7458 ieee transactions on power …pefcl.snut.ac.kr/upload/study01/194789bbe98cd1a65eb… · ·...

7458 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 32, NO. 10, OCTOBER 2017

Reduced Active Switch Front-End MultipulseRectifier With Medium-Frequency

Transformer IsolationJose Juan Sandoval, Student Member, IEEE, Harish Sarma Krishnamoorthy, Member, IEEE,

Prasad N. Enjeti, Fellow, IEEE, and Sewan Choi, Senior Member, IEEE

Abstract—This paper presents a reduced switch count multi-pulse rectifier with medium-frequency (MF) transformer isolation.The proposed topology consists of a three-phase push–pull basedac to dc rectifier with a MF ac link employing two active switches. Athree-phase, five-limb, multiwinding MF transformer is employedfor isolation. The secondary side of the transformer is connected ina zig-zag configuration and is fed to two six-pulse diode rectifiers,achieving 12-pulse rectifier operation. The primary advantage ofthe proposed system is reduction in size/weight/volume comparedto the conventional 60 Hz magnetic transformer isolation rectifiersystem. Operating the transformer at 600 Hz is shown to result inthree times reduction in size. Furthermore, the proposed systememploys only two active semiconductor switching devices operatingunder a simple pulse width modulation scheme. Also, the zig-zagtransformer connection helps to balance leakage inductance on thesecondary side. Detailed analysis, simulation, and experimental re-sults on a 208Vl−l , 3.15 kW laboratory prototype are presented tovalidate the performance of the proposed approach.

Index Terms—Medium-frequency (MF) isolation, multipulserectifier, power density, three-phase ac–dc power conversion.

I. INTRODUCTION

MULTIPULSE rectifier systems are used in a wide varietyof applications in the industry [1], [2]. Both isolated and

nonisolated transformer configured multipulse rectifier systemshave been in use [3]–[6]. The primary advantage of multipulserectifier systems is high-quality dc-output voltage with simulta-neous elimination of low-frequency harmonic currents at the in-put utility terminals. In particular, 12-pulse and 18-pulse rectifiersystems result in input current total harmonic distortion (THD)less than 16%, thereby facilitating compliance with the IEEE

Manuscript received September 28, 2016; accepted November 4, 2016. Dateof publication November 24, 2016; date of current version May 9, 2017. Thiswork was supported in part by the National Research Foundation of Koreafunded by Korea Government under Grant 2014R1A2A2A01003724 and in partby the Overseas Research Professor Program of Seoul Tech. Recommended forpublication by Associate Editor T. Shimizu.

J. J. Sandoval, H. S. Krishnamoorthy, and P. N. Enjeti are with the Departmentof Electrical and Computer Engineering, Texas A&M University, College Sta-tion, TX 77843 USA (e-mail: [email protected]; [email protected]; [email protected]).

S. Choi is with the Department of Electrical and Information Engineering,Seoul National University of Science and Technology, Seoul 139-743, SouthKorea (e-mail: [email protected]).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TPEL.2016.2632717

519 harmonic current limits [7]. Nevertheless, much of thesesystems employ low-frequency (50/60 Hz) magnetics that con-tribute to large size/weight/volume particularly in high-powerapplications.

Reduction in size and weight is achieved with the half-powertransformer-based 12-pulse rectifier system as explained in [6].In this approach, a line frequency transformer processing halfthe power is employed and thus size remains a concern. Fur-thermore, this system is not suitable in applications requiringto step-down or step-up the voltage level [6]. Autotransformer-based multipulse rectifier systems are detailed in [7]–[10]. Re-duction in size and weight is achieved due to reduced kVA ratingof the autotransformer configuration. Autotransformer rectifierconfigurations with 0.18Po and 0.38Po ratings are reported in[7] and [8], respectively, compared with the 1.03Po rating ofthe conventional 12-pulse multiwinding transformer. However,these autotransformer configurations do not have galvanic iso-lation and employ 60 Hz magnetics. Thus, the use of line fre-quency magnetics continues to have a negative impact on thesize/weight of the rectifier system.

Modular three-phase power factor correction (PFC) ac to dcrectifier systems with high-frequency magnetics are detailed in[11]–[14]. Despite the improvement in input current quality,these systems employ multiple power conversions and employa high number of semiconductor devices. Furthermore, activePFC schemes require a significant sensing effort and are com-plicated to control. Also, electromagnetic interference (EMI)is a concern in these topologies due to their high switchingfrequency operation.

In contrast, the proposed topology, shown in Fig. 1, seeks toimprove over the existing 12-pulse ac to dc rectifier systems(isolated and autoconnected) by reducing size/weight/volumeand improving the performance. The advantages of the proposedsystem architecture are as follows:

1) the approach employs medium-frequency (MF) (600 Hz)magnetics that is shown to improve power density byreducing the size/weight of the system [15], [16];

2) the approach employs only two active semiconductordevices, this contributes to the system simplicity andreduced cost;

3) the 5th and 7th harmonics are eliminated in the input linecurrent over a wide range of output voltage control therebyresulting in reduction in input current THD;

0885-8993 © 2016 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

SANDOVAL et al.: REDUCED ACTIVE SWITCH FRONT-END MULTIPULSE RECTIFIER WITH MEDIUM-FREQUENCY TRANSFORMER 7459

Fig. 1. Proposed 12-pulse ac–dc rectifier system with MF transformer isolation employing two active switches. An example adjustable speed drive system isshown at the output.

4) output voltage can be controlled by varying the duty cycleof the pulse width modulated signal;

5) the system offers galvanic isolation between the in-put and output thereby minimizing the interference andcontributing to safety;

6) the approach is suitable for applications where the powerdensity, performance, and simplicity in control are ofparamount importance;

7) the secondary side of the MF transformer can be config-ured to higher pulse operation (i.e., 18-pulse and 24-pulse)to further improve input current quality.

These benefits make the topology suitable for operation up to480-V three-phase systems for powering loads up to 150 kW.The paper details the analysis and design of the proposed multi-pulse rectifier system along with a design example. Simulationand experimental results are discussed on a scaled-down labo-ratory prototype.

II. PROPOSED FRONT-END RECTIFIER WITH MFTRANSFORMER ISOLATION

The proposed system employing MF isolation with a mul-tiwinding transformer is shown in Fig. 1. The operationof the proposed topology can be divided in the followingstages:

1) diode rectifiers with clamp circuit;2) MF multiwinding transformer;3) 12-pulse diode rectifier;4) modulation scheme; and5) input current analysis.

A. Diode Rectifiers With Clamp Circuit

This part of the system is composed of two three-phasediode rectifiers, each connected to a high-voltage active switch(S1/S2) and a clamp circuit, which consists of a capacitor and ableeding resistor. The three-phase ac link across the transformerwindings is achieved by switching S1 and S2 complementarilywith 50% duty cycle as first described in [17] and as shownin Fig. 2(a) and (b). The overall switching function for 50%duty cycle is shown in Fig. 2(c). The primary windings of thezig-zag transformer can be divided into two sets that are 180°phase shifted in magnetic coupling, namely windings (Wa1 ,Wb1 , Wc1) and windings (Wa2 , Wb2 , Wc2). As shown in Fig.1, the center tap of each primary winding is connected to theutility grid. In addition, the switching terminals of windings(Wa1 , Wb1 , Wc1) are connected to a diode rectifier whose out-put is in turn connected to S1 . Similarly, the switching terminalsof windings (Wa2 , Wb2 , Wc2) are connected to a diode rectifierwhose output is in turn connected to S2 .

When S1 is gated ON and S2 is gated OFF, the switching termi-nals of windings (Wa1 , Wb1 , Wc1) are shorted through the dioderectifier while the switching terminals of windings (Wa2 , Wb2 ,Wc2) are open. In essence, the switching terminals of windings(Wa1 , Wb1 , Wc1) are shorted to the utility’s neutral point. There-fore, at this instant, the line-to-neutral voltages Van , Vbn , andVcn appear across windings Wa1 , Wb1 , and Wc1 , respectively.The voltages across windings (Wa2 , Wb2 , Wc2) have oppositepolarity compared to the voltages across windings (Wa1 , Wb1 ,Wc1) because they are 180° in magnetic coupling. Meanwhile,the induced voltages on the secondary side have the same po-larity as the voltages across windings (Wa1 , Wb1 , Wc1).


Fig. 2. (a) Gating function for S1 ; (b) gating function for S2 ; (c) overallsystem switching function; (d) Van , input line-to-neutral voltage; and (e) MFac link is the multiplication of Ssw and Van created by switching S1 and S2complementarily with 50% duty cycle.

When S1 is gated OFF and S2 is gated ON, the switching termi-nals of windings (Wa1 , Wb1 , Wc1) are open while the switchingterminals of windings (Wa2 , Wb2 , Wc2) are shorted and are at thesame potential as the utility’s neutral point. At this instant, theline-to-neutral voltages Van , Vbn , and Vcn appear across wind-ings Wa2 , Wb2 , and Wc2 , respectively. The induced voltageson the secondary side and the voltages across windings (Wa1 ,Wb1 , Wc1) have opposite polarity compared to the utility gridline-to-neutral voltages. The voltage polarity across each wind-ing changes as S1 and S2 are switched. Therefore, by switchingS1 and S2 at MF a three-phase ac link is created.

The three-phase ac link is simply a multiplication of theline-to-neutral voltages with a square wave switching function.Fig. 2(e) shows the mathematical ac voltage across winding Wa1when the system operates at 50% duty cycle. With a line-to-neutral voltage as in (1) and a square wave switching functiondescribed by (2), the resulting voltage across the transformerwinding Wa1 can be expressed as in (3). It is evident that thefrequency of the square wave switching function determines thefundamental frequency of the ac link created across the trans-former windings. The voltages across windings Wb1 and Wc1have a similar expression as in (3) but are 120° and 240° phaseshifted, respectively. The expression in (3) is valid for 50% dutycycle operation

Van =

√23VLLsin (ωst) (1)

Ssw =4π

∞∑n=1,3,5,...

1n

sin (nωsqrt) (2)

VWa1 =

√23VLL

∞∑n=1,3,5,...

2nπ

sin ({nωsqr ± ωs} · t) . (3)

Fig. 3. Multiwinding three-phase, five-limb transformer.

When the switching terminals of windings (Wa1 , Wb1 , Wc1)or (Wa2 , Wb2 , Wc2) are open, the clamp circuit provides apath for the energy stored in the leakage inductance of thewindings. For example, when S1 is OFF, the energy stored in theleakage inductance of windings (Wa1 , Wb1 , Wc1) is transferredto the capacitor, which clamps to the highest line-to-line voltage.Furthermore, in order to avoid overlap (instances in which bothS1 and S2 are ON) a dead time between S1 and S2 is necessary.During this dead time, the clamp circuit also provides a path forthe energy stored in the leakage inductance of the windings. Theenergy stored in the capacitor can be used to power a switchedmode power supply (SMPS). This SMPS can power gate drivecircuitry.

B. MF Multiwinding Transformer

The switching frequency of S1 and S2 determines the op-erating frequency of the transformer. The well-known trade-off between power density and efficiency must be consid-ered when selecting the switching frequency. Operating at MF(600–1000 Hz) enables the transformer to be reduced in size andprovides a good efficiency tradeoff, especially in high-power ap-plications [15]. Increasing the switching frequency to the kHzrange increases transformer core loss and switching losses. Fur-thermore, operating in the kHz range increases the input EMIand introduces the need for additional EMI filtering at the input[8]. Thus, the transformer is designed to operate at 600 Hz.

Selection of the appropriate magnetic materials is also criticalto achieve high-power density. For high-power MF applications,magnetic core materials such as ferrite, amorphous, and siliconsteel should be considered [18]. Due to its high saturation fluxdensity and relatively low cost [19], a silicon steel core materialwas selected to build the MF transformer for the scaled-downlaboratory prototype. The transformer can be built using threesingle-phase multiwinding transformers, or it can be a singlethree-phase multiwinding transformer. To achieve a more com-pact design, a single five-limb transformer is employed for isola-tion. The primary and secondary windings are wound around theinterior three limbs of the transformer as depicted in Fig. 3. Theexterior limbs can carry any unbalanced flux in the transformeravoiding core saturation [20].

Generally, in 12-pulse applications a star-delta winding con-nection is used in the secondary side to generate a net 30o

phase difference. However, the leakage inductances of the ter-minals feeding the diode bridge rectifiers are not equal becausethe turns-ratio is different in the star-delta connected windings.


Fig. 4. Phasor diagram to obtain two sets of three-phase voltages in thesecondary side with a net 30° phase shift.

Fig. 5. Zig-zag connection of secondary windings. Interior limbs of three-phase transformer, each limb has four secondary windings.

This issue leads to unequal current sharing between the diodebridges [7]. To mitigate this problem, the secondary side of theMF transformer in the proposed system is connected in zig-zag.With zig-zag arrangement, the leakage inductances on the sec-ondary side are balanced because the turns-ratio of the windingsper phase is the same.

The secondary windings are connected such that two sets ofthree-phase voltages with a net 30° phase difference are fedto the 12-pulse diode rectifier. One set of three-phase voltagesis displaced by +15° with respect to the primary windings,while the second set of three-phase voltages is displaced by–15° with respect to the primary windings. A phasor diagramof the primary side and secondary side voltages is shown inFig. 4. To accomplish a net 30o phase difference, the windingsturn-ratio must be set as defined by (4) and connected as shown

in Fig. 5

NP 1 : NP 2 : NS1 : NS2 : NS3 : NS4

= 1 : 1 :√

3 − 1√6

:

√23

:√

3 − 1√6

:

√23. (4)

The transformer’s turn-ratio can be obtained by performingphasor operations. As shown in Fig. 4, the secondary side outputvoltage Vat is desired to have a unity magnitude with a phaseshift of +15o with respect to the primary line-to-neutral voltageVan . Such a phasor can be obtained by adding a portion of phasorVan and a negative portion of phasor Vbn . The magnitudes of thephasor Van and Vbn correspond to the turn-ratio of the windingsand can be found by solving (5). Breaking (5) into its real andimaginary components yields a system of two equations andtwo unknowns and therefore the magnitudes of Van and Vbncan be obtained. Similar phasor operations can be performed toobtain the magnitudes of the phasors giving a secondary sideoutput voltage Vas with unity magnitude and a phase shift of–15o with respect to the primary line-to-neutral voltage. Thiszig-zag connection yields two set of voltages (Vabs , Vbcs , Vcas)and (Vabt , Vbct , Vcat), which have a 30o phase shift with respectto each other

Van∠0◦ + Vbn∠ − 120◦ = 1∠15◦. (5)

The voltampere (VA) rating of the multiwinding MF trans-former can be calculated using the rms voltage and rms currentof each winding under the assumption that the output current Idhas negligible ripple. The relation between the line-to-line rmsinput voltage VLL and the output dc voltage is expressed as

VLL =Vdc

1.35 · 2 . (6)

The rms value of the current through the primary windings is

Iwa1 = 1.11 · Id . (7)

The rms value of the current through the secondary windingsis

Ias =

√23· Id = 0.816Id . (8)

The rms value of the voltage across the primary windings is

Vwa1 =VLL√

3= 0.577VLL = 0.214Vdc . (9)

The rms voltage across the windings with turns-ratio NS1 is

Vs1 =√

3 − 1√6

· VLL√3

= 0.17VLL = 0.064Vdc . (10)

Similarly, the rms voltage across the windings with turns-ratioNS2 is

Vs2 =

√23· VLL√

3= 0.47VLL = 0.175Vdc . (11)


Then, the sum total of the VA product of the MF transformeris

V Atot = 6 · Iwa1 · Vwa1 + 6 · Ias (Vs1 + Vs2)

= 6.96VLLId = 2.58VdcId . (12)

Thus, the equivalent VA rating of the MF transformer is

V Aeq =12V Atot = 1.29VdcId = 1.29Po. (13)

Although the VA rating of the proposed transformer configu-ration is slightly higher than the conventional 12-pulse isolationtransformer (1.03Po ), the proposed transformer configurationdoes not carry the line frequency component. Instead, the trans-former operates at MF enabling a size reduction.

C. 12-Pulse Diode Rectifier

Two sets of three-phase voltages with a net 30o phase differ-ence are created with the zig-zag arrangement. Each set is fed toa six-pulse diode rectifier achieving 12-pulse rectification. Theoperation of the bridge rectifiers is similar to the conventionalline frequency 12-pulse configuration. The main difference isthat the diodes should be able to switch at the operating fre-quency. The dc link output voltage Vdc of the 12-pulse dioderectifier is calculated by (6). The design of the output inductorLout and the output capacitor Cout is similar to the conven-tional line frequency diode rectifier and depend primarily on therequirements of the load.

D. Modulation Scheme

A major advantage of the proposed system is that no closedloop control is required because the topology intends to replicatethe performance of a conventional line frequency transformer.Therefore, no sensing is required in the proposed scheme. Ifoutput voltage regulation is desired, the proposed system canoperate with simple variable duty cycle operation. This wouldrequire sensing the output dc voltage. Operation at 50% dutycycle provides the maximum MF ac link rms voltage. At dutycycles less than 50%, zero states are introduced in the MF ac linkdecreasing the overall output dc voltage. If the duty cycle is in-creased beyond 50%, short circuits occur across the transformerwindings. Consequently, the modulation becomes simple androbust in contrast to other systems, which employ complicatedmodulation strategies.

E. Input Current Analysis

By the virtue of the net 30o phase shift created by the zig-zagarrangement in the secondary side, the 5th, 7th, 17th, 19th, etc.,harmonics are eliminated in the utility line currents. Mathemati-cal analysis is provided for the line current in phase “a” throughFourier series and under the assumption of negligible ripple inthe output dc current Id . Ideally, the input current Ia dividesequally through the center tap windings Wa1 and Wa2 and canbe expressed as

Ia = Iwa1 + Iwa2 . (14)

The turns-ratio of the center-tap windings NP 1 and NP 2 isthe same. Thus, by VA balance the input current is expressedas

NP 1 (Iwa1 + Iwa2) = NP 1 · Ia = NS1Ias1 + NS2Ias2

+ NS3Ias3 + NS4Ias4 (15)

where NS1 , NS2 , NS3 , and NS4 are the turns-ratio of the sec-ondary windings associated with phase “a” and are determinedby (4) and (5). Similarly, Ias1 , Ias2 , Ias3 , and Ias4 are the cur-rents flowing through the secondary windings associated withphase “a”; these currents can be expressed in terms of the cur-rents flowing through the output six-pulse diode rectifiers asfollows:

Ias1 = −Isec1 B (16)

Ias2 = Isec1 A (17)

Ias3 = −Isec2 C (18)

Ias4 = Isec2 A . (19)

Similarly, the currents flowing through the output six-pulsediode rectifiers can be expressed as

Isec1 A = S1

(ωst − π

12

)· Ssw (20)

Isec1 B = S1

(ωst − π

12− 2π

3

)· Ssw (21)

Isec2 A = S1

(ωst +

π

12

)· Ssw (22)

Isec2 C = S1

(ωst +

π

12+

2π

3

)· Ssw . (23)

The –15o and +15o phase shift is evident from (20) and(22), respectively. Ssw corresponds to the switching functiondescribed by (2) and S1(ωst) corresponds to the quasi-squarewave nature of the current in six-pulse rectifiers and is expressedas

S1 (ωst) =∞∑

n=1,3,5,7

(4Id

nπcos

(nπ

6

))· sin (nωst) . (24)

Thus, the input current Ia is determined by (25), shown at thebottom of the next page. The switching function Ssw is a squarewave with duty cycle 0.5. Thus, squaring this function yields aconstant 1. After simplification Ia is described by (26), shownat the bottom of the next page. From (26) it is observed thatharmonics 5th, 7th, 17th, and 19th are eliminated as in conven-tional 12-pulse operation. This analysis demonstrates that theproposed front-end rectifier system with MF isolation can bea retrofit replacement of bulky line frequency transformers inconventional 12-pulse systems. The input current performanceis maintained while improving power density with a reducedactive switch count and simple modulation scheme. The the-oretical THD value of the input current (16%) is the same asin the conventional 12-pulse rectifier. To improve the currentperformance, an input passive filter must be included as shownin Fig. 1. Since the line input current in the proposed system has


TABLE IOPERATING CONDITIONS FOR THE SYSTEM IN FIG. 1

Grid voltage (line-to-line rms) 208 VGrid frequency 50 HzRectifier output voltage 560 Vd c

Rated power 10 kWSwitching frequency (fs q r ) 600 HzOutput inductor (Lo u t ) 2 mHOutput capacitor (Co u t ) 200 μF

Fig. 6. (a) Transformer winding voltage due to 50% duty cycle operation ofS1 and S2 at 600 Hz. Note the voltages across windings Wa1 , Wb1 , and Wc1are displaced by 120°; (b) line-to-line voltages Vabs ; and Vabt are 30° phaseshifted; and (c) rectifier input currents on the secondary side of the transformer.

identical harmonic spectrum as the line current in the conven-tional 12-pulse rectifier, the requirements of the passive filter interms of size remain the same.

III. SIMULATION RESULTS

A 208 VLL , 10-kW design example is considered to demon-strate the operation of the proposed front-end rectifier systemin Fig. 1. The parameters in Table I were used for simulation inPSIM. The simulations were performed without an input passivefilter to compare with a conventional 12-pulse rectifier. Addinga passive filter with Lf = 1.7 mH (0.13 p.u) and Cf = 140 μFresults in a line current with THD <5% and a system powerfactor >0.98.

As stated in Section II, a three-phase MF ac link is createdacross the transformer windings by switching S1 and S2complementarily. The three-phase MF ac link can be observedin Fig. 6(a); it is evident that the voltages across the windings(Wa1 , Wb1 , Wc1) are displaced by 120° from each other. Thevoltages across the windings (Wa2 , Wb1 , Wc2) are also a setof three-phase voltages with 120° phase shift; however, theyare opposite in polarity with respect to the voltages across the

Fig. 7. FFT of the voltage across transformer winding Wa1 . Fundamentalfrequency of operation is 600 ± 50 Hz enabling the use of MF transformers.Other components appear at 3fsq r ± 50 Hz.

Fig. 8. DC output voltage at 560 V. Individual rectified voltages have 30°phase shift as in conventional 12-pulse operation.

Fig. 9. (a) Input line current for phase “a”; 12-pulse operation is evident and(b) FFT of the input current verifies 12-pulse operation. Note: the dominantharmonics are 11th and 13th (550 Hz and 650 Hz). Simulated current THD is16%.

first set of windings. Fig. 6(b) shows the line-to-line voltages(Vabs and Vabt), which feed the 12-pulse diode rectifier. Thevoltages are 30° phase shifted with respect to each other asin conventional 12-pulse operation. The input currents of thediode rectifiers Isec1 A and Isec2 A as shown in Fig. 6(c) alsodemonstrate the 30° phase shift. The fast Fourier transform(FFT) of the voltage across winding Wa1 confirms MF oper-ation as shown in Fig. 7. The fundamental voltage frequencyoccurs at 600 ± 50 Hz; this enables the use of MF transformersthereby reducing the weight/size of the system [15]. The outputdc voltage and the individual rectified voltages Vrec1 and Vrec2are depicted in Fig. 8. The 30° phase shift is also noticeable inthe individual rectified voltages ensuring 12-pulse operation.

Fig. 9(a) shows the input current for phase “a”; 12-pulseoperation can be observed in the input current. The simulated

Ia =

[√23

{S1

(ωst − π

12

)+ S1

(ωst +

π

12

)}− −1 +

√3√

6

{S1

(ωst − π

12− 2π

3

)+ S1

(ωst +

π

12+

2π

3

)}]· Ssw

2 (25)

Ia =4√

3 · Id

π

[sin (ωst) − 1

11sin (11 · ωst) − 1

13sin (13 · ωst) +

123

sin (23 · ωst) +125

sin (25 · ωst) + · · ·]

(26)


Fig. 10. Magnetic field density plot of the proposed five-limb, three-phasetransformer when excited with an MF ac link. The 2-D dimensions of thetransformer are (34 cm × 24 cm). The model has a depth of 5 cm.

Fig. 11. Magnetic flux lines of the proposed five-limb, three-phase transformerwhen excited with an MF ac link. The flux lines concentrate in the interior threelimbs of the transformer.

THD is 16%. The FFT of the line current is shown in Fig. 9(b);the dominant harmonics are 11th and 13th as in conventional 12-pulse configuration. This is in agreement with the input currentanalysis and demonstrates that the proposed topology can be aretrofit replacement of the bulky line frequency multiwindingtransformer in conventional 12-pulse systems.

In addition to the simulations in PSIM, the magnetic behav-ior of the proposed MF transformer was simulated using AnsysMaxwell finite element analysis (FEA) software. The primarywindings of the modeled transformer were excited using thevoltage expression derived in (3). The material of the core wasselected to be M19 silicon steel, which has a saturation fluxdensity of 1.4 T at MF. A plot of the magnetic field densityfor a two-dimensional (2-D) simulation of the proposed trans-former is given in Fig. 10. From the simulation, it is shownthat the flux density is higher in the interior three-limbs of thetransformer as expected. The core of the transformer is shownto operate at 0.8 T, which is below the saturation region. InFig. 11, the corresponding magnetic flux lines plot is presented.The flux lines also concentrate along the interior three-limbsof the transformer. Using the M19 core loss data at 600 Hz,the FEA software is used to simulate the core losses of the MF

Fig. 12. Simulated core losses using FEA software. The average core loss is69 W.

TABLE IISEMICONDUCTOR DEVICES USED FOR POWER LOSS ANALYSIS

System component Part number Manufacturer

S1 /S2 IRG4PH50SPbF InfineonDiode clamp circuit C4D40120D Cree12-pulse diode rectifier C3D20060D Cree

Fig. 13. System power loss breakdown for a 10 kW design example. Theefficiency of the system is 96.5%.

transformer. The simulated core losses shown in Fig. 12 arecompared with the actual losses to evaluate the efficiency of theproposed topology as discussed in the next section.

IV. EFFICIENCY ANALYSIS OF MULTIPULSE AC–DCCONVERSION STAGE

The efficiency of the proposed reduced active switch mul-tipulse rectifier can be calculated by analyzing switching andconduction losses, and transformer core and winding losses. Theswitching/conducting characteristics of the semiconductor de-vices in Table II were used for power loss analysis. The currentsand voltages through the semiconductor devices are obtainedfrom the simulation results in Section III. From calculation, itis determined that the switching and conduction losses of theactive switches (S1/S2) account for 18% of the total losses.Similarly, the diode clamp circuit accounts for 42% of the sys-tem’s losses while the 12-pulse diode rectifier on the secondaryside accounts for 15% of the losses.


TABLE IIICOMPARATIVE EVALUATION OF THE PROPOSED RECTIFIER WITH OTHER SCHEMES

Topologies Conventional Half-power Active Three Proposed12-pulse 12-pulse 12-pulse [8] single-phase

PFC

Configuration ac–dc ac–dc ac–dc–dc ac–dc–dc ac–ac–dc

No. of active switches front-end – – – 3 2dc–dc – – 4 12 –Total – – 4 15 2

Galvanic Isolation Yes No No No YesSensing effort and modulation complexity None Low Low High LowPhase-shifting transformer VA rating (operation frequency) 1.03Po (line

frequency)0.5Po (linefrequency)

0.38Po (linefrequency)

– 1.29Po (MF)

Power density of phase-shifting transformer (output-watts/liters) 290 252∗ 352∗ – 980

∗.The volume for these topologies was obtained using the physical size of the phase-shifting transformer and the size of the matching inductors reported in [8].

An estimate of the transformer core losses was obtained us-ing FEA analysis. The FEA simulation yields an average coreloss of 69 W when the transformer primary windings are ex-cited with a three-phase MF ac link. The FEA simulation resultswere experimentally verified through an open circuit test of theMF transformer. Through experiments, a core loss of 80 W wasobtained. The difference between the simulated and tested corelosses occurs because the FEA model cannot account for allphysical effects in a core with laminations [21]. The total trans-former losses account for 25% of the system’s losses. Using finergrades of steel or amorphous materials would decrease the trans-former losses but the increase in cost must be considered [15].The breakdown of the system losses is shown in Fig. 13. Overall,the efficiency of the proposed system is calculated to be 96.5%.

V. COMPARATIVE EVALUATION OF THE PROPOSED

MULTIPULSE RECTIFIER

In this section, the proposed multipulse front-end rectifier iscompared with other existing schemes. The multipulse schemesconsidered for the evaluation include the conventional 12-pulserectifier, the half power 12-pulse rectifier, and two active tech-niques. The results of the comparison are shown in Table III.The proposed scheme uses only two active switches reducing thegate drive circuitry allowing for a compact system but the activeswitches must be rated for 2VLL . Due to the low number of activeswitches and simple modulation scheme, the proposed schemehas a low realization complexity compared to the three-phasemodular PFC scheme. Similar to the active 12-pulse scheme,the sensing effort and modulation complexity of the proposedscheme is low, which is attractive in industrial settings.

Among the five compared topologies, only in the proposedtopology the phase-shifting transformer (1.29Po ) operates atMF with galvanic isolation. Operating at MF enables the powerdensity (W/L) of the proposed phase-shifting transformer tobe the highest among the topologies employing phase-shiftingtransformers. Power density is defined in (27). For powerdensity calculation, the power rating and physical size of thetransformers/matching inductors reported in [8] were used. Theproposed zig-zag MF isolation transformer rated at 4 kW hasa 980 W/L power density, which is nearly 3.4 times larger than

the power density of the conventional line frequency 12-pulseisolation transformer.

Compared to the half-power 12-pulse scheme described in[6], the power density of the proposed transformer configura-tion is nearly 3.9 times larger. Similarly, compared to the active12-pulse scheme in [8] the power density of the proposed trans-former configuration is about 2.8 times larger. This advantagein power density makes the proposed topology very attractive inapplications, where size is a constraint and isolation is required.

Power Density =Output Power(W )

Volume(L). (27)

In addition, the volume of the proposed MF transformer iscompared to the volume of a line frequency transformer throughAnsys Maxwell FEA modeling. Two three-phase transformers,one operating at line frequency and the other at MF frequency,were modeled for the same output load (4 kW), input voltage(208Vl−l), and efficiency (∼98%) requirements. Also, the sameM19 silicon steel core material was considered for comparison.Considering the B–H curves of the M19 material at differentoperating frequencies, a peak flux density of 1.6 T with 105 pri-mary turns was used for the line frequency transformer designwhile a peak flux density of 0.8 T with 68 primary turns was usedfor the MF transformer design. A comparison of the size betweenthe three-phase line frequency transformer and the three-phaseMF transformer is shown in Fig. 14. For the same output load,input voltage, and transformer efficiency requirements, the linefrequency transformer has a volume of 13.8 L (51 cm × 36 cm× 7.5 cm) while the MF transformer has a volume of 4 L (34 cm× 24 cm × 5 cm). Thus, the volume of the MF transformer is30% of the volume of the line frequency transformer.

VI. EXPERIMENTAL RESULTS

In order to validate the proposed topology a scaled-downlaboratory prototype rated at 3.15 kW is built and tested.The input three-phase line-to-line voltage is 208 Vrms withfundamental frequency of 50 Hz. A small input passive fil-ter with Lf = 100 μH was used. The switching frequency ofthe active devices, namely S1 and S2 , is set to 600 Hz. Theclamp circuit is composed of a film capacitor Ccl = 10 μF andRcl = 10 kΩ. The gate drive signals for the active switching


Fig. 14. Size comparison of three-phase transformers. Line frequency trans-former (left) has a volume of 13.8 L while the MF transformer (right) has avolume of 4.1 L.

Fig. 15. MF 600 Hz transformer with silicon steel core (volume = 4.08 L anddimensions: 34 cm × 24 cm × 5 cm).

Fig. 16. Experimental results. Ch.1: Voltage across winding Wa1 . Ch.2: Volt-age across winding Wb1 . Ch.3: Voltage across winding Wc1 . Note: voltagesare displaced by 120° with respect to each other.

devices are generated using a Texas Instruments microcontroller.Within the microcontroller, a dead time of 2 µs is assigned to thegating signals of S1 and S2 to avoid overlap operation. The zig-zag MF transformer is designed to operate at the desired switch-ing frequency and is built using silicon steel material. Fig. 15shows the MF transformer used for the hardware experiments.

The experimental results are shown to be similar to simula-tion results. Fig. 16 shows the three-phase MF ac link acrosswindings Wa1 , Wb1 , and Wc1 . It is evident that the set ofthree-phase voltages are displaced by 120° (6.67 ms) from eachother. Operation at MF is confirmed from Fig. 17. The FFTof the voltage Vwa1 shows fundamental components at 550 Hzand 650 Hz enabling the transformer to operate at MF. From

Fig. 17. Experimental results. Ch. 1: Primary side voltage Vwa1 . Ch. M: FFTof Vwa1 shows fundamental components at 550 Hz and 650 Hz enabling MFoperation.

Fig. 18. Experimental results. Ch.1: Line-to-line voltage on the secondaryside Vabs . Ch.2: Line-to-line voltage on the secondary side Vabt . Ch.3: DCoutput voltage. Note: Vabs and Vabt are displaced by 30° from each other toachieve 12-pulse rectification.

Fig. 19. Experimental results. Ch.4: Line input current Ia . Ch. M: FFT of Ia

shows dominant harmonics to be 11th (550 Hz) and 13th (650 Hz) as in 12-pulseoperation. The 5th, 7th, 17th, and 19th harmonics are eliminated. The measuredTHD is 17%.

Fig. 18, 12-pulse operation is observed; the secondary side volt-ages Vabs and Vabt show a net 30° phase shift (1.67 ms) withrespect to each other. This figure also shows a smooth dc outputvoltage as in 12-pulse operation. The line input current Ia isshown in Fig. 19 along with its FFT. The frequency spectrumshows that the 5th, 7th, 17th, and 19th harmonics have beeneliminated confirming 12-pulse operation. The measured THDof the current is 17% but can be improved with an input passivefilter with Lf = 0.13 p.u. The measured 11th and 13th harmonicof the input current are about 10% and 7%, respectively, from


Fig. 20. Experimental results. Ch.1: Utility line-to-neutral voltage Van . Ch.4:Line input current Ia .

the fundamental. From Fig. 20, high displacement power factoris shown between the line-to-neutral voltage Van and the lineinput current Ia . Furthermore, Fig. 20 shows that the effect ofthe switching frequency on the utility voltage is minimal.

VII. CONCLUSION

This paper proposes a reduced switch multipulse rectifier withMF transformer isolation employing two active semiconductordevices. It has been shown that operating at a MF of 600 Hzthe transformer size is 1/3 of the equivalent 60 Hz design.A 10 kW design example has been shown to achieve 96.5%efficiency. Simulation and experimental results on a laboratoryprototype demonstrate the 12-pulse operation with high inputcurrent quality. Overall, the advantages of the system includehigh power density, reduced active switch count, and simplepulse width modulation scheme.

REFERENCES

[1] R. A. Hanna and S. Prabhu, “Medium-voltage adjustable speed drives-users’ and manufacturers’ experiences,” IEEE Tran. Ind. Appl., vol. 33,no. 6, pp. 1407–1415, Nov./Dec. 1997.

[2] G. Song, M. Heldwein, U. Drofenik, J. Minibock, K. Mino, and J. W.Kolar, “Comparative evaluation of three-phase high-power-factor AC-DCconverter concepts for application in future more electric aircraft,” IEEETrans. Ind. Electron., vol. 52, no. 3, pp. 727–73, Jun. 2005.

[3] Toshiba, Toshiba MV Drives, (2012, Apr. 20). [Online] Avail-able: http://www.toshiba.com/ind/data/tag_files/MTX_NEMA_3R_MV_Brochure_3002.pdf

[4] GE, TM-GE MV Drives, Apr. 20, 2015. [Online] Available: http://www.wmea.net/Technical%20Papers/GE%20Medium%20Voltage%20Drives.pdf

[5] M. Swamy, T. Kume, and N. Takada, “A hybrid 18-pulse rectificationscheme for diode front-end rectifiers with large DC bus capacitors,” IEEETrans. Ind. Appl., vol. 46, no. 6, pp. 2484–2494, Nov/Dec. 2010.

[6] M. M. Swamy, “Uncontrolled and controlled rectifiers,” in Power Systems,3rd ed. Boca Raton, FL, USA: CRC Press, 2012.

[7] S. Choi, P. N. Enjeti, and I. J. Pitel, “Polyphase transformer arrange-ments with reduced kVA capacities for harmonic current reduction inrectifier-type utility interface,” IEEE Trans. Power Electron., vol. 11, no. 5,pp. 680–690, Sep. 1996.

[8] M. M. Swamy, “An electronically isolated 12-pulse autotransformer rec-tification scheme to improve input power factor and lower harmonic dis-tortion in variable-frequency drives,” IEEE Trans. Ind. Appl., vol. 51,no. 5, pp. 3986–3994, Sept./Oct. 2015.

[9] S. Choi, “A three-phase unity-power-factor diode rectifier with activeinput current shaping,” IEEE Trans. Ind. Electron., vol. 52, no. 6,pp. 1711–1714, Dec. 2005.

[10] G. R. Kamath, D. Benson, and R. Wood, “A compact autotransformerbased 12-pulse rectifier circuit,” in Proc. 27th Annu.Conf. IEEE Ind. Elec-tron. Soc., 2001, Denver, CO, USA, 2001, pp. 1344–1349.

[11] M. Swamy and A. Balakrishnan, “Three, single-phase power factor cor-rection (PFC) boost converters for use with three-phase, 3-wire variablefrequency drive systems,” in Proc. 2015 IEEE Energy Convers. Congr.Expo., 2015, pp. 6993–7000.

[12] G. Spiazzi and F. C. Lee, “Implementation of single-phase boost power-factor-correction circuits in three-phase applications,” IEEE Trans. Ind.Electron., vol. 44, no. 3, pp. 365–371, 1997.

[13] Y. K. E. Ho, S. Y. R. Hui, and Y.-S. Lee, “Characterization of single-stage three-phase power-factor-correction circuits using modular single-phase PWM DC-to-DC converters,” IEEE Trans. Power Electron., vol. 15,no. 1, pp. 62–71, Jan. 2000.

[14] H. S. Kim, W. Baek, M. H. Ryu, J. H. Kim, and J. H. Jung, “The high-efficiency isolated AC-DC converter using the three-phase interleavedLLC resonant converter employing the Y-connected rectifier,” IEEE Trans.Power Electron., vol. 29, no. 8, pp. 4017–4028, Aug. 2014.

[15] M. Kang, P. N. Enjeti, and I. J. Pitel, “Analysis and design of electronictransformers for electric power distribution sytem,” IEEE Trans. PowerElectron., vol. 14, no. 6, pp. 1133–1141, Nov. 1999.

[16] H. Krishnamoorthy, P. Garg, and P. Enjeti, “Simplified medium/high fre-quency transformer isolation approach for multi-pulse diode rectifier front-end adjustable speed drives,” in Proc. 2015 IEEE Appl. Power Electron.Conf. Expo., 2015, pp. 527–534.

[17] R. Gupta, K. K. Mohapatra, N. Mohan, G. Castelino, K. Basu, andN. Weise, “Soft switching power electronic transformer,” United StatesPatent 8 446 743 B2, pp. 446–743, May 21, 2013.

[18] W. M. Colonel and T. Mclyman, Transformer and Inductor Design Hand-book, 3rd ed. Boca Raton, FL, USA: CRC Press, 2004.

[19] X. She, A. Q. Huang, and R. Burgos, “Review of solid-state transformertechnologies and their application in power distribution systems,” IEEE J.Emerg. Sel. Topics Power Electron., vol. 1, no. 3, pp. 186–198, Sep. 2013.

[20] C.-k. Leung, S. Dutta, S. Baek, and S. Bhattacharya, “Design consid-erations of high voltage and high frequency three phase transformer forSolid State Transformer application,” in Proc. 2010 IEEE Energy Convers.Congr. Expo., 2010, pp. 1551–1558.

[21] (2012). ANSYSMaxwellTutorial. Example (2D/3D) transient-core loss.

Jose Juan Sandoval (S’11) received the B.S. degreein electrical engineering in 2011 from Texas A&MUniversity, College Station, TX, USA, where he iscurrently working toward the Ph.D. degree in electri-cal engineering as a Gates Millennium Scholar.

He received the Texas A&M Dissertation Fel-lowship and the Thomas Powell’62 Fellowship. Hisresearch interests include high-power density three-phase ac–dc conversion, electric/hybrid vehicle bat-tery chargers, and renewable energy grid integration.

Harish Sarma Krishnamoorthy (S’12–M’15) re-ceived the B.Tech. degree in electrical and electronicsengineering from the National Institute of Technol-ogy, Tiruchirappalli, India, in 2008 and the Ph.D.degree in electrical engineering from Texas A&MUniversity, College Station, TX, USA, in 2015.

From 2008 to 2010, he was an Engineer in the Gen-eral Electric Energy, India. He holds a U.S. patent andseveral journal publications. His research interests in-clude high power density converter design, renewableenergy conversion, etc.


Prasad N. Enjeti (M’85–SM’88–F’00) received theB.E. degree from Osmania University, Hyderabad,India, in 1980, the M.Tech. degree from the In-dian Institute of Technology, Kanpur, India, in 1982,and the Ph.D. degree from Concordia University,Montreal, QC, Canada, in 1988, all in electricalengineering.

Since 1988, he has been a member of Texas A&MUniversity Faculty, College Station, TX, USA. He iswidely acknowledged to be a Distinguished Teacher,Scholar, and Researcher. He currently holds the

TI-Professorship in analog engineering and is the Associate Dean for academicaffairs in the College of Engineering. His research emphasis on industry-basedissues, solved within an academic context, has attracted significant externalfunding. He has graduated 29 Ph.D. students and 11 of them hold academicpositions in leading universities in the world. He along with his students havereceived numerous best paper awards from the IEEE Industry Applications andPower Electronics Society. His primary research interests include advancingpower electronic converter designs to address complex power management is-sues, such as active harmonic filtering, adjustable speed motor drives, wind andsolar energy systems, and designing high temperature power conversion systemswith wide band-gap semiconductor devices.

Dr. Enjeti received the Distinguished Achievement Award for teaching fromTexas A&M University in 2004. In 2012, he received the IEEE PELS R. DavidMiddlebrook Technical Achievement Award from the IEEE Power ElectronicsSociety.

Sewan Choi (S’92–M’96–SM’04) received the B.S.degree in electronic engineering from Inha Univer-sity, Incheon, South Korea, in 1985 and the M.S. andPh.D. degrees in electrical engineering from TexasA&M University, College Station, TX, USA, in 1992and 1995, respectively.

From 1985 to 1990, he was a Research Engineerwith Daewoo Heavy Industries. From 1996 to 1997,he was a Principal Research Engineer with Sam-sung Electro-Mechanics Co., South Korea. In 1997,he joined the Department of Electrical and Informa-

tion Engineering, Seoul National University of Science and Technology (SeoulTech), Seoul, South Korea, where he is currently a Professor. His research inter-ests include power conversion technologies for renewable energy systems andenergy storage systems, and dc–dc converters and battery chargers for electricvehicles.

He is an Associate Editor of the IEEE TRANSACTIONS ON POWER ELECTRON-ICS and the IEEE JOURNAL OF EMERGING AND SELECTED TOPICS IN POWER

ELECTRONICS.

7458 ieee transactions on power …pefcl.snut.ac.kr/upload/study01/194789bbe98cd1a65eb… · ·...

Documents