Keynote paper, First International Conference on Energy, Power, and Motion Control
May 5-6, 1997, Tel Aviv, Israel
Some Topologies of High Quality Rectifiers
Robert W. Erickson
Department of Electrical and Computer Engineering
University of Colorado, Boulder 80309-0425
Abstract— Several basic classes of three-phase
high-quality rectifiers are described. Both single-switch
and six-switch three-phase rectifier topologies can be
derived from parent dc-dc converters. Single-switch
rectifiers are compared with the basic six-switch PWM
rectifiers performing similar power conversion functions,
using the measures of total semiconductor stress and
active semiconductor utilization. The single-switch
approach is shown to utilize the semiconductor devices
more effectively. Zero current switching and multi-
resonant approaches are found to exhibit low switch
stress over a wide range of operating points, with low
THD.
1. Introduction
High-quality low-harmonic rectification is
becoming increasingly required, to meet regulations
which limit ac line current harmonic content, such as
IEC-555 and IEEE-519. The application requirements
of three-phase high-quality rectifiers are varied. In
some simple cases, it may be necessary only to
produce a dc output voltage nearly equal the peak input
line-to-line voltage. In other cases, an output voltage
of variable and controlled magnitude, substantially
smaller or larger than the input line-to-line voltage,
may be required. Isolation of the dc load from the ac
line is also a requirement in applications such as
battery charging in the telecommunications or electric
vehicle areas. High-frequency EMI and common-mode
currents, generated by non-isolated low-harmonic
rectifiers, is also a major concern.
Figure 1 illustrates the desired properties of an
ideal three-phase rectifier, which presents a balanced
resistive load to the utility system. A three-phase
converter system is controlled such that resistor
emulation is obtained in each input phase. The rectifier
three-phase input port can then be modeled by per-
phase effective resistances R e , as illustrated in Fig. 1.
The harmonic content of the ac input currents therefore
match the harmonic content of the applied ac input
voltages, and hence are correspondingly low. The
instantaneous powers apparently consumed by these
effective resistors are transferred to the rectifier dc
output port. The rectifier output port can therefore be
1
3øac
input
dc output
ia
øa Re
+
ib Re ptot =
øb pa + pb + pc R v
ic –
øc Re
Fig. 1. Functional equivalent circuit of a three-phase
high-quality rectifier: input resistor emulation and
dc output power source characteristic.
modeled by a power source equal to the total three-
phase instantaneous input power as shown in Fig. 1
[1-3].
The various active approaches to high-quality
three-phase rectification fall into two classes. The first
comprises PWM converters operating in the
continuous conduction mode (CCM), with the three ac
line currents independently and actively regulated.
These converters usually contain six or more active
devices, and are capable of bidirectional power flow.
The second class may contain as little as one active
device, and input resistor emulation is obtained via the
natural response of the converter reactive elements to
the high-frequency switching of the active device. In
this paper, both approaches are discussed, and the
utilizations of their active semiconductor devices are
compared.
2. Three-phase PWM rectifiers
operating in CCM
A variety of 3øac-dc PWM rectifiers are known; a
few of the many references on this subject are listed
here [4-10]. The most well-known topology is the
three-phase ac-to-dc boost rectifier, illustrated in Fig.
2. This converter requires six SPST current-
bidirectional two-quadrant switches. The inductors and
capacitor filter the high-frequency switching
harmonics, and have little influence on the low-
frequency ac components of the waveforms. The
switches of each phase are controlled to obtain input
resistor emulation, either with a multiplying controller
 dc output output voltage V that is less than the peak line-line ac
3øac i1(t)
input i2(t) i3(t) input voltage. This converter resembles what is
ia (t) L + known as the current-source inverter, except that the
Q1 Q2 Q3
D1 D2 D3
øa
converter is operated as a rectifier, and the converter is
ib (t) L
C load v(t) controlled via high-frequency pulse-width modulation.
øb
Two-quadrant voltage-bidirectional switches are
ic(t) L required in this converter. A disadvantage of the 3ø
øc Q4 Q5 Q6
D4 D5 D6
– buck rectifier is the higher conduction losses induced
by the series connection of devices. Also, the rms
Fig. 2. Boost-type 3øac-dc rectifier. transistor currents are greater than in the 3ø boost
rectifier; this further increases the conduction loss.
scheme employing average current control, or with
The converter is capable of operation in inverter mode
some other approach. To obtain undistorted line
by reversal of the polarity of the output voltage v(t). A
current waveforms, the dc output voltage V must be
substantial input filter is usually required to smooth
greater than or equal to the peak line-to-line ac input
the pulsating ac line currents.
voltage V L,pk. In a typical realization, V is somewhat PWM 3øac-dc rectifiers which resemble most
greater than V L,pk. This converter resembles the well- other dc-dc converter topologies are also possible. Two
known voltage-source inverter, except that the examples are the 3øac-dc rectifier circuits of Figs. 3(b)
converter is operated as a rectifier, and the converter is and 3(c), based on the dc-dc buck-boost [9] and Cuk
controlled via high-frequency pulse-width modulation. converters. These converters can be viewed as being
The three-phase boost rectifier of Fig. 2 has derived from parent dc-dc converters via a
several attributes which make it the leading candidate transformation in which the dc input and switch
for most 3øac-dc rectifier applications. The ac input network are replaced by a three-phase input and six-
currents are nonpulsating, and hence
very little additional input EMI a)
filtering is required. As in the case of L dc output
3øac
the single-phase boost rectifier, the input iL(t)
Q Q Q 1 2 3
rms transistor currents and also the ia(t) +
øa
conduction losses of the three-phase D D D 1 2 3
boost rectifier are low relative to other ib(t)
øb C load v(t)
3øac-dc topologies such as those of
Fig. 3. The converter is capable of ic(t) Q Q Q 4 5 6
øc
bidirectional power flow. A input filter –
D D D
disadvantage is the requirement for six 4 5 6

active devices: when compared with a b)

dc-dc converter of similar ratings, the dc output
active semiconductor utilization is 3øac input
Q Q Q D
low. Also, since the rectifier has a i (t) a
1 2
Q
3
+
7
7

ø a i (t)
boost characteristic, it is not suitable D D D1 2 3
L

i (t)
for direct replacement of traditional ø b
b
L C load v(t)
buck-type phase-controlled rectifiers.
i (t) Q c
Q 4
Q 5 6
Three-phase ac-to-dc rectifiers ø c

input filter –
having buck, buck-boost, or other D D D4 5 6

characteristics, are possible, but find

much less use than the boost c)
topology. An example is the 3øac-dc 3øac dc output

buck rectifier illustrated in Fig. 3(a). input L2

L 1 ia(t) Q Q Q C1 +
Unlike the single-phase case, in three- øa D D 1 D 1
2
2
3
3

phase applications the buck topology L 1 ib(t)

D load
can supply constant power to a dc øb Q
C2 v(t) 7
7

load, with negligible distortion of the L 1 ic(t)

øc Q Q 4 Q 5 6
ac line current waveforms. When the D D D 4 5 6

voltage of one phase is zero, the other
two phases have nonzero voltage and
can supply the dc load power. This
converter can produce a controlled dc
–
Fig. 3. Several other non-isolated three-phase ac-dc rectifiers: (a) based
on the buck dc-dc converter, (b) based on the buck-boost dc-dc
converter, (c) based on the Cuk dc-dc converter.

2
switch bridge network. High-frequency isolation a)
transformers can be incorporated into most of these van(t) VM
converters, in a manner similar to that used to obtain
isolation in the parent dc-dc converters. t
3. Some other approaches to three-
phase high-quality rectification
The CCM three-phase rectifier approaches
described in section 2 require six or more active b) DCM boost
devices. Compared with conventional low-power-factor ia(t)
passive rectifier approaches, the increased active silicon
area and reduced semiconductor utilization of the six- t
switch approach can be expensive. In view of this, one
might ask what is the minimum active silicon area
required to perform the desired functions of the ideal
rectification application. It is well known that low-
harmonic 3øac-dc rectification can be performed using c) DCM flyback
a conventional passive six-diode rectifier and a i a(t)
harmonic trap filter. Hence fundamentally, no
semiconductor devices other than diodes are required.
t
When control of the output voltage is requisite, at
least one active device is needed. If it is desired to
avoid the use of low-frequency filter elements, then a
source of high-frequency switching harmonics is
needed, again necessitating inclusion of at least one d) ZCS buck
active device. So a single active device is the i a(t)
minimum needed to synthesize low-harmonic 3ø
rectifiers containing no low-frequency reactive
elements, having control of the output voltage, and t
having unidirectional power flow. Several single-
switch approaches to three-phase rectification are
known. Depending on the application, some of these
approaches may exhibit better active switch utilization e) Multiresonant ZCS buck
and reduced semiconductor area than six-switch vcra(t)
approaches. On the other hand, these single-switch
approaches generally require additional high-frequency
reactive elements. t
A single-switch 3øac-dc rectifier based on the
DCM boost converter [11,12] is illustrated in Fig. 4.
The input current waveform ia(t) is illustrated in Fig.
5(b). Transistor Q1 is controlled in the same manner as Fig. 5. Input current waveforms of various single-
switch three-phase rectifier circuits: (a) input line-
a dc-dc boost converter. Inductors L 1, L 2, and L 3 are of
equal small value, such that they operate in the to-neutral voltage v an(t); (b) 3ø DCM boost; (c) 3ø
discontinuous conduction mode in conjunction with DCM flyback; (d) zero-current-switching quasi-
resonant buck; (e) phase a tank capacitor voltage,
diodes D1 - D6. At the end of the transistor Q1 multi-resonant zero-current-switching buck.
conduction subinterval, the inductor currents reach
peak values which are also proportional to the applied emulation is obtained. The three-phase DCM boost
three-phase line-to-neutral voltages. When transistor rectifier does generate a modest amount of low-
frequency input current harmonics; the THD can be
Q1 turns off, then diode D7 becomes forward-biased and
reduced by increasing the dc output voltage.
the inductors release their stored energies to the dc
The three-phase DCM boost rectifier has the
output. Since the peak input currents are proportional
advantage of very simple control. The transistor can
to the applied input line-to-neutral voltages, then the
operate at constant switching frequency. Variation of
average values of the input currents are also
the duty cycle allows control of the dc output power.
approximately proportional to the input line-to-neutral
Only a single active device such as a MOSFET or
voltages. Approximate three-phase input resistor

3
3øac dc output
input are therefore approximately proportional to the
L1 ia (t) D7 +
øa D1 D2 D3 respective applied line-to-neutral voltages. At full load,
L2 ib (t) Q1
a THD of approximately 13-14% is observed; nearly
øb C v(t)
all of the THD can be attributed to the fifth harmonic.
øc
L3 ic (t) The THD can be reduced to less than 10%, by use of
D4 D5 D6
input filter
– any control scheme that leads to constant
instantaneous power flow.
Fig. 4. Single-switch three-phase DCM boost rectifier.
Transistor Q1 operates with zero current
IGBT is needed. A disadvantage is the need for an input switching. An IGBT, inverter-grade SCR, or other
filter to remove the high-frequency components of the device can be used. The peak voltage stress on Q1 is
pulsating input currents. As in all single-switch three- equal to the applied peak input line-to-line voltage,
phase rectifiers, bidirectional power is not possible. while the peak current is approximately twice the dc
A similar scheme, based on the DCM flyback output current.
converter [13,14], is illustrated in Fig. 6. A typical Transistor Q1 operates with an approximately
input current waveform is given in Fig. 5(c). This
constant on-time, equal to the length of the resonant
converter is effectively three independent single-phase
current pulse. The output power is controlled by
DCM flyback rectifiers, which share a single transistor
variation of the transistor off-time, and hence the
switch. The peak transformer magnetizing currents
converter operates with a variable switching frequency.
directly follow the applied ac phase voltages. This
The rectifier requires an input filter, to remove the
causes the average input currents to directly follow the
high-frequency components of the pulsating input
applied line-to-neutral voltages, without generation of
current waveforms.
low-frequency current harmonics. The rectifier can both
Another buck-type single-switch 3øac-dc rectifier
increase and decrease the voltage magnitude, and is
[16,17] is illustrated in Fig. 8. This is a multiresonant
inherently capable of inrush current limiting. In low-
power applications, this is a simple way to obtain rectifier, in which diodes D1 - D7 operate with zero
low-harmonic three-phase rectifier which incorporates voltage switching, while transistor Q1 operates with
high-frequency isolation transformers. It has the zero current switching. Input capacitors C r1 and dc-side
disadvantage of requiring an
3øac dc output
input filter for removal of the input
high-frequency components of ø T 1 i a(t) T 1
D D
1
D 2 3
D D D 7 8 9 +
a
the pulsating input current
T2 ib(t) T2
waveforms. øb
Q 1
C v(t)
A buck-derived single-
T3 ic(t) T3
switch rectifier [15] is øc D D
4
D 5 6
D D D 10 11 12 –
illustrated in Fig. 7. This input filter
converter is based on the zero-
current-switching (ZCS) Fig. 6. Single-switch three-phase DCM flyback rectifier.
quasi-resonant dc-dc buck 3øac L dc output
Q 1
input
converter. A resonant inductor +
Lr ia (t)
L r is placed in each input øa D D D 1 2 3

phase. As illustrated in Fig. Lr ib (t)

øb D Cr C v(t)
5(d), when transistor Q1 7

conducts, the resonant Lr ic (t)

øc D D D 4 5 6
inductors Lr ring in input filter
–

conjunction with resonant

Fig. 7. Single-switch three-phase zero-current-switching quasi-resonant buck rectifier.
capacitor C r. Input currents
3øac Q Ld dc output
ia(t), ib(t), and ic (t) are input 1

ia(t) L a +
approximately sinusoidal øa D D D
1 2 3
+ Lr
pulses, having peak amplitude C r1
–
vcra (t)
ib(t) L a
proportional to the applied øb C v(t)
C r1
line-to-neutral voltages v an(t), ic(t) L a D C r2 7
øc D D D
v bn(t), and v cn (t), C r1
4 5 6
–

respectively. The average

values of ia(t), ib(t), and ic (t) Fig. 8. Single-switch three-phase multiresonant zero-current-switching buck rectifier.

4
capacitor C r2 form a resonant network, in conjunction
with inductor L r. Inductors L a and L d operate in the
continuous conduction mode, with small switching
ripple. This converter exhibits nonpulsating input and
output currents, and requires minimal additional
filtering of the input current waveforms. This
converter exhibits low EMI and low switching loss.
The phase a resonant capacitor voltage waveform
v cra(t) is illustrated in Fig. 5(e). This voltage is
approximately sinusoidal, with peak amplitude
proportional to the input current ia(t). Approximate
input resistor emulation is therefore obtained. The
input current THD is a function of the value of L a;
THD less than 4% can be obtained at full load.
Transistor Q1 can be realized using an IGBT, SCR, or
other device. The peak transistor voltage is typically
twice the input line-to-line peak voltage. The dc load
power is controlled by variation of the switching
frequency.
A variety of other three-phase rectifier schemes
having a reduced number of switches are known. In
[18], a CCM PWM approach is described which
requires only three active devices. An approach
requiring two active devices which operate with zero-
current switching is described in [19].
4. Comparison of six-switch vs.
single-switch approaches
Consider an application in which it is desired to
replace a phase-controlled rectifier system with a high-
quality rectifier. It is therefore required that the output
voltage be of variable and controllable magnitude, less
than the peak input line-to-line voltage. Any of the
approaches of Fig. 3 could be employed in this
application; the buck topology of Fig. 3(a) is the
simplest. The single-switch converters of Figs. 6-8
could also be employed. Let us compare the buck-
derived single-switch rectifiers of Figs. 7 and 8 with
the six-switch buck rectifier of Fig. 3(a).
The active switch blocking voltages and peak
currents of the single-switch ZCS quasi-resonant buck
rectifier (Fig. 7) and six-switch CCM buck rectifier
(Fig. 3(a)) are compared in Tables 1 and 2, for a 25kW
application [15]. The input voltage is 440Vac, and the
load voltage and current are 370V and 67.5A,
Table 1. Comparison of per-switch stresses, single-
switch ZCS vs. six-switch PWM approaches
Voltage Peak current
ZCS rectifier 622V 100A
6-switch PWM 622V 46A
Table 2. Comparison of total active switch stress
Total stress Silicon utilization
ZCS rectifier 62.2kVA 0.4
6-switch PWM 171.7kVA 0.14
5
respectively. The total switch stress is also listed; this
is defined as the product of the switch blocking
voltages and peak currents, summed over all active
switches in the converter. The switch stress is a
measure of the total active silicon area required for
realization of the converter. Also shown is the silicon
utilization, defined as the converter output power
divided by the total switch stress.
In the dc-dc case, the quasi-resonant switch
approach is commonly thought of as restricted to low
power applications, because of the increased peak
switch stresses, poor switch utilization, and increased
conduction losses. However, these arguments do not
apply to the single-switch ZCS rectifier of Fig. 7,
because the 2:1 increase in peak current due to resonant
switching is more than offset by the 1:6 reduction in
total active semiconductor area arising from the single-
switch approach. The six-switch bridge network leads
to poor silicon utilization, since the semiconductor
devices are effectively utilized only near the peaks of
the applied ac phase current waveforms. For example,
if the six IGBT devices of a three-phase bridge module
were reconnected in parallel (forming a single device)
and then operated in an equivalent single-switch ZCS
rectifier, then the peak current density in each silicon
device would be reduced by a factor of one-third.
Table 3. Comparison of per-switch stresses, single-
switch multiresonant vs. six-switch PWM approaches
Voltage Peak current
multiresonant 685V 77.4A
6-switch PWM 340V 40.8A
Table 4. Comparison of total active switch stress
Total stress Silicon utilization
multiresonant 53kVA 0.113
6-switch PWM 83kVA 0.072
The single-switch multiresonant rectifier of Fig. 8
is compared with the six-switch buck rectifier of Fig.
3(a) in Tables 3 and 4. The ac input voltage is 240V,
and the load is 147V at 6kW - 0.6kW. The
multiresonant approach has the disadvantage of
increased peak switch voltage stress (by a factor of
approximately two); nonetheless, the total stress and
silicon utilization are superior to the six-switch
approach. The multiresonant rectifier exhibits the
advantages of (1) non-pulsating input currents, (2) very
low input current THD (less than 4% has been
demonstrated [17]), and zero voltage switching of all
diodes.
5. Conclusions
Three-phase high-quality rectifiers can be derived
from known dc-dc converter topologies. As a result,
rectifiers sharing the properties of their parent dc-dc
converters can be derived, including buck, boost, and
buck-boost conversion ratios, as well as high-
frequency transformer isolation. Both single-switch and
six-switch 3ø inputs can be obtained. The single-
switch approach utilizes the active semiconductor
devices more effectively. Zero current switching of the
active semiconductor devices, and zero voltage
switching of diodes, can be obtained.
R EFERENCES
[1] R. Erickson, M. Madigan, and S. Singer, “Design
of a Simple High Power Factor Rectifier Based on
the Flyback Converter,” IEEE Applied Power
Electronics Conference, 1990 Record, pp. 792-
801.
[2] S. Singer and R.W. Erickson, “Power Source
Element and Its Properties,” IEE Proceedings-
Circuits Devices Systems, vol. 141, no. 3, pp.
220-226, June 1994.
[3] S. Singer, “Realization of Loss-Free Resistive
Elements,” IEEE Transactions on Circuits and
Systems, vol. CAS-36, no. 12, January 1990.
[4] H. Mao, D. Boroyevich, A. Ravindra, and F. Lee,
“Analysis and Design of a High Frequency Three-
Phase Boost Rectifier,” IEEE Applied Power
Electronics Conference, 1996 Record, pp. 538-
544.
[5] B. T. Ooi, J. C. Salmon, J. W. Dixon, and A. B.
Kulkarni, “A Three-Phase Controlled-Current PWM
Converter with Leading Power Factor,” IEEE
Transactions on Industry Applications, vol. 23,
no. 1, pp. 78-84, 1987.
[6] P. Tenti and L. Malsani, “Three-Phase AC/DC
PWM Converter with Sinusoidal AC Currents and
Minimum Filter Requirements,” IEEE Transactions
on Industry Applications, vol. 23, no. 1, pp. 71-
77, 1987.
[7] A-M. Majed, T. C. Green, and B. W. Williams,
“Dynamic Properties of a Step-Down Sinusoidal
Current AC/DC Converter Under State-Feedback
Control,” IEEE Applied Power Electronics
Conference, 1993 Record, pp. 161-167.
[8] M. Rastogi, N. Mohan, and C. Henze, “Three-
Phase Sinusoidal Current Rectifier with Zero
Current Switching,” IEEE Applied Power
Electronics Conference, 1994 Record, pp. 718-
724.
[9] K. D. T. Ngo, S. Cuk, and R. D. Middlebrook, “A
New Flyback Dc-to-Three-Phase Converter with
Sinusoidal Outputs,” IEEE Power Electronics
Specialists Conference, 1983 Record, pp. 377-
388.
[10] M. J. Kocher and R. L. Steigerwald, “An Ac to Dc
Converter with High Quality Input Waveforms,”
IEEE Power Electronics Specialists Conference,
1982 Record, pp. 63-75.
[11] A. R. Prasad, P. D. Ziogas, and S. Manias, “An
Active Power Factor Correction Technique for
Three-Phase Diode Rectifiers,” IEEE Power
Electronics Specialists Conference, 1989 Record,
pp. 58-66.
[12] J. Kolar, H. Ertl, and F. Zach, “Space Vector Based
Analysis of the Input Current Distortion of a Three
Phase Discontinuous Conduction Mode Boost
Rectifier System,” IEEE Power Electronics
Specialists Conference, 1993 Record, pp. 696-
703.
[13] R. Itoh and K. Ishitaka, “Three-Phase Flyback Ac-
Dc Converter with Sinusoidal Supply Currents,”
IEE Proceedings, vol. 136, part B, no. 4, pp. 143-
151, 1991.
[14] O. Apeldoorn and P. Schmidt, “Single Transistor
Three-Phase Power Conditioners with High Power
Factor and Isolated Output,” IEEE Applied Power
Electronics Conference, 1994 Record, pp. 731-
737.
[15] E. H. Ismail and R. W. Erickson, “A Single
Transistor Three-Phase Resonant Switch for High
Quality Rectification,” IEEE Power Electronics
Specialists Conference, 1992 Record, pp. 1341-
1351.
[16] Y. Jang and Robert W. Erickson, “New Single-
Switch Three-Phase High Power Factor Rectifiers
Using Multi-Resonant Zero Current Switching,”
IEEE Applied Power Electronics Conference, 1994
Record, pp. 711-717.
[17] Y. Jang and R. Erickson, “Design and
Experimental Results of a 6kW Single-Switch
Three-Phase High Power Factor Rectifier Using
Multi-Resonant Zero Current Switching,” IEEE
Applied Power Electronics Conference, 1996
Record, pp. 524-530.
[18] J. Kolar, H. Ertl, and F. Zach, “Design and
Experimental Investigation of a Three-Phase High
Power Density High-Efficiency Unity-Power-
Factor PWM (VIENNA) Rectifier Employing a
Novel Integrated Power Semiconductor Module,”
IEEE Applied Power Electronics Conference, 1996
Record, pp. 514-523.
[19] M. Rastogi, N. Mohan, and C. Henze, “Three-
Phase Sinusoidal Current Rectifier with Zero
Current Switching,” IEEE Applied Power
Electronics Conference, 1994 Record, pp. 718-
724.
Robert W. Erickson received the B.S., M.S., and
Ph.D. degrees from the California Institute of
Technology, Pasadena, CA, U.S.A., in 1978, 1980,
and 1982, respectively. He then joined the Department
of Electrical and Computer Engineering at the
University of Colorado, Boulder, U.S.A., where he
currently holds the rank of Associate Professor. He is
the author of numerous conference and journal papers,
as well as the soon-to-be-published textbook
Fundamentals of Power Electronics.