2
2
2
Abstract: The modular development of a three-phase, single-stage AC-DC power converter based on
single-phase AC-DC full-bridge converters is described. Three-wired and four-wired modular
converter systems are examined and compared. The three-wired system is found to have inherent
power factor correction and is better than the four-wired system in terms of harmonic content. The
fluctuation of the converter neutral voltage in the three-wired system and the voltage stress on the
power switches are analysed. A simulation based on a PSPICE model is included. The modular
converter does not have a minimum component count when compared with non-modular converters.
But the modular converter offers attractive features such as simple DC-DC switching control,
electrically isolated output, inherent power factor correction and single-stage power conversion
without using power devices with excessively high voltages and current ratings. It also has flexible
power expandability and the potential to simplify the design and production of high-power AC-DC
power converter products. The operation of the modular full-bridge converter has been confmed
with a 1.5kW prototype. Both predictions and measurements show that the power quality of three-
wired modular converter system is good and well above IEC-1000 standards.
Also
“0
hence
-
Rearrange eqn. 2 to give
Fig. 1 S c h t i c of fhree-phaemodulurfull bridge (PB) converter svstem (3)
VDlp
VD1 8n
failure analysis may be examined. A four-wired system can that of the modular flyback converter system [12], in which
be configured simply with the converter neutral (NJ con- the coupled windings are used as coupled inductors rather
nected to the supply common. than transformers. Thus, a sinusoidal current waveform
can easily be achieved in the modular flyback converter
2.4 Simulation results system but not in the modular full-bridge converter system.
The schematic shown in Fig. 3 is used to study the per- The harmonic spectrum of the phase current is shown in
formance of a three-wired and four-wired system. The test Fig. 5.
conditions are input phase vofiage 200V (50Hi) and
switching frequency = 25kHz. The circuit parameters are
Lf = 5 m H , Cf = 0.47@, L = 20*, Ce = 9900pF (3 x
3300@), RL = 1.6Q. The turns ratio N equals two. The
duty ratio D Iis vaned in each case to obtain an output
voltage of 48V DC. A small capacitor of 2.5pF is put
across the output of the diode rectifier in each converter
module to absorb the voltage transient in a practical circuit.
frequency, Hz
Fig.5 Harmonic spectrwn ofphase current in four-wired modulmfill-bridge
system
Iin, = 0 (9)
X
Case (1) When one and only one of the input current is
3! I/ zero: Suppose Iin, = 0, Iinb < 0 and Iin, > 0, using eqns.
7-9 we have
- D2G(Vib- N.Vo) + D 2 G ( V i ,- N.V,) = 0
(Vib + Vi,) = 0
V6+ V c- 2 V N , = 0
VNc = ( V6 + V c 7 Vu )
= -2
A . 1 (10)
200 300 400 500
frequency, Hz Case (2) When all input currents are non-zero: Suppose I,
Fig.7 Harmonic spectrum ofphase m e n t in three-wired nwddar full-brdge > 0, Ib < 0 and I, > 0. Subsituting eqns. 7 and 8 into eqn. 9
convert@' yields
D2G(V/i,- N.V,) - D2G(Vib- N.Vo)
+ D2G(Vi, - N.Vo)= 0
( V i , - Vib + Vi,) + N.V, = 0
+
V a + Vb + V c - ~ V N ,N.Vo = 0
v,, = --N.Vo 3
3.7.2 CCM: In the CCM, line current can be given by the
-400; ........................................................................................... i following expression:
40 50 60 70 80 90
time, ms
Fig.8 S i m u h i e d p h e voltage and Converter neutral voltage in three-wired
Jystem
d 1 D
3 Voltage stress analysis and control -dt( l o x ) = -
L (-*sgn(Iinx).VxN,
N - Vo)
characteristics
X = a,6 and c (13)
In the previous Section it was seen that the converter neu- Differentiating eqn. 9 and using eqns. 12 and 13, we have
tral voltage fluctuation is beneficial. In this Section, this
voltage fluctuation is studied. In addition, the voltage stress
on the power electronic components is analysed.
X
3.7 Determination of converter neutral point In summary, the converter neutral voltage VNc under
voltage VNc DCM and CCM operation can be described as in Tables 1
and 2.
3.7.7 DCM: Let the three-phase system be described as
Table 1: DCM operation
V u = E m . sin(wt)
V b = E m . sin (wt - $) Conditions
When one and only one input
vNc
410 IEE Proc-Electr. Power Appl., Vol. 146, No. 4, July I999
Table 2 Heavy CCM operation Within this range of duty cycle, the modular converter
operates in discontinuous current mode. For tlus particular
Conditions VNC designed output voltage range up to 60V, Fig. 10 shows
When one and only one input V N=~-Va/2 for /in, = 0 that the transfer characteristic of the AC-DC modular full-
current is zero VNc= -VbJ2 for /in, = 0 bridge converter is similar to that of a DC-DC converter.
VNc= -Vd2 for /inc= 0
70
All modules under CCM V,, = -N.Vd3DXX sgn(/in,)
60
3.2 Maximum input voltage for full-bridge >a 50
converter module or
a,
20
across the switches when they are in the OFF state in each 10
full-bridge converter module is equal to the rectified volt-
age, i.e. \Vu - VNJ.Therefore the maximum voltage that 0.1 0.2 0.3 0.4 0.5
the switches have to withstand is the maximum value of duty ratio D1
\Vu - V,,l given by: V,,,,, = max(1Em sin(wt) - V,,I). Fig. 10 Variationof output voltage with ahty cycle D,
Consider the extreme cases when D1 = 0 and Dl = 1. +- Yo (DC Source)
-A- V,, (AC Source)
(4 D1 = 0,
Now V, = 0, it follows that VNc= 0 and V,t,.ess= Em
(b) D1 = 1
In this case 1Em Si(wt) - VNclreaches its maximum when
wf = d3 and VstreSs = 4312 Em.
Therefore, the highest voltage stress (excluding transients)
in a switch in this three-phase modular PFC occurs with
low duty ratio when a low output voltage is required. How-
ever, the maximum voltage stress is limited to Em, i.e. the
peak value of the phase voltage as in a single-phase con-
verter. T h ~ is
s much lower than the typical voltage stress of
the switches (> 43 Em) in some three-phase power factor
correction circuits [8]. Fig. 9 shows the simulated voltage
waveforms under the condition of D = 1, the neutral point
voltage becomes a triangular waveform and the rectified
voltage, i.e. ]Vu- VNclreaches its maximum value of 43/2
Em.
I
Fig. 11 Memured voltage ami w e n t wavefomv in four-wired system
Upper trace: phase voltage 200Vidiv; lower traces: phase currentS SAidiv
1: Sms, 200V; A, B and C: 5ms, l00mV
4 Experimental results
4. I Four-wired system
The modular FB converter system was initially tested as a
four-wired system. The test conditions are input phase volt-
.............................
-4004............................... .............................. i age V,, = 135V (RMS), output DC voltage V, = 48V and
40 50 60 70 80 90 100 switchng frequency = 25kHz. The output power is 1kW.
time, ms
Fig. 11 shows the measured input phase voltage (upper
Fi .9 Sbnubted voltage wave om of phase voltage Va converter neutral trace) and the three phase currents of the four-wired sys-
vo&e v N c anti mput voltage 0J.r reciijii va - v,, lauirr t / z conrzirion of
D=l tem. As explained in the preceding Section and predicted in
the simulation, the phase currents are highly distorted. The
3.3 DC-DC characteristic of AC-DC modular harmonic spectrum (100Wdiv) of the phase current is
converter shown in Fig. 12. The first component is the 50Hz funda-
In [12], we showed that the modular flyback converter has mental component. The phase current in the four-wired
an overall DC-DC type transfer function. With the help of system consists of sigrllficant odd harmonics such as the
the computer model for the modular full-bridge converter third, fifth, seventh and ninth harmonics. The thu-d har-
(Fig. 3), we examined ths. Two simulation tests using the monic is almost 30% of the fundamental component. It is
same model were carried out. The first test uses a three- clear that the four-wired system is far from satisfactory.
phase AC voltage supply with a phase voltage of 200V
(RMS). In the second test, the three AC phase voltage 4.2 Three-wired system
sources are replaced with 3 DC voltage sources. The DC The modular full-bridge converter system has also been
voltage is V,, = VMs = 200V. Other circuit parameters tested as a three-wired system. The test conditions are input
are identical in both tests. Fig. 10 shows the variation of phase voltage = 200V (50Hz), V, = 48V and switching fre-
the output voltage with the duty cycle D1 for both tests. quency = 25kHz. The output power is 1.5kW. A voltage
IEE Proc -Electr Power Appf , Vol 146,No 4, July 1999 411
mode control with a PI compensator is used to regulate the firm the single-stage AC-DC power conversion using sim-
DC output voltage to about 48V with a load current of ple DC-DC type PWM control.
30A. Circuit parameters are Lf = 5 m H , C, = 0.47$, L =
20pH and C, = 9900pF (3 x 3300pF). A small capacitor of
2.5pF is put across the output of the diode rectifier in each
converter module to absorb the voltage transient.
412 IEE Proc.-Electr. Power Appl., Vol. 146, No. 4, July 1999
listed in Table 3. These harmonics are well below the IEC-
1000-3-2 Class A limits as shown in Fig. 17. The energy
effciency, THD and PF are also plotted against the load
power and are shown in Fig. 18. At full load, an efficiency
of about 89% was achieved without using soft-switching.
In summary, the power factor correction feature and the
DC-DC type characteristic are inherent in the three-wired
modular converter system. Experimental results confirm
that the three-wired system is better than the four-wired
system in this modular approach.
5 Conclusions
-.-
IEEIow-.+~Z L yA limits dictions of the modular system in a short time (typically
-4- 478 W 5.9s per 50Hz cycle). This modular approach enhances the
-A-956W
3- 1434 w flexibllity and power expandability of power converters and
class A l i i t s
has the potential of simplifying the production and mainte-
nance procedures of such power products. The operating
”03
principles of the modular system based on the full bridge
converters have been confiied. The three-wired system
performs better than the four-wired counterpart. The float-
ing converter neutral voltage fluctuates at three times the
0.8
fundamental frequency and automatically cancels the third
harmonics and harmonics at multiples of three. The power
factor correction feature is inherent in the modular
0.6 I I I I
approach. Simple DC-DC type PWM control can be
O478
I 956 1434
mental results show that harmonic content of the three-
wired system is well below the IEC-1000-3-2 standards. As
output power, W mentioned in the introduction, the modular approach does
Fig..18, Vai-btwns of energy @ciency power factor and total hmmonic &- not necessarily minimise the component counts. However,
tortwn wzth loadpower it does provide a simple and convenient way to parallel
-4- PF
-A- efficiency three single-phase converter modules for power expansion.
a-THD Further research is being carried out to optimise the con-
verter performance.
Table 3: Analytical data obtained from Voltech 3300 power
analyser 6 Acknowledgment
Load(W) 478 956 1434 The authors would like to thank the Hong Kong Research
RMS 1.OOA 1.885A 2.796A Grant Council for financial support. ~s project started in
Fundamental 0.983A 1.877A 2.778A 1994. The initial support from Sydney University and dis-
3rd 0.042 4.2% 0.023 1.1% 0.050 1.8% cussions with Mr. Andrew Levido (now manager of GEC-
5th 0.114 11.4% 0.157 8.3% 0.189 6.8%
Plessey Communication Pty. Ltd., Australia) are also grate-
fully acknowledged.
7th 0.066 6.6% 0.086 4.6% 0.211 7.6%
9th 0.031 3.1% 0.012 0.65% 0.007 0.25% References
1I t h 0.051 5.1% 0.056 2.3% 0.053 1.9%
MANIAS, S., ZIOGAS, P., and OLIVER, G.: ‘An ac-dc converter
13th 0.015 1.5% 0.022 1.2% 0.057 2.0% with improved input power factor and high power density’, IEEE
15th 0.022 2.2% 0.015 0.8% 0.015 0.5% Trans., 1986, U-22, (6), pp. 1073-1081
KAZERANI, M., ZIOGAS, P.D., and JOOS, G.: ‘A novel active
THD 16.3% 10.3% 11.1% current waveshaping technique for solid state input power factor con-
PF 0.955 0.985 0.989 ditioners’, IEEE Trans. Ind. Electron., 1991, 38,(l), pp. 72-78
BOYS, J.T., and GREEN, A.W.: ‘Current forced single phase reversi-
ble rectifier’, ZEE Proc. B, Electr. Power Appl., 1989, 136, (5), pp. 201-
The performance of the 1.5kW prototype was also evalu- 212
ITOH, R., and ISHIZAKA, K.: ‘Single-phase sinusoidal rectifier with
ated using a Voltech PM3300 power analyser under differ- stepidown characteristics’, IEE Proc. B, Electr. Power Appl., 1989, 136,
ent load conditions. Tests were carried out at about 0.5kW, (9,pp. 237-242
1kW and 1.5kW. The measured current harmonics, total GREEN, AW., BOYS, J.T., and GATES, G.F.: ‘3-phase voltage
sourced reversible rectifier’, IEE Proc. B, Electr. Power Appl., 1988,
harmonic distortion (THD) and power factor (PF) are 135,(6), pp. 362-370
IEE Proc.-Electr. Power Appl., Vol. 146, No. 4, July 1999 413
6 PAN, C.T., and CHEN, T.C.: ‘Modelling and analysis of a three .SUBCKT FBAVG 1 2 3 4 5 9 12 PARAMS: FS=100K L=lOOU
phase PWM AD-DC converter without current sensor’, IEE Proc. B, N=l ILO=O
Electr. Power Appl., 1993, 140, (3), pp. 201-208
7 VLATKOVIC, V., BOROJEVIC, D., and LEE, F.C.: ‘Soft-transition * I=IN+; 2=TN-; 3=0UT+; 4zOUT-i 5=DUTY CYCLE;
three-phase PWM conversion technology’. Proceedings of IEEE 9=D3; 12=D2
PESC‘94, 1994, pp. 79-84
8 WU, R., DEWAN, S.B., and SLEMON, G.R.: ‘Analysis of a PWM *h*************************************************
AC-DC voltage source converter under predicted current control with
a futed switching frequency’, IEEE Trans. Ind. Appl.. 1991, 27, (4), pp. L1 6 3 {L} IC={ILO}
756-764
9 GAUGER, D., and FROESCHLE, T.: ‘A three-phase off-line switch- RS 7 6 0.001
ing power supply with unity power factor and low TIF‘. INTELET, E01 7 8 VALUE={V(3,4)*(l-V(1 l)-V(12))}; D3*VO
1986, pp. 115-121
10 CHAPMAN, D., and TUCK, C.J.: ‘A high-densiy 48V 200A rectifier E02 84 VALUE={V(1,2)*V(ll)/N) ; Dl*Vi/N
with power factor correction - an engineering overview’. INTELET.
1993, pp. 118-125 * Iin=IL*Dl@l +D2)/N
11 SPIAZZI, G., and LEE, F.C.: ‘Implementation of single-phase boost
power factor correction circuits in three-phase applications’, IEEE GIN 1 2 VALUE={V(7,6)*1000*V(I I)/(V(lI)+V(12))/N}
Trans. Ind Electron., 1997, 44, (3), pp. 365-371 RD501G
12 HUI, S.Y.R., and CHUNG, H.: ‘Parallehg power converter for AC-
DC step-down power conversion with inherent power factor correc- RDl 11 0 1G
tion’, IEE Proc. B, Electr. Power Appl, 1999, 146,(2), pp. 247-252
13 WALLER, P.H.: ‘Forward converters operating directly off the three ED1 11 0 TABLE{V(5)}=O.Ol,O.O11,l
phase rectified mains suitable for exchange use’. Proceedings of the R19 19 0 IG
IEE international Telecommunication energy conference, 1981, pp.
234239 El9 19 0 TABLE {V(1,2)/N-V(3,4))0.01,O.Ol 1000,1000
14 MIDDLEBROOK, R.D., and CUK, S.: ‘A general d i e d approach
to modeling switching converter power stages’. Proceedings of IEEE RD12 11A 0 1G ;DUTY Dl+D2
PESC, 1976, pp. 18-34
15 LEE Y.S.: ‘Computer-aided analysis and design of switch-mode power ED12 11A0 TABLE {4*V(7,6)*1000*FS*L/V(19)N(11)}=0,0 1,l
supply’ (Marcel Dekker Inc. 1993), Chap. 3 RD2 12 0 1G
9 Appendix: Average model for isolated full bridge ED2 12 0 TABLE {V(11A~V(ll)}=O,O 1,l ;DUTY D2
converter ED3 9 0 VALUE={1-v(1I)-V(l2)} ;D3
The equivalent average circuit model of the FB converter is .ENDS FBAVG
shown in Fig. 19. The input port is a current generator lN10-1 7OUTl
(Gin) as described by eqn. 6. The output port implements
eqn. 2 by two voltage controlled voltage sources Eol, Eo2,
an inductor L and a current sensing resistor Rs where both
DCM and CCM are handled. The other generators imple-
ment eqns. 3-5 are self-explanatory in the list.
...................................................
* AVERAGE MODEL FOR FULL BRIDGE ISOLATED
CONVERTER
* FOR BOTH CCM AND DCM Fig. 19 Average circuit mdel offull-bridge converter
................................................... Eo1 = Vo x D3; E02 = Vi” x DI; Gin = iwN. DI/(DI + D2)
414 IEE Proc-Electr. Power Appl., Vol. 146. No. 4, July I999