„This material is posted here with permission of the IEEE. Such permission of the IEEE does not in any way imply IEEE endorsement of any of ETH Zürich’s products or services. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promo-tional purposes or for creating new collective works for resale or redistribution must be obtained from the IEEE by writing to pubs-permission@ieee.org. By choosing to view this document you agree to all provisions of the copyright laws protecting it.”

Novel Concepts for Integrating the Electric Drive

and Auxiliary Dc-Dc Converter for Hybrid Vehicles

H. Plesko1), J. Biela1), J. Luomi2), J. W. Kolar1) 1) Power Electronic Systems Laboratory 2) Power Electronics Laboratory Swiss Federal Institute of Technology (ETH) Helsinki University of Technology ETH Zentrum, ETL/H23 P.O. Box 3000, FI-02015 TKK, Finland 8092 Zurich, Switzerland Email: plesko@lem.ee.ethz.ch Email: luomi@lem.ee.ethz.ch

Novel Concepts for Integrating the Electric Drive

and Auxiliary Dc-Dc Converter for Hybrid Vehicles

H. Pleskol), J. Bielal), J. Luomi2), J. W. Kolarl)1) Power Electronic Systems Laboratory

Swiss Federal Institute of Technology (ETH)ETH Zentrum, ETL/H238092 Zurich, Switzerland

Email: plesko@lem.ee.ethz.ch

Abstract- Cost, volume and weight are three major drivingforces in the automotive area. This is also true for hybrid vehicleswhich attract more and more attention due to increasing fuelcosts and air pollution.

In hybrid vehicles the energy distribution system causes asignificant share of the volume and the costs. In order toreduce the costs and the volume of this system two new conceptsfor integrating the dc-dc converter, which transfers the powerbetween the low and the high voltage bus, functionally in theinverter/drive system are presented in this paper. These conceptsare verified by simulations and experimental results.

I. INTRODUCTION

Growing fuel costs are just one factor that push the de-velopment of hybrid vehicles. Today, there are many carmanufacturers that produce hybrid vehicles and the researchactivities are growing. Figure 1 shows common voltage levelsand interconnections for the electric power system of a hybridvehicle.

Conventional Hybrid Electric Vehicles usually have twodifferent voltage levels. The 14 V dc bus is often supplied by a12 V battery over a battery charge/discharge unit. Traditionalloads such as a heater and lighting systems are connected tothis low voltage bus. In addition, there is a high voltage200 V...600 V dc bus which provides the necessary powerfor the propulsion system and which is connected to thelow voltage bus via a bidirectional dc-dc converter. Thisconverter needs to be galvanically isolated for safety reasons.The machine voltages are produced through a dc-ac converterconnected to the high voltage bus.

In recent years, the growing number of additional electricalloads forced the car industry to think about the introduction ofan additional 42 V dc bus powered by a 36V battery. Thereare other popular possibilities. The high voltage level could be

200 V .. 600 V Dc Bus

Figure 1: One possibility of a conventional electrical power distribu-tion system architecture for hybrid-electric vehicles.

2) Power Electronics LaboratoryHelsinki University of Technology

P.O. Box 3000, FI-02015 TKK, Finland

Email: luomi@lem.ee.ethz.ch

omitted and the propulsion system could be supplied with 42 V.In [1] this system is compared to the conventional variant. Asthe advantages of the high voltage solution outweigh that ofthe low voltage variant, it is assumed in the following that thepropulsion system is fed via a high voltage bus.

In order to reduce the costs of the electric system andincrease the power density the dc-dc converter, linking thelow and the high voltage bus, and the inverter system couldbe functionally integrated into the inverter/machine. Forexample, in [2] a possible integration of the dc-dc converterin the inverter system is presented, which is, however, notgalvanically isolated.

In [2] a galvanic isolation is not required since the highvoltage bar is omitted and the rotating system is fed from42 V. The zero-sequence voltage at the machine's star point isused to bidirectionally transfer power between the high andthe low voltage dc bus. The output voltage is controlled bychoosing an appropriate switching scheme. Connecting thezero-sequence system, the inverter operates as a buck-boostconverter and the machine's zero-sequence inductance servesas the buck-boost inductor.

There are similar concepts without galvanic isolation pre-sented in [3], [4], [5] and [6].As mentioned above, it would be desirable to feed the

propulsion system from the high voltage bus. The conceptproposed in [2] cannot be used here because the high voltageto low voltage converter needs to have a galvanic isolation forsafety reasons. In addition the output voltage VOtl of the dc-dc converter cannot be controlled in every operation regionof the drive system. If the inverter operates for example inthe six-step mode, the switching state of the inverter is totalydetermined. There is no degree of freedom to control the

j jnP-3

P21 Z

V jv~~~~Vin~ ~~MahnPX _1

s s

Figure 2: Concept of integration without isolation

1-4244-0714-1/07/$20.00 C 2007 IEEE. 1025

leg, leg2 leg3 seg5 ICg94 leg,, Iegt,

SIn 2F : i2S3 gtl -C

2%i jii2 Machinle Jki 1 iii2(SaIt d c v

(a) Separated dc-dc converter and inverter

leg, leg2 leg3 le94 leg,j leg

S], S42 S34 StSl

I~~~~P1 =IS| * MachinLe 14A

(b) Proposed integrated dc-dc converter: MFCS-I

vou

= I'

leg1 leg2 leg3 Ieg4 legf] legt2, _4. S. -S -2SIS SgS S2

s r Macin St XE

(c) Proposed integrated dc-dc converter: MFCS II

Figure 3: Multi Functional Converter Systems (MFCS).

output voltage. To overcome these problems an integrated highvoltage to low voltage dc-dc converter with galvanic isolationis proposed in this paper. Depending on the actual distributionsystem this converter can be used to convert 200...600V to14V or 42V.The starting point for the following considerations is a

conventional full bridge converter feeding a high frequencytransformer (see figure 3a), whose high frequency output isrectified to the output voltage V,,,t. The secondary siderectifier could be a conventional full bridge configuration or a

diode rectifier.In a first step, the fifth leg (leg5 in figure 3a) of the

conventional system is replaced by the zero-sequence voltage.The transformer is thereby connected to the star point and themidpoint of the primary side leg four (see figure 3b). Thisconcept is referred to as Multi Functional Converter System I

(MFCS-I) [9].In a second step the motor's iron could be used for inte-

grating the transformer as shown in figure 3c. This concept isreferred to as Multi Functional Converter System II (MFCS-II)[10]. Other possibilities to use the motor's iron or the inputinverter to integrate the transformer are presented in [7] and[8], but the proposed concept saves space because it does notneed a three phase rectifier.

In this paper, the basic functions of these two concepts are

explained and it will be shown by simulations and experimen-tal results that these concepts work well. The first variant, themulti functional converter system I (MFCS-I), is discussed insection II. In subsection IIA the basic operation principle for

the MFCS-1 is explained, followed by a detailed mathematicalflux analysis in subsection IIB. The theoretical results willbe verified experimentally by means of a 3 kW prototype insubsection IIC. In section III the results for the MFCS-11 arepresented. Finally, a conclusion is given in section IV.

II. MFCS-IIn the MFCS-1 the transformer is connected to the star

point and the output of the fourth leg (leg4, see figure 3b)thus eliminating one leg of the full bridge compared to theconventional system.

This section starts with an explanation of the basic operationprinciple of the MFCS-1 in subsection IIA by studying thezero phase voltage and introducing an equivalent circuit forthe dc-dc converter. After discussing the control in subsectionIIB, the flux is analyzed in detail in subsection IIC.

A. Basic Principle of Operation - MFCS-I

Both concepts are based on the utilization of the zero-sequence voltage whose definition and calculation are sum-marized before the details of the concepts are explained. Thezero-phase voltage of a three phase system is defined as

1Vz -(VP1+VP2+ VP3),3 (1)

where vpl, Vp2 and Vp3 are the voltages applied to themachine. Depending on the switching state of the inverter,the zero-sequence voltage related to the negative bus bar canbe calculated as

vz = 3 (on, + on2 + on3),

where on, is '1' if s. is open and s.1 closed, or else '0'.

(2)

50V4Y717

3.2ms 3.3n1s 3.4n1s 3.5nis 3.6s

(a) ConventionalV ,

50V-

OV3 2ms 3.3rs ms 3.6lst(b) Flat top

Figure 4: Simulation of the zero-sequence voltage for M 0.5,f= 10 kHz and Vi, = 50V for (a) conventional and (b) flat

top modulation.

Figure 4a shows the characteristic of the zero-sequencevoltage for the conventional modulation, where the modulationsignals are

mpi (t) = M cos(wt)- 6 cos(3wt)6 (3)

with Vi, = 50V and a modulation index M = 0.5. Thesimulation was repeated for the flat top modulation, where thephase with the highest absolute current value is clamped to thepositive bus bar if this current is positive, else to the negative(see figure 4b).

1026

The fourth leg (leg4, see figure 3b) is required because the The only degree of freedom is the relative time offset betweenzero-sequence voltage remains constant if the inverter operates the zero phase voltage and the fourth leg voltage, representedin six-step mode. Thus,7 an alternating voltage across the through the angle y5,4 (see figure 6).transformer can only be achieved by proper pulsing of S4 ands4. vin-t-F =i : :

Lo Ro

itl gVs4Figure 5: Equivalent circuit for the dc-dc converter.

For the dc-dc converter, the only important part of theinverter-machine subsystem is the zero-sequence voltage v,which can be calculated by (2). Thus, the integrated dc-dc converter can be modelled as shown in figure 5. Thetime dependent voltage source v, represents the zero-sequencevoltage and V,4 describes the voltage across SS4 (see figure3b). These two voltage sources are connected to an LRcircuit, modelling the machine's zero sequence impedance. vtrepresents the transformer voltage (see figure 3b).

It has to be mentioned that v , as it is defined in (2), canonly be measured if no zero sequence current flows, otherwiseit is an analytical parameter. The voltage measured betweenthe star point and the negative bus bar is equivalent to v -vlr.

This system is similar to a full bridge - rectifier system,but as the voltage v, applied to the system is neither aver-age free (related to the switching frequency) nor free fromeven harmonics, a conventional modulation scheme cannot beadopted.

In the actual implementation an active rectifier is used toguarantee that the voltage time product across the transformerremains below a predetermined level. The described dc-dcconverter is similar to a dual active bridge [11].

B. Control of MFCS-I

In the further discussion it is without loss of generality as-sumed, that the motor is driven with space vector modulation.Neglecting the required dead time, there is always one switchof the leg closed.

It is clear that the switching period average of the zerosequence voltage is not always :i related to the negativebus bar but may depend on wt, which is equivalent to themotor's angle. It can be calculated as

Vzavg = U (*P1 + P2 + 6P3), (4)3

where 6p, is the relative off time of si etc. depending onthe motor's angle and the actual modulation factor. To avoidsaturation of the transformer, it is necessary that the appliedvoltage does not show a local average value. This means forthe fourth leg that the relative off time 6,4 of SS4 is determinedby the zero sequence voltage as

6s4 = I( 1 + 6P2 + 6P3)- (5)

ov I----------

o 2it cot2iS4a

v P1nV,.,-ov -

0vSt 2 x

E~t

Figure 6: Converter waveforms: zero sequence voltage v,voltage V54 across 84 ,voltage vtl across stl, voltage Vt2 across

st2 and transformer voltage vt

Further the voltage on the other side of the transformer hasto be average free which means that the relative on time ofst, and St2 is 50 percent in stationary operation. The angularoffset of the voltage vtl across st, is called fotj and the offsetof the voltage Vt2 across St2 is called Y5t2.The Fourier series of the three voltages for the switching

frequency fs can be described as00

vz az(O)+ z(i) cos(iW5t)i=l00

Vs4 sa4(0) + j as4(i) cos(iwst -Os4)i=

00

Vt at(j) [cos(iwst -yt)1 cos(iwst -

i=l

with ws= 27f5 and

a2

{Vi [sin(i-W6a) + sin(i60b) + sin(iWF6c)]az(i) l 2Vi_ [sin(i-W6a) + sin(i60b) + sin(iWF6,)]

{ 2Vi. sin(i-F654) if i odd

a54(1) = l 2Vj4'sinr(iw6s4) if i even

2nVout if i odd

at(i) = i7r0 if i even

(6)

(Ot2)1

if i odd

if i even

(7)

As the system in figure 5 is an LTI system, the current itcan be calculated trough

00

it = E G(jiws) {a.() cos(iwst + ZG(jiws))i=1

-as4(i) cos(iwst -(0s4 + ZG(jiws))-at(i) cos(iwst -Lt + ZG(jiws))

+at(i) cos(iwst -t2 + ZG(jiws))},where G(jiw,) is the transfer function from vlr to t , i.e. theimpedance, at the frequency i,.

1027

(8)

The transferred power is

ntVotGt(jZ{ s 2sin( 2tl- t

i=l

tY5ti + i~Ot2[a j sin ( 2 + G(jiws))

sin ( ifPtl + ift2 + ZG(jiws))2

sin iti + ift2 t + ZG(jiw5))

+at() sin i(;ti + i(t2 _t2 + ZG(jiw5))]}. (9)

When the turns ratio n is increased the power that canbe transferred through the converter increases too, but whenincreasing n above a certain value the reactive power growsas well, which results in higher device stresses and higherlosses. Therefore the turns ratio is selected to be 14 to getan acceptable efficiency and on the other hand to get enoughpower in every possible converter state.As mentioned above s 4oti and Ot2 can be selected

independently. In the actual implementation y5s4 = w andOt2 = fot + w is selected. Independent of the modulationfactor, which is defined through the machine's operating point,it is possible to control the power through variation of o asshown in Figure 7. The output power is also slightly dependenton the machine's angle but this influence is small and it can beeliminated with the variation of o too. It is worth noticing thatfor the given parameter values the maximum power that canbe transferred from Vi, to VOtl (positive power) is higher thanthe power that can be transferred from VOtl to Vi, (negativepower). In a normal dual active bridge for ideal conditionsthe positive and the negative maximum power are equal asstated in [11]. As the resistor in the conventional dual activebridge is quite small the reality matches the ideal equationquite well. For the motor we used, the zero phase resistanceis significantly higher (see figure 17). Thus, the voltage dropacross this resistance causes an unbalance in the maximums.Depending on the choice of n, Vi, and VOtl the positive orthe negative power maximum is higher.

Figure 8 shows the actual controller. Other control schemesthat use the other degrees of freedom will be described infuture publications. There will be an analysis of the bestchoice of the degrees of freedom specified in table I.

Controller degrees of freedom:fs4: phase-shift angle of VA4Sotj: phase-shift angle of vtlOt2: phase-shift angle of Vt2

Converter degrees of freedom:n: turns ratio

Table I: Degrees of freedom.

The model in figure 3b, where the machine model of figure12 was inserted (which is explained in the next section), was

Output power

F 1.2kW

OM 9) I-~~~01.15 -X

-0

L-1.2kW

Figure 7: Power's dependency of o and M for n = 14,Vi½ = 400V, t = 0 and SOs4 = F.

~~~~~~~~~si]

VI"utij1T12~/I

T12 T Ot)

Figure 8: Output voltage control of the MFCS I.

simulated with the controller from figure 8. The load is aresistor. Figure 9 shows the simulation for the parametersgiven in table II. As can be seen, the output voltage andthus the output power is constant over the motor period.Furthermore, the voltage time product across the transformeris below the desired level and saturation is prevented.

14 _I.....,,,,1lll........z"'0101lll

.....

,, I° 3|. I11 1R1

-14-i1 Oms 15ms 20ms

(a) Phase currents and output voltage

400V

ov200V

IOAOA

-1OA200V

-200V._1O.OOms 10.05ms

(b) Zoomed voltages of figure 51 0.20ms

Figure 9: Simulation results for the MFCS-I with parametersgiven in table II.

In table III the characteristic values of the simulation arelisted. As the zero sequence resistance of the actual machineis quite large, only an efficiency of about 70 percent isreached. Since the actual machine is not intended for the use inautomotive applications, we measured an asynchronous motor

1028

Parameter Value# poles p 8

motor speed 1500 rpmswitching frequency f 20 kHz

input voltage Vin 400 Vback EMF 102 Vrms

output voltage V0tlt 14 Vmotor power 3 kW

converter output power 500Winductances see table IVresistances see table IVturn ratio n 14

Table II: Parameters of the MFCS.

that was used in a parallel hybrid car. The measurement showsthat the impedance is significantly smaller. For a relevantmachine like this the efficiency of the dc to dc converteris about 85 percents. Additionally, the efficiency can beincreased by choosing a new control schema and includingthe other degrees of freedom specified in table I, as will beshown in future publications.

L[uH] R[Q]A1-_40_302010

X,gX X,. X X \\

,,.v \\\XJ c ,>r#,,, . .\. - N

Ia ...............(a) Phase current Sux

M. .--,,E,,...e

(b) Zero-scquencc current flux

Figure 11: COMSOL simulation of the magnetic flux for (a)symmetrical, sinusoidal currents and (b) zero-sequence current.

This means that a zero-sequence current excites only a leakageflux.

Although the local flux is influenced by the zero sequencecurrent, the total torque is not. Because of the spacialdistribution of the zero-sequence current, the forces cancelto zero, so the total torque of a machine with zero sequencecurrent is equal to the one of a machine without zero sequencecurrent.

To model these effects, the equivalent motor circuit depictedin figure 12 is used.

1

0

1 10 f [kHz]

Figure 10: Zero phase inductance and resistance for a 50kWmotor.

Parameter ValueIROIrms 0.538 AIR02rms 0.310AIR02rms 0.727 AIROIrms 0.331 AItlrmS 32A

,s4,m,s 2.3 A

Table III: Simulation results of the MFCS.

C. Detailed Flux Analysis of MFCS-I

The three voltages vpl, Vp2 and Vp3 excite a magnetic fluxin the machine. One part of this flux, called the main fluxi',m, is linked with all three phases. The other part of the flux,consisting of the slot leakage flux, the end winding leakageflux etc., is represented through the leakage flux T.

Figure 1la shows a COMSOL simulation for symmetrical,sinusoidal currents. Due to the spacial distribution the mainfluxes of the individual windings sum up to zero if a zero-

sequence current flows through the machine (see figure 1lb).

M=-1 L2Lm

Figure 12: Simplified equivalent circuit of the motor.

As the losses due to skin and proximity effect and the core

losses are frequency dependent, the resistance Ro also changeswith the frequency. The same applies to the inductance Lo.Figure 13 shows the four-step R-L ladder circuit for modellingthe frequency dependence of the zero-sequence impedance.The parameters were calculated to fit the measured impedanceof the machine shown in figure 17.

R04

L03 Ro

Lo, Rol 4

Figure 13: Equivalent circuit representing the frequency dependentRo and Lo in figure 5.

1029

0.1 1 10 f [kHz] 0.1

ParameterLo,RolLo,R02L02R03L04R04

Value1.0132 mH150.29 Q

0.60183 mH32.252 Q0.2094mH6.921 Q

72.861,uH1.4852 Q

Table IV: Parameters for the equivalent circuit in figure 5.

It is clear that the mutual inductances are frequency depen-dent too, but as for the dc-dc converter only the zero phasecurrent is relevant, this effect is omitted in our equivalentcircuit.

D. Design and Experimental Results of MFCS-I

To verify the functionality of the proposed MFCS-1, asynchronous motor with accessible star point (and integratedsecondary windings in order to also allow the evaluation ofMFCS-11) with the parameters given in table V has beendesigned. Figure 14 shows the motor with the extra windingsas described in the next section.

Figure 15: Active secondary side rectifier.

The results of the zero-sequence voltage measurement infigure 16 and the simulation in figure 4b match very well.The measurement of the phase inductance Lp = Lm+ Lo andthe zero sequence inductance L- show the large zero sequence3resistance, but as already stated this problem will disappear asin hybrid cars machines with lower resistances are used (seefigure 10).

Parameter Value# poles 8

nominal speed 1500rpminput voltage Vin 250 V..400 V

motor power 3 kWback EMFs eR,es and eT 67.8 Vrms at 1000 rpm

Table V: Parameters of the prototype.

synchronousmachine

secondarywindings

load machine

Figure 16: Measurement of the zero-sequence voltage for amodulation index ofM = 0.5 and an input voltage Vin = 50 V.

L[uH]700600500

R[Q]30 --T20

10

0.1 1 10 f[kHz][ 0.1 1 10 f [kHz]

measurement primary windings withwindings accessible star point

Figure 14: 3 kW permanent magnet synchronous motor withaccessible star point and secondary windings for MFCS-II.

The motor is driven by the 6.8 kW three phase back-to-back converter. The output stage, fed with an external dc-linkvoltage, is used to drive the motor. One leg of the input stageserves as the fourth leg in figure 3b).The output rectifier in figure 15 was designed for an output

power of 1 kW.

Figure 17: Zero phase inductance and resistance of the syn-

chronous motor (figure 14).

III. MFCS-II

In a second integration step the galvanic isolation of VOtlcan also be integrated into the motor (see figure 3c) inorder to reduce the costs and the weight further. To do so,

additional windings must be placed in the slots together withthe conventional winding.

A. Basic Principle of Operation MFCS-II

Figure 18 shows a simplified sketch of the integration ofthe transformer proposed for the MFCS-II. The conventional

1030

40(

N

motor windings act as the primary windings of the threetransformers. The additional windings are forming the sec-ondary ones. The machine stator iron is used as magneticcore. As the machine has 120 primary turns per pole pairand the winding geometry allows not less than 12 secondarywindings per phase, the secondary windings of the four polepairs will be connected in series while the primaries areconnected in parallel. So the desired turn ratio of 14 (asexplained in subsection IIA) is achieved. Additionally, there isone measurement winding per slot to proove if the COMSOLsimulations match with the actual fluxes.

~~~measuremrenlt winldingstator additional / secondary winding

; =/ -~norl-nal / primary winldingrotor

Figure 18: Transformer integrated into the motor.

but again, the output power is constant over the motor period.

IV. CONCLUSION

By integrating the dc-dc converter, which interfaces the lowand the high voltage buses of a hybrid vehicle. In this paper,two new concepts for integrating the converter are presentedand verified by simulations. With the first concept MFCS-I,the inverter and the machine are used to implement a primarybridge leg of an isolated full bridge dc-dc converter. Inthe second concept MFCS-II, the transformer is additionallyintegrated into the machine so that only one auxiliary bridgeleg and a secondary rectifier are required to realize an isolateddc-dc converter.

In a next step, the hardware for the MFCS-I and the MFCS-II are taken into operation and the simulations are verified.Further different control schemes with the degrees of freedomdescribed in table I are analyzed.

If there is a voltage applied to phase P1, a flux Tpibegins to build up, inducing a voltage vplp in the secondarywinding P12 proportional to vpl, (ignoring losses). The sameapplies to phases P2 and P3. If the secondary windings areconnected in series, the voltage appearing on the secondaryside is without load just v2 = 1 (vz -V4) as the remainingvoltages cancel to zero. If a load is connected to the secondaryside, the voltage is slightly changed because of the losses andthe reactive voltage drop (see figure 19, V2l1ad). Again anactive rectifier is used to obtain the desired dc voltage V,"t.The control can be implemented in a similar way as for MFCS-I.

400V

ov200Vov vt

-200V--5AOA i-5A

400V0v

-400V10.OOms 10.05ms 1 0.20ms

Figure 19: Simulation results tor the MFCS-II with parametersgiven in table II.

Figure 19 shows the simulation for figure 3b with theparameters given in table II. The phase currents and the outputvoltage are similar to figure 9a so they are not depicted here,

REFERENCES

[1] A. Emadi, M. Ehsani, J.M. Miller, "Advanced Silicon Rich AutomotiveElectrical Power Systems," Proc. 18th Digital Avionics Systems Con-ference, St. Louis, Missouri, Oct. 1999.

[2] K. Moriya, H. Nakai, Y. Inaguma, H. Ohtani, S. Sasaki, "A Novel Multi-Functional Converter System Equipped with Input Voltage Regulationand Current Ripple Suppression," Industry Applications Conference,2005. Fourtieth IAS Annual Meeting. Conference Record of the 2005.

[3] Seung-Ki Sul, Sang-Joon Lee, "An Integral Battery Charger for Four-wheel Drive Electric Vehicle," IEEE Transactions on Industry Applica-tions, Vol. 31, No. 5, September/October 1995.

[4] Bayrische Motoren Werke AG, "Switching Device for Linking VariousElectrical Voltage Levels in a Motor Vehicle," European Patent WO2006/105840 Al.

[5] Shigenori Kinoshita, Koetsu Fujita, Junichi Ito; Fuji Electric Co."Electric System for Electric Vehicle," United States Patent 6,066,928.

[6] Shoichi Sasaki; Toyota Jidosha Kabushiki Kaisha "Multiple PowerSource System and Apparatus, Motor Driving Apparatus, and HybridVehicle with Multiple Power Source System Mounted Thereon," UnitedStates Patent 6,476,571 Bi.

[7] D.S. Carlson, C.C. Stancu, J.M. Nagashima, S. Hiti, K.M. Rahman;General Motors Corporation "Auxiliary Power Conversion by Phase-Controlled Rectification," United States Patent 6,617,820 B2.

[8] C.C. Stancu, S. Hiti, J. Nagashima; General Motors Corporation"Auxiliary Power Conversion for an Electric Vehicle Using High Fre-quency Injection into a PWM Inverter," United States Patent 6,262,896B1.

[9] ETH Zurich, "Drehstromantriebssystem mit motorintegriertem Hochfre-quenztrafo zur bidirektionalen Kopplung der Versorgungsspannungen,"Schweizerische Patentanmeldung, July 27, 2006.

[10] ETH Zurich, "Drehstromantriebssystem mit hochfrequent potential-getrennter bidirektionalen Kopplung der Versorgungsspannungen,"Schweizerische Patentanmeldung, July 27, 2006.

[11] M.H. Kheraluwala, R.W. Gascoigne, D.M. Divan, E.D. Baumann,"Performance Characterization of a High-Power Dual Active Bridgedc-to-dc Converter," Power Electronics in Transportation, pp. 39-44,Oct. 2002.

1031

-