A Novel Scheme For Sliding-Mode Control of DC-DC Converters With A Constant Frequency Based On The Averaging Model
A Novel Scheme For Sliding-Mode Control of DC-DC Converters With A Constant Frequency Based On The Averaging Model
A Novel Scheme For Sliding-Mode Control of DC-DC Converters With A Constant Frequency Based On The Averaging Model
JPE 10-1-1
A Novel Scheme for Sliding-Mode Control of
DC-DC Converters with a Constant Frequency
Based on the Averaging Model
Yiwen He
, Weisheng Xu
SMC is simple, cost effective and dynamically fast. In order to improve the dynamic characteristics of the reaching
phase and to alleviate chattering, the second approach called reaching law SMC is also presented. Analyses and simulation results
demonstrate the feasibility of the proposed scheme.
Key Words: DC-DC converter, Sliding mode control, Constant switching frequency, Averaging model, Reaching law
I. INTRODUCTION
The use of sliding-mode control (SMC) techniques in vari-
able structure systems (VSS) makes these systems very robust
to parameter variations and external disturbances. Switching
converters constitute an important case of VSS, and various
sliding mode strategies to control this class of circuits have
been reported in the past [1][7].
The main advantages of SMC over conventional Pulse
Width Modulation (PWM) control are stability (even for
large supply and load variations), robustness, good dynamic
response and simple implementation.
An ideal sliding-mode (SM) requires an innite switch-
ing frequency in order to maintain the state on the sliding
manifold. Obviously, an innite switching frequency is not
acceptable in practice and therefore one must implement
a technique that can ensure a nite and possibly constant
switching frequency [4]. In many applications, having a con-
stant switching frequency is desirable and sometimes it is
mandatory [5]. Due to this fact, methods to limit the switching
frequency have been proposed [4][10]. In Section 2, these
methods are discussed and a new scheme for the SMC of
DC-DC converters with a constant switching frequency is
proposed.
In Section 3, two approaches using the new scheme for a
buck converter are proposed and the design procedures are
Manuscript received Jun. 26, 2009; revised Oct. 22, 2009
1 =
on
if s(x) > k
0 =
o
if s(x) < k
unchanged otherwise
(3)
Thus, by introducing a region k < s(x) < k where
no switching occurs, the maximum switching frequency of
the SM controller can be controlled. Additionally, it is now
possible to control the frequency of the operation by varying
the magnitude of k. However, if k is constant, SM con-
trolled converters generally suffer from signicant switching
frequency variation when the input voltage or output load is
varied [6].
In order to ensure a constant switching frequency for all
operating conditions, there are two common approaches that
can be used. One approach is to incorporate a constant ramp or
timing function directly into the controller [7][9]. The main
advantage of this approach is that the switching frequency
is constant under all operating conditions, and can be easily
controlled through varying the ramp/timing signal. However,
due to the imposition of a ramp or timing function onto the
SM switching function, the resulting converter system suffers
from deteriorated transient response [6]. The other approach
utilizes an adaptive hysteresis band, but this leads to a more
complicated controller implementation [4].
B. Equivalent control based indirect SMC
Indirect SMC is realized by changing the modulation
method of the SM controller from hysteresis modulation to
PWM, also known as duty cycle control. The advantage
of PWM is that by xing the frequency of the ramp, the
frequency of the output switching signal will be constant,
regardless of how the duty cycle varies with the variation of
the control signal [10]. A converter controlled by an indirect
SM controller is shown in Fig. 2. To achieve such a controller,
a relationship between direct SMC and duty cycle control is
required.
If a direct SM controller works in the ideal case of innite
switching frequency, at some time, the system trajectory will
Fig. 2. Block diagram of the indirect SMC scheme.
evolve on the surface s(x) = 0 . When this happens, the
system trajectory is described by means of the equivalent
dynamics [5]:
x = Ax +Bu
eq
+D, (4)
where u
eq
, referred to as the equivalent control, is the solution
for u in the equation s(x) = 0.
It has been proven in [11] that u
eq
is the average of u.
Hence, the performance of a SM controller u should be similar
to the performance of an average controller d provided that
d = u
eq
[12]. Consequently, it is possible to design a SM
controller like the one dened in Eq. (2) and implement it
with a PWM by making d = u
eq
.
In practice, indirect SMC is similar to classical PWM
control schemes in which the control signal is compared to
the ramp waveform to generate a discrete gate pulse signal.
However, the implementation of u
eq
is usually far more
difcult than s(x) [5] and it is not applicable to all kinds of
sliding surfaces. To simplify SM controller implementation,
an alternative method is proposed in [13]. The advantages of
this method are that it does not need additional hardware cir-
cuitries since the switching function is performed by the PWM
modulator, and that its transient response is not deteriorated.
However, the implementation is nontrivial in order to preserve
the original SM control law, especially when both current and
voltage state variables are involved [6].
C. Proposed SMC scheme
Using the state space averaging method, Eq. (1) can be
transformed to:
x
1
x
2
0
1
L
1
C
1
RC
x
1
x
2
V
i
L
0
u. (6)
where x
1
and x
2
are the inductor current i
L
and the capacitor
(output) voltage v
c
(v
0
) respectively; u is a scalar control tak-
ing values from the discrete set {0,1} which denes the switch
positions; C, L and R are the capacitance, inductance and
instantaneous load resistance respectively; V
i
is instantaneous
input voltage.
The state space averaging technique is an approximation
technique used to analyze switch mode power converters. The
behavior of a circuit during each switching state is taken
into account and an average is obtained. This method can be
used for any of the converter topologies by monitoring the
state variable waveforms for each state and then averaging
them over time. Using the state space averaging method, the
averaging model of a buck converter in CCM can be obtained
by:
x
1
x
2
0
1
L
1
C
1
RC
x
1
x
2
V
i
L
0
d (7)
where x
1
and x
2
are the moving averages of x
1
and x
2
respectively; d is the duty cycle.
B. Design of a constant d
+
and d
SM controller
SMC have been used to improve the robustness and the
dynamic response in different converters. Many papers have
presented a variety of SM controller design steps, which could
be summarized as follows:
1) To relate the steady state and the dynamic requirements
with an appropriate sliding surface.
2) To establish a structure-control law with reference to
the selected sliding surface, such that the conditions of the
reaching and the existence of a sliding regime are satised.
1) Design of the sliding surface: Typically, the sliding
surface is a linear combination of the state variables. This
is due to its easy implementation and theoretical analysis. The
SM controller forces the system to be held in the mentioned
surface and then the system is driven to the equilibrium point.
The sliding surface must include the equilibrium point.
The inductor current x
1
and the output voltage x
2
are both
continuous variables. The derivative of the output voltage x
2
is proportional to the capacitor current, which is also a con-
tinuous variable. Let the desired output voltage be V
ref
. Then
the moving averaging of the output voltage error is dened
as z
1
= V
ref
x
2
and its derivative is z
2
= z
1
=
i
C
/C.
As a result the averaging state-space model in phase canonical
form, derived from Eq. (7), is:
z
1
= z
2
z
2
=
1
LC
z
1
1
RC
z
2
V
i
LC
d +
V
ref
LC
(8)
A buck converter is a second order system with one control
input. Therefore the dynamic response obtainable under a
SMC is of the rst order. The desired steady state and dynamic
4 Journal of Power Electronics, Vol. 10, No. 1, January 2010
Fig. 5. Phase trajectories of d = d
+
for different starting positions.
Fig. 6. Phase trajectories of d = d
d
+
> V
ref
/V
i
s(z) > 0
d
< V
ref
/V
i
s(z) < 0
, (10)
For d = d
+
and the system described by Eq. (8), the
equilibrium point is given by: (V
ref
d
+
V
i
, 0). The phase
trajectories of d = d
+
for different starting positions are shown
in Fig.5. For d = d
V
i
, 0). The phase
trajectories for d = d
should satisfy
the following inequalities:
1 > d
+
> V
ref
/V
i min
0 < d
< V
ref
/V
i max
, (11)
where V
i min
and V
i max
are the possible minimum and
maximum values of the input voltage.
3) Existence condition of sliding regime: The control law
in Eq. (10) only provides the general requirement that the
trajectories will be driven towards the sliding line. To ensure
that the trajectory is maintained on the sliding line, the
existence condition, which is derived from Lyapunovs direct
method to determine asymptotic stability, must be obeyed [10]:
lim
s0
s s < 0 (12)
Substituting Eq. (9) into (12), Eq. (12) becomes:
s =
z
1
+ z
2
< 0 for 0 < s <
z
1
+ z
2
> 0 for < s < 0
(13)
where is an arbitrarily small positive quantity. From Eqs.
(8), (10) and (13) we have:
1
=
1
LC
z
1
+ (
1
RC
)z
2
V
i
LC
d
H
+
V
ref
LC
< 0, 0 < s <
2
=
1
LC
z
1
+ (
1
RC
)z
2
V
i
LC
d
L
+
V
ref
LC
> 0, < s < 0
(14)
To compare the sliding region with the direct SMC, the
existence condition of the direct SMC is rewritten as follows
[10]:
1
=
1
LC
z
1
+ (
1
RC
)z
2
V
i
LC
+
V
ref
LC
< 0, 0 < s <
2
=
1
LC
z
1
+ (
1
RC
)z
2
+
V
ref
LC
> 0, < s < 0
(15)
A Novel Scheme for Sliding-Mode Control of DC-DC Converters with 5
Fig. 8. Regions of existence of the SM for the direct SMC and the
constant d
+
and d
SMC.
The above conditions are depicted in Fig.8 for the proposed
approach and the direct SMC. For the proposed approach, the
intersections of lines
1
= 0,
2
= 0, and axis z
1
are (V
ref
d
+
V
i
, 0) and (V
ref
d
V
i
, 0). The sliding mode will only
occur on the portion of the sliding line s = 0. This portion
is within A and B, where A is the intersection of s = 0 and
1
= 0; and B is the intersection of s = 0 and
2
= 0.
For direct SMC, the intersections of lines
1
= 0,
2
=
0 and axis z
1
are (V
ref
V
i
, 0) and (V
ref
, 0). The sliding
mode will only occur on the portion of the sliding line s = 0
(provided that the sliding line is the same as the former). This
portion is within A
and B
, where A
is the intersection of
s = 0 and
1
= 0; and B
2
= 0 . Since d
+
< 1 and d
and .
C. Design of a reaching law SM controller
Although the constant d
+
and d
approach is easy to
analyze, synthesize and implement, some problems still exist.
First, because a SMC itself has a non-continuous switch
feature, it can cause high frequency chattering of a system,
which can excite non-modeled high frequency components and
cause a remarkable disturbance and even make the system
unstable. Second, in the stage from the initial states to the
sliding surface (i.e. the reaching stage), the system is only
an ordinary feedback control, which has no self-adaptation
to parameter variation or external disturbance. It weakens the
robustness of the system to a certain degree.
In order to improve the character of the reaching segment
and alleviate the chattering, the concept of a reaching law was
introduced in [16].
The reaching law is a differential equation which species
the dynamics of the switching function s(x). The differential
equation of an asymptotically stable s(x) is itself a reaching
condition. In addition, by the choice of the parameters in the
differential equation, the dynamic quality of a VSC system in
the reaching mode can be controlled [18]. A practical general
form of the reaching law for a single input system is:
s = sgns f(s) (16)
where > 0, f(0) = 0, sf(s) > 0.
It is known from (16) that the existence condition and the
reaching condition of the sliding mode are satised automati-
cally, that is:
s s = |s| sf(s) < 0 (17)
In the following we will show the constant rate reaching
and the constant plus proportion rate reaching given in [16].
1) Constant rate reaching:
Constant rate reaching is dened as:
s = sgn(s) (18)
Suppose the initial value of the switching function is s(0),
after the time interval t
r
=
1
s(0), the system trajectory
will hit the switching surface at a constant rate . The merit
of this reaching law is its simplicity. But if is too small, the
reaching time will be too long. On the other hand, an that
is too large will cause severe chattering.
2) Constant plus proportion rate reaching:
Constant plus proportion rate reaching is dened as:
s = sgn(s) ks, > 0, k > 0 (19)
After the time interval t
r
= k
1
[ln(s(0) + k
1
)
ln(k
1
)], the system trajectory will hit the switching surface,
then move across the switching surface and nally reach
the equilibrium of the system. Clearly, when adding the
proportional rate term ks, the state is forced to approach
the switching manifolds faster when s is large. Therefore, a
shorter reaching time t
r
and a smaller chattering amplitude
can be obtained by choosing appropriate parameters for
and k. Thus, constant plus proportion rate reaching has better
dynamic characteristics than constant rate reaching for SMC
systems. As a result, the constant plus proportion rate reaching
law is adopted in the proposed SMC scheme.
Having selected the reaching law equation, the control law
can now be determined. Computing the time derivative of s(x)
along the reaching mode trajectory, we have:
s = z
1
+ z
2
=
1
LC
z
1
+(
1
RC
)z
2
V
i
LC
d+
V
ref
LC
(20)
Then from Eqs. (19) and (20), we can get:
1
LC
z
1
+(
1
RC
)z
2
V
i
LC
d+
V
ref
LC
= sgn(s)ks (21)
This equation is solved for the control law, resulting in:
d =
LC
V
i
sgn(z
1
+z
2
) +
V
ref
V
i
+ (
kLC
V
i
1
V
i
)z
1
+
LC
V
i
(
1
RC
+k)z
2
(22)
For designing the PWM modulator, the feed-forward adap-
tive method proposed in [10] can also be utilized to make
6 Journal of Power Electronics, Vol. 10, No. 1, January 2010
the system robust to input voltage. In terms of PWM-based
controlled systems, the instantaneous duty cycle d is expressed
as:
d =
V
c
v
ramp
(23)
where V
c
is the control signal for the PWM and v
ramp
is the
peak magnitude of the constant frequency ramp signal. From
Eqs. (22) and (23), we have:
V
c
= LCsgn(z
1
+z
2
) +V
ref
+ (kLC 1)z
1
+LC(
1
RC
+k)z
2
(24)
and:
v
ramp
= V
i
(25)
for the practical implementation of the reaching law SM
controller. Therefore, the line regulation robustness can be
maintained by varying the peak magnitude of the ramp signal
with the input voltage.
It is worthwhile to mention that the control signal V
c
is
actually load dependent. As a result, for the controller to be
regulation-robust to load changes, the instantaneous value of
R should be fed back. However, this would require additional
sensors and computations, which complicate the controller.
Fortunately, the dependence and sensitivity of V
c
on the load
can be minimized by the proper design of , and k such
that ( + k) 1/RC. Therefore, the design value of the
load resistance can be made a constant parameter R
nom
, the
nominal value of load resistance [10].
A nal but important note is that the control law (22),
obtained via a reaching law, automatically leads to a free-
order switching scheme. From a practical point of view, this
scheme appears to be the most efcient [16].
IV. SIMULATION RESULTS AND DISCUSSIONS
In this section, a simulation is conducted to verify the
validity of the analysis results given above. The parameters
of buck converter are: L = 0.11mH, C = 100F, R = 6,
V
i
= 24V , V
ref
= 12V and f
s
= 200kH
z
.
For the constant d
+
and d
SM
controller because of the improved dynamic characteristics in
the reaching phase. Both of the gures indicate controllers
with rst order responses and no overshoot of output voltage.
It shows order reduction which is one of the SMC properties.
Fig. 9. Response in start up and load change (constant d
+
and d
SMC).
Fig. 10. Response in start up and load change (reaching law SMC).
In order to show the robustness of the controllers, a sudden
load change is introduced at t = 4ms. Fig.11 shows a com-
parison of the two controllers responses to a load resistance
decrease from 6 to 3 . Fig.12 presents a comparison of
a load resistance increase from 6 to 12. Moreover, the
state trajectories on the phase plane of the two controllers
in start up and load change are shown in Fig.13 and 14,
respectively. It can be seen that the controllers are highly
robust under load variations and that the reaching law SM
controller has better performance than the constant d
+
and
d
SM controller
and a reaching law SM controller for buck converters are
given in detail. To show the feasibility of the proposed control
methodologies, both of them are applied to buck converters in
simulation. The simulation results show that the reaching law
SMC exhibits superior static and dynamic properties than the
A Novel Scheme for Sliding-Mode Control of DC-DC Converters with 7
Fig. 11. Comparison of the response under load increase.
Fig. 12. Comparison of the response under load decrease.
Fig. 13. Phase plane trajectory of start up and load change (constant d
+
and d
SMC).
Fig. 14. Phase plane trajectory of start up and load change (reaching law
SMC).
constant d
+
and d