Development of Tools for the Study of Chaotic Behavior in Power Electronics

Eduardo Colon Uriel Contreras . Felix Rodriguez Efi-ain O’Neill-Canillo

Electrical & Computer Engineering Department
University of Puerto Rico-Mayagiiez
P.O.Box 9042, Mayagiiez, PR 00681-9042
oneill@ieee.org
Abstraa - This paper describes the development of software devices. Chaotic systems have time responses that appear to
tools for the analysis of chaos in power electronics systems. be noise, but the dynamical representation of such processes
This project was motivated by potential contributions of is deterministic [3]. Previous research work has emphasized
chaos theory in the design, analysis and control of power the chaotic characteristics in the Buck converter [4] - [6].
electronics circuits. Nonlinear analysis software and com- Nevertheless, there are chaotic components in other power
puter programs were used for the detection and analysis of electronics circuits [7]. Chaos theory can make a significant
chaotic components. Simulations of power electronics &- contribution in control schemes, noise reduction and other
vices were performed using commercially available circuit power electronics applications [7]. Chaotic dynamics pro-
analysis packages. The voltage and current time series ob- vides an alternate modeling approach that can lead to better
tained from the circuit analysis were studied using the nonlin- designs and new ways to explain some of the nonlinear phe-
ear analysis tools developed for this project as well as existing nomena in power electronics circuits.
nonlinear analysis programs. The results of the nonlinear The objective of this paper was to compile available non-
analysis can be used to design and implement better mitigat- linear analysis and simulation tools for use in power electron-
ing and control techniques for the circuits under study. Cha- ics. The motivation for this work was the potential contribu-
otic dynamics provide an alternate representation that can tion of chaos theory in the analysis and control of power elec-
lead to better designs and new ways to explain nonlinear be- tronics circuits. Algorithms for the calculation of the largest
havior in power electronics circuits. There is also a great Lyapunov exponent [SI, [9] were used to study of chaotic
pedagogical value in complementing the traditional represen- behavior in simple systems such as the Lorenz system [3].
tation of power electronic circuits with alternate models. Tools for the study of more complex systems are currently
being developed. The simulations of power electronics &-
I. WTRODUCTIOE vices were performed using Spice.
Power electronics devices have revolutionized the way 11. CHAOS IN POWER ELECTRONICS
electric energy is processed and used. These devices provide
new ways to control and utilize energy more efficiently [I]. A . Theoretical Background
Nowadays, half the electric energy in the U.S. is processed by Voltages and currents in an electronic circuit can behave
some kind of power electronics device. However, the nonlin- in one of the following ways: a) settle to a constant value; b)
ear characteristics of most power electronics circuits may pm- increase exponentially until limited by the power supply; c)
duce power quality problems, e.g., harmonics [2]. Mitigating vary with a defined period; d) behave chaotically [IO]. Chaos
strategies need to be developed to deal with such undesirable is only found in nonlinear systems (of order three or greater),
effects. The use of power electronics models in computer however a common philosophy in circuit analysis is to line-
-
simulations enables the researcher to experiment with a wide arize before analysis. This approach does not account for
variety of operating conditions and identify the critical I- experimentally observed nonlinear phenomena that can be of
pects of the device operation that create power quality con- paramount importance in power electronics. For example, for
cerns. Later, the researcher can build a scaled version of the certain sets of parameters of dc-dc converters the steady state
system under study based on the software simulations. trajectory does not settle to a desired value but appears to
Many traditional methods to model power electronics cir- approach an aperiodic waveform. This kind of erratic behav-
cuits include the linearization of components [I]. Although, ior could be explained using nonlinear models based on cha-
using idealized characteristics simplifies the analysis, leaving otic dynamics [IO].
out the nonlinearity inherent in power electronics may affect Chaotic signals show irregular, non-periodic but bounded
the results of the simulation. Perhaps more important, unex- behavior. Sensitivity to initial conditions is another charac-
pected difficulties may appear in the implementation of the teristic of chaos; even small changes in initial conditions may
circuit or mitigating strategy. Chaotic dynamics provide an cause the dynamics of a chaotic system to change signifi-
alternate way to model nonlinearities in power electronics cantly. Chaotic time series appear to be random even though

0-7803-6561-5/00/$10.00 0 2000 IEEE 177

the system dynamics are known to be deterministic. The pre- B. Conditions for Chaos
sent state of a chaotic system has some relationship with pre- Chaos in power electronics is sometimes treated as noise,
vious and future states. Some authors prefer to use the term other times it is regarded as a difficult subject with no appar-
deterministic chaos to emphasize this point. These charac- ent effect on circuit operation. However, power converters
teristics make chaotic systems difficult to predict in the long are rich in chaotic dynamics. When the regions of operation
term. If the equations that govem the dynamics of a system are pushed towards stability limits, it is worthwhile to explore
are known, the presence of chaos can be mathematically if the so-called unstable modes are in fact chaotic. Some of
proven, even in the presence of discontinuities [ 1 11. Chaotic the first references of chaos in power electronics date back to
dynamics can also be identified from system measurements the early 1980s [17]. Even though many references present
for some systems [3], [l 13. theoretical studies, chaos has also been reported in practical
These seemingly contradictory characteristics of chaos operating ranges of power electronic devices [7]. Some drive
can be understood using stability concepts. A system may systems as well as several dc-dc converter configurations
show unstable properties within a bounded region, while exhibit chaotic behavior [181 - [22].
showing stable behavior globally (to an observer outside the The interaction of nonlinear components with certain
bounded region) [3], [12]. Trajectories in a chaotic system are range of operating parameters can cause qualitatwe changes
contained within a region of state space, called a strange at- or bifurcations in a power converter that lead to multiple
tractor for its peculiar shape. Variations in initial conditions steady states. Which steady state emerges will depend on
will cause nearby trajectories to diverge, following one of the the initial conditions [lo]. If bifurcations continue to occur,
orbits contained in the attractor, but the trajectories will E- the converter could start exhibiting chaotic behavior. Corn-
main inside a bounded region (the attractor). In other words, mon sources of nonlinearities in power electronics include:
a chaotic system is an aperiodic limit cycle [3], [13]. In a cha- Non-linear snubber capacitors, the capacitance: of diodes,
otic system, it is difficult to predict actual values far ahead. BJTs, and mosfets; non-linear inductance in transformers,
However, the attractor determines global features useful in the ferroresonant controllers and saturable snubber inductors.
analysis of the system. Fig. 1 shows the xy phase portrait of a Load changes, noise and external disturbances create the
well-known chaotic system, the Lorenz system [14]. need for the control of the converter duty cycle [7]. Most of
the literature of chaos in power converters deals with instabili-
--I ; I-. I ties caused by the operation of switches controllled by pulse
width modulation (PWM). Switch-mode systems tend to de-
velop instability and chaos because switches cause the
topological structure of the converter to vary with time [lo].
In other words, the operation of switches and other nonlinear
components makes the converter piecewise-linear in time,
introducing discontinuities in the circuit operation [ 1 I].
Chaos may appear in current-mode or voltage feedback
controller topologies [7], [23]. The use of a switching fre-

-- ; ‘----+--- . .
I
quency much higher than the system charactertstic frequen-
cies is another cause for chaos in power electronics 1241.
-60 I I Some designs might use a simple circuit to deal with protec-
40 -20 0 20 40 tive modes; however operation of a converter in a protective
State X mode can also lead to chaos [20]. The interaction of a satu-
Fig. 1. Attractor for the Lorenz system (xy-state plane)
rated transistor and a diode-resistor combination creates a
The presence of chaos can be viewed from two perspec- lack of synchronization between a circuit and its trigger
tives. Chaotic signals as the cause of undesirable behavior in mechanism that may cause chaos [ 131.
the system, e.g., non-desired “noise” and unreliability of sys-
tem operation. The goal on these studies is to detect chaos 111. SIMULATION
TOOLS
and find ways of reducing its impact on the system. On the
Tools commonly used for power electronics simulations
other hand, the chaotic properties of a process can be used in
can be classified into three groups: Power electronics simula-
a positive way as a mechanism of identification, a tool for
tion tools; general transient simulation tools or EMPT type of
analysis or even to mitigate other problems [15]. For example,
programs; general harmonics simulation tools or frequency
if chaotic behavior is part of a mode of operation, it will limit
domain simulation tools [25]. For this work, the first classifi-
the ability to predict future states of a system. Forecasting
cation was considered.
windows can be created to identify how far in the future can
Students developed simulation tools as part of an under-
predictions be made reliably [16]. graduate research project guided by a faculty advisor. Com

178
mercially-available packages were used to simulate dc-dc
converters. As a first step, a simple program that simulates
laboratory components was used to study currents and volt-
ages in the circuit components. Because of the ease of its
use, this software was used to introduce students to power
electronics fundamentals. Spice was used for the actual i n
plementation of power electronic topologies. It was also used
to obtain current and voltage waveforms for analysis of cha-
otic properties.
The goal was to develop modules that could be used to
study chaos in various operating modes of the converters. I I
i
i
These simulations provided the necessary data for nonlinear
analysis. These modules are the first modules of an environ- Fig. 2. Buck converter with PWM control
ment for the study of chaos in power electronics. At this
early stage, emphasis was given on topologies that the Parameters used were: L = 20 mH, C = 47 @, R = 22 R,
authors considered present fundamental concepts in power amplifier gain = 8.4, Vrel= 11.3 v, initial ramp voltage = 3.8 v,
electronics, the Buck and boost converters. Furthermore, upper ramp voltage = 8.2 v, T= 400 p.
there is ample literature coverage on these two topologies.
The time step for the simulations should be carefully B. BOOST CONVERTER
selected. It should be small enough to avoid convergence This topology is also known as a step up converter be-
problems and to ensure that chaotic dynamics are not f cause the output voltage is greater or equal to the input volt-
fected. Guidelines to avoid truncation and rounding errors in age. There are two modes of operation: Continuous and dis-
the simulation of chaos in Spice are presented in [ 131. continuous. In continuous mode, the inductor current is al-
ways positive. If the inductor current reaches zero before the
A. BUCK CONVERTER end of a period, the mode of operation is discontinuous. Dis-
This topology is also known as a step down converter be- continuous mode occurs when the inductance is too small or
cause its principal function is to transform a non-regulated the switching period is relatively long. The value of induc-
voltage (input) to a lower level regulated voltage (output). tance should be greater than in ( 2 ) to avoid the discon-
The inductor current is positive, meaning that it is operating tinuous mode.
in a continuous mode. Applications of the buck topology
include resonant converters (reduce switching losses), in ac- Lmjn= [D*(1-D)"2* R]I[2fl
ac converters (regulate bus voltages, isolation), and in elec-
where D is the duty cycle, R is the load, and f is the switching
tronic ballasts (used for power factor correction).
frequency.
The simulations in this section were based on information
The PWM control of the switching in the Boost is similar
cited in [7]-[lo], [13] and [22]. Fig. 2 shows the configuration
to the structure discussed for the Buck converter. The fre-
of the Buck converter implemented in Spice. The input-output
quency of the sawtooth waveform was 3 kHz, and the refer-
relationship of this converter is given by
ence sinusoid frequency used was 60Hz. The input-output
V, =Kn*D, relationship is:

Controlling the switch's duty cycle, D, the value of Vo,,,

can be V, =Vj,,l(l-D). (3)
kept within a desired range. The converter has PWM control The simulations in this section were based on 11.51, 1201 and
of the switching. A reference voltage is compared to the out-
[21]. Fig. 3 shows the Boost topology as well as the values
put voltage (capacitor voltage), and then the result is c o m used for the simulations.
pared to a sawtooth (ramp) waveform so that corrections can The duty cycle of the circuit studied was 0.2874, the value
be made on the value of D. ofR = 125 Q. Using these values in (2) results in an Lmin =3
A first order Buck converter (no capacitor in the output) mH. The inductor value used was 2 mH, therefore simulations
was simulated to familiarize students with the topology. Stu- of the Boost converter were made in discontinuous mode. For
dents tested a range of voltages to identify those values that the intervals where the inductor current is zero, the diode and
did not cause a quick change in the system while the switch switch do not allow conduction to the rest of the circuit while.
was closed (multiple pulsing). Students also experimented The capacitor discharges through the load. Analysis of the
with values of circuit parameters. The second order buck
chaotic properties of the boost converter will be presented in
converter (as shown in Fig. 2) exhibits chaotic dynamics as the next section.
discussed in Section IV.

179
 where R/is the variable resistor in Fig. 3, R, is the resistor at
the top of the voltage divider in Fig. 3 and V, is the height of
the ramp voltage applied to the second comparator. The con-
verter exhibited chaos for values over 0.14. Fig. 5 shows the
capacitor voltage waveform for &=O. 14. There is bounded,
":
non-periodic behavior emerging from a deterministic system
I.. x without any random stimulus. Thus, there is qualitative evi-
dence of chaotic behavior.
II 1.11 -
Fig. 3. Boost converter with PWM control

IV. ANALYSIS TOOLS

Within Spice, there are various analytical tools that were

used to study the characteristics of the topologies discuss:d
in Section 111. Current and voltage time series were also ana-
lyzed using Matlab. Time series and frequency domain ob- Fig. 5. Time series for capacitor voltage in a boost converter
servations were carried out. The attractor of both converters
Table 1 presents the results of the analysis carried out using
was studied to identify chaotic properties.
time series and attractors from the topologies simulated.
A . Buck Converter
TABLE 1
Chaos occurred because the inductor current entered in a SUMMARY OF RESULTS FOR THE BUCK AND BOOST CON-
non-periodic region as the input voltage of the buck (see Fig. VERTERS
2) was increased. Input voltage was varied from 30 to 35 Converter Type BUCK
volts. If the loop gain is high, the ripple that is fed-back inter-
feres with the PWM process. The voltage is kept close to
reference, but the ripple is chaotic [IO]. Nevertheless, since
chaos is bounded behavior, the circuit can be made to work
reliably [13]. Thus, chaotic operation of the converter does
behavior
not necessarily mean unreliable operation. The capacitor
voltage followed the changes in inductor current.
C. Lyapunov Exponents
Fig. 4 shows voltage and current in the state space. It
The Lyapunov exponents are a measure of the rate of di-
shows non-periodic, but bounded behavior, and the dynamics
vergence (or convergence) of state trajectories whose initial
are known to be deterministic. Thus, we have qualitative evi-
conditions are infinitesimally close together. A positive Ly-
dence of chaos in the buck converter. Results obtained from
apunov exponent indicates net average divergence from initial
the simulations agree with previous studies in the area.
conditions. This in turn implies sensitive dependence on ini-
> ."$"
tial conditions, and therefore the presence of chaos.
In general, there are k Lyapunov exponents associated
with a dynamic process in a k-dimensional phase space. For a
discrete dynamical system, the k" exponent is defined as

hk =h[ lim (Sn,k>""] (5)

n-m
whenever the limit exists, s,,k can be regarded as the length
of the kh semi-axis of the n'* iterate of an infinitesimally small
ellipsoid of initial conditions [3]. The Lyapunov exponents
. "*,,"-.-.-.---...
__. ^.... ._ .___
____..__--. .-..._
_____"*
. U "2
~

U im:..f
E...
*I :m"C n -7.- rmra
mT n.n.2- j are a useful diagnostic tool, since the magnitude of the largest
Fig. 4. Attractor obtained from a second order Buck converter positive Lyapunov exponent determines the time scale over
B. Boost Converter which the system dynamics become unpredictable.
In this simulation, the closed loop gain was varied be- If the dynamics of a system are known, one could apply
tween 0.09 and 0.15. The close loop gain is given by ( 5 ) to obtain all k exponents. If there is no mathematical de-
scription of the system, then the largest positive Lyapunov
exponent, k,,can be estimated using a method based on art

180
bedding dimensions [8]-[9]. Chaotic systems that do not con- ear dynamics in a practical setting. This paper also demon-
tain discontinuities can be easily handled analyzed using e- strates the value of interdisciplinary work, since we would be
ther method. However, switching in power electronic devices applying a seemingly complex mathematical subject to the
create discontinuities that complicate the calculation of Ly- solution of an engineering problem. These tools can be used
apunov exponents [ I l l . to familiarize electrical engineering students with nonlinear
Software was developed in Matlab to model and obtain analysis, and also mathematics and physics students with the
the Lyapunov exponent of simple chaotic systems. The ex- engineering applications of nonlinear dynamics.
ponents could be regarded as “theoretical” values since they
were obtained directly from a model. A program was also writ- VI. FUTUREWORK
ten in Fortran to determine the largest Lyapunov exponent
from experimental data [SI. The program was validated with The ultimate objective of this research is to develop a
the simple models, and its results closely matched the “theo- software environment in which to study, not only chaos, but
retical” values. However, due to the discontinuities of the nonlinear phenomena in general. This project is the founda-
simulated data from the buck and boost converters, the algo- tion for the subsequent development of power electronics
rithm did not converge. Other algorithms are being evaluated experiments (hardware) that would support undergraduate
to address this problem [I 13. and graduate courses in power electronics, power quality and
Chaos in power electronics should receive attention k- possibly mathematics or physics. The tools discussed in this
cause it can create undesirable nonlinear effects in some de- paper constitute a first step in the compilation of resources to
vices. If the regions of chaotic operation are identified, one be used in an applied nonlinear systems laboratory at the
can avoid those modes of operation. However, chaos might University of Puerto Rico-Mayaguez (UPRM).
be an inherent part of the operation of a circuit that cannot be Work presently under way includes the implementation of
filtered out or disregarded. Since chaos is bounded behavior, other topologies such as the buck-boost and the Cuk con-
the chaotic operation of a dc-dc converter does not necessar- verters. In regard to the calculation of the Lyapunov expo-
ily mean unreliable operation. The practical merit of studying nents, a variational approach discussed in [l 11 is being con-
chaos is a better understanding of the nonlinear dynamics of sidered to deal with the discontinuities of power electronics
dc-dc converters, which will lead to more reliable designs, and measurements.
new possibilities of operating regimes that can help optimize Further research using these tools could result in: Poten-
design [J 01. tial reduction of EMC (by chaotic spread spectrum) [15][26];
better understanding of the stability of circuits [ 151; mitigation
v. OUTCOMES and control of chaos [27], [28]; controlled perturbations of a
chaotic system to obtain a wider range of operating modes
Although the study of non-linear systems was not an (flexibility of chaotic systems) [28]. However, there are certain
easy task for the students involved in this project, the experi- guidelines to follow when purposely using the chaotic proper-
ence motivated them to study harder and perform close to ties of a converter. In [29], the authors have identified the
graduate level. They engaged in literature search, technical regions of operation to ensure reliable chaotic operation (ro-
writing, and presentation of results to their peers and industry bust chaos) and a boost topology was tested satisfactorily.
people. It was also an opportunity to work closer with Fac-
ulty. For the advisor it was a chance to integrate his research VII. CONCLUSIONS
interests into teaching (undergraduate research is an elective
course at the University of Puerto Rico). The study of chaos may yield a better understanding of
Students used the simulation and analysis tools to iden- the nonlinear dynamics of dc-dc converters, which may result
tify mitigating strategies for chaos. The first recommendation in reliable designs, and new possibilities of operating regions
is to reduce switching frequency to diminish the impact of that can help optimize design. The analysis tools developed
parameter changes due to chaos. A designer should also in this project will be used to demonstrate power electronics
strive to keep circuits as simple as possible because complex principles in the classroom. Simulations were developed us-
switching schemes result in more nonlinearities and potential ing Spice. The calculation of Lyapunov exponents in power
sources of chaotic behavior. Finally, it is very important to electronics was also considered.
test components before implementation in order to identify The development and testing of software tools to study
possible imperfections that may cause nonlinear problems. chaos is a significant contribution to research efforts in power
Besides the research value of these tools, there is great electronics at the University of Puerto Rico-Mayaguez
pedagogical value in complementing the traditional represen- (UPRM). This approach provides an altemate representation
tation of power electronic circuits with altemate models. This that can lead to new ways to describe some of the nonlinear
work provides an excellent teaching tool to demonstrate not phenomena in power electronics circuits.
only power electronics principles, but also the use of nonlin-

181
 ACKNOWLEDGMENT [16] E. O’Neill-Carrillo, G.T. Heydt and E.J Kostelich, “Chaotic Phe-
nomena in Power Systems: Detection and Applications,” vol. 27, no. 1,
This work was supported by UPRM’s Industrial Affiliates January 1999, pp. 79-91.
Program and by the National Science Foundation’s ERC Pro- [ 171 J. Baillieul, R. Brockett, R. Washbum, “Chaotic Motion in Non-
gram (Award Number EEC-9731677). linear Feedback Systems,” IEEE Transactions on Circuits and Systems,
vol. 27, no. 1 1 , 1980, pp. 990-997.
REFERENCES
[ 181 K. Chau, J. Chen, C. Chan, J. pong, D. Chan, “Chaotic Behavior
[I ] N. Mohan, T. Undeland, W. Robbins, Power Electronics: Convert- in a Simple DC Drive,” Proceedings of the 1997 Power Electronics
ers, Applications and Design, Wiley, New York, 1995. Specialists Conference, St. Louis, vol. 1, pp. 473-479.

[2] J. Arrillaga, N.Watson, S. Chen, Power System Quality Assessment, [19] Z. Suto, I., Nagy, E. Masada, “Avoiding Chaotic Processes in
Wiley, Chichester, England, 2000. Current Control of AC Drive,” Proceedings of the I998 Power Elec-
tronics Specialists Conference, Fakuoka, Japan, vol. 1, pp. 255-261.
[3] K. Alligood, T. Sauer, J. Yorke, Chaos: A n Introduction to &-
namical Systems, Springer-Verlag, New York, 1997. [20] Isaac Zaffrany, Sam Ben-Yaakov, ‘‘ A Chaos of Subharmonic Os-
cillations in Current Mode PWM Boost Converters,” I’roceedings of
[4] J.B. Deane, D. Hamill, “Instability, Subharmonics and Chaos in the I995 Power Electronics Specialists Conference, Atlanta, June 1995,
Power Electronics Systems,” IEEE Transactions on Power Electronics, vol. 2, pp. 1 1 11-11 17.
vol. 5, no. 3, July 1990, pp. 260-268.
[21] C.K. Tse, “Flip Bifurcation and Chaos on Three-State Boost
[5] E. Fossas, G. Olivar, “Study of Chaos in the Buck Converter,” Switching Regulator,” IEEE Transactions on Circuits and Systems-I,
IEEE Transactions on Circuits and Systems I, vol. 43, January 1996, vol. 41, no. 1, Jan. 1994, pp. 605-608.
pp. 13-25.
[22] C.K. Tse, S.C. Fung, M.W. Kwan; “Experimental :I Current- Pm-
[6] G. Poddar, K. Chakrabarty, S. Banerjee, “Experimental Control of grammed Cuk Converter,” IEEE Transactions on Circuirs and Systems-
Chaotic Behavior of Buck Converter,” IEEE Transactions on Circuits I, ~01.43,no.7, Ju1.1996, pp. 605-608.
and Systems, vol. 42, no. 8, August 1995, pp. 502-504.
[23] M. di Bemardo, F. Garofalo, L. Glielmo, F. Vasca, “Analysis of
[7] M. di Bemardo, F. Garofalo, L. Glielmo, F. Vasca, “Analysis of Chaotic Buck, Boost and Buck-Boost Converters through Switching
Chaotic Buck, Boost and Buck-Boost Converters through Switching Maps,” Proceedings of the I997 Power Electronics Specialists Confer-
Maps,” Proceedings of the 1997 Power Electronics Specialists Confer- ence, St. Louis, vol. 1, pp. 751-760.
ence, St. Louis, vol. 1, pp. 751-760.
[24] J. Marrero, J. Font, G. Verghese, “Analysis of Chaotic Regime for
181 A. Wolf, J. Swift, H. Swinney, J. Vastano, “Determining Lyapunov DC/DC Converters under Current-Mode Control,” Proceedings of the
Exponents From a Time Series,” Physica D. vol. 16, 1985, pp. 285- 1996 Power Electronics Specialists Conference, Baveno, Italy, vol. 2,
317. pp. 1477-1483.

[9] H. Kantz, “A Robust Method to Estimate the Maximal Lyapunov [25] Modeling Task Force & Digital Simulation Working Group,
Exponent of a Time Series,” Physics Letters A, vol. 185, 1994, pp. “Guidelines for Modeling Power Electronics in Electric Power Engi-
77-87. neering Applications,“ IEEE Transactions on Power Delivery, vol. 12,
no 1, jan. 1997, pp 505-5 14.
[IO] D.C. Hamill, J. Deane, D.J: Jeffries; “Modeling of Chaotic DC-DC
Converters by Iterated Nonlinear Mappings,” IEEE Transactions on [26] J. Deane, P. Ashwin, D.C. Hamill, , D.J: Jeffries: “Calculation of
PowerElectronics, vol. 7 no. 1, Jan 1992, pp. 25-36. Periodic Spectral Components in a Chaotic DC-DC Converter,” IEEE
Transactions on Circuits and Systems-I, vol. 46, no. 11, Nov. 1999,
[ l 11 Y. Lim, D. Hamill, “Problems of Computing Lyapunov Expo- pp. 1313-1319.
nents in Power Electronics,” Proceedings of the 1999 International
Symposium on Circuits and Systems, Orlando, May 1999, vol. 5, pp. [27] J. Marrero, R . Santos, G. Verghese, ‘‘Analysis and Control of
V-297 - V-301. Chaotic DUDC Switching Power Converter,” Proceedings of the 1999
International Symposium on Circuits and Systems, Orlando, May 1999,
[ 121 J. R. Wood, ‘’ Chaos: A Real Phenomenon in Power Electronics”, -
vol. 5 , pp. V-287 V-292.
Proceedings ofthe I989 Power Electronics Specialists Conference, vol. (281 G.Poddar, K. Chakrabarty, S. Banerjee, “Control of Chaos in DC-
1, March 1989, pp. 115-124. DC Converters,” IEEE Transactions on Circuits and Sytems-I, vol. 45,
[I31 D.C. Hamill, “Leaming About Chaotic Circuits with Spice,” I€E€ no. 6, June 1998, pp. 672-676.
Transactions on Education, vol. 36, no. 1, February 1993, pp. 28-35.
[29] S. Banerjee, D. Kastha, S. Das,G. Vivek, “Robust Chaos - The
[ 141 E. Lorenz, “Deterministic Non-periodic Flow,” Journal of the Theoretical Formulation and Experimentatal Eviden’ze,” Proceedings
Atmospheric Sciences, vol. 20, 1963, pp. 130-141. of the I999 International Symposium on Circuits iind Sysrems, Or-
lando, May 1999, vol. 5, pp. V-293- V-296.
[ 151 D.C. Hamill, 1. Deane, P. Aston, “Some Applications of Chaos in
Power Converters,” IEE Colloquium on New Power Electronics Tech-
niques, May 1997, no. 1997/091, pp. 511-515.

182