A Comparison Fo Three Level Converters Versus Two Level Converters

download A Comparison Fo Three Level Converters Versus Two Level Converters

of 11

description

A Comparison Fo Three Level Converters Versus Two Level Converters

Transcript of A Comparison Fo Three Level Converters Versus Two Level Converters

  • IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 41, NO. 3, MAY/JUNE 2005 855

    A Comparison of Three-Level Converters VersusTwo-Level Converters for Low-Voltage Drives,

    Traction, and Utility ApplicationsRalph Teichmann and Steffen Bernet, Member, IEEE

    AbstractThis paper evaluates three-level topologies as alterna-tives to two-level topologies in converters for low-voltage applica-tions. Topologies, semiconductor losses, filter aspects, part count,initial cost, and life-cycle cost are compared for a grid interface, aconventional drive application, and a high-speed drive application.

    Index TermsLife-cycle cost, multilevel converters, semicon-ductor losses.

    I. INTRODUCTION

    THE low-voltage power conversion market (nominalline-to-line voltage up to 690 V (IEC); 575 V(ANSI)) is almost exclusively satisfied by the conventionaltwo-level dc-voltage link hard-switching converter topology.Alternatives such as matrix converters or soft-switching con-verters have failed to penetrate this market. Technologicalprogress has been slow, relying mainly on semiconductordevice improvements and control intelligence refinements, aswell as on better integration and packaging techniques.

    In contrast, the medium-voltage power conversion marketoffers some diversity with the presence of a dc-current linkconverter and various dc-voltage link multilevel converter struc-tures. In particular, the multilevel converters became very suc-cessful in the past decade. While the development of the multi-level structures was mainly driven by the limited semiconductorvoltage blocking capability in conjunction with static and dy-namic voltage-sharing concerns, it turned out that the multileveltopology offered some additional benefits. Among those area superior harmonic spectrum for a given gate switching fre-quency, a lower overvoltage stress at cables and end windingsof transformers/motors, a lower common-mode voltage, andsubstantially lower semiconductor switching losses.

    The objective of this paper is the evaluation of the poten-tial benefits of a three-level topology in low-voltage power

    Paper IPCSD-04-075, presented at the 2003 Industry Applications SocietyAnnual Meeting, Salt Lake City, UT, October 1216, and approved for publica-tion in the IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS by the IndustrialPower Converter Committee of the IEEE Industry Applications Society. Manu-script submitted for review March 1, 2004 and released for publication February3, 2005. This work was supported by ABB Corporate Research, Germany.

    R. Teichmann is with GE Global Research, Niskayuna, NY 12309 USA(e-mail: [email protected]).

    S. Bernet is with the Fakultt IV Elektrotechnik und Informatik, Institutfr Energie- und Automatisierungstechnik, Fachgebiet Leistungselek-tronik, Technische Universitt Berlin, D-10587 Berlin, Germany (e-mail:[email protected]).

    Digital Object Identifier 10.1109/TIA.2005.847285

    converters. With the emerging need for higher switching fre-quencies in applications such as high-speed drives, a desire forsmaller and lighter (filter) components in grid-connected andtraction converters, a three-level-based low-voltage converterfamily appears to be one technologically attractive solution.A first indication of a commercial interest was stated in [1]and [2]. References [3] and [4] explore the utilization ofhigh-volume automotive components to build cost-effectivethree-level low-voltage drive converters. Reference [5] showsthe efficiency gains of three-level converters over two-levelconverters in wind power applications with a high share ofpartial load operation. A three-level converter technology isdiscussed for three typical application cases. Insulated gatebipolar transistor (IGBT) semiconductor losses and switchingfrequency boundaries are discussed for state-of-the-art IGBTtechnology. Initial cost and life-cycle cost comparisons willconclude the evaluation.

    II. TOPOLOGIES

    DC-voltage link multilevel converters were proposed inseveral configurations. The diode-clamped multilevel con-verter (DCMLC) [6], also known as the neutral-point-clampedmultilevel converter, the flying-capacitor multilevel converter(FCMLC) [6], also known as the imbricated cell converter, andthe series-connected/cascaded multilevel converter (SCMLC)are distinguished [6]. A comparison of these topologies waspresented in [6] and [7]. The common roots of DCMLC andthe FCMLC topologies were shown in [8]. It should be addedthat the total semiconductor losses and the terminal voltagespectrum of a three-level diode-clamped converter and athree-level flying-capacitor converter are exactly the same forany given operating point in terms of modulation index anddisplacement angle. Unidirectional and partially bidirectionalconverters as shown in [9] also feature exactly the same totalsemiconductor losses and terminal voltage spectrum if operatedwithin functional boundaries. Despite using different currentpaths, the device number and device types conducting duringeach switching state and their duty cycles remain the same.Only the distribution of the semiconductor losses among thesemiconductor modules is different. This assumes a comparablemodulation technique, a neutral point clamp/auxiliary diodetechnology similar to that found in the inverse diodes of themain switches, and disregards the influence of parasitic circuitelements.

    0093-9994/$20.00 2005 IEEE

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • 856 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 41, NO. 3, MAY/JUNE 2005

    Fig. 1. Three-level NPC converter (DCMLC).

    Fig. 2. Conventional two-level converter.

    A useful implementation of the SCMLC starts with afive-level topology featuring four times the semiconductors/gatedrives compared to a conventional two-level converter. This isbased on the desire for the optimum utilization of the inevitableisolation transformers. The DCMLC is best implemented asa three-level converter; higher order levels typically require aseries connection of the clamp devices. The FCMLC is subjectto a more complex startup routine and a higher number ofcapacitors with their detrimental effects on reliability. Basedon these considerations a three-level DCMLC (Fig. 1) is setagainst a two-level converter (Fig. 2) for the low-voltage powerconverter tradeoff analysis.

    III. DEFINITION OF EVALUATION CRITERIALow-voltage dc/ac power conversion for drives and grid con-

    verters is usually required for nominal three-phase rms voltagelevels of 200230 V, 380460 V, or 575690 V. These voltageranges correspond to IGBT voltage classes (two-level structure)of 600, 1200, and 1700 V, respectively. A three-level converterfor similar terminal voltages is theoretically to be fitted with300-, 600-, and 850-V modules. The topological evaluationwas based on the 1200-V/600-V IGBT technology (two/threelevel) due to the widespread availability of semiconductors ofthe same device technology and similar packaging. IGBTs with300- and 900-V ratings are available as discrete devices [10],and can easily be packaged in modules if a general technologytransition to three-level converters is desired. Fitting both

    converters with the same device technology eliminated devicetechnology dependency. Representing the main competingtechnologies, conventional and trench-gate nonpunch-through(NPT) IGBTs (EUPEC) [11] and trench-gate punch-through(PT) IGBTs (MITSUBISHI) [12] were included. With a ma-jority of three-phase low-voltage converters in a power levelbetween 10 kVA1 MVA, the evaluation is based on a 100-kVAconverter. Common to both the three-level and two-leveltopologies is a nominal grid/motor voltage of 400 Vrms anda dc-bus voltage of V. The same displacementpower factor (DPF) is set in both topologies. Two-level andthe three-level converters are controlled by an asynchronous,continuous, sine-triangle modulation scheme with added thirdharmonics. Both converter topologies feature the same carrierfrequency yielding the first carrier frequency band at the samefrequency in the ac voltage spectrum.

    The following converter applications were compared: 1) a100-kVA 400-V 50-Hz grid converter; 2) an inverter for a con-ventional 75-kW 400-V NEMA B induction motor; and 3) aninverter for a 90-kW 400-V 1.2-kHz high-speed drive.

    IV. EVALUATION OF SEMICONDUCTOR LOSSES

    Semiconductor losses are a central evaluation criterion fora topology due to their direct correlation with virtually allother electrical and mechanical converter parameters. Apartfrom gate control characteristics and commutation circuitparameters, the semiconductor device losses in three-phasevoltage-source dc/ac converters depend on the device tech-nology, the dc voltage and ac current levels ( ),the junction temperature , the switching frequency , thedisplacement phase angle (displacement power factor), and themodulation index . The modulation index is defined as peakline-to-line voltage over dc voltage with a definition range of01. A displacement phase angle of zero (unity displacementpower factor) designates a real power flow from dc to ac, i.e.,inverter operation.

    A. Device Technology/ Switching FrequencyFor a given dc-link voltage the devices in the three-level

    converter require half the blocking voltage capabilities of thatof a two-level converter, featuring superior conduction andswitching characteristics for any given current. For example,600-V IGBTs feature on-state voltages that are roughly 10%lower than that of a 1200-V IGBT at the same rated current andtechnology (Table I). Similarly, the switching loss energies ofa 600-V IGBT with the same device technology and currentare smaller by a factor of 35. This implies that, theoretically,even a series connection of two 600-V devices features only40%60% of the switching losses of one 1200-V IGBT whileincreasing the conduction losses by less than twice. Similarhigh-loss energy differences exist between the voltage classes1200 and 1700 V.

    To compare a two-level converter and a three-level converterthe semiconductor losses were calculated using a MATLAB-based simulation tool calculating the switch states and the in-stantaneous device voltages and currents over one period of theoutput frequency at a sampling rate of s. Conduction

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • TEICHMANN AND BERNET: COMPARISON OF THREE-LEVEL CONVERTERS VERSUS TWO-LEVEL CONVERTERS 857

    TABLE IIGBT LOSS DATA FOR MITSUBISHI (PT-IGBT) AND EUPEC (NPT-IGBT) DEVICES [10],[11]

    TABLE IISEMICONDUCTOR LOSS CHARACTERISTICS AND COEFFICIENTS

    losses are attributed to devices being in the on-state. Switchinglosses are distributed to the devices involved in a commutationwhenever a state transition takes place. The losses during thesampling period and the switching loss energies were calculatedby

    (1)(2)(3)

    (4)

    (5)

    whereactive device on-state loss energy in sampling period;passive device on-state loss energy in samplingperiod;active device turn-on loss energy;active device turn-off loss energy;

    passive device turn-off loss energy;instantaneous commutation voltage;device manufacturers test voltage;instantaneous device current;sampling period;curve-fitting parameter.

    On-state voltages and loss energies as a function of the de-vice current for the IGBT and diode were obtained from man-ufacturer data sheets [11], [12]. The curve-fitting parametersare summarized in Table II. For simplicity, the variation of thesemiconductor losses as a function of junction temperature isneglected; the values specified for a junction temperature of

    C were assumed. Details of the semiconductor lossevaluation tool can also be found in [13] and [14]. For the semi-conductor loss and switching frequency boundary evaluation theconverters were assumed to be operated in steady state at a con-stant case temperature of C and a maximum junctiontemperature of C.

    Fig. 3 shows the total semiconductor losses in a three-phasetwo- and three-level inverter as a function of device technology(PT, NPT, trench/conventional, fast/low ) and carrier fre-quency at one operating point ( V, ,

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • 858 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 41, NO. 3, MAY/JUNE 2005

    Fig. 3. Total semiconductor losses as a function of carrier frequency andparameter device technology (sine-triangle modulation with added thirdharmonic, V = 400 V, V = 700 V, T = 80 C, T = 125 C,DPF = 0:9, I = 147 A, 200-A IGBTs).

    Fig. 4. Equi-loss line (two-level/three-level converter) as a function of loadcurrent and parameter semiconductor technology (sine-triangle modulation withadded third harmonic,m = 0:95, V = 700V,DPF = 0:9, I = 147A).

    , A). Both 1200- and 600-V moduleshave the same nominal current rating. A device developed forfast switching operation (FF200R12KS4, mJ,

    mJ, V at A) was included for com-parison purposes.

    Despite the higher conduction losses (two devices in currentpath) the three-level topology quickly features lower totalsemiconductor losses than conventional and fast IGBTs ina two-level topology as the switching frequency increases. Atthis operating point the three-level topology features fewer totalsemiconductor losses at switching frequencies above 4/5 kHzfor trench-gate PT as well as planar NPT IGBTs. Clearly,the recent introduction of the trench-gate technologies shiftsthis crossover point to smaller switching frequencies in favorof the three-level converter. Some of the loss curves are notextended to the upper switching frequency range, indicatingthat the losses cannot be dissipated from the device withinthe thermal specifications ( C, C).

    Fig. 5. Semiconductor loss distribution of two-level converter as a function ofdisplacement power factor (sine-triangle modulation with added third harmonic,V = 400 V, V = 700 V, T = 125 C, I = 147 A, f = 10 kHz, 3IGBT: CM200DU24F).

    B. Load Current

    The magnitude of the load current in relation to the devicecurrent rating has a substantial influence on the switching fre-quency above which the three-level converter becomes superior.This crossover switching frequency is shifted to lower values asthe load current becomes smaller. This is due to the fact thaton-state losses decline faster ( quadratic dependency) thanswitching losses ( linear dependency) with falling device cur-rent and both loss categories are of roughly the same magni-tude. Fig. 4 shows an equi-loss line (identical semiconductorlosses of two- and three-level converters) of a dc/ac inverter( V, , , A) as a func-tion of the load current. For any given load current, switchingfrequencies above the lines are preferably implemented with athree-level converter. Most converters are typically operated ina load range between 0.40.7 per unit. A three-level topologytherefore becomes a very attractive solution for switching fre-quencies as low as kHz.

    C. Modulation Depth, Displacement Power Factor

    A variation of the modulation depth and the terminal dis-placement angle changes the total semiconductor losses andloss distribution among active and passive semiconductors.Figs. 58 depict the loss characteristics with the acronymsPconD, PconT, PonT, PoffT, and PoffD describing diode con-duction losses, IGBT conduction losses, IGBT turn-on losses,IGBT turn-off losses, and diode turn-off losses, respectively.

    In a two-level converter feeding a three-phase balanced loadthe losses of all diode chips are the same and the losses of allIGBT chips are the same [15], [16]. Fig. 5 shows that the totalconduction losses vary with the displacement power factor withmaximum conduction losses in the IGBT and the inverse diodesat and , respectively. The switching lossesof each individual chip are not affected by the displacementpower factor. References [15] and [16] have shown that the con-duction losses of the IGBT chips increase almost linearly with

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • TEICHMANN AND BERNET: COMPARISON OF THREE-LEVEL CONVERTERS VERSUS TWO-LEVEL CONVERTERS 859

    Fig. 6. Semiconductor loss distribution of three-level converter as a function ofdisplacement power factor (sine-triangle modulation with added third harmonic,V = 400 V, V = 700 V, T = 125 C, I = 147 A, f = 10 kHz, 6IGBT: CM200DU12F).

    Fig. 7. Semiconductor loss distribution of three-level converter in invertermode as function of modulation index (sine-triangle modulation with addedthird harmonic, V = 400 V, V = 700 V, DPF = 0:9, T = 125 C,I = 147 A, f = 10 kHz, 6 IGBT: CM200DU12F).

    increasing modulation depth while the conduction losses of theinverse diodes are declining.

    Diode and IGBT losses exhibit a similar pattern in the three-level converter. The sum of each type of switching losses, e.g.,diode switching losses, is also constant. However, the distribu-tion of both switching and conduction losses among the devicesis a function of the displacement angle (Fig. 6). Inverse diodes

    ( 1, 2, 3; 14) are subject to the same conductionlosses irrespective of the displacement angle. The loss depen-dency on the modulation depth is displayed in Figs. 7 and 8 forinverter and rectifier mode, respectively. The conduction lossesof the IGBTs in switches and remain constant while theconduction losses of the NPC diodes and the conduction lossesof the IGBTs in and vary with the modulation index in in-verter mode. In rectifier mode the conduction losses of the mainswitch inverse diodes increase with increasing while the con-duction losses of the NPC diodes and the IGBTs in and

    Fig. 8. Semiconductor loss distribution of three-level converter in rectifiermode as a function of modulation index (sine-triangle modulation with addedthird harmonic, V = 400 V, V = 700 V, DPF = 0:9, T = 125 C,I = 147 A, f = 10 kHz, 6 IGBT: CM200DU12F).

    Fig. 9. Common-mode voltages of two-level converter and three-levelconverter (f = 10 kHz, V = V ).

    decline. The influence of different continuous and discontinuousPWM schemes on the losses of a three-level DCMLC was dis-cussed in [17].

    V. PASSIVE COMPONENT CONSIDERATIONSPassive components are a substantial contributor to weight,

    cost, and losses in power converters. In fact, a substantial shareof the converter cabinet of industrial converters operating atswitching frequencies between 25 kHz is taken up by passivefilter components. The design implications for the ac filter, dcfilter, filter, and common-mode filter are discussed.

    A. Common-Mode FilterCommon-mode currents are considered to be the main

    reasons for bearing currents/failures and electromagnetic inter-ference (EMI) problems. Fig. 9 depicts the magnitudes of thecommon-mode voltages, which cause the common-modecurrents, impressed on the ac system for a two- and athree-level converter with a sine-triangle modulation. Clearly,the three-level converter has a much smaller common-modevoltage. Absolute values depend on the operating point and

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • 860 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 41, NO. 3, MAY/JUNE 2005

    modulation scheme. For the conventional sine-triangle modula-tion, a reduction of the switching frequency component of thecommon-mode voltage of the three-level converter by roughly25%30% was found. Furthermore, [18] proposes a three-levelconverter modulation scheme with which the common-modevoltages can be suppressed completely.

    B. FilterTransient overvoltages caused by the pulsewidth-modu-

    lation (PWM) switching operation are a major concern fortransformers/generators/motors connected with long cables.The overvoltage phenomenon due to fast switching transients(high ) in long cables has been discussed in severalpapers [19][26]. The overvoltage at the machine/transformerterminals is caused by a reflection of the voltage pulse of theinverter due to a mismatch in load and cable surge impedance.If the voltage pulse takes longer than one-third of the voltagerise time to travel from the inverter to the machine/transformera full reflection will occur at the machine/transformer terminalsand the voltage pulse amplitude will approximately double[22]. A 50% reduction in the pulse magnitude in the three-levelconverter will obviously reduce the overvoltage stress at theload terminals in a design without additional filter.

    The drastically reduced switching losses of the three-levelconverter can also be used to reduce the using the IGBTgate units. A potential avoidance of the filter at the con-verter terminals may justify the increased switching losses.

    C. AC FilterThe high-frequency content in the terminal voltage of the

    three-level converter is substantially smaller due to the three-level characteristic with an effective commutation voltage of50% of that of the two-level converter. The resulting harmoniccurrents are reduced by the same factor [24]. Because the high-frequency losses are a function of the square of the high-fre-quency content of the current, the application of a three-levelconverter at a given inductance, i.e., motor/generator, will re-sult in substantially smaller losses. Alternatively, the ac filterinductance can be smaller/less heavy for a three-level topologyto meet a given specification in terms of voltage or current har-monics.

    Specifically, for a three-phase system with a modulationscheme without a phase difference in the carrier signals, thecarrier switching frequency is a common-mode signal andis not present in the line-to-line voltages. Only the typicallysmaller side bands around the switching frequency need to beattenuated, additionally reducing filter size and losses.

    D. DC FilterTwo- and three-level converters show ripple current com-

    ponents on the dc side at side bands that are centered aroundthe switching frequency. The three-level converter additionallyshows a third harmonic (and other triplen harmonics) at the dcside [25]. The three-level converter requires a dc-link centerpoint management. Despite the fact that these points are intrin-sically stable, device asymmetries and variations in componentvalues will offset this balance.

    TABLE IIICOMPONENT COUNT/RATING COMPARISON

    VI. EVALUATION OF PACKAGING AND CONTROLS

    A. Packaging

    The design of the packaging is highly application dependent.Until higher production volumes justify an integration of com-plete converter legs, the three-level topology requires at least theadditional NPC diode modules. In cases where single 1.2-kVmodules can be replaced by a dual 600-V module (typically,when device current rating is set by thermal considerations) themain switch arrangement must not be modified.

    Specifically for an introduction of a three-level converter intoa dc/ac converter with 1.2-kV IGBTs, the arrangement of thedc-link capacitors must not be modified. Other voltage levelsrequire a re-design of the dc capacitor arrangement (Table III).

    The total size and weight of the converter will certainly ben-efit from the smaller cooling unit and smaller passive compo-nents. In certain applications a transition of the cooling methodto simpler convection cooling might reduce maintenance effortsand increase reliability.

    B. Sensors/Controls

    The number of dc-link voltage sensors, in some cases assimple as a resistive divider, is increased by a factor of two.The number of -processors remains the same. For example,the widely used fixed-point TMS320F24x series offers between716 dedicated PWM channels. An additional benefit existsfor larger manufacturers that can reuse their medium-voltageconverter control platform, producing a unified control platformfor low- and medium-voltage converters.

    The stabilization of the center point potential in the three-level diode-clamped converter imposes one additional controlcriteria. Control of the dc-link center point can be achieved un-less the converter is operated in overmodulation [26], [27]. Un-doubtedly, the biggest disadvantage of the three-level converteris the higher number of gate units with its accompanying isola-tion requirements. The number of gate drivers must be increasedby a factor of two. However, since each gate unit is subjectto only half the switching frequency on average, the total gatedrive power remains the same. Simplified gate drive units forthree-level structures were presented in [4].

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • TEICHMANN AND BERNET: COMPARISON OF THREE-LEVEL CONVERTERS VERSUS TWO-LEVEL CONVERTERS 861

    VII. RELIABILITY EVALUATION

    The three-level converter features a higher part count(Table III), which tends to compromise reliability. However,a closer analysis reveals other effects that partly balance thecomponent-count-based reliability numbers. Assuming a fullyintegrated converter leg and a dedicated control processor, theunavoidable additional components are the gate units, cabling,and sensors.

    The failure in time (FIT; FIT failure inoperating hours) rate of the semiconductors (

    ) is very small compared to other components in thesystem. Cabling, control electronics, and fan failures are thepredominant causes of failure in power electronics [28].

    All causes of failure are/can be positively influenced by ade-quate design choices. In particular, the lower losses of the three-level converter can reduce the ambient temperature of the gateunits (e.g., 1015 K) such that the number of failures is sta-tistically equal to that of a two-level converter, despite doublingthe number of units. The power and number of cooling fans, acomponent contributing highly to total system failures, can bereduced, in particular, in high-frequency applications. Similarly,the capacitor life expectancy can be increased due to the lowerripple current stress.

    A complete reliability assessment is case dependent; athree-level converter is not automatically inferior to a two-levelconverter.

    VIII. EVALUATION OF APPLICATION EXAMPLES

    A. Grid Converter

    Grid converters operate in rectifier and inverter modedepending on the application. Both types of operation arecharacterized by a relatively constant modulation index andpower factor. Variations are to be expected in the current dueto varying load conditions. Typical mains frequencies are50/60 Hz. Certain utility interconnect requirements regardingthe current quality and the behavior during grid irregularitiesmust be fulfilled, e.g., IEEE 519-1992. A switching frequencybetween 24 kHz is typically chosen at this power level, whichhas traditionally been regarded as a good compromise betweenfilter size/losses and converter efficiency.

    The following parameters were set for the purpose of theevaluation of a grid-connected load and a grid-connected gen-eration unit: kVA, V, V,

    , and kHz. Fig. 10 depicts thesemiconductor losses of the two- and three-level converter as afunction of the load current for the PT trench-gate technologyand a switching frequency of kHz. While for inverter op-eration the three-level converter is superior over the entire loadrange, in rectifier mode the three-level converter is superior upto a load current of about 0.8 p.u. at this switching frequency.The loss savings amount to roughly 30%35% at 0.2 p.u. currentfor both rectifier and inverter operation. During rectifier opera-tion at rated load the losses of the three-level converter are 8%higher than that for the two-level converter.

    Fig. 10. Total semiconductor losses during grid operation of converter (V =750 V, f = 4 kHz, m = 0:85, DPF = 0:98=0:98, T = 125 C, 3CM200DU24F, 6 CM200DU12F).

    Fig. 11. Total semiconductor losses during grid operation of converter (V =750 V, f = 8 kHz, m = 0:85, DPF = 0:98=0:98, T = 125 C, 3CM200DU24F, 6 CM200DU12F).

    For a switching frequency of kHz, the semiconductorlosses are displayed in Fig. 11. In this case, the three-level con-verter is clearly superior in both applications over the entire loadrange. For both rectifier and inverter operation the loss savingsamount to roughly 50%55% and 25% at 0.2 and 1 p.u., respec-tively. It should be noted that the two-level converter cannot beoperated at a switching frequency of kHz and the thermalrequirements of C and C with theMITSUBISHI 200-A devices. In contrast, the three-level con-verter can be operated with the 200-A devices up to a switchingfrequency of 28 and 10 kHz in inverter and rectifier mode, re-spectively.

    An LCL filter is designed to meet grid current quality criteriaset forth in the recommendations of IEEE 519-1992. The designrules are outlined in [16]. In order to attenuate the switchingharmonics below the limits required, the inductance values andfilter capacitor values (Y-connected) as shown in Table IV are

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • 862 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 41, NO. 3, MAY/JUNE 2005

    TABLE IVFILTER INDUCTANCE VALUES

    Fig. 12. Relative semiconductor loss difference for field-oriented operation ofinduction machine with two-/three-level converter (V = 700V, f = 4 kHz,T = 125 C, 3 CMX00DU24F, 6 CMY00DU12F, IM: P = 75 kW,V = 400 V, 50 Hz, X = X = 0:1, X = 2:7, R = 0:021, R =0:025).

    required. At this current level the weight for a three-phase inputinductor amounts to roughly 1 kg/10 H.

    B. Conventional Drives ConverterIn contrast to the grid application, the modulation depth, dis-

    placement power factor, and load current are highly variable. Astandard 75-kW 400-V 50-Hz induction motor (

    p.u., p.u., p.u., p.u.)is assumed. The current at rated operation amounts to

    A. The losses of the two- and three-level converter basedon PT trench-gate devices were calculated at distinct operatingpoints in the speedtorque plane. Field-oriented motor controlis assumed. For the carrier frequency of kHz the relativeloss difference

    (6)

    is shown in Fig. 12. The loss difference of the three-level andtwo-level converter is given relative to the losses of the two-levelconverter. The losses of the three-level converter are smaller inall operating conditions. Variation of torque has a significantimpact on the loss savings. The loss savings range from 3%(40 W) at rated torque and to 30% (120 W) at low torque.Variations in speed modify the relative losses only marginally.If the switching frequency is increased to 8 kHz, the three-level converter features substantially fewer losses in the entirespeedtorque plane. The loss savings amount to

    W and W for and ,

    respectively. For a switching frequency of 8 kHz, 200 A, 600 V,and 400 A, 1200-V PT trench-gate IGBTs are assumed. As dis-cussed earlier, the 1.2-kV devices are not able to dissipate thelosses at a switching frequency of kHz and the thermalparameters of C, C. In this case,400-A modules are needed and the installed switch power of thethree-level converter is only 62.5% compared to the two-levelconverter (see Table III).

    Apart from the converter losses, the losses in the sine filter orthe harmonic losses in the motor are smaller due to the lowerharmonic voltages of the three-level converter.

    C. High-Speed Drives ConverterThe elimination of the mechanical gear in applications such

    as compressors and small gas turbines is the main advantage ofhigh-speed drive technology. Both inverter and rectifier func-tionality are, thus, required. High-speed drives in the 100-kVApower class are typically implemented as permanent-magnetmachines operating at a fundamental frequency between0.53 kHz. For moderate harmonic losses in the machine,typically featuring a low leakage inductance, the minimumswitching frequency should be substantially above 912 timesthe fundamental frequency. A similar switching frequencyrange is desired for servo drives [2].

    For the purpose of an evaluation of a two- and three-level con-verter the switching frequency shall be set to kHz.The thermal boundaries were set to C and

    C. The analysis in the previous section has re-vealed that the two-level converter with 200-A devices cannotachieve this switching frequency within the thermal specifica-tions given. The chip area of a 400-A 1.2-kV module was neededto achieve the switching frequency while safely dissipating thesemiconductor losses. The investigation will be limited to ratedoperation in both rectifier and inverter mode ( V,

    , A, C, C).Fig. 13 shows the semiconductor loss distribution of the two-

    level and three-level converter with smallest possible IGBT cur-rent rating for inverter and rectifier operation. The lower lossesof the three-level converter permit the application of 200-A de-vices for the inverter ( kHz) and 300-A devices forthe rectifier application ( kHz).

    Additionally, the cooling effort needed for the three-levelconverter is substantially lower since the losses are roughly50% of that of the two-level converter. The semiconductor losssavings in actual terms amount to 2000 and 2100 W for inverterand rectifier operation, respectively. The losses of thefilter, in this case a highly damped system required by the highswitching frequency, will be substantially smaller.

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • TEICHMANN AND BERNET: COMPARISON OF THREE-LEVEL CONVERTERS VERSUS TWO-LEVEL CONVERTERS 863

    Fig. 13. Losses of two-level and three-level converter with smallest possibleIGBT current rating (given in brackets) for inverter and rectifier operation(V = 700 V, f = 20 kHz, m = 0:95, DPF = 0:98=0:98,T = 125 C, 3 CMX00DU24F, 6 CMY00DU12F).

    IX. COST EVALUATION

    Initial cost analyses and life-cycle cost analysis are funda-mentally different cost assessment methods. With increasingintegration of power electronics components in overall en-ergy supply solutions and continuing trends toward businessmodels in which sole entities are manufacturer/owner/operatora cost analysis over the entire product lifetime becomes ofhigher importance [14], [29]. A detailed cost analysis for thegrid-connected bidirectional converter operated at 4 and 8 kHzis presented.

    Before the initial costs of the two- and three-level convertersare to be assessed, the cost savings potential of 1 kW of savedlosses will be determined. The net cost (excluding recoverabletaxes) per kilowatthour paid by an industrial client in year 2000ranges from roughly 13 ct/kWh (Japan), 6 ct/kWh (Germany),4 ct/kWh (USA), down to 2 ct/kWh (South Africa) [15]. Thepresent worth of the future energy cost savings can be deter-mined as

    (7)

    whereaccumulated savings in present worth terms;energy cost savings in month ;annual interest rate;number of years.

    Disregarding potential energy price escalations or additionalcosts of debts, a converter system with 1 kW fewer losses oper-ated 8000 h per annum achieves a cost saving in each individualyear of roughly , , and in themajor markets of Japan, Germany, and the USA, respectively.For a converter operated over eight years with 8000 h per yearand an interest rate of 4% ( ), the energy cost saving per

    kilowatthour at present value are , , andfor Japan, Germany, and the USA, respectively.

    For the three-level converter technology to be attractive ona life-cycle cost basis the present worth savings must behigher than the sum of additional initial and development costsof the three-level technology. Additional benefits such as costsavings due to a weight/volume reduction in mobile applica-tions, simpler cooling method, reduced installation cost, etc.,are not included.

    Higher initial costs of the three-level converter are mostly at-tributed to the semiconductor cost. A market survey of semi-conductor prices (dual modules, year 2000) indicated that theprices for 10 000 units would roughly amount to 17 ct/A and9 ct/A for 1200- and 600-V IGBTs, respectively. This shows thattwo 600-V IGBTs are only slightly more expensive than one1200-V IGBT. Additionally, the 600-V NPC diodes were ac-counted for with 6 ct/A per diode. The total IGBT/diode modulecosts amount to $204 and $288 for the two-level and three-levelconverters fitted with 200-A devices, respectively. For an op-eration at 8 kHz, the two-level converter must be fitted with400-A devices, bringing the two-level converter cost to $408.Gate units were accounted for by $10 per channel. The com-plete controller was included at $50 independent of the numberof channels.

    The Austerlitz heat sinks RLS 250.16-500 and RLS 300.14 500 [30] were assumed, which were quoted at 400 USD andat 450 USD, respectively. Based on the loss calculationsdiscussed in the previous section the converters operating ata switching frequency of 4 kHz and the three-level converteroperated at 8 kHz were fitted with the RLS 250.16-500 heatsink. The two-level converter operated at 8 kHz was fitted withthe larger RLS300.14 type to achieve a comparable heat-sinktemperature at the higher losses.

    For a grid converter application, the immediate effect is areduction of the filter cost. Prices for a three-phase coil werequoted at roughly 60 ct H, and the price for three capacitorsat this voltage level is roughly 1 USD F. At 60 ct H in thiscurrent class the inductor cost amounts to $432 and $300 for thetwo-level and three-level converter (Table IV) operated at 4 kHz,respectively. The cost for the individual converter componentsare summarized in Table V for operation at 4 and 8 kHz. FromTable V, it is clear that the initial cost of the three-level converteris equal to or substantially lower than that of the two-level con-verter of the same power rating despite the higher cost for semi-conductors and gate/control units. This is mainly due to savingsin the filter elements and the heat sink.

    For a grid interface chosen to operate at 4 kHz the lossesduring rectifier operation close to rated load are 120 W higherin a three-level converter in comparison to a two-level con-verter (Fig. 10). However, the losses of the three-level converterare lower than that for the two-level converter by 40 W and50120 W for inverter operation and rectifier operation at par-tial load, respectively. From life-cycle cost perspective, a clearunderstanding of the operation schedule of the grid interface isneeded to make a decision on the optimum converter topology.

    Fig. 11 shows that the absolute loss savings amount to400 W for a switching frequency of kHz and rated

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • 864 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 41, NO. 3, MAY/JUNE 2005

    TABLE VCOMPONENT COST ANALYSIS

    load for both rectifier and inverter operation. This indicates thatsubstantial cost savings can be achieved throughout the entireproduct life cycle. Even if the initial costs of the three-levelconverter are slightly above those of the two-level converterthe three-level converter is paying off in high energy pricemarkets like Japan. At a switching frequency of kHz,the application of the three-level converter is also justifiedin all markets due to the higher initial semiconductor costin comparison to the two-level converter which needs 400-Adevices. Note that potential loss savings in the filter and themachine are not yet included. The same statement is valid forthe conventional converter driving an induction machine. A

    filter will, if needed, improve the cost balance in favorof the three-level converter. The utilization of a three-leveltopology for high-speed drives and other high-switching-fre-quency applications is recommended due to significant costreduction in both initial and operating costs.

    X. CONCLUSION

    Three-level topologies are highly attractive for low-voltagepower converters, specifically for applications with medium tohigh switching frequencies.

    1) The difference in switching loss energies between IGBTsof the same current class and commonly adjacent voltageclasses is a factor of 35. This is the key device character-istic enabling low switching losses.

    2) Due to the specific correlation of switching and conduc-tion losses with load current, a three-level topology is su-perior in terms of total semiconductor losses at switchingfrequencies as low as and beyond 23 kHz in practicalapplications. At switching frequencies above 5 kHz, thethree-level converter always features lower losses.

    3) Compared to the two-level converter, the three-level con-verter enables substantially higher switching frequencies( 4 ) applying devices of the same technology and cur-rent class in both converters.

    4) The introduction of low on-state voltage trench-gateIGBTs has a higher leverage effect on three-level con-verters than on two-level converters.

    5) Substantial reductions in filter size/weight and cost (30%) are possible due to the lower voltage harmonics

    in the three-level converter.6) Reliability concerns can be mitigated by adequate design

    choices. The reduction of semiconductor losses will re-duce the average temperature at the components and, thus,decrease the failure rate.

    7) The three-level converter is economically viable in high-energy-cost markets, even at low switching frequencies.

    REFERENCES[1] T. Tanaka. (2003) Yaskawa Tech. Rev. (Special Issue on Inverter Drive).

    [Online] Available: http://www.yaskawa.co.jp/en/technology/gihou/65-2/t01.htm

    [2] A. Buente, T. Oezgen, and T. Ortmann, Dreipunkt-wechselrichter fuerhochgeschwindigkeitsantriebe, in Conf. Rec. PCIM Europe, Nurem-berg, Germany, 2003.

    [3] B. A. Welchko, M. B. R. Correa, and T. A. Lipo, A cost effective three-level MOSFET inverter for low power drives, in Proc. IEEE IECON02,vol. 2, 2002, pp. 12671272.

    [4] S. Park and T. M. Jahns, A self-boost charge pump topology for agate drive high-side power supply, in Proc. CPES Seminar, 2002, pp.267272.

    [5] O. Al-Naseem, R. W. Erickson, and P. Carlin, Prediction of switchingloss variations by averaged switch modeling, in Proc. IEEE APEC00,vol. 1, 2000, pp. 242248.

    [6] Y. Shakweh and E. A. Lewis, Assessment of medium voltage PWMVSI topologies for multi-megawatt variable speed drive applications,in Proc. IEEE PESC99, vol. 2, 1999, pp. 965971.

    [7] R. Teodorescu, R. Blaabjerg, and J. K. Pedersen, Multi-level converters A survey, in Conf. Rec. EPE99, 1999, pp. 408418.

    [8] F. Z. Peng, A generalized multilevel inverter topology with self voltagebalancing, in Conf. Rec. IEEE-IAS Annu. Meeting, vol. 3, 2000, pp.20242031.

    [9] K. A. Corzine, J. R. Baker, and J. Yuen, Reduced parts-count multi-level rectifiers, in Conf. Rec. IEEE-IAS Annu. Meeting, vol. 1, 2001,pp. 589596.

    [10] (2000) IGBT Datasheet. IXYS Corp. [Online]. Available: www.ixys.com[11] (2000) IGBT Data Sheets: BSM200GB120DLC, BSM200GB06DLC,

    FF200R12KS4, FF200R12KE3, FF200R17KE3. EUPEC Corp. [On-line]. Available: www.eupec.com

    [12] (2000) IGBT Data Sheets: CM200DU24F, CM200DU12F. MIT-SUBISHI Corp. [Online]. Available: www.mitsubishichips.com

    [13] F. Blaabjerg, U. Jaeger, S. Munk-Nielsen, and J. K. Pedersen, Powerlosses in PWM-VSI inverter using NPT or PT IGBT devices, IEEETrans. Power Electron., vol. 10, no. 3, pp. 358367, May 1995.

    [14] R. Teichmann, Auxiliary resonant commutated pole converters, Ph.D.dissertation, Dept. elect. Eng. Inf. Technol., Dresden Univ. Technol.,Dresden, Germany, 2002.

    [15] S. Bernet and R. Teichmann, Potential and risks of the matrix converterfor modern AC-drives, in Conf. Rec. COBEP, Belo Horizonte, Brazil,1997, pp. A1A10.

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

  • TEICHMANN AND BERNET: COMPARISON OF THREE-LEVEL CONVERTERS VERSUS TWO-LEVEL CONVERTERS 865

    [16] S. Bernet, S. Ponnaluri, and R. Teichmann, Design and loss compar-ison of matrix converters and voltage source converters for modern ACdrives, IEEE Trans. Ind. Electron., vol. 49, no. 2, pp. 304314, Apr.2002.

    [17] T. Brckner and D. G. Holmes, Optimal pulse width modulationfor three-level inverters, in Proc. IEEE PESC03, vol. 1, 2003, pp.165170.

    [18] K. R. M. N. Ratnayake and Y. Murai, A novel PWM scheme to elim-inate common mode voltage in three-level voltage source inverter, inProc. IEEE PESC98, vol. 1, 1998, pp. 269274.

    [19] E. Persson, Transient effects in applications of PWM inverters to in-duction motors, IEEE Trans. Ind. Appl., vol. 28, no. 5, pp. 10951101,Sep./Oct. 1992.

    [20] P. van Poucke, R. Belmans, W. Geysen, and E. Ternier, Over-voltages ininverter fed induction machines using high frequency power electronicscomponents, in Proc. IEEE APEC94, vol. 1, 1994, pp. 542548.

    [21] A. H. Bonnett, Analysis of the impact of pulse-width modulated in-verter voltage waveforms on AC induction motors, IEEE Trans. Ind.Appl., vol. 32, no. 2, pp. 386392, Mar./Apr. 1996.

    [22] A. von Jouanne, D. Rendusara, P. Enjeti, and W. Gray, Filtering tech-niques to minimize the effect of long motor leads on PWM inverter fedAC motor drive systems, IEEE Trans. Ind. Appl., vol. 32, no. 4, pp.919926, Jul./Aug. 1996.

    [23] P. T. Finlayson, Output filters for PWM drives with induction motors,IEEE Ind. Appl. Mag., vol. 4, no. 1, pp. 4652, Jan./Feb. 1998.

    [24] P. Guggenbach, Pulse removal in sinusoidal PWM with constantswitching frequency to reduce switching frequency or current har-monics, in Proc. IEEE PESC98, 1998, pp. 294301.

    [25] F. Jenni and D. Wuest, Steuerverfahren Fuer Selbstgefuehrte Strom-richter. Zurich, Switzerland: vdf Hochschulverlag Zurich, 1995.

    [26] S. Ogasawara and H. Akagi, Analysis of variation of neutral point po-tential in neutral point clamped voltage source PWM inverters, in Conf.Rec. IEEE-IAS Annu. Meeting, 1993, pp. 965970.

    [27] C. Newton and M. Summer, Neutral point control for multi-levelinverters: Theory, design, and operational limitations, in Conf. Rec.IEEE-IAS Annu. Meeting, 1995, pp. 24512458.

    [28] O. V. Thorsen and M. Dalva, A survey of the reliability with an analysisof faults on variable frequency drives in industry, in Conf. Rec. EPE95,1995, pp. 10331038.

    [29] J. J. Stroker, What is the real cost of higher efficiency, IEEE Ind. Appl.Mag., vol. 9, no. 3, pp. 3237, May/Jun. 2003.

    [30] (2004). Austerlitz Electronic GmbH. [Online]. Available: www.auster-litz-electronic.de

    Ralph Teichmann received the MS. and Ph.D.degrees in electrical engineering from DresdenUniversity, Dresden, Germany, in 1997 and 2002,respectively.

    He was a Visiting Student at the University of Wis-consin, Madison, in 19951996 and a Research As-sistant at Nagasaki University, Japan, in 19981999.Since 1997, he has been an independent consultantfor various companies, including ABB Corporate Re-search and Philips Medical Systems. He is currentlywith GE Global Research, Niskayuna, NY. His re-

    search interests include high-power conversion, renewable energies, distributedgeneration, hard- and soft-switching converter topologies, as well as convertercontrols.

    Dr. Teichmann has been the recipient of several awards, including two schol-arships, one industrial fellowship, an IEEE paper award, and a German electricalengineering society publication award.

    Steffen Bernet (M97) was born in Ilmenau, Ger-many, in 1963. He received the Diploma degreefrom Dresden University of Technology, Dresden,Germany, in 1990, and the Ph.D. degree from the Il-menau University of Technology, Ilmenau, Germany,in 1995, both in electrical engineering. The subjectof his Ph.D. dissertation was the investigation ofpower semiconductors in soft-switching converters.

    He was a Development Engineer in the Depart-ment of Private Communication Systems at Siemensfrom 1994 to 1995. During 1995 and 1996, he was

    a Postdoctoral Researcher in the Electrical and Computer Engineering Depart-ment, University of Wisconsin, Madison. In 1996, he joined ABB Corporate Re-search, Heidelberg, Germany, where he led several strategic power electronicsand drives research projects for low-voltage and medium-voltage applications.He led the Electrical Drives Group at ABB Corporate Research from 1998 to2001. From 1999 to 2000, he was responsible for ABB research worldwide inthe areas of power electronic systems, electric drives, and electric machines. In2001, he joined the Berlin University of Technology, Berlin, Germany, as a Pro-fessor of Power Electronics. His main research areas are high-power convertertopologies, power semiconductors, and motor drives.

    Authorized licensed use limited to: UNIVERSITAT POLITCNICA DE CATALUNYA. Downloaded on January 26, 2010 at 12:33 from IEEE Xplore. Restrictions apply.

    tocA Comparison of Three-Level Converters Versus Two-Level ConverteRalph Teichmann and Steffen Bernet, Member, IEEEI. I NTRODUCTIONII. T OPOLOGIES

    Fig.1. Three-level NPC converter (DCMLC).Fig.2. Conventional two-level converter.III. D EFINITION OF E VALUATION C RITERIAIV. E VALUATION OF S EMICONDUCTOR L OSSESA. Device Technology/ Switching Frequency

    TABLE I IGBT L OSS D ATA FOR MITSUBISHI (PT-IGBT) AND EUPEC (NPTTABLE II S EMICONDUCTOR L OSS C HARACTERISTICS AND C OEFFICIENTSFig.3. Total semiconductor losses as a function of carrier freqFig.4. Equi-loss line (two-level/three-level converter) as a fu

    Fig.5. Semiconductor loss distribution of two-level converter aB. Load CurrentC. Modulation Depth, Displacement Power Factor

    Fig.6. Semiconductor loss distribution of three-level converterFig.7. Semiconductor loss distribution of three-level converterFig.8. Semiconductor loss distribution of three-level converterFig.9. Common-mode voltages of two-level converter and three-leV. P ASSIVE C OMPONENT C ONSIDERATIONSA. Common-Mode FilterB. $dv/dt$ FilterC. AC FilterD. DC Filter

    TABLE III C OMPONENT C OUNT /R ATING C OMPARISONVI. E VALUATION OF P ACKAGING AND C ONTROLSA. PackagingB. Sensors/Controls

    VII. R ELIABILITY E VALUATIONVIII. E VALUATION OF A PPLICATION E XAMPLESA. Grid Converter

    Fig.10. Total semiconductor losses during grid operation of conFig.11. Total semiconductor losses during grid operation of conTABLE IV F ILTER I NDUCTANCE V ALUESFig.12. Relative semiconductor loss difference for field-orientB. Conventional Drives ConverterC. High-Speed Drives Converter

    Fig.13. Losses of two-level and three-level converter with smalIX. C OST E VALUATION

    TABLE V C OMPONENT C OST A NALYSISX. C ONCLUSIONT. Tanaka . (2003) Yaskawa Tech. Rev. (Special Issue on InverterA. Buente, T. Oezgen, and T. Ortmann, Dreipunkt-wechselrichter fB. A. Welchko, M. B. R. Correa, and T. A. Lipo, A cost effectiveS. Park and T. M. Jahns, A self-boost charge pump topology for aO. Al-Naseem, R. W. Erickson, and P. Carlin, Prediction of switcY. Shakweh and E. A. Lewis, Assessment of medium voltage PWM VSIR. Teodorescu, R. Blaabjerg, and J. K. Pedersen, Multi-level conF. Z. Peng, A generalized multilevel inverter topology with selfK. A. Corzine, J. R. Baker, and J. Yuen, Reduced parts-count mul

    (2000) IGBT Datasheet . IXYS Corp. [Online] . Available: www.ixy(2000) IGBT Data Sheets: BSM200GB120DLC, BSM200GB06DLC, FF200R12(2000) IGBT Data Sheets: CM200DU24F, CM200DU12F . MITSUBISHI CorF. Blaabjerg, U. Jaeger, S. Munk-Nielsen, and J. K. Pedersen, PoR. Teichmann, Auxiliary resonant commutated pole converters, Ph.S. Bernet and R. Teichmann, Potential and risks of the matrix coS. Bernet, S. Ponnaluri, and R. Teichmann, Design and loss compaT. Brckner and D. G. Holmes, Optimal pulse width modulation forK. R. M. N. Ratnayake and Y. Murai, A novel PWM scheme to eliminE. Persson, Transient effects in applications of PWM inverters tP. van Poucke, R. Belmans, W. Geysen, and E. Ternier, Over-voltaA. H. Bonnett, Analysis of the impact of pulse-width modulated iA. von Jouanne, D. Rendusara, P. Enjeti, and W. Gray, Filtering P. T. Finlayson, Output filters for PWM drives with induction moP. Guggenbach, Pulse removal in sinusoidal PWM with constant swiF. Jenni and D. Wuest, Steuerverfahren Fuer Selbstgefuehrte StroS. Ogasawara and H. Akagi, Analysis of variation of neutral poinC. Newton and M. Summer, Neutral point control for multi-level iO. V. Thorsen and M. Dalva, A survey of the reliability with an J. J. Stroker, What is the real cost of higher efficiency, IEEE (2004). Austerlitz Electronic GmbH. [Online] . Available: www.au