1160 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO.

3, MARCH 2008

General Design Issues of Sliding-Mode

Controllers in DC–DC Converters
Siew-Chong Tan, Member, IEEE, Y. M. Lai, Member, IEEE, and Chi K. Tse, Fellow, IEEE

Abstract—This paper examines the practical design issues of lot of interest in the use of SM controllers for dc–dc converters
sliding-mode (SM) controllers as applied to the control of dc–dc [16]–[66].
converters. A comprehensive review of the relevant literature is However, despite being a popular research subject, SM
first provided. Major problems that prevent the use of SM control
in dc–dc converters for industrial and commercial applications are control is still rarely applied in practical dc–dc converters.
investigated. Possible solutions are derived, and practical design There are various reasons accommodating this. First, un-
procedures are outlined. The performance of SM control is com- like PWM controllers, SM controllers are not available in
pared with that of conventional linear control in terms of transient integrated-circuit (IC) forms for power-electronic applications.
characteristics. It has been shown that the use of SM control can Second, there is a lack of understanding in their design prin-
lead to an improved robustness in providing consistent transient
responses over a wide range of operating conditions. ciple by power-supply engineers. No systematic procedure is
available for the design of SM controllers. Third, there is
Index Terms—DC–DC converter, design issues, nonlinear a strong reluctance to the employment of SM controllers in
controller, sliding-mode (SM) control.
dc–dc converters because of their inherently high and variable
switching frequency, which causes excessive power losses,
I. I NTRODUCTION electromagnetic-interference (EMI) generation, and filter-
design complication. Fourth, all discussions regarding the use-
T HE SLIDING-MODE (SM) controller is a kind of nonlin-
ear controller which was introduced for controlling vari-
able structure systems (VSSs) [1]–[5]. Its major advantages are
fulness and advantages of SM controllers have been theoretical.
The practical worthiness of using SM controllers is generally
the guaranteed stability and the robustness against parameter, unproven. In essence, SM controllers are not used in practical
line, and load uncertainties [1]. Moreover, being a controller dc–dc converters because of the inconvenience of using them,
that has a high degree of flexibility in its design choices, the as well as the lack of strong evidence to support the need for us-
SM controller is relatively easy to implement as compared with ing them. These explain why the application of SM controllers
other types of nonlinear controllers. Such properties make it in dc–dc converters has only been of academic/research interest
highly suitable for control applications in nonlinear systems. but of little industrial value. Specifically, as the theoretical
This explains the wide utilization of SM controllers in var- groundwork of SM control is fairly matured, it is timely to
ious industrial applications, e.g., automotive control, furnace direct more research efforts toward developing practical SM
control, etc. [3]. Some recent works on SM control and its controllers for dc–dc converters. This will enable the industry to
application can be found in [6]–[10]. truly benefit from the advantages of designing power supplies
Incidentally, characterized by switching, dc–dc converters based on the SM paradigm.
are inherently variable-structured. It is, therefore, appropriate Hence, the objective of this paper is to present an overall
to use SM controllers for the control of dc–dc converters [11]. discussion to clarify the different aspects of the application of
This seems more naturally so considering the excellent large- SM controllers in dc–dc converters. A comprehensive literature
signal handling capability that the SM control can offer. Since review is conducted to update practicing engineers about the
the design of conventional pulsewidth-modulation (PWM) con- previous and ongoing works in the area. The various problems
trollers in power electronics is small-signal-based, the system of applying SM controllers to dc–dc converters are discussed.
being controlled operates optimally only for a specific condition The practical issues related to its implementation are also
[12] and often fails to perform satisfactorily under large para- addressed. The aim is to provide practicing engineers with
meter or load variations (i.e., large-signal operating condition) feasible solutions that can be adopted for the development
[13]–[15]. By substituting the linear PWM controllers with SM of practical SM controllers for dc–dc converters. Finally, the
(nonlinear) controllers in power converters, better regulation advantages of using the SM controllers as compared with
can be achieved for a wider operating range. This has aroused a the conventional linear controllers are discussed through some
case-study examples.

II. L ITERATURE R EVIEW

Manuscript received July 22, 2005; revised September 12, 2007.
The authors are with the Department of Electronic and Information Engi- The literature review is organized under several categories of
neering, Hong Kong Polytechnic University, Kowloon, Hong Kong (e-mail:
[email protected]). discussions. This is to provide a clear historical overview of the
Digital Object Identifier 10.1109/TIE.2007.909058 major stages of development in this area.

0278-0046/$25.00 © 2008 IEEE

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
TAN et al.: GENERAL DESIGN ISSUES OF SLIDING-MODE CONTROLLERS IN DC–DC CONVERTERS 1161

A. Earliest Works In addition, in 1998, López et al. [32] proposed the use of SM
control for interleaving parallel-connected dc–dc converters.
The earliest works on developing SM controllers for
However, the illustration was only limited to the buck-type
dc–dc converters were reported in 1983 [16] and 1985 [17].
modular converters. Then, in 2000, Giral et al. [33] demon-
Bilalović et al. [16], the earliest group that worked in this
strated the application of SM controllers for interleaved boost
area, demonstrated how SM controllers can be applied to dc–dc
converters.
converters. However, their work was preliminary and focused
In 2002, Mazumder et al. [34] combined the integral-
only on the buck converter. Later, Venkataramanan et al. [17]
variable-structure and multiple-sliding-surface control schemes
presented a more complete description of how SM control can
to nullify the bus-voltage error and the error between the load
be applied to all different basic topologies of second-order
currents of the parallel-connected converter modules. In 2004,
dc–dc converters. They also showed the idea of relating the
López et al. [35] presented a detailed paper on the analysis and
equivalent control method of SM-control theory to the duty-
design of an SM-controlled parallel-connected boost-converter
ratio control method of PWM technique to achieve constant-
system. In comparison with the papers [29], [30], [31], and [34],
frequency SM controllers. This will be discussed in-depth later
which address the basic theory, the work of López et al. [35] is
in this paper.
more relevant to practical power electronics.

B. Higher Order Converters D. Theoretical Works

After Venkataramanan et al. [17] published their work on Due to simplicity and ease of analysis, simple second-order
second-order dc–dc converters, Huang et al. [18] experimented dc–dc converters have been the subject of investigation in much
the SM controller on a more complex fourth-order Ćuk con- of the previous works. The main focus was the theoretical
verter in 1989. This spurred a series of related works on the Ćuk derivation of SM-control methodologies. These earlier studies
converter. In 1992, Fossas et al. [19] examined the audiosuscep- have played an important role in the development of practical
tibility and load-disturbance properties of the SM-controlled SM controllers for dc–dc converters.
Ćuk converter. In 1995, Malesani et al. [20] systematized In 1994, Sira-Ramirez and Rios-Bolivar [36] proposed the
the design approach for the SM-controlled Ćuk converter. In use of an extended linearization method in SM-controller de-
1996, Oppenheimer et al. [21] implemented the SM con- sign that results in SM controllers that have excellent self-
troller on a third-order-type Ćuk converter. Moreover, in 1996, scheduling properties. In 1995, they proposed to incorporate
Mahdavi and Emadi [22] developed the first PWM-based SM- passivity-based controllers into the SM-controlled dc–dc con-
controlled Ćuk converter. Then, in 1997, Calvente et al. [23] verter to enhance its robustness properties [37], [38]. In 1996,
introduced a method of designing locally stable SM controller Fossas and Biel [39] presented a new approach for the design
for the bidirectional coupled-inductor Ćuk converter, and in of SM-controlled dc–dc converters for generating an ac signal.
1998, they provided a thorough analytical evaluation of the In 1997, Carrasco et al. [40] proposed to incorporate neural
converter [24]. networks with SM controllers for power-factor-correction
At the same time, the work of Huang et al. [18] also applications.
generated new interest on other types of higher order dc–dc In 2001, Bock et al. [41] proposed a design procedure for the
converters. In 1993, Mattavelli et al. [25] proposed a general- selection of high-pass-filter parameters of the SM-controlled
purpose SM controller which is applicable for both Ćuk and bidirectional dc–dc converters. In 2002, Fossas and Pas [42]
single-ended primary inductance (SEPIC) converter. In 1997, applied a second-order SM controller to a buck converter to
the same group extended the research by proposing a method reduce the chattering. Furthermore, in 2002, Shtessel et al.
of deriving the small-signal models of these converters, which proposed two SM-control strategies for boost and buck–boost
allows the investigation of stability and the selection of the con- converters: one using the method of stable system center [43]
trol parameters [26]. In 1994, Dominguez et al. [27] performed and the other using the dynamic sliding manifold [44].
a stability analysis of an SM-controlled buck converter with an In 2003, Vazquez et al. [45] proposed a new sliding surface
input filter. In 2000, Castilla et al. [28] presented the design that eliminates the use of a current sensor, and Gupta and Patra
methodology of the SM-control schemes for quantum resonant [46] proposed a hybrid SM controller that employs a com-
converters. bined form of voltage and current sliding surfaces to improve
robustness. In addition, in 2003, Sira-Ramirez [47] proposed
a hysteresis modulation type of SM controller to achieve a
C. Parallel-Connected Converters
generalized proportional integral (GPI) continuous control of
The interest in applying SM control to more complex types a buck converter. He also presented a tutorial that revisits and
of dc–dc converters has also covered the class of parallel- evaluates the performances of direct and indirect SM-controller
connected dc–dc converters. In 1996, Donoso-Garcia et al. schemes and proposed the use of GPI control technique to
[29] and Shtessel et al. [30] proposed the use of SM control improve the robustness of the system with respect to unmodeled
to achieve current equalization and output-voltage regulation load-resistance variations [48].
of modular dc–dc converters. In 1998, Shtessel et al. [31] Apart from developing SM-control theory for power-
suggested that dynamic sliding surface can be employed for converter applications, evaluations of performance and com-
better stabilization and control of such modular converters. parisons with other control methods were also reported. In

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
1162 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008

1997, Raviraj and Sen [49] made a comparative study of the trollers [64]. In 2004, Perry et al. [65] proposed a digital fuzzy-
buck converter’s performance when controlled by PI, SM, and logic SM-like controller that has a fixed switching frequency
fuzzy-logic controllers. They concluded that there are some and provides zero steady-state error, and Iannelli and Vasca
similarities in the system’s behavior between fuzzy-logic and [66] proposed a method of dithering to maintain a finite and
SM controllers. In 2002, Cortes and Alvarez [50] revisited the constant switching frequency. In addition, in 2004, Mazumder
work on the SM control of boost converters by comparing and and Kamisetty [67] gave an experimental validation of their
analyzing the performances using different proposed schemes proposed control scheme [34] for the parallel converters, which
and sliding surfaces. In 2002, Morel et al. [51]–[53] studied the not only optimizes the transient and steady-state responses but
nonlinear behavior exhibited by the conventional current-mode- also achieves a constant switching frequency at steady state.
controlled boost converter and proposed the alternative use of SM control has, in most cases, been studied in continuous
SM controller to eliminate the chaotic behavior of the converter. time. A few attempts to study its discrete-time counterparts
In summary, the objectives of most previous works are to have been reported. In 2000, Matas et al. [68] proposed a
simplify design procedures, enhance robustness, reduce the discrete-time SM-controlled boost converter for output-voltage
number of components, and prevent chattering. tracking, and Orosco and Vazquez [69] provided a complete
analysis of the discrete-time SM-controlled dc–dc converters.
They argued that the discrete-time implementation of the SM
E. Practical Works
controller can overcome the inherent drawbacks of variable
A few experimental evaluations of SM-controlled dc–dc switching-frequency operation in the conventional continuous-
converters have been reported in the literature. Most of the time implementation.
reported experimental works, however, focused their attention
on performance evaluation rather than on developing design
G. Remarks
procedures.
In 1999, Escobar et al. [54] performed experiments to com- In concluding the literature survey, the major developments
pare five different control algorithms, including the SM-control of SM control in dc–dc converters are summarized as follows,
scheme, on a dc–dc boost power converter. They concluded that along with some comments on the likely directions for current
nonlinear controllers provide a promising alternative to the lin- and future research developments.
ear average controller, which performs poorly in tracking time- First, the idea of applying SM (nonlinear) control in high-
varying references. Furthermore, in 1999, Chiacchiarini et al. order converter systems is acceptable by large since linear
[55] conducted an experiment to compare the performances of a controllers are incapable of providing good control over such
buck converter controlled by digital and analog SM controllers. systems. Higher design and implementation costs can be easily
They concluded that the SM-controlled system gives consistent justified in such circumstances. On the other hand, the idea of
responses despite variation in load conditions. applying SM controllers to basic second-order dc–dc converter
In 2001, Alarcon et al. [56] reported the first analog IC SM- topologies is often challenged and is by and large unpopular
control prototype for dc–dc converters. They concluded that even within the research community. The main complaint is the
the megahertz operating range of the controller fits the require- conceptual/implemental complexity of the control scheme as
ments supported by modern power-electronic technologies. In compared with the existing linear controllers, which are already
2003, Ahmed et al. [57], [58] first implemented and then pro- offering acceptable control properties in such converters for
vided an experimental evaluation of the dynamic performance most applications. Moreover, the notion of using relatively
of an SM-controlled buck converter. In another paper [59], they higher cost digital means to implement the SM controllers,
also implemented the SM controller on a buck–boost converter as illustrated in many previous attempts, has been deemed
using the control desk dSPACE. unrealistic for commercial applications. Hence, it is important
to find out what the benefits and drawbacks of using SM
controllers are, as compared to the existing linear PWM con-
F. Constant-Frequency SM Controllers
trollers, assuming that it is possible to implement the former in
Some researchers have noted the importance of maintaining a a comparable form and cost with the latter.
constant switching-frequency operation in their SM-controlled Second, it can be concluded that a large part of the previous
converter systems. effort has been devoted to the development of the theoreti-
In 1992, Cardoso et al. [60] proposed several methods of cal/mathematical framework of SM control for dc–dc convert-
reducing the switching frequency of the SM-controlled dc–dc ers. Practical work, however, has been very much neglected due
converters. In 1995, Nguyen and Lee [61] proposed an adaptive to the conventional belief that the derived SM-control schemes
hysteresis type of SM controller to ensure constant switch- can be easily realized in digital forms. Thus, the interest in
ing frequency. In 1996, they proposed an indirect method its development normally halts at the theoretical/mathematical
of implementing SM controllers in buck converters so that stages. However, considering that if SM controllers are to be
constant switching frequency can be achieved [62]. In 1997, implemented for commercial applications, thorough studies
Mahdavi et al. [63] proposed a method of deriving PWM- of practical problems are necessary. Specifically, the different
based SM-controlled dc–dc converters that have a constant means of developing analog SM controllers, which operate at
switching frequency. Later, in 2000, they extended the work by a constant switching frequency in compliance with the usual
incorporating neural networks into their PWM-based SM con- industrial standard, should be explored. Yet, under such design

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
TAN et al.: GENERAL DESIGN ISSUES OF SLIDING-MODE CONTROLLERS IN DC–DC CONVERTERS 1163

criteria, a contradiction exists between the ideal operation of the same formula of switching between multiple structures
SM controllers at infinitely high frequency and the practical to achieve a desired output. Ultimately, they are all tracking
constraint of physical converter systems that are only allowed some kind of sliding surfaces, each uniquely defined by its
to operate at limited frequency ranges. Thus, while aiming at control strategy, to achieve equilibrium. Therefore, the dif-
competitive pricing and compatible standard with the existing ference between the conventional control methodologies and
PWM controllers, there must also be considerations on how the the actual SM-control methodology can be distinguished by
nonideality of finite switching frequency can be compromised the way in which the controllers are being designed. For the
in practical SM controllers without sacrificing the large-signal conventional control methodologies, the sliding surfaces are
properties of SM control. Essentially, more investigations into indirectly formulated through the design of the control para-
such aspects are still required. meters and controller’s type, which are determined by some
In conclusion, before SM controllers can be commercially stability analysis. The controller designers are neither informed
viable in common dc–dc converter applications, the aforemen- nor have they any direct control on how the sliding surfaces are
tioned issues should first be overcome. This may prove to be constructed. For the SM control, however, the design procedure
difficult considering that many still regard the work on the starts with the formulation of the sliding surfaces. The con-
application of SM controllers in simple second-order converters troller designers always determine the type of sliding surfaces
as redundant and impractical. Hence, before it is possible to they desire, and this allows them to exercise a direct control
proceed with further investigation, the viability of using SM over the dynamic response of the system. The way in which
controllers for simple converters has to be clearly demonstrated. the control parameters are chosen is then purely defined by
For instance, it has been shown in some recent works [25], [55], the hitting condition, the existence condition, and the stability
[56], [70] that SM controllers can be implemented in terms of condition of the SM-control law.
simple analog circuits. These viewpoints have brought new insights to how dif-
ferent control objectives can be achieved in power-electronic
converters.
III. S LIDING -M ODE C ONTROLLERS
This section gives a summarized overview of the various C. Principle of Operation
aspects of the SM controllers, particularly for applications in
dc–dc converters. The basic principle of SM control is to design a certain
sliding surface in its control law that will direct the trajectory
of the state variables toward a desired origin when coincided.
A. Ideal Controller In the case of a single switch dc–dc converter, it is appropriate
The basic principle of SM control is to employ a certain to have a control law that adopts a switching function such as
sliding surface as a reference path such that the controlled
1
state variables’ trajectory can be directed toward the desired u= (1 + sign(S)) (1)
equilibrium. Theoretically, such ideology of the SM control 2
can be fully achieved only with the absolute compliances of where u is the logic state of the converter’s power switch, and
certain conditions, namely, the hitting condition, the existence S is the instantaneous state variable’s trajectory which, in the
condition, the stability condition, and the condition that the case of a second-order controller, is described as
system operates at an infinite switching frequency.
In such respect, what is derived is an idealized controlled S = α1 x1 + α2 x2 + α3 x3 (2)
system, whereby no external disturbances or system’s uncer-
where α1 , α2 , and α3 represent the control parameters, usually
tainties can affect the ideal control performance of having zero
referred to as sliding coefficients, and x1 , x2 , and x3 denote the
regulation error and very fast dynamic response. Hence, in a
desired state feedback variables to be controlled. By enforcing
certain sense, the SM controller is actually a type of ideal
S = 0, a sliding surface (plane), as shown in Fig. 1, can be
controller for the class of VSSs.
obtained.
A detailed discussion of the SM-control principle can be
B. Applications to DC–DC Converters found in [1]. In brief, the entire SM-control process can be
divided into two phases. In the first phase (reaching phase),
“Practically all design methods for VSSs are based on
regardless of the starting position, the controller will perform
deliberate introduction of sliding modes which have played,
a control decision that will drive the trajectory of the state
and are still playing, an exceptional role both in theoretical
variables to converge to the sliding surface [see Fig. 1(a)].
developments and in practical applications.”
This is possible through the compliance of the so-called hitting
This statement is extracted from the preface of [1] by
condition1 [3].
Utkin et al. If such a reasoning is elaborated in the con-
When the trajectory is within a small vicinity of the sliding
text of power electronics, a logical deduction follows that
surface, it is said to be in SM operation, which is the second
since all power-electronic converters are intrinsically variable-
structured, their control methodologies, be it linear or nonlinear, 1 Satisfaction of the hitting condition assures that, regardless of the initial
are all based on some form of SM control. This is true since condition, the state trajectory of the system will always be directed toward the
the objectives of all controllers in the dc–dc converters share sliding surface.

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
1164 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008

the SM-operation phase can be obtained by equating (2) as

S = 0, i.e.,

d2 x1 α1 dx1 α3
+ · + · x1 = 0 (3)
dt2 α2 dt α2
when operated at the nominal load resistance RL(nom) (t). How-
ever, in the case when the load deviates from the nominal design
condition, the system dynamics will be changed to


d2 x1 α1 1 1 dx1 α3
2
+ + − · + · x1 = 0
dt α2 rL (t)C RL(nom) (t)C dt α2
(4)
where rL (t)
= RL(nom) is the instantaneous load resistance.
It is possible by choosing (α1 /α2 )  (1/rL C) to ensure an
almost constant dynamical characteristic for the SM-operation
phase throughout all operating conditions. Note that such prop-
erty only applies to the SM-operation phase and not to the
reaching phase, which has a different dynamic characteristic for
each operating condition. Yet, considering that the time taken to
complete the SM-operation phase is usually much greater than
the reaching phase, it is sufficient to consider only the dynamics
of the former for the controller design. Hence, it is possible to
have a system that can attain constant dynamical behavior using
the SM controller.
This is not the case for the conventional types of linear PWM
Fig. 1. Graphical representations of state variables’ trajectory behavior in SM controllers. It is well known that the small-signal structure of
control process. (a) Phase 1—illustrating trajectory converging the sliding plane the linear controllers can allow the dynamics of the system to be
irrespective of its initial condition. (b) Phase 2—illustrating trajectory being
maintained within a small vicinity from the sliding plane and concurrently
characterized at only one specific operation condition. Hence,
being directed to converge to the origin O. when the operating condition differs from the nominal condi-
tion, the dynamical response of the system will be different. A
phase of the control process. The controller will give a series comparison of these properties is given in the later part of the
of control actions via switching such that the trajectory is paper.
maintained within a small vicinity of the sliding surface and is
concurrently directed toward the desired reference at origin O E. Quasi-Sliding-Mode (QSM) Control
[see Fig. 1(b)]. In other words, the SM controller is performing
its control decision by utilizing the sliding plane as a reference Ideally, to achieve a perfect SM-control operation, the sys-
path, on which the trajectory will track and eventually converge tem must be operated at an infinite switching frequency so
to the origin to achieve steady-state operation. This is possible that the controlled variables can exactly follow the reference
by satisfying the so-called existence condition2 and stability track to achieve the desired dynamic response and steady-
condition3 [3]. state operation [1]. This requirement for operation at an in-
Hence, when the system enters into SM operation, its equiv- finite switching frequency, however, challenges the feasibility
alent trajectory can be ideally described as S = 0. This also of applying SM controllers in dc–dc converters. This is be-
defines the dynamic characteristic of the system, which can cause extreme high-speed switching in dc–dc converters results
be designed by the proper choice of control parameters, i.e., in excessive switching losses, inductor and transformer core
sliding coefficients [71]. losses, and EMI-noise issues. Hence, for SM controllers to
be applicable to dc–dc converters, their switching frequencies
must be constricted within a practical range. Different methods
D. Constant Dynamics (hysteresis, constant sampling frequency, constant ON time,
An interesting property of the SM controller is its capa- constant switching frequency, and limited maximum switching
bility of providing constant system’s dynamics during SM frequency) were proposed to limit the switching frequency [60].
operation. Using the same example, the transient dynamics for However, they fall short of a set of systematic design methods
and implementation criteria.
Nevertheless, this constriction of the SM controller’s switch-
2 Satisfaction of the existence condition ensures that the state trajectory at
ing frequency transforms the controller into a type of QSM
locations near the sliding surface will always be directed toward the sliding controller, which operates as an approximation of the ideal
surface.
3 Satisfaction of the stability condition ensures that the state trajectory of the SM controller. The consequence of this transformation is the
system under SM operation will always reach a stable equilibrium point. reduction of the system’s robustness and the deterioration of

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
TAN et al.: GENERAL DESIGN ISSUES OF SLIDING-MODE CONTROLLERS IN DC–DC CONVERTERS 1165

the regulation properties. Clearly, the proximity of QSM to the

ideal SM controller will be closer as the switching frequency
increases. Since all SM controllers in practical dc–dc converters
are frequency-limited, they are, strictly speaking, QSM con-
trollers. Typically, the term SM controller has been adopted to
represent QSM controller in many research papers.
Fig. 2. Simplified HM and PWM structure.
F. Conventional Hysteresis-Modulation (HM)-Based
Sliding-Mode Controller According to the literature, there are basically two ap-
proaches in keeping the switching frequency of the HM-based
The conventional method of implementing the SM controller SM controller constant. One approach is to incorporate a con-
is based on the control law described in (1) and (2). The stant ramp or timing function directly into the controller [25],
former is a type of signum function and is easily realized using [60], [66]. The main advantage of this approach is that the
a switch relay. The latter, which computes the instantaneous switching frequency is constant under all operating conditions
state-variable trajectory S, is realized through an analog or and can be easily controlled through varying the ramp/timing
digital computer. However, the direct implementation of this signal. However, this comes at the expense of additional hard-
control law results in converters that are switched at very high ware circuitries, as well as deteriorated transient response in the
and uncontrolled frequency [1], [70]. As mentioned previously, system’s performance caused by the superposition of the ramp
this makes it unsuitable for the converter’s control purpose. function upon the SM switching function.
Hence, it is necessary to suppress the operating frequency The second approach is to include some forms of adap-
into a controllable form. Of the many methods proposed, the tive control into the HM-based SM controller to contain the
most popular is the HM method. This method does not require switching-frequency variation [61]. For line variation, the fre-
additional computation or auxiliary circuitries, and its imple- quency variation is alleviated through the adaptive feedforward
mentation is easily accomplished by refining (1) into control, which varies the hysteresis band with the change
 1 = “ON, ” when S > κ
of the line input voltage. For load variation, the frequency
variation is alleviated through the adaptive feedback control,
u= 0 = “OFF, ” when S < −κ (5)
which varies the control parameter (i.e., sliding coefficient)
previous state, otherwise
with the change of the output load. Conceptually, these methods
where κ is an arbitrarily small value. The introduction of a of adaptive control are more direct and less likely to suffer
hysteresis band with the boundary conditions S = κ and S = from deteriorated transient response. However, the architecture
−κ provides a form of control to the switching frequency of of the resulting controller is relatively more complex and may
the converter, thus solving the practical problem of a very high increase the implementation cost of the controller.
frequency switching operation. However, due to the lack of On the other hand, constant switching-frequency SM con-
systematic design methods and implementation criteria, the im- trollers can also be obtained by employing PWM instead of
plementation of HM-based SM controllers for dc–dc converters HM [62]. In practice, this is similar to classical PWM con-
still relies on the trial-and-error tuning of the κ magnitude to trol schemes in which the control signal is compared to the
achieve the desired switching frequency for a specific operating ramp waveform to generate a discrete gate pulse signal [13].
condition. A method of formulating this design equation for The advantages are that additional hardware circuitries are not
buck converter has recently been proposed in [70]. needed 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
IV. T HE N EED FOR F IXED -F REQUENCY SM C ONTROLLER
the original SM control law, particularly if both the current and
Despite the imposition of the hysteresis band upon the voltage state variables are involved. Hence, this approach is not
switching relay to suppress the switching frequency, the prob- always implementable for some SM-controller types. As this is
lem of variable switching-frequency operation still exists. It has an unconventional and relatively new topic, more discussion is
been shown that HM-based SM-controlled converters gener- provided in the following section.
ally suffer from significant switching-frequency variation when
the input voltage and the output load are varied [25], [70].
A. Pulsewidth-Modulation-Based Sliding-Mode Controller
This complicates the design of the input and output filters.
Obviously, designing the filters under a worst case (lowest) Fig. 2 shows the idea of the PWM-based SM controller,
frequency condition will result in oversized filters. Moreover, where PWM is used in lieu of HM without destroying the SM-
the variation of the switching frequency also deteriorates the control properties. This requires the relationship of the two
regulation properties of the converters. Furthermore, it is well control techniques to be established. Two key results are useful
known that switching converters are severe noise generators. here. First, in the SM control, the discrete control input (gate
The task of containing noise can be made easier with fixed- signal) u can be theoretically replaced by a smooth function
frequency operation. It is therefore necessary to have the SM- known as the equivalent control signal ueq , which can be
controlled converters operate at a constant switching frequency formulated using the invariance conditions by setting the time
for all operating conditions. differentiation of (2) as Ṡ = 0 [1]. Second, at a high switching

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
1166 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008

Fig. 3. Graphical representation of high- and low-frequency components of the state trajectory.

frequency, the equivalent control is effectively a duty-cycle the high-frequency trajectory component, and the switching
control [72]. Since a duty cycle is also basically a smooth action of ulow produces only the low-frequency trajectory
analytic function of the discrete control pulses in PWM, a component.
PWM-based SM controller can be obtained by mapping the Now, ignoring the high-frequency component, which is often
equivalent control function onto the duty-cycle function d of filtered out by the plant,4 the motion of the trajectory is then
the pulsewidth modulator, i.e., d = ueq . solely determined by the low-frequency component. Hence, it
Interestingly, the derived PWM-based SM controllers can is reasonable to consider only the low-frequency continuous
also be viewed as a type of nonlinear state feedback controllers switching action ulow as the desired switching action that will
designed from some nonlinear “per-cycle averaged models” produce a trajectory that is of near equivalence to an ideal SM-
of the converters. However, it should be emphasized that a operation trajectory. This is the so-called equivalent control.
main difference between the two approaches is that, while an Therefore, the often used terminology, the equivalent control,
“average model” is assumed, the PWM-based SM controller i.e., ueq , is actually the low-frequency continuous switching
approach, which only performs averaging during the controller action ulow described earlier. The method of obtaining the
implementation, retains much of the converter dynamics. This equivalent control is given as follows.
results in a set of design restrictions: the existence condition, First, consider a system ẋ = f (x, u), where u = (0, 1). In
which arises from the instantaneous dynamics of the converter, an ideal SM operation, the state trajectory S is always mov-
as required by the SM control. Such a design restriction is ing along the sliding plane, i.e., S = 0, and without any
absent from the conventional nonlinear PWM controller design high-frequency oscillation, it is also true that dS/dt = Ṡ = 0.
approach. Next, consider that the equivalent control produces a trajectory
An insight to the theory of the equivalent control and the whereby its motion is exactly equivalent to the trajectory’s
duty-ratio control is provided next. motion of an ideal SM operation. Under such assumptions, the
state trajectory equation Ṡ = G · f (x, u), where G = ∂S/∂x,
can be rewritten as Ṡ = G · f (x, ueq ). Then, the solution of
B. Equivalent Control the equivalent control ueq can be obtained by solving G ·
As discussed, to achieve an ideal SM-control operation, the f (x, ueq ) = 0. Finally, if ueq is substituted back into the origi-
system must be operated at an infinite switching frequency nal system
so that the state variables’ trajectory is oriented precisely on
the sliding surface. However, in the practical case of finite ẋ = f (x, ueq ) (6)
switching frequency, the trajectory will oscillate in some vicin-
ity of the sliding surface while moving toward the origin which is the motion equation of the converter operating in
[see Fig. 1(b)]. It is possible to identify the movement of the SM operation, can be obtained. This method of deriving the
trajectory as a composition of two isolated components: a fast- equivalent control signal ueq , as well as the insertion of
moving (high-frequency) component and a slow-moving (low- the signal into the original system to formulate the motion
frequency) component. equation as shown in (6), is known as the equivalent control
Fig. 3 shows a graphical representation of the state trajectory method [1].
being separated into a high-frequency component and a low-
frequency one. It can be seen from the diagram that the high- C. Duty-Ratio Control
frequency component is actually a discontinuous trajectory that
alternates between the +ve and −ve direction, whereas the In conventional PWM control, which is also known as the
low-frequency component is actually a continuous trajectory duty-ratio control, the control input u is switched between 1 and
that moves along the sliding plane. Since the movement of 0 once every switching cycle for a fixed small duration ∆. The
the trajectory is a consequence of the switching action u, it is time instance at which the switching occurs is determined by
therefore possible to relate the corresponding low- and high- the sampled value of the state variables at the beginning of each
frequency components of the trajectory to a low-frequency switching cycle. Duty ratio is then the fraction of the switching
continuous switching action ulow , where 0 < ulow < 1, and to cycle in which the control holds the value of one. It is normally
a high-frequency discontinuous switching action uhigh , where
uhigh = −ulow or 1 − ulow , and that u = uhigh + ulow . Under 4 Note that, in the case of dc–dc converters, the high-frequency ac signals are
such assumptions, the switching action of uhigh produces only filtered out by the output filter capacitor.

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
TAN et al.: GENERAL DESIGN ISSUES OF SLIDING-MODE CONTROLLERS IN DC–DC CONVERTERS 1167

a smooth function of the state vector x(t), and it is denoted ever, more sensing and/or computations of the state vari-
by d(x), where 0 < d(x) < 1. Then, for each switching cycle ables are required.
interval ∆, the control input u can be written as 2) Which state variables to control?
 Comment: The use of voltage state variables is eas-
1, for t ≤ τ < t + d(x)∆ ier because of the simplicity of implementing voltage
u= (7)
0, for t + d(x)∆ ≤ τ < t + ∆ . sensors. However, the choice for their derivatives may
require the use of noise sensitive differentiators. Alter-
It follows that a system ẋ = f (x, u) can be expressed as natively, indirect means of sensing such state variables
is possible, e.g., dvo /dt can be found by sensing the

t+d(x)∆ 
t+∆ current of the output capacitor. Yet, this will lead us to
x(t + τ ) = x(t) + [f (x(τ ))] dτ + 0 dτ. (8) the subsequent question as to whether it is convenient to
t t+d(x)∆
perform such sensing.
On the other hand, an integral term of the controlled
The ideal average model of the PWM-controlled system re- variable x is often required to reduce the steady-state
sponse is obtained by allowing the duty-cycle frequency to tend error of the system, which may otherwise exceed the
to infinity, i.e., ∆ to approach zero. Under such consideration, regulation requirement.
the earlier equation becomes 3) Is the location accessible or feasible for such sensing?
Comment: The sensing of capacitor current to find
 
lim∆→0
t+d(x)∆
f (x(τ )) dτ the derivative of voltage variable is not always the best
lim∆→0 [x(t + ∆) − x(t)] t
option. This is particularly true for the case where the
=
∆ ∆ capacitor is a filter capacitor. The addition of a current
(9)
sensor in its current flow path will increase the impedance
i.e., and therefore deteriorate the filtering process.
4) What types of sensors are required?
dx Comment: For the sensing of the filtering capacitors’
= ẋ = f (x, d) (10)
dt currents, very low impedance current transformers are
normally required. However, for sensing currents that
which is referred to as the average PWM-controlled system.
have dc average components, resistors or sophisticated
Therefore, it is shown that, as the duty-cycle frequency tends
Hall-effect sensors may be required. These may affect the
to infinity, the ideal average behavior of the PWM-controlled
overall efficiency or cost of the dc–dc converters.
system is represented by the smooth response of the system
constituted by the duty ratio d(x). It should also be noted Moreover, for the PWM-based SM controllers, the indirect
that the duty ratio d(x) replaces the discrete function u in the implementation of the original SM control law may result in
same manner as the equivalent control ueq of the SM-control unexpected complications in the signal computation and is
scheme to obtain (6). Hence, the relationship d(x) = ueq (x) is therefore not always implementable for some SM controller
established [72]. types. Hence, the choice for the state variable is critical for the
successful implementation of an analog SM controller.
Finally, similar to conventional controllers, the physical limi-
V. P RACTICAL I SSUES IN I MPLEMENTATION : tation of the analog devices, e.g., bandwidths, propagation time
A N A NALOG V IEWPOINT delays, slew rates, and saturation limits, of the SM controller
should be properly noted. They are the key factors affecting
There are a few works addressing the practical aspects of the healthy operation of the SM controller. More work is still
implementing the SM controller for dc–dc converters, specif- required in such areas of investigation.
ically in the area of analog implementation. This results in
the lack of understanding in its design principle. Issues like
the requirement for constant switching-frequency operation VI. P RACTICAL D ESIGN G UIDES
in the SM controller and the need to redefine the sliding Since SM control can achieve order reduction, it is typically
coefficients to meet the practical component constraints have sufficient to have an SM controller of n − 1 order for the stable
been covered in [60]–[70]. Here are a few aspects related to the control of an n-order converter. However, if fixed-frequency
analog implementation of the SM controller, which also deserve SM controllers are to be employed, the robustness and reg-
consideration. ulation properties of the converter system under the order-
First, the choice of system’s state variables, i.e., voltage, reduced SM controller will be deteriorated. A good method
current, and their derivatives and/or integrals, is important in of alleviating these deteriorations is to introduce an additional
that it affects not only the control performance but also the integral control variable into the fixed-frequency SM controller.
complexity of the implementation. These are a few questions This is known as integral or full-order SM control since the SM
that should be considered. controller is now of the same order as the converter [1]. The
1) How many state variables to control? function of making the SM controller full-order is to improve
Comment: The higher the number of state variables, the robustness, as well as the regulation, of the system. Hence,
the higher the degree of system’s controllability. How- in terms of ease of implementation and control performance, a

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
1168 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008

good option is to adopt an SM controller that employs a linear comply with the condition for SM existence. This is possible
combination of the system states, which has the same order as by inspecting the local reachability condition of the state trajec-
the converter. tory, i.e.,
Now, assuming that a full-order sliding-mode controller is
designed for a basic second-order converter, the control func- lim S · Ṡ < 0. (14)
S→0
tion of such a second-order controller is
 + In the case of a buck converter, the substitution of the con-
u , when S > 0 verter’s description into the aforementioned condition gives
u= (11)
u− , when S < 0

α3 α1 1
where S takes the form as described in (2). The task of the 0 < LC (Vref − βvo ) − βL − iC + βvo < βvi
α2 α2 rL C
designer is to determine the state of u+ /u− and to select proper (15)
parameters for α1 , α2 , and α3 such that the controller meets the
hitting, existence, and stability conditions for all of the system’s where C, L, and rL denote the capacitance, inductance, and
operating input and loading conditions. instantaneous load resistance of the converters, respectively;
Vref , vi , and vo denote the reference, instantaneous input, and
instantaneous output voltages, respectively; β denotes the feed-
A. Step 1: To Meet Hitting Condition back network ratio; and iC denotes the instantaneous capacitor
The design of the SM controller to meet the hitting condition current.
is rather straightforward in the case of power converters. As- Here, C, L, β, and Vref are the known parameters of the
sume that the output voltage is the control variable. Then, the converter system, and their exact values can be substituted di-
state variables of the full-order SM controller, which are to be rectly into the inequality for inspection. However, for vi and rL
controlled, may be expressed in the following form: which typically represent a range of input and output operating
    conditions that the converter may engage, it is necessary to
x1 Vref − βvo consider the boundary points of these operating conditions.
 x2  =  d(V ref −βv o )  (12) The compliance of either the maximum or minimum point of
 dt
x3 (Vref − βvo )dt these operating conditions is generally sufficient for ensuring
the abidance of the existence condition for the entire range of
where Vref and βvo denote the reference and sensed instan- condition. As for vo and iC , which are the instantaneous state
taneous output voltages, respectively; and x1 , x2 , and x3 are variables, the consideration of the time-varying nature of these
the voltage error, the voltage-error dynamics (or the rate of components undesirably complicates the evaluation. In the case
change of voltage error), and the integral of voltage error, re- of designing an SM controller with a static sliding surface,
spectively. For the design of the hitting condition, it is sufficient a practical approach is to ensure that the existence condition
to consider only the immediate state variable x1 , which is is met for the steady-state operation [26], [48]. With such
predominant in the composition of S during the reaching phase. considerations, the state variables iC and vo can be substituted
Apparently, if the sensed output voltage is much lower than with their expected steady-state parameters, i.e., iC(SS) and
the reference voltage, i.e., S is positive, the intuitive switching vo(SS) , which can be derived from the design specification. This
action required for the compensation is to turn on the power assures the compliance of the existence condition at least in
switch so that the energy is transferred from the input source the small region around the origin. Taking into consideration
to the inductor. Conversely, if the sensed output voltage is all the discussed presumptions, the resulting existence con-
much higher than the reference voltage, i.e., S is negative, the dition is
intuitive switching action is to turn off the power switch so

that the energy transfer between the source and the inductor is α3  α1 1
0 < LC Vref − βvo(SS) − βL −
discontinued. This forms the basis for the formulation of the hit- α2 α2 rL(min) C
ting condition. The resulting control function under the config-
× iC(SS) + βvo(SS) < βvi(min) . (16)
uration is
 The selected sliding coefficients must comply with the stated
1 = “ON,” when S > 0
u= (13) inequality. Similar existence conditions can be derived for other
0 = “OFF,” when S < 0.
types of converters using the same approach.
Clearly, the method of ensuring the hitting condition of the SM
controller is similar to the way in which the switching states of
C. Step 3: To Meet Stability Condition
the hysteresis controller are designed. Thus, the same approach
may be adopted for ensuring the hitting condition of an SM In addition to the existence condition, the selected sliding
controller which employs the current-mode type of control. coefficients must concurrently abide by the stability condition.
This is to ensure that, in the event of SM operation, the created
sliding surface will always direct the state trajectory toward a
B. Step 2: To Meet Existence Condition
point where a stable equilibrium exists. Interestingly, this can
With the switching states u+ /u− determined, the next stage be inherently accomplished through the design of the sliding
is to ensure that the selected sliding coefficients α1 , α2 , and α3 coefficients to meet the desired dynamical property [71]. This is

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
TAN et al.: GENERAL DESIGN ISSUES OF SLIDING-MODE CONTROLLERS IN DC–DC CONVERTERS 1169

Fig. 4. Experimental waveforms of output-voltage ripple vo and inductor current iL of the buck converter, with (a)–(c) the classical PWM voltage-mode
controller and (d)–(f) the 20-kHz-bandwidth PWM-based SM controller, operating at 5-kHz step-load-resistance change. (a) RL between 6 and 12 Ω (PWM
controller). (b) RL between 3 and 6 Ω (PWM controller). (c) RL between 3 and 12 Ω (PWM controller). (d) RL between 6 and 12 Ω (SM controller). (e) RL
between 3 and 6 Ω (SM controller). (f) RL between 3 and 12 Ω (SM controller).

possible by using the invariance property. Since it is known that, VII. C ASE -S TUDY E XAMPLES AND D ISCUSSIONS
in SM operation, the state trajectory S will track the path of the
In this section, experimental results of converters with the
sliding surface to a point of stability, an equation describing the
conventional PWM controllers and the PWM-based SM voltage
dynamical property of the system can be obtained by equating
controllers are provided as case-study examples to illustrate
the state trajectory to this sliding surface, i.e., by setting S = 0.
the difference between a large-signal-controlled system and a
The proper selection of the sliding coefficients will tailor the
small-signal-controlled system.
system to respond as desired by the designer. It follows that the
stability condition is inherently met.
In our example, the equation relating the sliding coefficients
A. PWM-Based SM Controller Versus Conventional Voltage
to the dynamic response of the converter during SM opera-
PWM Controller in Buck Converter
tion is
 The dynamic behavior of the PWM-based SMVC buck con-
dx1 verter is compared with that of the classical type of PWM
α1 x1 + α2 + α3 x1 dt = 0. (17)
dt voltage-mode-controlled buck converter. In the experiment, the
former employs a 20-kHz-bandwidth critically damped PWM-
This equation can be rearranged into a standard second-order based SM controller, whereas the latter employs a TL494
system form in which the design of the sliding coefficients α1 , PID PWM voltage-mode controller that is tuned to operate
α2 , and α3 can result in a system adopting one of these three with a critically damped response at a step-load change from
possible types of responses: underdamped, critically damped, RL = 3 Ω to RL = 12 Ω. Fig. 4(a)–(f) shows the experimental
and overdamped, with a desired convergency rate [74]. Hence, waveforms with both converters operating at 5-kHz step-load
designers can easily select the sliding coefficients based on their change.
converters’ response time and voltage overshoot specifications. With the classical PWM voltage-mode controller, the dy-
It should be noted that the selected sliding coefficients should namic behavior of the system is dissimilar at different operating
comply with the existence condition described in the previ- settings. Specifically, the response becomes more oscillatory
ous section. Additionally, if (α1 /α2 )  (1/rL C), an almost at lower currents, i.e., the output-voltage ripple waveform in
constant dynamical characteristic can be achieved throughout Fig. 4(a) exhibits more oscillation, Fig. 4(b) with less oscilla-
all operating conditions for the SM-operation phase. Likewise, tion, and Fig. 4(c) with no oscillation (when entered momen-
similar steps can be adopted for designing controllers of higher tarily into discontinuous conduction mode). This is expected
order. since the PWM controller is designed for a specific operating

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
1170 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008

Fig. 5. Experimental waveforms of output- voltage ripple vo and output current ir of the boost converter with the peak-current-mode controller operating at
input voltages of 20 V (minimum), 24 V (nominal), and 28 V (maximum) and alternating between load resistances of 24 Ω (minimum), 48 Ω (half), and 240 Ω
(maximum). (a) vi = 20 V (0.2/1.0 A). (b) vi = 20 V (1.0/2.0 A). (c) vi = 20 V (0.2/2.0 A). (d) vi = 24 V (0.2/1.0 A). (e) vi = 24 V (1.0/2.0 A). (f) vi = 24 V
(0.2/2.0 A). (g) vi = 28 V (0.2/1.0 A). (h) vi = 28 V (1.0/2.0 A). (i) vi = 28 V (0.2/2.0 A).

condition, which leads to changes in the response behavior response of the latter will only comply with the design at a
when a different operating condition is engaged. specific operating condition.
On the other hand, with the PWM-based SM controller,
the dynamic behavior of the output-voltage ripple is basically
B. PWM-Based SM Controller Versus Peak-Current-Mode
similar (i.e., critically damped) for all three operating load con-
Controller in Boost Converter
ditions, even when it enters momentarily into a discontinuous
conduction mode and even when it experiences a change in The dynamic behavior of the PWM-based controller is
converter’s description. This demonstrates an advantage of the compared with that of a UC3843 peak-current-mode PWM
SM controller in terms of robustness in the dynamic behavior at controller that is optimally tuned to operate at the step-load
different operating conditions and uncertainties. Additionally, change from rL = 240 Ω to rL = 24 Ω for the input condition
the example also illustrates a major difference between a large- vi = 24 V. Fig. 5(a)–(i) shows the experimental waveforms
signal-controlled system (SM) and a small-signal-controlled of the peak-current-mode-controlled boost converter operating
system (PWM), i.e., the former complies with the design with at a load resistance that alternates between rL = 24 Ω and
a similar response for all operating conditions, whereas the rL = 240 Ω for various input voltages.

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
TAN et al.: GENERAL DESIGN ISSUES OF SLIDING-MODE CONTROLLERS IN DC–DC CONVERTERS 1171

Fig. 6. Experimental waveforms of output-voltage ripple vo and output current ir of the boost converter with a 1.5-krad/s bandwidth PWM-based SM controller
operating at input voltages of 20 V (minimum), 24 V (nominal), and 28 V (maximum) and alternating between load resistances of 24 Ω (minimum), 48 Ω (half),
and 240 Ω (maximum). (a) vi = 20 V (0.2/1.0 A). (b) vi = 20 V (1.0/2.0 A). (c) vi = 20 V (0.2/2.0 A). (d) vi = 24 V (0.2/1.0 A). (e) vi = 24 V (1.0/2.0 A).
(f) vi = 24 V (0.2/2.0 A). (g) vi = 28 V (0.2/1.0 A). (h) vi = 28 V (1.0/2.0 A). (i) vi = 28 V (0.2/2.0 A).

It can be seen that, with the peak-current-mode PWM con- since the peak-current-mode controller is designed under a
troller, the dynamic behavior of the system differs for different linearized small-signal model that is only optimal for a specific
operating conditions. Specifically, the response becomes less operating condition. Thus, when a different operating condition
oscillatory at higher input voltages. Moreover, the dynamic be- is engaged, the response varies.
havior and transient settling time are also different between the On the other hand, with the PWM-based SM controller,
various cases of operating conditions. Specifically, at a lower the dynamic behavior of the output-voltage ripple is basically
step-current change, i.e., from 0.2 to 1.0 A, the response of similar (i.e., slightly underdamped) for all operating input and
the system becomes critically damped, instead of its optimally load conditions. This is shown in Fig. 6(a)–(i), which shows
designed response which is slightly underdamped, as shown in the experimental waveforms of the PWM-based SM-controlled
Fig. 5(f). Furthermore, in the worst case operating condition: boost converter operating at the same set of operating condi-
vi = 20 V and step output-current change of 0.2–2.0 A, the tions as the peak-current-mode-controlled boost converter.
system has a settling time of 5.8 ms and a relatively high- Furthermore, the transient settling time, which is around
voltage ripple swing of 5.6 V [see Fig. 5(c)], which are much 4.4 ms, is also independent of the direction and magnitude
in deviation from the optimally designed value of 2 ms and a of the step-load change and the operating input voltages. This
voltage ripple swing of 2.6 V [see Fig. 5(f)]. This is expected coincides with the design which, being a 1.25-krad/s bandwidth

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
1172 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008

controller, is expected to have a settling time of 5τ = 5/1.25 = [17] R. Venkataramanan, A. Šabanović, and S. Ćuk, “Sliding mode control of
4.0 ms. Moreover, in the worst case operating condition: vi = DC-to-DC converters,” in Proc. IEEE Conf. IECON, 1985, pp. 251–258.
[18] S. P. Huang, H. Q. Xu, and Y. F. Liu, “Sliding-mode controlled Ćuk
20 V and step output-current change of 0.2–2.0 A, the settling switching regulator with fast response and first-order dynamic character-
time is still around 4.4 ms, and the voltage ripple swing is 3.4 V istic,” in Proc. IEEE PESC Rec., Jun. 1989, pp. 124–129.
[see Fig. 6(c)]. This is close to the optimally designed system, [19] E. Fossas, L. Martínez, and J. Ordinas, “Sliding mode control reduces
audiosusceptibility and load perturbation in the Ćuk converter,” IEEE
which has a voltage ripple swing of 3.0 V [see Fig. 6(f)]. This Trans. Circuits Syst. I, Fundam. Theory Appl., vol. 39, no. 10, pp. 847–
illuminates again the strength of the SM controller in terms 849, Oct. 1992.
of robustness in the dynamic behavior at different operating [20] L. Malesani, L. Rossetto, G. Spiazzi, and P. Tenti, “Performance opti-
mization of Ćuk converters by sliding-mode control,” IEEE Trans. Power
conditions and uncertainties. It also reinforces our belief that Electron., vol. 10, no. 3, pp. 302–309, May 1995.
the main advantage of the large-signal controller (SM) over the [21] M. Oppenheimer, M. Husain, M. Elbuluk, and J. A. De Abreu Garcia,
small-signal controller (PWM) is that it gives more consistent “Sliding mode control of the Ćuk converter,” in Proc. IEEE PESC Rec.,
Jun. 1996, vol. 2, pp. 1519–1526.
dynamic performances for operating conditions requiring wide [22] J. Mahdavi and A. Emadi, “Sliding-mode control of PWM Ćuk con-
ranges. verter,” in Proc. 6th Int. Conf. Power Electron. Variable Speed Drives,
Sep. 1996, vol. 2, pp. 372–377.
[23] J. Calvente, L. Martinez, and R. Giral, “Design of locally stable slid-
VIII. C ONCLUSION ing modes in bidirectional switching converters,” in Proc. 40th Midwest
Symp. Circuits Syst., Aug. 1997, vol. 1, pp. 615–618.
The various aspects concerning the application of sliding- [24] L. Martinez-Salamero, J. Calvente, R. Giral, A. Poveda, and E. Fossas,
“Analysis of a bidirectional coupled-inductor Ćuk converter operating
mode controllers in dc–dc converters are presented. A com- in sliding mode,” IEEE Trans. Circuits Syst. I, Fundam. Theory Appl.,
prehensive literature review on the area is provided. The vol. 45, no. 4, pp. 355–363, Apr. 1998.
issues related to the analog implementation of the sliding- [25] P. Mattavelli, L. Rossetto, G. Spiazzi, and P. Tenti, “General-purpose
sliding-mode controller for DC/DC converter applications,” in Proc. IEEE
mode controller are discussed. The advantage of using sliding- PESC Rec., Jun. 1993, pp. 609–615.
mode controller is highlighted. It is shown that the sliding-mode [26] P. Mattavelli, L. Rossetto, and G. Spiazzi, “Small-signal analysis of
controllers generate more consistent transient responses for a DC–DC converters with sliding mode control,” IEEE Trans. Power Elec-
tron., vol. 12, no. 1, pp. 96–102, Jan. 1997.
wide operating range as compared with the conventional linear [27] F. Dominguez, E. Fossas, and L. Martinez, “Stability analysis of a buck
controllers. converter with input filter via sliding-mode approach,” in Proc. IEEE
Conf. IECON, Sep. 1994, pp. 1438–1442.
[28] M. Castilla, L. C. de Vicuna, M. Lopez, O. Lopez, and J. Matas, “On
R EFERENCES the design of sliding mode control schemes for quantum resonant con-
[1] V. Utkin, J. Guldner, and J. X. Shi, Sliding Mode Control in Electro- verters,” IEEE Trans. Power Electron., vol. 15, no. 6, pp. 960–973,
mechanical Systems. London, U.K.: Taylor & Francis, 1999. Nov. 2000.
[2] V. Utkin, Sliding Modes in Control Optimization. Berlin, Germany: [29] P. F. Donoso-Garcia, P. C. Cortizo, B. R. de Menezes, and M. A. Severo
Springer-Verlag, 1992. Mendes, “Sliding mode control for current distribution in DC-to-DC
[3] C. Edwards and S. K. Spurgeron, Sliding Mode Control: Theory and converters connected in parallel,” in Proc. IEEE PESC Rec., Jun. 1996,
Applications. London, U.K.: Taylor & Francis, 1998. pp. 1513–1518.
[4] W. Perruquetti and J. P. Barbot, Sliding Mode Control in Engineering. [30] Y. B. Shtessel, O. A. Raznopolov, and L. A. Ozerov, “Sliding mode
New York: Marcel Dekker, 2002. control of multiple modular DC-to-DC power converters,” in Proc. IEEE
[5] J. J. E. Slotine and W. Li, “Sliding control,” in Applied Nonlinear Control. Int. Conf. Control Appl., Dec. 1996, pp. 685–690.
Englewood Cliffs, NJ: Prentice-Hall, 1991, ch. 7. [31] Y. B. Shtessel, O. A. Raznopolov, and L. A. Ozerov, “Control of multiple
[6] M. Comanescu and L. Xu, “Sliding-mode MRAS speed estimators modular DC-to-DC power converters in conventional and dynamic sliding
for sensorless vector control of induction machine,” IEEE Trans. Ind. surfaces,” IEEE Trans. Circuits Syst. I, Fundam. Theory Appl., vol. 45,
Electron., vol. 53, no. 1, pp. 146–153, Feb. 2006. no. 10, pp. 1091–1100, Oct. 1998.
[7] S. Janardhanan and B. Bandyopadhyay, “Output feedback sliding-mode [32] M. López, L. G. De-Vicuòa, M. Castilla, and J. Majo, “Interleaving of
control for uncertain systems using fast output sampling technique,” IEEE parallel DC–DC converters using sliding mode control,” in Proc. IEEE
Trans. Ind. Electron., vol. 53, no. 5, pp. 1677–1682, Oct. 2006. Conf. IECON, Sep. 1998, pp. 1055–1059.
[8] F. Betin, A. Sivert, A. Yazidi, and G. A. Capolino, “Determination [33] R. Giral, L. Martinez-Salamero, R. Leyva, and J. Maixe, “Sliding-mode
of scaling factors for fuzzy logic control using the sliding-mode ap- control of interleaved boost converters,” IEEE Trans. Circuits Syst. I,
proach: Application to control of a DC machine drive,” IEEE Trans. Ind. Fundam. Theory Appl., vol. 47, no. 9, pp. 1330–1339, Sep. 2000.
Electron., vol. 54, no. 1, pp. 296–309, Feb. 2007. [34] S. K. Mazumder, A. H. Nayfeh, A. Borojevic, P. Gaya, and O. López, “Ro-
[9] A. V. Topalov, G. L. Cascella, V. Giordano, F. Cupertino, and O. Kaynak, bust control of parallel DC–DC buck converters by combining integral-
“Sliding mode neuro-adaptive control of electric drives,” IEEE Trans. Ind. variable-structure and multiple-sliding-surface control schemes,” IEEE
Electron., vol. 54, no. 1, pp. 671–679, Feb. 2007. Trans. Power Electron., vol. 17, no. 3, pp. 428–437, May 2002.
[10] A. M. Lienhardt, G. Gateau, T. A. Meynard, and B. Bandyopadhyay, “Dig- [35] M. López, L. G. De-Vicun̆a, M. Castilla, P. Gaya, and O. López, “Current
ital sliding-mode observer implementation using FPGA,” IEEE Trans. distribution control design for paralleled DC/DC converters using sliding-
Ind. Electron., vol. 54, no. 4, pp. 1865–1875, Aug. 2007. mode control,” IEEE Trans. Ind. Electron., vol. 51, no. 2, pp. 419–428,
[11] R. Venkataramanan, “Sliding mode control of power converters,” Ph.D. Apr. 2004.
dissertation, California Inst. Technol., Dept. Elect. Eng., Pasadena, CA, [36] H. Sira-Ramirez and M. Rios-Bolivar, “Sliding mode control of DC-
May 1986. to-DC power converters via extended linearization,” IEEE Trans. Cir-
[12] R. D. Middlebrook and S. Ćuk, “A general unified approach to model- cuits Syst. I, Fundam. Theory Appl., vol. 41, no. 10, pp. 652–661,
ing switching-converter power stages,” in Proc. IEEE PESC Rec., 1976, Oct. 1994.
pp. 18–34. [37] H. Sira-Ramirez, R. Ortega, R. Perez-Moreno, and M. Garcia-Esteban,
[13] D. M. Mitchell, DC–DC Switching Regulator Analysis. New York: “A sliding mode controller-observer for DC-to-DC power converters:
McGraw-Hill, 1998. A passivity approach,” in Proc. 34th IEEE Conf. Decision Control,
[14] A. J. Forsyth and S. V. Mollow, “Modelling and control of DC–DC Dec. 1995, pp. 3379–3384.
converters,” Power Eng. J., vol. 12, no. 5, pp. 229–236, Oct. 1998. [38] H. Sira-Ramirez, G. Escobar, and R. Ortega, “On passivity-based sliding
[15] J. G. Kassakian, M. F. Schlecht, and G. C. Verghese, Principles of Power mode control of switched DC-to-DC power converters,” in Proc. 35th
Electronics. Reading, MA: Addison-Wesley, Jun. 1992. IEEE Conf. Decision Control, Dec. 1996, pp. 2525–2526.
[16] F. Bilalović, O. Mus̆ić, and A. S̆abanović, “Buck converter regulator [39] E. Fossas and D. Biel, “A sliding mode approach to robust generation on
operating in the sliding mode,” in Proc. 7th Int. Conf. PCI, Apr. 1983, DC-to-DC nonlinear converters,” in Proc. IEEE Int. Workshop Variable
pp. 331–340. Struct. Syst., Dec. 1996, pp. 67–71.

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
TAN et al.: GENERAL DESIGN ISSUES OF SLIDING-MODE CONTROLLERS IN DC–DC CONVERTERS 1173

[40] J. M. Carrasco, J. M. Quero, F. P. Ridao, M. A. Perales, and [63] J. Mahdavi, A. Emadi, and H. A. Toliyat, “Application of state
L. G. Franquelo, “Sliding mode control of a DC/DC PWM converter space averaging method to sliding mode control of PWM DC/DC con-
with PFC implemented by neural networks,” IEEE Trans. Circuits Syst. verters,” in Proc. Conf. Rec. IEEE IAS Annu. Meeting, Oct. 1997, vol. 2,
I, Fundam. Theory Appl., vol. 44, no. 8, pp. 743–749, Aug. 1997. pp. 820–827.
[41] S. A. Bock, J. R. Pinheiro, H. Grundling, H. L. Hey, and H. Pinheiro, [64] J. Mahdavi, M. R. Nasiri, and A. Agah, “Application of neural networks
“Existence and stability of sliding modes in bi-directional DC–DC con- and state-space averaging to a DC/DC PWM converter in sliding mode
verters,” in Proc. IEEE PESC Rec., Jun. 2001, vol. 3, pp. 1277–1282. operation,” in Proc. IEEE Conf. IECON, 2000, vol. 1, pp. 172–177.
[42] E. Fossas and A. Pas, “Second order sliding mode control of a buck [65] A. G. Perry, F. Guang, Y. F. Liu, and P. C. Sen, “A new sliding mode like
converter,” in Proc. 41st IEEE Conf. Decision Control, Dec. 2002, vol. 1, control method for buck converter,” in Proc. IEEE PESC Rec., Jun. 2004,
pp. 346–347. vol. 5, pp. 3688–3693.
[43] Y. B. Shtessel, A. S. I. Zinober, and I. A. Shkolnikov, “Boost and [66] L. Iannelli and F. Vasca, “Dithering for sliding mode control of DC/DC
buck–boost power converters control via sliding modes using method converters,” in Proc. IEEE PESC Rec., Jun. 2004, vol. 2, pp. 1616–1620.
of stable system centre,” in Proc. 41st IEEE Conf. Decision Control, [67] S. K. Mazumder and S. L. Kamisetty, “Experimental validation of a
Dec. 2002, vol. 1, pp. 340–345. novel multiphase nonlinear VRM controller,” in Proc. IEEE PESC Rec.,
[44] Y. B. Shtessel, A. S. I. Zinober, and I. A. Shkolnikov, “Boost and Jun. 2004, pp. 2114–2120.
buck–boost power converters control via sliding modes using dynamic [68] J. Matas, L. G. De-Vicuňa, O. López, and M. López, “Discrete slid-
sliding manifold,” in Proc. 41st IEEE Conf. Decision Control, Dec. 2002, ing mode control of a boost converter for output voltage tracking,” in
vol. 3, pp. 2456–2461. Proc. 8th Int. Conf. Power Electron. Variable Speed Drives, Sep. 2000,
[45] N. Vazquez, C. Hernandez, J. Alvarez, and J. Arau, “Sliding mode control pp. 351–354.
for DC/DC converters: A new sliding surface,” in Proc. IEEE Int. Symp. [69] R. Orosco and N. Vazquez, “Discrete sliding mode control for DC/DC
Ind. Electron., Jun. 2003, vol. 1, pp. 422–426. converters,” in Proc. IEEE Int. CIEP Power Electron. Congr., Oct. 2000,
[46] P. Gupta and A. Patra, “Hybrid sliding mode control of DC–DC power pp. 231–236.
converter circuits,” in Proc. IEEE Region Ten Conf. Convergent Technol. [70] S. C. Tan, Y. M. Lai, M. K. H. Cheung, and C. K. Tse, “On the practical
Asia-Pac. Region, Oct. 2003, vol. 1, pp. 259–263. design of a sliding mode voltage controlled buck converter,” IEEE Trans.
[47] H. Sira-Ramirez, “Sliding mode-∆ modulation control of a “buck con- Power Electron., vol. 20, no. 2, pp. 425–437, Mar. 2005.
verter”,” in Proc. 42nd IEEE Conf. Decision Control, Dec. 2003, vol. 3, [71] J. Ackermann and V. Utkin, “Sliding mode control design based on
pp. 2999–3004. Ackermann’s formula,” IEEE Trans. Autom. Control, vol. 43, no. 2,
[48] H. Sira-Ramirez, “On the generalized PI sliding mode control of DC- pp. 234–237, Feb. 1998.
to-DC power converters: A tutorial,” Int. J. Control, vol. 76, no. 9/10, [72] H. Sira-Ramirez, “A geometric approach to pulse-width modulated con-
pp. 1018–1033, 2003. trol in nonlinear dynamical systems,” IEEE Trans. Autom. Control,
[49] V. S. C. Raviraj and P. C. Sen, “Comparative study of proportional- vol. 34, no. 2, pp. 184–187, Feb. 1989.
integral, sliding mode, and fuzzy logic controllers for power converters,” [73] L. Martinez, A. Poveda, J. Majo, L. Garcia-de-Vicuna, F. Guinjoan,
IEEE Trans. Ind. Appl., vol. 33, no. 2, pp. 518–524, Mar./Apr. 1997. J. C. Marpinard, and M. Valentin, “Lie algebras modeling of bidirectional
[50] D. Cortes and J. Alvarez, “Robust sliding mode control for the boost switching converters,” in Proc. ECCTD, Sep. 1993, vol. 2, pp. 1425–1429.
converter,” in Proc. IEEE Int. CIEP Power Electron. Congr., Oct. 2002, [74] S. C. Tan, Y. M. Lai, C. K. Tse, and M. K. H. Cheung, “A fixed-frequency
pp. 208–212. pulse-width-modulation based quasi-sliding mode controller for buck
[51] C. Morel, J. C. Guignard, and M. Guillet, “Sliding mode control of DC- converters,” IEEE Trans. Power Electron., vol. 20, no. 6, pp. 1379–1392,
to-DC power converters,” in Proc. 9th Int. Conf. Electron., Circuits Syst., Nov. 2005.
Sep. 2002, vol. 3, pp. 971–974.
[52] C. Morel, “Slide mode control via current mode control in DC–DC con-
verters,” in Proc. IEEE Int. Conf. Syst., Man, Cybern., Oct. 2002, vol. 5,
pp. 6–11.
[53] C. Morel, “Application of slide mode control to a current-mode-controlled
boost converter,” in Proc. IEEE Conf. IECON, Nov. 2002, vol. 3,
pp. 1824–1829.
[54] G. Escobar, R. Ortega, H. Sira-Ramirez, J. P. Vilain, and I. Zein,
“An experimental comparison of several nonlinear controllers for power Siew-Chong Tan (S’00–M’06) received the B.Eng.
converters,” IEEE Control Syst. Mag., vol. 19, no. 1, pp. 66–82, (with honors) and M.Eng. degrees in electrical and
Feb. 1999. computer engineering from the National University
[55] H. Chiacchiarini, P. Mandolesi, and A. Oliva, “Nonlinear analog controller of Singapore, Singapore, in 2000 and 2002, respec-
for a buck converter: Theory and experimental results,” in Proc. IEEE Int. tively, and the Ph.D. degree from Hong Kong Poly-
Symp. Ind. Electron., Jul. 1999, pp. 601–606. technic University, Kowloon, Hong Kong, in 2005.
[56] E. Alarcon, A. Romero, A. Poveda, S. Porta, and L. Martinez-Salamero, He worked briefly as a Research Associate and
“Sliding-mode control analog integrated circuit for switching DC–DC then as a Postdoctoral Fellow with the Department
power converters,” in Proc. IEEE Int. Symp. Circuits Syst., May 2001, of Electronic and Information Engineering, Hong
pp. 500–503. Kong Polytechnic University, where he is currently a
[57] M. Ahmed, M. Kuisma, K. Tolsa, and P. Silventoinen, “Implementing Lecturer. His research interests include motor drives
sliding mode control for buck converter,” in Proc. IEEE PESC Rec., and power electronics.
Jun. 2003, vol. 2, pp. 634–637.
[58] M. Ahmed, M. Kuisma, P. Silventoinen, and O. Pyrhonen, “Effect of im-
plementing sliding mode control on the dynamic behavior and robustness
of switch mode power supply (buck converter),” in Proc. 5th Int. Conf.
Power Electron. Drive Syst., Nov. 2003, vol. 2, pp. 1364–1368.
[59] M. Ahmed, M. Kuisma, O. Pyrhonen, and P. Silventoinen, “Sliding
mode control for buck–boost converter using control desk dSPACE,” in
Proc. 5th Int. Conf. Power Electron. Drive Syst., Nov. 2003, vol. 2,
pp. 1491–1494. Y. M. Lai (M’92) received the B.Eng. degree in
[60] B. J. Cardoso, A. F. Moreira, B. R. Menezes, and P. C. Cortizo, electrical engineering from the University of Western
“Analysis of switching frequency reduction methods applied to sliding Australia, Perth, Australia, in 1983, the M.Eng.Sc.
mode controlled DC–DC converters,” in Proc. IEEE APEC, Feb. 1992, degree in electrical engineering from the University
pp. 403–410. of Sydney, Sydney, Australia, in 1986, and the Ph.D.
[61] V. M. Nguyen and C. Q. Lee, “Tracking control of buck converter us- degree from Brunel University, Uxbridge, U.K.,
ing sliding-mode with adaptive hysteresis,” in Proc. IEEE PESC Rec., in 1997.
Jun. 1995, vol. 2, pp. 1086–1093. He is an Assistant Professor with Hong Kong
[62] V. M. Nguyen and C. Q. Lee, “Indirect implementations of sliding-mode Polytechnic University, Kowloon, Hong Kong. His
control law in buck-type converters,” in Proc. IEEE APEC, Mar. 1996, research interests include computer-aided design of
vol. 1, pp. 111–115. power electronics and nonlinear dynamics.

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.
1174 IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 55, NO. 3, MARCH 2008

Chi K. Tse (M’90–SM’97–F’06) received the

B.Eng. (Hons.) degree (with first class honors) in
electrical engineering and the Ph.D. degree from the
University of Melbourne, Melbourne, Australia, in
1987 and 1991, respectively.
He is currently a Chair Professor with and the
Head of the Department of Electronic and Informa-
tion Engineering, Hong Kong Polytechnic Univer-
sity, Kowloon, Hong Kong. He is a Guest Professor
with Wuhan University, Wuhan, China, and South-
west University, Chongqing, China. His research in-
terests include power electronics, complex networks, and nonlinear systems. He
is the Author of Linear Circuit Analysis (Addison-Wesley 1998) and Complex
Behavior of Switching Power Converters (CRC Press, 2003), the Coauthor
of Chaos-Based Digital Communication Systems (Springer-Verlag, 2003) and
Chaotic Signal Reconstruction with Applications to Chaos-Based Communica-
tions (TUP, 2005), and the Coholder of a U.S. patent and two pending patents.
In 1987, he was awarded the L.R. East Prize by the Institution of Engineers,
Australia, the IEEE TRANSACTIONS ON POWER ELECTRONICS Prize Paper
Award in 2001, and the International Journal of Circuit Theory and Appli-
cations Best Paper Award in 2003. In 2007, he is awarded the Distinguished
International Research Fellowship by the University of Calgary, Calgary, AB,
Canada. While with Hong Kong Polytechnic University, he received twice the
President’s Award for Achievement in Research, the Faculty’s Best Researcher
Award, the Research Grant Achievement Award, and a few other teaching
awards. From 1999 to 2001, he served as an Associate Editor for the IEEE
TRANSACTIONS ON CIRCUITS AND SYSTEMS PART I—FUNDAMENTAL
THEORY AND APPLICATIONS, and since 1999, he has been an Associate Editor
for the IEEE TRANSACTIONS ON POWER ELECTRONICS. In 2005, he served
as an IEEE Distinguished Lecturer. Currently, he also serves as the Editor-in-
Chief of the IEEE Circuits and Systems Society Newsletter, the Associate Editor
for the International Journal of Systems Science, and the Guest Editor of a few
other journals.

Authorized licensed use limited to: IEEE Xplore. Downloaded on December 17, 2008 at 22:14 from IEEE Xplore. Restrictions apply.