2004 351h A n n u l IEEE Power Electronics Specialists Conference Aachen.

Germany, 2004

Closed-Loop SPWM Control for Grid-Connected Buck-Boost Inverters

Yaosuo Xue and Liuchen Chang, IEEE Senior Member

Department of Electrical and Computer Engineering
University ofNew Brunswick
PO Box 4400, D36 Head Hall, 15 Dineen Drive
Fredericton, NB, Canada E3B 5A3
Email: LChang@unb.ca

single-stage buck-boost inverter, the output current usually

Abslracf-DC input voltage of inverters fluctuates
dramatically in distributed generation applications such as in a
wind energy system. Yet, a high quality ac output is required
for grid interconnection under variable source conditions.
Previously developed control strategies mainly focused on
improvements under load variations, a de input ol’ relatively
small ripples, etc. This paper proposed a closed-loop sinusoidal
PWM control method with real-time waveform feedback
techniques for a grid-connected buck-boost inverter. The
control-to-output function was derived through steady state
modeling based on the power balance condition, which
provides an approach when the output cannot easily he
characterized in a single-stage buck-boost inverter. The closed-
loop control model was studied with a newly-invented single-
stage buck-boost inverter circuit. Simulations verified the
method provided fast dynamic response and robustness under
large de voltage variations, non-ideal grid voltage, and
component parametric uncertainties. The controlled inverter
achieved a low-THD sinusoidal output with a small ac filter
and without a dc link capacitor. Therefore, it is concluded that
the proposed method can he a preferred choice for grid-
connected buck-boost inverters in distributed generation
systems.
I. INTRODUCTION
Myriads of voltage- and current-mode strategies have
devoted to the control of inverters in power electronics. Of
those voltage control strategies, space vector modulation
(SVM) was widely used in three-phase inverters. However,
it requires extensive calculations, thereby limiting the
bandwidth of control systems. Function approximation can
simplify the execution, but also reduces the accuracy. Some
improved SVM methods based on artificial neural networks
[I] [2] can reduce computations while still retain accurate
results. Random PWM [3] addresses the problems of
acoustic noise and radio interference. Optimal switching
pattern PWM methods [4] [5] were proposed to either
reduce the output filter size or compensate for a non-ideal
dc-link voltage. When large disturbances and uncertainties
exist, these methods need to be combined with complicated
model reference adaptive systems [6], H , loop-shaping [7],
and fuzzy logic or neural network controllers [XI-[IO], to
track desired trajectories or reference models.
Current control strategies, like hysteresis current control
(HCC) [ I I]-[14], provide the tightest control of output
current in a traditional full-bridge inverter. However, for a
cannot be directly controlled and these methods introduce
lower-order harmonics in the output current.
Variable structure control [ 151 was introduced and
applied to power electronics [I61 with the strengths of large-
signal stability and robustness. Although this method is
extremely useful for tracking problems, it is not popular due
to its theoretical complexity and difficulty in finding a
satisfactory sliding surface [17]. Furthermore, sensing o f all
state variables and generating of references [I81 adds to the
cost for a low power inverter system.
Single-stage buck-boost inverters can operate under wide
dc input voltage ranges and variations, thus presenting a
low-cost solution for small wind energy distributed
generation systems. This paper is intended to establish a
simple yet effective control method for such inverters
tolerant of large dc voltage fluctuations and ac grid
variations. Specifically, a new closed-loop sinusoidal PWM
(SPWM) control method is proposed through the evolution
from open-loop, reference feedforward, to closed-loop
control analyses. Advantages of this method are
demonstrated by simulation of a grid-connected single-stage
buck-boost inverter.
11. CLOSEDLOOPSPWM CONTROL WITH REAL-TIME
WAVEFORMFEEDBACK
SPWM, as illustrated in Fig. I , is a carrier-based voltage
control method. As known in a SPWM inverter, the
spectrum of harmonic frequencies in the output is shifted
towards the high frequency, and the lower-order harmonics
are reduced significantly. Thus, the output filter elements
can be quite smaller to attenuate only higher-order
harmonics, which are carrier-frequency-related with
modulating-frequency sidebands.
Averaging method [19] [20] is widely used to analyze
PWM switching power converters [21] [22] based on the
principles of inductor volt-second balance and capacitor
charge (ampere-second) balance with % ( I ) and v,(t) being
state variables. But this method is used for a limited class of
converters, mainly for dc-dc converters. Low frequency
describing method was proposed in [23] and applied to
analyze inverters [24]. Generalized describing function
method was proposed in [25] by performing frequency
analysis with complexity of derivations. For grid-connected

0-7803-8399-0/04/$20.0002004 IEEE. 3366

 2004 351h Annual IEEE Power Elecrronics Specialisrs Conference Aachen. Germany. 2004

1
& i t =- (5)
r f,
ztSlI a I ,
nnnnn With a unity carrier signal ( A, = 1 ), the reference signal
lid I, !I I , I

d
gw; I' '
'' t is defined as:
4 r: I- -I&J- T D VT y,(t) = A, sin wf = M sin OT (6)
Fig. I . Sinusoidal pulsewidth modulation.
First, the maximum charging pulse width, qrn, can be
single-stage buck-boost inverters, the differential equations determined. Considering the where N is an even
of the Power circuits are usually at least 3 Or 4 orders, number, there will be only one maximum charging pulse at
higher than those of dc-dc converters and traditional the centre of each haif period, as shown in ~ i 1. F~~~
~ . the
inverters, thus it is hard to model the inverter with trigonometry, we have
aforementioned methods. A simple analytic method, based
on the input-output power balance condition, is proposed to 2
tga = -
characterize this class of inverters. 1s

A. Control-to-output funcfion, open-loop SP WM modeling Msin(---)=-

z7 WTrn Trn Trn
2 2 2 Iga=- ts
For a flyback buck-boost inverter, the typical input
current wave is a series of discontinuous charging triangular where a is the angle between the carrier wave and t-axis as
pulses as presented in Fig. 2, when S P W M switching shown in Fig. I .
function g is applied and the circuit is working in the
discontinuous-conduction-mode (DCM) operation with a Since A. >>Yo, sin(---) 4 r n i 1 , and
fixed switching frequency. 2 2
Trn it: Mt, (7)
Because tl = dt3, Eqs. 2,3, and 7 result in

d ( t ) = M sin 01 (8)
Secondly, the maximum peak point of the charging
Fig. 2. lnpul current of SPWM invener. current I, can be solved.
di
Considering that the switching frequency is much higher L-=V,
dl
than the modulating frequency, i.e. f,>>A,, we can define
the peak value of input current i,(t), charge time t,(f) , and VATm -
I Irn =--- (9)
charging duty cycle d(t) as the following: L LA
where L is the inductance of energy-storage component.
& ( I ) = I, sin wf (I) Finally, integrating i,(t) will yield the average value I , ,
t,(r) = Trnsin wf (2) Refer to Fig. 2 and perform the following integration:

(3)

where U = 2i&, f, is the output frequency, Irm, T,,, and

0, are the maximum peak charging current, maximum
charge time, and maximum charging duty cycle,
respectively. (Let x=wf and w T = 2 n )
In each half period, the number of carrier signal is
I.

4
For a large N, the switching period t.: is approximately
uniform and For a grid-connected inverter, applying the input-output

3367
 2004 35Ih Annual IEEE Power Elecrronics Speciolirrs Conference
power balance condition and combining Eqs. 9 and IO will
generate therms output current.
Furthermore, with the charge time f l proportional to the
modulation index M, a maximum modulation index M,,
can be determined by the critical DCM condition in the
switching cycle with maximum charging current, where
I, = T,, + T2m. Assume a constant output capacitor voltage
during the discharging interval, V, = &V ,then
Also T2,,, = I, - qm= /,T -MI, = (I - M ) / , .
Thus the maximum permissible modulation index. for
given dc voltage and grid can be expressed as,
A V
M,, =-
v,+ A V
As a result, the full control-to-output function is given by
Eq. 13 accounts for the effect of switching frequency on
the output power within a valid frequency range, while
averaging methods cannot give this property. It also
specifies the maximum modulation index applying to the
inverter beyond which an overmodulation will cause
“current jump” problems where the desired energy cannot
be transferred to the output in some switching cycles.
In addition, the output current is demodulated from the
discharging current which has the same peak pulses as the
charging current. Interest is taken in the ratio of output
current’s fundamental component, to the peak value of
unfiltered output current, which can be computed as
follows.
B. Reference feedforward SP WM control
A simple inverse feedforward compensator (FFC)
combined with SPWM as in Fig. 3 can be “good” enough to
reject the effect of dc voltage variations. Fed with a wide
range of fluctuated dc voltage, FFC can automatically adjust
the instantaneous modulation index to keep a desired output.
However, it does not compensate for the effect from a
distorted grid voltage waveform.
3368
Aochen, Gennany, 2004
Fly 3 SPWLI with IbdionarJ &nnF.nsalor
(S2H?IIS a single-stage huck.bom mener.1
The feedforward compensator takes D demanded current
as reference and derives the required modulation index
either from a look-up table or a function approximation of
M = [-]” subject to M-2
JjlV
V, + f i r
(15)
C. Closed-loop SP WM control
A real-time waveform feedback closed-loop SPWM
control model is proposed in Fig. 4, where the dc voltage
fluctuation, grid voltage variation, and other constant
component uncertainties are modeled as disturbances. The
grid voltage variation contains lower-order harmonics; the
dc voltage fluctuates half-wave sinusoidally at 6 times of
output frequency when obtained from a three-phase bridge
rectifier; and the constant parametric uncertainties are
usually inherent in the buck-boost inverters. In this control
model, these disturbances are not measured and only the
output current is sampled through a lowpass filter.
Fig. 4. Closed-loop SPWM control block diagram.
Both PI controller and PID controller have been
investigated. For a PI controller, there exists a tradeoff
between the response time and overshoot. By introducing
the rate control of D controller, both overshoot and rise time
can be reduced together along with additional insensitivity
to grid variations. The PID controller with the form of Eq.
16 is designed or tuned to reject the aforementioned
disturbances.
C ( s )= K + Kf + K,s (16)
‘ T
Ill. EXAMPLE:A SINGLE-STAGE BUCK-BOOS1
INVERTER
Fig. 5 illustrates the topology of a newly-developed
single-stage buck-boost inverter with 4 power switches [26].
 2004 35lk Annual IEEE Power Elecrronics Specialists Conference
It is composed of one energy charge loop ( T I , L I , and T,)
and two energy discharge loops (T,, Ll, C, Dj, and T, for the
positive half cycle, T,, L1, D1,T1,and C for the negative half
cycle). The energy-storage components, Ll and L2, are the
primary and secondary windings of a flyback transformer
and have the identical inductances ( L ) and NmS.
Fig. 5. A single-stage buck-boost invener
With the switch gating signals generated by the SPWM
control scheme shown in Fig. 6 , the above-derived control-
to-output function is validated by the simulations of
different dc input voltage and modulation indexes, as
marked in Fig. 7.
Fig. 6 . SPWM generator
Cantml-to-OutputCUM. of S P W M Conlml
Oo 03 0.2 0.3 a.4 0.5 os 0.7
Modulation Index. M
Fig. 1.Control-to-output Curves of SPWM control.
(L=9.6 k M , L = 150 x I04H, V = 12011
The leakage inductance of flyback transformer can lead to
an asymmetric output waveform between each half cycle.
For the pure open-loop SPWM control, a constant
coefficient K (14 > 1) in SPWM generator has to be
specified to compensate this asymmetry. However, with
Ancken, Germany, 2004
feedfonvard control or closed-loop control, this coefficient
can be simply specified as - 1 and the reference wave will be
adjusted dynamically.
A . Referencefeedfirward SPWM control results
Simulation waveforms with a feedforward compensator
are presented in Fig. 8, where the look-up table pre-
computed from Eq. 15 and linear interpolation are applied.
Although the dc input comes directly from rectified three-
phase generator without using a dc filter, a constant 1 kW
output power is obtained under the large variable dc-link.
The output THD is about 2% with a 9.6 kHz switching
frequency.
Vdc
0 00 0 02 0.W 0.08 0.06 0.10
Time ( 5 )
Fig. 8. Simulation waveforms of FFC-SPWM.
B. Closed-loop SP WM control results
Closed-loop SPWM control with a PID controller (KP=
0.01, Kl = 333; KO = 5x10") was applied to reject the
constant unknown parametric uncertainty (LdL) and low
frequency sinusoidal disturbances, such as 3d, 5', and 7'-
order harmonics in the grid voltage and a 6 w 0 rectified sine
variation in the dc voltage. Fig. 9 illustrates the state space
block diagram for the example inverter. Fig. 10 shows the
simulation results, where the THD of output current is about
3 .A% and the output capacitor used is only 10 9.
iC
0.8 0.9 I
Fig. 9. State space block diagram.
3369
 2004 35th A n n u l IEEE Power Electronics Specialists Conference Aachen, Germany, 2004

Vdc

................

e@)
3.00
2.00
1 .oo
0.00
-1.oo
-2.00

15.00
...

0.00 0.05 0.10 0.15

Time (s)
Fig. IO. Simulation waveforms of closed-loop SPWM control

Simulation demonstrates that the closed-loop system has
fast dynamic response and good tracking performance in the
following cases: 1) The grid voltage contains 5 1 0 % 3”-,
5*-, orland 7*-0rder harmonics; 2) The dc-link voltage, as
obtained from a three-phase rectifier, contains large ripples
and the dc link capacitor is completely eliminated; 3) The
high frequency flyback transformer contains certain
leakages. The PWM reference signal is automatically
adjusted to compensate output asymmetry dynamically
without pre-assigning a fixed compensation coefficient as in
the open-loop control; 4) Output current setpoint suddenly
changes at any time.
Computer simulation shows both current control methods
demonstrate similar results as in Fig. I I . Compared to
SPWM control, although the output current can be increased
with regards to the peak value of unfiltered output current, it
is distorted with lower-order harmonics, even with a 3‘d-
order harmonic compensation in the current reference.
10,OO .................. ,.. ........-.................
5.00 .................. :.... ......-... .....-.........

0.00 ........ .:...... ...; ..........L.. ......;. ........

C. Current control results
-5.00 ......... :..... ____~______ ....:. .......:.. ......
For the purpose of comparison, current control methods
were also conducted with the proposed single-stage buck- -qo,oo ..........: ................. ..’ .........:.... . . .
boost inverter. For a current-mode controller, with the -15.00
magnitude limitation of the sampled inductor current by a 100.00 110.00 120.00 130.00 140.00 150
sinusoidal reference current and the detection of its zero Time (ms)
point, the circuit can be controlled to operate on the critical Fig. I I . Simulation waveform ofhysteresis current control
DCM mode, i.e. no idle time in each switching period.
(V,= 190V,IEy21A.
THD=6%)
In addition, hysteresis current control was also applied to
implement a continuous-conduction-mode (CCM)
IV. CONCLUSION
operation. The inductor current,
iL = + in, in the positive half cycle; A real-time waveform feedback closed-loop SPWM
control method is proposed, analyzed and simulated through
i, = - iT,, in the negative half cycle, a newly-invented single-stage buck-boost inverter. A simple
is controlled continuously within a specified envelope of a steady state modeling method is also proposed for the
reference signal by proper control logics. inverters where the output current is hard to characterize. It

3370
 2004 35th A n n u l IEEE Power Electronics Speciolists Conference
was shown that the proposed control method with a PID
controller provided fast dynamic response, robustness and
insensitivities to dc and ac variations as well as parametric
uncertainties. The controlled inverter can achieve a high
quality sinusoidal output with small ac filters and without dc
filters. Therefore, this control method is preferred for grid-
connected buck-boost inverters, which regularly work under
a wide range- of dc voltage - variations in small distributed
generation applications.
REFERENCES
A. Bakhshai, G. Joas, and H. Jin, “Application of the Kohonen’s
competitive layer in the implementation of the space vector
modulation,” Pror IEEE Elcct. Comput. Eng. Can. Coni
(CCECE’97).SI. John’s, Canada, 1997, pp. 415-418.
F. Blaabjerg, S. Freysson, H.-H. Hansen, and S. Hansen, “A new
optimized space-vector modulation strategy for a companent-
minimized voltage source inverler;’ IEEE Trans. Power Electron.,
vol. 12, pp. 704-714, July 1997.
S. Y. R. Hui, S. Sathiakumar, and K.-K. Sung “Navel random PWM
schemes with weighted switching decision,” IEEE Tram. Power
Electron., vol. 12, pp. 945-952,Nov. 1997.
M. Pande, G. Jws, and H. Jin, “Output voltage integral contml
technique for compensating nonideal dc buses in voltage source
inverters,” E E E Trons. Power Electron.. vol. 12, pp. 302-310, Mar.
1997.
A. M. Trzynadlowski and S. Legowski, “Application of neural
networks to the optimal control of three-phase voltagewntrolled
inverters,” IEEE Trans.Power Electron., vol. 9, pp. 397-404, July
1994.
G. Jws, P. D. Ziogas, and D. Vincenti, “A model reference adaptive
PWM Technique,” IEEE Trans. Power Electron., vol. 5 , pp. 485-494,
Oct. 1990.
T:S. Lee, S:J. Chiang, and J:M. Chang, “ H , loop-shaping
controller designs far the single-phase UPS inverters,” IEEE Trans.
Power Electron., vol. 16, pp. 473-481, July 2001.
P. 1. Werbas. “Neumwntrol and elastic fuzzy logic: capabilities,
concepts, and applications,’’ IEEE Trons. lnd. Electron.. vol. 40. pp.
170-180. Apr. 1993.
X. Sun, M. H. L. Chow, F. H. F. Leung, D. Xu, Y. Wang, and Y:S.
Lee, “Analogue implementation of a ieural network cintraller for
UPS inverter applications,” IEEE Tram. Power Electmn., vol. 17, pp.
305-313, May 2002.
1101 A. M. Trzynadlowski and S. Legowski, “Application of neural
networks to the optimal control of three-phase voltage-controlled
inverters,” IEEE Trans. Power Electron., vol. 9, pp. 397-404, July
3371
Aarhen, Germany. 2004
1994.
[ I I] D. W. Cyma, “A novel control method to minimim distortion in ac
inverters,” IEEE Trans.Power Electron., vol. 12, pp. 862-866, Sept.
1997
1121 A. I. Maswood and M. A. Rahman, “Boosting the fundamental output
voltage in delta modulated inverters,” IEEE Power Eng Rev., vol. 20.
pp. 51-53, Dec. 2000.
1131 K, M, Rahman, M, R, Khan, M, A, Choudhury, and M, A, Rahman,
“Variable-band hysteresis cumnt controllers for PWM voltage-source
inverters.” IEEE Tram. Power Electron., vol. 12. ..
DD. 964-970, Nov.
1997.
1141 M. Ohshima and E. Masada, “A single-phase PCS with a novel
wnstantly sampled current-regulated PWM scheme,” IEEE Tram
Power Electron., vol. 14, pp. 823-830, Sept. 1999.
(151 J. Y. Hung, W. Gao, and J. C. Hung, “Variable structure control: a
suwey,” IEEE Tram. I d . Electron., vol. 40, pp. 2-22, Feb. 1993.
1161 R. 0.Caceres and 1. Barbi, “A bwst dc-ac converter: analysis, design,
and experimentation,”IEEE Tons. Power Electron., vol. 14, pp. 134-
141,Jan. 1999.
(171 Y.-Y. Tzou, R.-S. Ou, S.-L. Jun& and M.-Y. Chang, “High-
performance programmable ac power source with low harmonic
distortion using DSP-based repetitive control technique,” IEEE Tram.
Power Electmn., vol. 12, pp. 715-725, July 1997.
1181 P. Manavelli, L. Rosseno, and G. Spiazzi, ”Small-signal analysis of
dc-dc converters with sliding mode control,” IEEE Trans. Power
Electron., vol. 12, pp. 96-102, Jan. 1997.
[I91 S. R. Sanders, J. M. Noworolski, X. 2. Liu, and G. C. Verghese,
“Generalized averaging method for power conversion circuits: 1EEE
Tram. Power Electron., vol. 6, pp. 251-259, Apr. 1991
1201 1. Sun, D. M. Mitchell, M. F. Greuel, P. T. Krein, and R. M. Basis,
“Averaged modeling of PWM wnverters operating in discontinuous
conduction mode,” IEEE Trans. Power Electron., vol. 16, pp. 482-
492, July 2001
1211 Y.-F. Liu and P. C. Sen, large-signal modeling ofhysteretic current-
programmed converters,” IEEE Tram Power Electron., vol. I I, pp.
423430, May 1996.
[22] F. Guinjoan, J. Calvente, A. Poveda, and L. Martinez, “Large-signal
modeling and simulation of switching d c d c converters,” IEEE Tram.
Power Electron., vol. 12, pp. 485-494, May 1997.
(231 K. D. T. Nga, “Low frequency characterization of PWM converters,”
IEEE Trons. Power Electron., vol. PE-I,pp. 223-229, Oct. 1986.
(241 R.-S. Lai and K. D. T. Ngo, “A PWM method for reduction of
switching loss in a full-bridge inverter,” IEEE Tram. Power
Electron., vol. IO, pp. 326-332, May 1995.
1251 R. Tymenki, “Frequency analysis of time-intervalmodulated
switched networks,” IEEE Trans. Power Electron., vol. 6, pp. 287-
295,Apr. 1991.
1261 Y. Xue, ‘‘Analysis, simulation, and test of a novel buck-boost
inverter,” M.S. thesis, Dept. Elect. & Comp. Eng., University o f New
Brunswick, Fredericton, Canada, Jan. 2004