplification and auxiliary switch stress reduction. The modified PWM zero voltage switching pole inverter represents an interesting alternative for the ARCPI circuit in high power advanced applications. REFERENCES i,.... IoOV/Div: I0,4/Div:ZuS/Div i. i i... :....:... i... Fig. 1 I. Relevant experimental waveforms when SI is turned on carrying the load current....,..,. v i s2 [..+...! i :. IOOVIDiV: IOA/Div;2uS/Div j..... :.... :... ; Fig. 12. Relevant experimental waveforms when S? is turned off from carrying the load current. Fig. 11 shows the diode to switch commutation (DI to S,) process. With load current of 30A, the peak resonant inductor current is 53A and the commutation duration is IOuS, corresponding to the predicted peak resonant inductor current of 55A according to Fig. 9(a) and the predicted commutation duration of 9.4uS according to Fig. 8(a). On the other hand, Fig. 12 shows the switch to diode commutation (S2 to DI) process. With load current of 36A, the peak resonant current is 5A and the commutation duration is 1.6uS, corresponding to the predicted peak resonant current of 7A and the predicted commutation duration of 1.9uS. The analysis results are well verified. The error has been generated mainly due to the resistance existing in the resonant loop which has been neglected in the analysis in section V. VI. CONCLUSIONS The analysis and the experimentation presented demonstrate that the modified PWM zero voltage switching pole inverter solves successfully the autotransformer resetting problem of the TruePWMPole [ 11 D. M. Divan, LowStress Switching for Efficiency, IEEE Spectrum, Dec. 1996, pp [2] G. Bingen, Utilisation de Transistors a Fort Courant et Tension Elevee, Record of EPE Conference, 1987, pp [3] W. McMurray, Resonant Snubbers with Auxiliary Switches, IEEE Trans. on Ind. App., Vol. 29, No. 2, MarcNApril 1993, pp [4] R. W. De Doncker and J. P. Lyons, The Auxiliary Resonant Commutated Pole Inverter, Record of IEEE IAS, 1990, pp [5] H. J. Beukes, J. H. R. Enslin and R. Spee, Integrated Active Snubber for High Power IGBT Modules, Record of IEEE AF EC, 1997, pp [6] P. P. Mok, R. Spee, H. J. Beukes and J. H. R. Enslin, Control Complexities Related to High Power Resonant Inverters, Record of IEEE PESC, 1996, pp F. R. Salberta, J. S. Mayer and R. T. Cooley, An Improved Control Strategy for a 50kII~ Auxiliary Resonant Commutated Pole Converter, Record of IEEE PESC, 1997, pp H. Matsuo, K. Iida and K. Harada, 210 kva Soft Switching PWM Type AC Auxiliary Power Supply System of the Electric Railway Rolling Stock Using an Improved ARCP Three Phase Inverter with SI Thyristor Switches, Record of IPECYokohama, 1995, pp [9] K. Iida, T. Sakuma, A. Mechi, H. Matsuo and F. Kurokawa, The Influence of the Conducting Inductance in the Auxiliary Resonant Commutated Pole Inverter, Record of IEEE PESC, 1997, pp [lo] I. Barbi and D. C. Martins, A True PWM Zero Voltage Switching Pole with Very Low Additional RMS Current Stress, Record of IEEE PESC, 1991, pp [ 111 Xiaoming Yuan, Soft Switching Techniques for Multilevel Inverters, Ph. D Thesis, INEPUFSC Brazil, May