Vous êtes sur la page 1sur 5

Design of Controller for Buck-Boost Converter

V. D. Yurkevich
Novosibirsk State Technical University, Novosibirsk, 630092, Russia, e-mail: yurkev@ac.cs.nstu.ru

Abstract The problem of output regulation with guaranteed transient performances for buck-boost converter
with inverting topology is discussed. The fast dynamical
controller with the relative highest derivative of output
signal in feedback loop is used. Consequently, two-timescale motions are induced in the closed-loop system. Stability conditions imposed on the fast and slow modes and
sufficiently large mode separation rate can ensure that the
full-order closed-loop system achieves the desired properties in such a way that the output transient performances
are desired and insensitive to external disturbances and
parameter variations in the system. The existence of stable
limit cycle in the fast motion subsystem gives the robustness
of the output transient performances in the presence
of external disturbance and parameter uncertainty. The
describing function method is used to analyze the existence
and parameters of stable limit cycle.

I. INTRODUCTION
There is a broad set of references devoted to analysis
and design of switching buck, boost, or buck-boost
converters. In the most of references the derivation of the
converter circuit topology is discussed really [1], [3][6],
[15], rather than methods of switching controller design.
The subject matter of this paper is the guaranteed
cost control for buck-boost converter with inverting
topology under uncertainties of parameters and external
disturbances represented by varying value of a load
resistance. Hence, optimization techniques can not be
applied for the discussed control problem solution. As
far as nonsmooth nonlinearities are inherent property of
such power converters, then the control system design
methodology based on sliding modes [2], [8], [9] is
widely used for this purpose in presence of uncertainties.
The control system with the highest derivative in
feedback [11], [12] applied to a buck-boost converter is
discussed in this paper as well as peculiarities caused
by fast oscillations in the system. Note that the analysis
of fast oscillations by the describing function method
in the control systems with the highest derivative and
differentiator in feedback was discussed in [10]. In the
recent paper the modified control law structure [13], [14]
in the form of the fast dynamical controller with the
highest derivative of output signal in feedback loop is
used. The proposed control law structure allows us to
include the integral action in the control loop without
increasing the controllers order.

The paper is organized as follows. First, a model of the


buck-boost converter with inverting topology is defined.
Next, the discussed design method, influence of fast
oscillations, and simulation results are presented.
II. BUCK-BOOST CONVERTER
A. Buck-boost converter with inverting topology
Let us consider the buck-boost converter with inverting
topology shown in Fig. 1 (see, e.g., [4], [6], [15]). The

Fig. 1.

Buck-boost converter circuit.

switched model of the buck-boost converter is given by


E
x2
x 1 =
u (1 u),
(1)
L
L
x1
1
x2 + (1 u),
(2)
x 2 =
RC
C
where x1 = IL , x2 = VC = Vout , and u takes values in
the set {0, 1}. If the transistor is ON (OFF), then u = 1
(u = 0).
B. Buck-boost converter control task
The control problem is to provide the following condition:
lim VC (t) = VCd
(3)
t

where VCd is the reference value (reference input) of voltage drop VC across a capacitor. Moreover, the controlled
transients VC (t) VCd should have desired transient
performance indices. These performance indices should
be insensitive to parameter variations of the buck-boost
converter and external disturbance represented by varying
value of the resistor R = R(t).
In the paper a two-step approach will be used: an inner
controller of the current IL through the inductor with
inductance L is desined such that
lim IL (t) = ILd ,

(4)

and then an outer controller is conctructed in order to


meet the requirement (3).

III. INNER CONTROLLER DESIGN


A. System with continuous control variable
Let x1 be the measurable output of the system (1)(2)
and consider the system given by
x2
E
u
(1 u
),
L
L
x1
1
x2 + (1 u
),
=
RC
C

x 1 =

(5)

x 2

(6)

where u
is the continuous control variable. Let u
(0, 1).
Hence, the system (5)(6) describes an average behavior
of the system (1)(2).
First, assume that
x1 (t) = r1 = const,

t [0, )

(7)

where r1 = ILd . Then from (5) and (7), we get


0=

E + x2
x2
u
.
L
L

(8)

Denote u
r1 = u
(t) x1 (t) = r1 as the solution of (8).
Hence, we obtain
u
r1 =

x2
,
E + x2

x2 = E

u
r1
.
1u
r1

(9)

Since u
r1 (0, 1), we obtain x2 (0, ). Then the
system (5)(6), having dimension 2, degenerates into the
system
x1 = r1 = const,


1
x2
1
x 2 =
1
x2 +
r1 ,
RC
C
E + x2

(10)

having dimension 1. The degenerated system (10) has the


unique asymptotically stable positive equilibrium point
xs2 given by
#
"r
4r1 R
E
s
1 .
(11)
1+
x2 =
2
E
Therefore, the internal stability of the system (5)(6) is
satisfied under condition that x1 = r1 (or, in other words,
the system (5)(6) is the minimum phase system).
Second, the variable x1 is considered as the output of
the system (5)(6). From (5), it follows that the relative
degree of this system equals one. Hence, let the desired
(1)
output behavior of x1 be assigned by x1 = F (x1 , r1 ),
where
1
(1)
x1 =
[r1 x1 ].
(12)
T1
The deviation between the desired dynamics F (x1 , r1 )
assigned by (12) and the actual value of the relative
(1)
highest output derivative x1 is denoted by
(1)

eF = F (x1 , r1 ) x1 ,

where eF is the error of the desired dynamics realization.


Then the control problem represented by (4) corresponds
to the insensitivity condition given by
eF = 0.

(13)

Third, the relative degree of the system (5)(6) equals


1 and its internal stability is satisfied. Therefore, the
control law with the 1st output derivative in feedback
31 u
(3) + d2 21 u
(2) + d1 1 u
(1) + d0 u

(1)

= k1 {T11 [r1 x1 ] x1 }

(14)

can be applied in order to meet the requirement (13),


where 1 is a small positive parameter.
We see that in the closed-loop system given by (5)
(6) and (14), the two-time-scale motions are induced as
1 0. Hence, we obtain the fast-motion subsystem
(FMS) given by


E + x2
3 (3)
2 (2)
(1)
1 u
+ d 2 1 u
+ d 1 1 u
+ d0 + k 1
u

L


r1 x 1 x 2
= k1
,
(15)
+
T1
L
where x1 , x2 are the frozen parameters during the transients in (15) and Tf ms = 1 /[d0 + k1 (E + x2 )/L]1/3 is
the time constant of the fast-motion subsystem.
Assume that the control law parameters
1 , d2 , d1 , d0 , k1 have been selected such that the
FMS (15) is stable as well as time-scale decomposition
is maintained in the closed-loop system. Then, letting
0 in (15), we obtain the steady state (more
precisely, quasi-steady state) of the FMS (15), where
u
(t) = u
s (t) and


k1 L
r1 x 1 x 2
s
.
(16)
u
=
+
d0 L + k1 (E + x2 )
T1
L
Substitution of (16) into (5)(6) yields the slow-motion
subsystem (SMS) given by


d0 L
r1 x1 x2
r1 x1

+
x 1 =
(17)
T1
d0 L+k1 (E + x2 )
T1
L
1
x1
x 2 =
x2 +
RC
C



k1 L
r1 x1 x2
1
. (18)
+
d0 L+k1 (E +x2 )
T1
L
The behavior of x1 in the SMS (17)(18) approximates
to (12) if d0 = 1 and k1 . If d0 = 0 then, (17)
is the same as (12) and by that the integral action is
incorporated in the control loop. Note that if d0 = 0
then, by letting x1 = r1 in (17)(18), we obtain the
degenerated system (10).
By linearization of (10) at the equilibrium point xs2 we
obtain z = aint z , where
2r1
1
i,
h
aint =
p
RC
C 1 + 1 + 4r R/E
1

and Tint = 1/aint is the time constant of the linearized


internal subsystem at the point xs2 . Take
E = 15 V, L = 0.02 H, C = 0.001 F, R = 200 ,
d0 = 0, T1 = 0.02 s, 1 = 0.002 s, k1 = 0.001.

Hence, from the above, we get xs2 3.28 V, Tint 0.03


s, and Tf ms 0.0021 s when r1 = 0.02 A. Note that
Tint  Tf ms where Tf ms 0 as 1 0. Similarly,
xs2 48 V, Tint 0.0036 s, Tf ms 0.0014 s when r1 =
1 A. The parameters T1 and 1 have been selected such
that T1  Tf ms and Tint  Tf ms in order to produce
slow-fast decomposition in the closed-loop system.
B. Switching regulator design
As the next step, let us consider the system (1),(2)
governed by the following switching regulator:
(3)

(2)

(1)

31 u1 + d2 21 u1 + d1 1 u1 + d0 u1

(19)

(1)

= k1 {T11 [r1 x1 ] x1 }, u = [1 + sgn(u1 )]/2.

From (1) and (19), we get the block diagram of the FMS
shown in Fig. 2, where
D(1 s) = 31 s3 + d2 21 s2 + d1 1 s + d0 .

Fig. 2. Block diagram of the FMS in the closed-loop system (1)(2)


controlled by (19).

Assume that a limit cycle exists in the FMS shown


in Fig. 2 and the nonlinearity input u1 (t) is given by
u1 (t) = u 01 + A sin(t) where u01 is the constant bias
signal. Take d0 = 0. Hence, due to the integral action
incorporated in feedback loop, we have
Z t+2/
eF (t)dt = 0
(20)
t

for the stationary oscillations in the FMS. So, the average


value of eF corresponds to the insensitivity condition
(13) and the desired behavior of x1 (t) with assigned
dynamics (12) is satisfied if sufficiently fast oscillations
take place. Therefore, the expression (20) represents the
insensitivity condition of x1 (t) with respect to external
disturbances and variations of parameters of the buckboost converter in the average sense. We can see that the
key element to reach the desirable behavior of x1 (t) is
the existence of the fast oscillations in the FMS.
Then, in accordance with the describing function
method, let us replace the relay switch by its quasi-linear
approximation. Let the sinusoidal transfer function
k1 /D(1 s)

in Fig. 2 displays a low-pass filtering property. Consider


the output u(t) of the nonlinearity represented by its
Fourier series

X
u(t) = u0 +
{bk sin(kt) + ck cos(kt)}
(21)
k=1

with coefficients u0 , bk , ck . The particular feature of


the discussed system is the nonsymmetric limits of the
nonlinearity. Hence, u01 6= 0, u0 6= 0, and it is known
that for the given nonlinearity we have (see, e.g., [7])
 0
1 1
1 u1
u0 = + sin
,
2
A
s
 0 2
u
2
b1 =
1 1 , c1 = 0,

A
where A |u01 | and y = sin1 (x) denotes the inverse
sine of x. Therefore, the sinusoid plus bias describing
function of the discussed nonlinear element has the gain
for the bias u0 /u 01 and the gain for the sinusoid
s
 0 2
2
u
Gn (j, A) =
(22)
1 1 .
A
A

Assume that F = 0 and d0 = 0. Then, by the block


diagram shown in Fig. 2, we get the balance equation
for the constant bias signal u01 of the discussed FMS:

 0 
1 1
x2
1 u1

+ sin
= 0.
(23)
2
A
E + x2
The 1st order harmonic balance equation for the FMS
shown in Fig. 2 yields
s
 0 2
2k1 (E + x2 )
u
(24)
1 1 = 0.
1+
LAD(j1 )
A
From (24) we obtain

where

m2 A4 A2 + [u 01 ]2 = 0,
m=

(d2 d1 d0 )L
.
2k1 (E + x2 )

d1 /1 ,

(25)
(26)

The oscillations in the FMS induce the oscillations in


x1 (t) and have an influence on accuracy of stabilization
for x1 (t). Let eosc be the amplitude of the stationary
oscillations of x1 (t) with frequency . In accordance
with Fig. 2, we get to a first approximation
(d2 d1 d0 )1

A
k1 d 1
given that is sufficiently large. Note that and
eosc 0 as 1 0. Hence, an acceptable level of
ripple for the output voltage Vout can easily be provided
by selection of 1 .
Take T1 = 0.02 s, 1 = 0.002 s, k1 = 0.001, d0 = 0,
d1 = 16, and d2 = 5. From (25) we get 2000
eosc

is represented by the varying resistance R = R(t)


of the buck-boost converter (BBC) and there are two
controllers: the designed above inner switching controller
C1 and an outer continuous controller C2 . Here, we
d .
denote r2 = Vout

2
x1 (t)
r1 (t)

1.5
1
0.5
0
0.5
45
40
35
30
25
20
15
10
5
0

0.05 0.1 0.15 0.2 0.25 0.3


Fig. 4. Block diagram of the closed-loop system with an inner
switching controller C1 and an outer continuous controller C2 .

x2 (t)

1.2
1
0.8
0.6
0.4
0.2
0

0.05 0.1 0.15 0.2 0.25 0.3


u(t)

Assume that x1 = r1 in the average sense for the


stationary oscillations in the FMS. Then the behavior of
x2 (t) can be approximately described by the degenerated
system (10) of the 1st order, where r1 is the new control
variable, x2 (t) is the new output variable, and the relative
degree of the degenerated system (10) equals 1. Let the
desired behavior of x2 be assigned by
(1)

x2 =

1
[r2 x2 ].
T2

(27)

Therefore, the structure of C2 can be selected in the form


(1)

(1)

0.05 0.1 0.15 0.2 0.25 0.3

0.04
u1 (t)

0.03
0.02
0.01
0
0.01
0

0.05 0.1 0.15 0.2 0.25 0.3

Fig. 3. Simulation results for the switched system (1)(2) controlled


by (19).

rad/s. Let R = 200 . The joint numerical resolution of


(23) and (25) yields u01 0.005, A 0.006, eosc
0.24 V when r1 = 0.02 A and, x2 = xs2 3.28 V.
If r1 = 1 A and x2 = xs2 48 V then we get u01
0.014, A 0.019, eosc 0.77 V. Simulation results for
the switched system (1)(2) controlled by the algorithm
(19) are displayed in Fig. 3, where the initial conditions
are x1 (0) = 0.02 A, x2 (0) = 3.26 V, u1 (0) = 0 and
R = 200 for all t [0, 0.3] s. The simulation results
confirm the analytical calculations.
IV. OUTER CONTROLLER DESIGN
Let us consider the block diagram of the control
system shown in Fig. 4 where the external disturbance

2 r1 = k2 {T21 [r2 x2 ] x2 }

(28)

and designed similar to (14). Hence, from the closedloop system given by (10), (28), where the two-timescale motions are induced as 2 0, we obtain the
fast-motion subsystem (FMS) given by


k2 E
r2 x 2 k 2 x 2
(1)
2 r1 +
,
r1 = k 2
+
C(E + x2 )
T2
C

where x2 is the frozen parameter. We get that Tf ms =


2 C(E + x2 )/(k2 E) is the time constant of this fastmotion subsystem, where Tf ms (0.61102 , 2.1102 )
when x2 (3.26, 48). The corresponding SMS is the
same as (27).
We can take T2 = 0.1 s, 2 = 0.01 s, k2 = 0.002
where T2 and 2 are selected such that T2  Tf ms
to produce slow-fast decomposition in the closed-loop
system given by (10), (28). Simulation results for the
switched model (1)(2) controlled by the algorithm (19),
(28) with the assigned above parameters are displayed
in Fig. 5 for the time interval t [0, 0.3] s, where
r2 = 49 V and the initial conditions are the following:
x1 (0) = 1.1 A, x2 (0) = 50 V, u1 (0) = 0, r1 (0) = 1.
The initial conditions are matched here in order to avoid
undesirable transients in the closed-loop system.
V. CONCLUSIONS
The presented method of switching regulator design
allows us to obtain the desired transients for buck-boost
converter under uncertainty in model description and in

the presence of unknown external disturbances. The discussed fast dynamical controller with the relative highest
derivative in feedback produces slow-fast decomposition
in the closed-loop system. It has been shown that if a
sufficient time-scale separation between the fast and slow
modes and stability of FMS are provided by selection
of controller parameters, then SMS equation has the
desired form, and, thus, we have the desired transient
performance indices of the current IL and the voltage
Vout in the closed-loop system.
REFERENCES
[1] B. Bryant and M.K. Kazimierezuk, Derivation of the buckboost PWM DC-DC converter circuit topology, in Proc. of the
IEEE Int. Symp. on Circuits and Systems, vol. 5, pp. V 841V
844, 2002.
[2] F. B. Cunha and D. J. Pagano, Limitations in the control of a
DC-DC boost converter, in Proc. of 15th IFAC World Congress,
Barcelona, Spain, 2002.
[3] R. Giral, E. Arango, J. Calvente, and L. Martinez-Salamero,
Inherent DCM operation of the asymmetrical interleaved dual
buck-boost, in Proc. of the 28th IEEE Annual Conf. of the
Industrial Electronics Society, IECON 02, 5-8 Nov. 2002, vol. 1,
pp. 129134, 2002.
[4] Jianping Xu, An analytical technique for the analysis of switching DC-DC converters, in Proc. of the IEEE Int. Symp. on
Circuits and Systems, vol. 2, pp. 12121215, 1991.
[5] D. Maksimovic and S. Cuk, Switching converters with wide
DC conversion range, IEEE Trans. Power Electronics, vol. PE16, no. 1, pp. 151157, 1991.
[6] N. Mohan, T.M. Undeland, and W.P. Robbins, Power Electronics: Converters, Applications and Design, Wiley, 1995, 802 p.
[7] I.P. Paltov, Transients performance and compensator design for
nonlinear control systems, Nauka, Moscow, 1975, 368 p.
[8] Y.B. Shtessel, A.S.I. Zinober, and I.A. Shkolnikov, Boost and
buck-boost power converters control via sliding modes using
dynamic sliding manifold, in Proc. of the 41st IEEE Conf. on
Decision and Control, vol. 3, pp. 24562461, 2002.
[9] H. Sira-Ramirez, Sliding modes, passivity, and flatness, in
Sliding Mode Control in Engineering, Perruquetti, W. and
Barbot, J.P (Eds.), New York: Marcel Dekker, pp. 163189,
2002.
[10] A. Suvorov, Synthesis of control systems based on the method
of localization of given fast oscillating process, in Proc. of
the Int. Workshop on Control System Synthesis: Theory and
Application, 27 May - 1 June, 1991, Novosibirsk, USSR, pp. 93
99, 1991.
[11] A.S. Vostrikov, On the synthesis of control units of dynamic
systems, Systems Science, Techn. Univ., Wroclaw, vol. 3, no. 2,
pp. 195205, 1977.
[12] A.S. Vostrikov, Synthesis of nonlinear systems by means of
localization method, Novosibirsk State University, Novosibirsk,
1990, 120 p.
[13] V.D. Yurkevich, Design of two-time-scale nonlinear timevarying control systems, Series: Analysis and Design of Nonlinear Systems, St.-Petersburg: Nauka, 2000, 287 p.
[14] V.D. Yurkevich, Design of nonlinear control systems with the
highest derivative in feedback, Series on Stability, Vibration and
Control of Systems, Series A - Vol. 16, World Scientific, 2004,
352 p.
[15] G.S. Zinoviev, Fundamentals of power electronics: textbook,
Novosibirsk, Novosibirsk State Technical University, 2002,
664 p.

5
4
3
2
1
0
1
5
4
3
2
1
0
1

x1 (t)

0.05 0.1 0.15 0.2 0.25 0.3


r1 (t)

0.05 0.1 0.15 0.2 0.25 0.3

55
50
45
x2 (t)
r2 (t)

40
35

1.2
1
0.8
0.6
0.4
0.2
0

0.05 0.1 0.15 0.2 0.25 0.3


u(t)

0
200
180
160
140
120
100
80
60
40

0.05 0.1 0.15 0.2 0.25 0.3

R(t)

0.05 0.1 0.15 0.2 0.25 0.3

Fig. 5. Simulation results for the switched system (1)(2) controlled


by the algorithm (19), (28).