Jour 92

Download as pdf or txt
Download as pdf or txt
You are on page 1of 16

INT. J. ELECTRONICS, 1994, VOL. 77, No.

5, 731- 746

Average simulation of PWM converters by direct implementation of


behavioural relationships

SAM BEN- Y AAKOVt

Average modelling of PWM converters is reexamined in the light of the


behavioural-dependent sources now included in modern versions of electronic
circuit simulators. A topology-independent behavioural model is developed to
emulate the operation of PWM converters in voltage and current modes, for
continuous and discontinuous inductor current cases. It is shown that the
operation of the switching part involves, in general, three behavioural blocks: the
generic switched inductor model, the duty cycle generator and the inductor current
generator. Explicit expressions and equivalent circuits are developed for all
possible modes of operation and a PSpice (MicroSim Inc., Irvine, CA) netlist
template is appended.

I. Introduction
Over the past several years, electronic circuit simulation has become highly
popular among electronic designers. In some areas, e.g. VLSI design for digital
systems, simulation has practically surpassed the classical breadboarding stage. In
other areas, such as analogue circuit design, simulation is gaining momentum as the
available simulators become more sophisticated, accurate and user friendly.
The switched inductor model (SIM) (Ben-Yaakov 1989, Ben-Yaakov et al. 1990,
Kimhi and Ben-Yaakov 1991, Amran et al. 1991, Ben-Yaakov and Gaaton 1992)
developed for PWM and quasi-resonant converters is based on the average behav-
iour of the switch mode circuit. This inherent feature and the fact that it is topology
independent makes the model highly compatible with general-purpose electronic
circuit simulators. Hence, it is potentially more 'engineer friendly' than the state
space averaging approach. As has already been shown, the SIM can be applied for
both small and large signal analysis of switch mode systems. Simulation results
obtained by running the SIM on general-purpose circuit simulators were found to be
in good agreement with previous studies and new experimental results for converters
operating in continuous and discontinuous modes (Ben-Yaakov 1989, Amran et al.
1991), cur:rent mode (Kimhi and Ben-Yaakov 1991, Amradn et al. 1991), quasi-
resonant converters (Ben- Yaakov et al. 1990) and average current mode (Ben-
Yaakov and Gaaton 1992).
The objective of this study is to reexamine the possibilities for simplifying
average simulation of PWM converters in the light of the new developments in
electronic circuit simulation technology.

2. Modern electronic circuit simulators


Most modern electronic circuit simulators are based on SPICE (Nagel 1975).
From a mathematical point of view, SPICE was designed to solve algebraic and

Received 7 September !993; accepted 20 February !994.


tDepartment of Electrica! and Computer Engineering, Ben-Gurion University of the
Negev, P.O. Box 653, Beer-Sheva, Israel.

0020-7217/94 SIO.OOif:) 1994 Taylor & Francis LId.


732 s. Ben-Yaakov

differential equations of networks that include passive and active devices and
polynomial-dependent sources. It turns out that average simulation of switch mode
systems requires the inclusion of more sophisticated dependent sources. This was
accomplished in previous studies by introducing complex equivalent circuits that
emulate the desired relationships (Ben-Yaakov 1989, Ben-Yaakov et al. 1990, Kimhi
and Ben-Yaakov 1991, Amran et al. 1991, Ben-Yaakov and Gaaton 1991).
The shortcoming of earlier simulators has been alleviated in modern packages
which now include generalized behavioural models. These dependent sources can be
described by any algebriac relationship and hence the restriction to polynomial
functions has been removed. For example, HSPICE (Meta-Software Inc., Campbell,
CA) permits the definition of a behaviour-dependent voltage source (EX) as

K x ~a. b)

EX=~

where K is a constant, J'(m.n)denotes the voltage between node m and n and ABS
denotes absolute value (note that the above form of the mathematical expression
does not conform to the syntax of HSPICE).
Once defined, the behavioural relationship is fully compatible with other
functions of the simulator That is, it can be applied in DC analysis, transient (large
signal) analysis and small signal analysis, which is performed after linearizing the
function around the quiescent operating point.

3. Behavioural PWM average model


A prerequisite for successful average simulation of PWM converters is the
substitution of the switching parts with an average model. Close examination of the
classical PWM topologies (Fig. I) reveals that they all share a common, topology-
independent switching module: a switched inductor. Although normally realized by
a switch and a steering diode, the switching action can be described by a toggle
switch which alternately connects one end of the inductor to two terminals b and c
(Fig. 2) (Ben- Yaakov 1989). The interaction of the switched inductor module with
the interface circuitry is described here by a generic switched inductor model
(GSIM) which includes three dependent current sources (Fig. 3). The dependent
sources of the GSIM represent the average "terminal currents of the switched
inductor assembly (Fig. 2). The behavioural description of these dependent current
sources, i.e. their magnitude as a function of other parameters of the converter, will
be a function of the operational mode of the converter. In general, the dependent
current sources will be a function of the average current of the inductor, the duty
cycle of the toggle switch and the average voltage across the terminals of the GSIM.
The proposed behavioural PWM average model includes three basic blocks
(Fig. 4): the GSIM, the inductor current generator (ICG) and the duty cycle
generator (DCG). Like the GSIM, the DCG function is algebriac, that is, it is the
mathematical expression for the duty cycle in voltage mode and current mode
converters. Since these functions are algebraic, the behavioural model of the DCG
will be algebraic.
The ICG is the subcircuit or algebraic expression that describes the average
inductor current as a function of the relevant parameters. Here, four cases have to
be distinguished (Fig. 5). In the continuous inductor current cases, the inductor
current is a dynamic function of the voltage across the GSIM terminals As will be
Simulation of PWM converters by behavioural relationships 733

shown below, this differential equation relationship can conveniently be emulated by


an equivalent circuit. For the discontinuous inductor current case, the relationship
between the inductor current and the voltage across the GSIM tenninals is algebraic
and therefore can be described by an algebraic behaviour model.

4. Behavioural modules for PWM converters


The algebriac and dynamic relationships for the basic PWM topologies have
been derived for all modes of operation and are summarized in Appendix A. The
behavioural modules represent the average response of the module as a function of
the relevant parameters. The modules which involve a differential equation, i.e. the
ICG in continuous mode, are represented by an equivalent circuit which is solved
dynamically during simulation. Algebraic modules, e.g. the DCG, are represented by
the corresponding mathematical relationships. These equations are applied by the
simulator to calculate the relevant parameters during a large signal simulation. In
the case of small signal (AC) simulation, the simulator will first linearize the
equations around the DC operating point before applying it in the (average) small
signal simulation.

(b) 1 Lr
r : d
~
:1 IL --+-:
Ic j .
:
~ + : :--
Tv~:t
m -: +1 : l ~ .R~
~ J ...f
1 1
'", ,

Basic PWM topologies: (a) buck; (b) boost; (c) buck-boost.

~
734 s. Ben-Yaakov

Figure 2. Switched inductor assembly.

4.1. ICG and GSIM-continuous mode case


The proposed topology-independent equivalent circuit is developed by consider-
ing the average voltage across the inductor. Examination of the switched inductor
assembly (Fig. 2) reveals that the average voltage VL is a function of the voltage
across the terminals {~a.b)' ~a.cJ and the time switch to the ON (TON) and OFF
(TOFFpositions:

(I)

where Ts is the period or the switching cycle, i.e.

VL= DON Jl(a.b) + DOFFJl(a.c/


(2)
Since the duty ratio signals will be produced, in the present approach, by dependent
voltage sources (see below), they are encoded into proportional voltage signals. The
definition of the dependent voltage source EL required to generate the average
inductor voltage is thus

EL=VDONJ'(..b)+VDOFFJ'(..C) (3)
The average inductor current in the continuous mode case is determined by the
average voltage impressed on the inductor. Hence, the inductor current can be
simulated in this case by the simple equivalent circuit given in Appendix A
(Ben-Yaakov 1989, Ben-Yaakov et al. 1990). The equivalent circuit includes the
inductor, its parasitic resistance and an average voltage source EL which, in general,
will be time dependent. Also included in the equivalent circuit is a voltage source of
zero voltage that is used to sense the current in the loop (Nagel 1975).

Figure 3. Generic switched inductor model (GSIM).

'" .
Simulation of PWM converters by behavioural relationships 735

a -b
--
c

=- IL' Ipk -1-

DUTY -CYCLE V(a.b)~ INDUCI'OR


GENERA TOR CURRENT
(DCG) ~a.c~ GENERA TOR
~onstants ~I (ICG)
t .

YE

Figure 4. Behavioural model of the switching section of PWM converters.

Once the inductor current is available, the dependent current sources of the
GSIM model (Fig. 3) can be readily defined by considering the way the current is
split between the terminals. Since the current of terminal a is identical to the
inductor current the dependent current source Go can be defined as

Ga=I(VsN) (4)
where I(VsN) is the current of the dummy voltage source (SN) of the equivalent
circuit (Appendix A). The current is then divided between terminals a and b (Fig. 2)
according to the fraction of time the inductor is connected to them, namely

Gb= VDoNI(VsN) (5)

G.= VDoFFI(VsN) (6)


Note that in the above equations the duty cycle is already encoded into the voltage.

4.2. ICG and GSIM-discontinuous mode case


Unlike the continuous inductor current case, the inductor current in the
discontinuous mode is an algebraic (and not a dynamic) function of the controlling
parameters. This is a consequence of the fact that the peak inductor current (I Pk)is a
linear function of the product { J'(a,b)TON}and that the terminal currents of the GSIM

Continuous Discontinuous
Inductor Inductor
CulTent Current

Voltage
Mode

Current
M(xje

Figure 5. Functional nature of the inductor current generator (ICG) for all PWM cases.

-:-L
736 s. Ben-Yaakov

ILt
Ipk ..

!1lME

Figure 6. Inductor current in the discontinuous case.

are related to the peak current by simple geometrical relationships (Fig. 6). The
GSIM dependent current sources for the discontinuous case are hence defined as:

IG.I=~IIpk.1 (7)

(8)

(9)

where (Fig. 6)

I = I J.(.,b}1 V DON 7:
pk L . (10)

4.3. DCG-voltage mode cases


For a given primary control voltage of the PWM modulator VE, the definition of
the dependent voltage source (EDoN) that generates the duty cycle signal (VDON)that
is

EDoN=KMVE (II)

..

~
Simu/ation of PWM converters by behavioura/ re/ationships

(a) (b)

Figure 8. GSIM orientation and physical current directions for: (a) buck; (b) boost.

where KM is the proportionality constant of the modulator.


In the continuous mode case DOFF is related to DoN by

DoFF= I -DoN

Consequently

EDOFF= 1- v DoN (13)

In the discontinuous mode case, the relationship between DOFF and DON is

,...;

/0

Figure 9. Benchmark circuit used to demonstrate the proposed simulation approach


(Middlebrook 1989).

~
738 s. Ben-Yaakov

0.105.0.@ GSIM

'.5 b

-:'
>Ro
@
Co
1.5Q
2
Rl
':
lK
@
g~ lOK.Q l~

A L @:L-::;=-l
l =- ~~F R2
\-/EL SN I~~ :lK

--~l<e) ~JF=~ :

@) DCG

-O

VOOF
( \
'ROOF
RYE fEOOF~

Figure 10. PSpice-compatible equivalent circuit for the converter of Fig. 9 when operated in
the continuous inductor current mode.

~
Simulation of PWM converters by behavioural relationships

Gain [dB]I Phase[Deg.]

100 lOk lM

Frequency [Hz]

Figure 12. Small signal control-to-output (vouJvJ transfer function of the power stage
shown in Fig. 9. Discontinuous inductor current case (loon load).

where L is the inductance of the inductor. Consequently

EDOFF = VDoNI "<..b)1


I "<..C)I

4.4. DCG-current mode cases


Applying the expression for the DCG for the continuous mode case (Middle-
brook 1989) the duty cycle-dependent voltage source can be defined as

YE-II(VsN)IKs
(
EDON= 7: M + O.5K sIJ':(., b) I) (16)
s c L

where YEis the primary control voltage (the output of the error amplifier), Ks is the
current loop gain and Mc is the slope of the compensating ramp. In the current
mode discontinuous inductor current case, the control voltage YE is found to be
(Fig. 7)

Output Voltage [V]


15.011

15.00.

14.991 ., , ,
° 0.2 0.4 0.6 0.8 1
Time [~J

Figure 13. Output voltage of the converter of Fig. 9 when subjected to a load change of
O.5A. Continuous inductor current case (1.5Q load).

~
740 s. Ben-Yaakov

Output Voltage [V]


15.011

15.00:

~
14.991 I
° 0.2 0.4 0.6 0.8 ]
Time [Soc]

Figure 14. Output voltage of the converter of Fig. 9 when subjected to a load pulse-change
of 0.5 A. Discontinuous inductor current case ( 100 .Q load).

EE= (MC+~ L ) ToN


Consequently, the duty cycle-dependent voltage source is defined as

YE

)
V DON=

1;( Mc+-z:-
Ksi~..b)l

5. Applying the behavioural model


The topology-independent behavioural modules developed above and summar-
ized in Appendix A can be used to develop input files for electronic circuit
simulators which include algebraic behavioural models, such as recent versions of
HSPICE and PSpice. To simulate a given topology, the model has to be placed in the
corresponding orientation and, in the discontinuous inductor current case, the
direction of the controlled current sources set according to the expected directions of
the currents (Fig. 8). Other PWM topologies can be simulated by reducing them to
one of the basic topologies (Fig. I ). For example, the forward converter is treated as
a buck converter with a scaled input voltage, etc. .
PSpice-compatible input film templates that cover all cases are given in
Appendix.B.

6. Example
The proposed simulation approach is demonstrated by considering a current
mode (CM) buck converter (Fig. 9). The power stage is the benchmark circuit
presented by Middlebrook (1989) and investigated by Kimhi and Ben-Yaakov
(1991). The circuit is reduced to a PSpice-compatible circuit by applying the
definitions of the GSIM, ICG and DCG (Fig. 3) given in Appendix A. The resulting
average equivalent circuit (Fig. 10) was translated to a PSpice-compatible file by
applying the template subcircuits given in Appendix B. Simulation results obtained
by applying the proposed behavioural models were found to be in good agreement
with previously published (Amran et a/. 1991, Ben-Yaakov and Gaaton 1992, Nagel
1975) analytical and experimental results.

~
Simu/ation of PWM converters by behavioura/ re/ationships 741

The power of the proposed direct simulation approach is exemplified by the


results of a number of key analyses. Figure II presents the small signal control-to-
output (voutVE)transfer function for a 1.5 .Q.load. When the load is changed to 100.Q.
the inductor current of the converter will be discontinuous and the equivalent circuit
has to be switched to the discontinuous case as per the relationships given in
Appendix A. The corresponding control to output (VouJVE)transfer function for the
discontinuous case is given in Fig. 12.
Closed-loop, large signal simulation is demonstrated by the results of transient
analysis when the converter is exposed to a load current change (Figs 13 and 14).
Clearly, the sluggish recovery (Fig. 14) is consistent with the narrower bandwidth
observed for the discontinuous inductor current situation (Fig. 12).

7. Discussion and conclusion


Simulation of the rather complex analogue circuits presented here could be
hampered by convergence problems. This is especially true in closed-loop simula-
tion, in which the dynamical behaviour of the system destabilizes the numerical
algorithm used by the simulator. Nonetheless, it was found that in most cases the
problem can be alleviated by following some simple procedures. The following
suggestions were found to be of great help in eliminating convergence deadlocks:

(a) Evaluate the quiescent point voltages by running a DC sweep in open-loop


configuration;
(b) apply the operating point data obtained in step (a) to define initial voltage
(by the .NODESET command) of critical nodes (e.g. output voltage, outputs
of operational amplifiers, etc.). It should be noted that the .NODESET
command imposes a voltage source of relatively low internal resistance.
Consequently, it is advisable to add output resistance to amplifiers which are
to be clamped by the .NODESET instruction, assuming of course that this
does not change the basic characteristics of the circuit;
(c) convergence problems in closed-loop simulation are aggravated by fast
devices and high-gain error amplifiers. Therefore, avoid 'ideal' devices
(diodes, transistors, etc.) of zero or very low internal capacitance, and reduce
the (open loop) gain of the error amplifier as much as possible;
(d) it should also be noted that the subcircuits given in Appendix B assume a
specific mode of operation. They do not include a toggle arrangement to
automatically switch from say, continuous inductor current mode to discon-
tinuous mode (Amran et a/. 1991). Consequently one has to make sure that
the converter remains in the assumed mode throughout a large signal
simulation. This can be conveniently done by examining the sum

{~ DON
+ VDoNI ~..b)l
I ~..c)1 } (19)

which should be smaller than one for the discontinuous mode cases and
equal to or larger than one for continuous mode cases. To help this
evaluation, the subcircuits were designed with extra ports which carry the
essential data (Appendix B). An extensive discussion and a practical solution
for implementing the automatic toggling was given by Edry (1992).
742 s. Ben-Yaakov

The major advantage of the proposed topology-independent models is the fact


that they can be defined as subcircuits and directly applied to run DC, small signal
(AC) and large signal (TRAN) analysis in open and closed-loop configurations. By a
simple modification (Ben-Yaakov and Gaaton 1992) the model can be applied to
simulate average current mode systems (Dixon 1990) and to explore other single or
multiple feedback configurations.
This paper has been confined to average simulation of processes that are well
below half the switching frequency. When the Nyquist sampling rate boundary is
approached, the effect of sampling on the dynamics of the system has to be taken
into account. This can conveniently be done by including an extra transfer function
that emulates the sampling effect. For example, in current mode, according to Tan
et al. 1993, an extra pole has to be included in the DCG. Similarly the current mode
DCG applied here can be replaced by the correction proposed by Verghese et al.
(1989) which refined the classical expression applied in this study. In fact. the
proposed behavioural approach will accommodate any refinement of the expressions
that describe the operation of PWM converters.

ACKNOWLEDGMENTS
The simulation results presented in this paper were obtained with an evaluation
version of PSpice (V.5). The author wishes to thank MicroSim Co. for making this
powerful demonstrator public domain. Special thanks are due to Daniel Edry for
helping to streamline the manuscript.

Appendix A
Behavioural relationships and equivalent circuits of PWM converters

Continuous inductor Discontinuous inductor


current current

Inductor current generator Algebraic.


(ICG) No need for an auxiliary
circuit.
Use expressions below.

Generic switched inductor


model (GSIM) sources

G. = I(VsN)

Gb= VDoNI(VsN)

Go = VDoppI(VsN)
Simulation of PWM converters by behavioural relationships 743

Duty cycle generators EDoN=KMVE EDoN=KMVE


(DCG) voltage mode
EDOFF= 1- VDON E -VDONIJ-(..b)1
DOPP-
I ~(..a) I

Duty cycle generators VE


EDoN = 7'( IIVE-II(YsN)IKs
-1-O.5Ksi "(..b)l\
(DCG) (current mode) EDON=7'f 1.1 -L~\
7;\ Mc+--r-- } 7;\ Mc+-r-- }

VDONIJ.(..b)1
E DOFF = -rY;;::;r-
EDOFF= 1- VDON

Definitions of dependent sources and generators

Notes to the Table:


I. VDON' VDopp,Ipk, J'(..b)' J'(..O)'I(VsN) are average, time-dependent variables.
2. DON, DOFF' are voltage coded (VDON' VDoFF' respectively).
3. For discontinuous inductor current case, set G., Gb, Go according to expected physical
directions per the given topology.
4. The inputs to the model are VE(output signal of error amplifier) and the voltages across
GSIM terminals a, b and c. All other variables are derived internally from the input signals
and constants: Ks=current gain, Mc=slope of compensating ramp (VS-I), KM=PWM
modulation constant (V-I), L=inductance (H), J;=switching frequency period (s).

Appendix B
p Spice-compatible templates
PWM DEMO SUB CIRCUITS .
* See Figs 7 & 8 of paper
*Please NOTE: <expression) of'VALUE' must be on one
*line. Correct if shown otherwise in this printout.
* Basic output circuit
RO OUT 0 1.5; Load resistor
CO C 0 2700U; Output capacitor
RC OUT C .012; Capacitor's ESR
* Input section
VIN RS 0 30; 30 Volt source
RS RS IN 0.105; Internal resistance of source
*Error amplifier and phase compensation network
*See Fig. 8 of main text for node labels
ERR E 0 PLUS MINUS IE4; Error Amplifier
*VERR E 0 DC .5; For open loop simulation
VREF PLUS 0 DC 7.5 AC I; Reference plus perturbation
RREF PLUS 0 IE5
CF E EE IOn
RF EE MINUS 10K
RI OUT MINUS IK
R2 MINUS 0 IK
DDIOD 0 DIOD DN4X; Steering diode

*For DISCONTINUOUS cases


*See SUBCKT "DISCONT" netlist for definition of ports
*XSIM OUT IN DIOD E D DOF PK DISCONT
744 s. Ben-Yaakov

* For CONTINUOUS cases


* See SUBCKT II CONT" netlist for definition of ports
XSIM OUT IN DIOD E D DOF CONT

*Simulating a step in load current


IOUT OUT 0 PULSE 0 500m O.lm 10u 10u .5m
.TRAN 50U IM
.AC DEC 101 IMEG
* .DC VERR 0 8 .5; For open loop simulation
*.NODESET v(out)= 15; to help convergence
* Modified diode model
.MODEL DN4X D(RS=.01 CJO=.005PF IS=I.38E-09 N=I.7
+VJ=.34V 1T=5U M=.38+BV=2V)
*Continuous conduction subcircuit
.SUBNCKT CONT A B C E D DOF
* A B C = > GSIM TERMINALS
*E = > CURRENT MODE- OUTPUT OF ERROR AMP
* VOLTAGE MODE: V(Don)
* D = > DUTY CYCLE INDICATOR (Don)
*DOF = > DUTY CYCLE INDICATOR (Doff)
* Change according to parameters of circuit
+PARAMS:LM=40U RL=.O08 TS=40U KS=0.1 MC=5E4
* INDUCTOR LM RES I/Fs CUR GAIN SL COMP
*Choose "EDD" for current mode, "ED" for voltage mode
EDOF DOF O VALUE= {V(E)*LM/ABS«V(A,C»*TS*KS+ IU)}; Doff indicator

*Voltage mode
*ED DD 0 VALUE= {V(E)}
*Current Mode
,. EDD DD 0 VALUE= {(V/E)-ABS(I(VSN»*KS)/
(TS*(MC + 0.5*KS* ABS(V(A,B»/LM» }

*inductor Current Generator (ICG)


VSN MI 0 0
RL MR MI {RL}; Inductor's resistance
LM M MR {LM}; inductor .
EM M 0 VALUE= {(VA,B)*V(D)+ V(A,C)*(I-V(D»)}
*Generic Switched Inductor Model (GSIM)
GOA A 0 VALUE={I/VSN)}
GOC 0 C VALUE={I(VSN)*(I-V(D»}
GOB 0 B VALUE={I(VSN)*V(D)}
/0 RVE E 0 IOEIO
RD D 0 IOEIO
RDOF DOF 0 IEIO
RDD DD D 10K
DLM 0 D DN4X
.ENDS
*Discontinuous conduction subcircuit
.SUBCKT DISCONT A B C E D DOF PK
* A B C = > GSIM TERMINALS

~
Simulation of PWM converters by behavioural relationships 745
* E = > VOLTAGE MODE: Don
* D = > DUTY CYCLE INDICATOR (Don)
* DOF = > DUTY CYCLE INDICATOR (Doff)
* PK = > INDUCTOR PEAK CURR INDICATOR
* Change according to parameters of circuit
+PARAMS:LM=40U RL=.OO8 TS=40U KS=O.l MC=5E4
* INDUCTOR LM RES l/Fs CUR GAIN SL COMP
* Current Mode
EDD DD O VALUE={V(E)/(TS*(MC+KS*ABS(V(A,B»/LM»}

*Voltage mode
*ED DD O VALUE= {V(E)
EPK PK O VALUE= {V(A,B)*V(D)*TS/LM}
EDOF DOF O VALUE= {ABS(V(A,B)*V(D)/V(A,C»}
*Generic Switched Inductor Model (GSIM)
GA A O VALUE= {(V(A,B)*V(D)*TS/LM)*«V(D)+ V(DOF»)/2}
GC O C VALUE= {(V(A,B)*V(D)*TS/LM)*V(DOF)/2}
GB O B VALUE= {(V(A,B)*V(D)*TS/LM)*V(D)/2}

RPK PK O I
RYE E O IOEIO
RD D O IOEIO
RDOF DOF O IEIO
RDD DD D IOK
DLM O D DN4X
.ENDS
.PROBE
.END

REFERENCES
AMRAN. Y.. HULIEHEL, F., and BEN-YAAKGV, S.. 1991. A unified SPICE compatible average
model of PWM converters. IEEE Transactions on Power Electronics, 6, 585-594.
BEN-Y AAKOV, S.. 1989, SPICE simulation of PWM DC-DC convertor systems: voltage
feedback. continuous inductor conduction mode. IEE Electronics Letters, 16,
1061-1063.
BEN-YAAKOV,S., EDRY. D.. AMRAN. U.. and SmMONY. 0., 1990, SPICE simulation ofquasi-
resonant zero-current-switching DC-DC convertors. IEE Electronics Letters. 26,

7 847-849.
BEN-YAAKOV, S.. and GAATON, Z., 1992. Unified SPICE compatible model of current
feedback in switch mode convertors. IEE Electronics Letters, 28, 1356-1357.
DIXON. L. H., 1990, Average current-mode control of switching power supplies. 5-1-5-14.
Unitrode Switching Regulated Power Supply Design Seminar Manual (Merrimack:
Unitrode Corp.).
EDRY. D., 1992, Capacitive-Ioaded push-pull parallel-resonant converters. M.Sc. thesis. Ben-
Gurion University of the Negev, Israel (in Hebrew).
KlMHI, D.. and BEN-YAAKOV. S.. 1991, A SPICE model for current mode PWM converters
operating under continuous inductor current conditions. IEEE Transactions on Power
Electronics, 6, 281-286.
MIDDLEBROOK.R. D., 1989, Modeling current programmed buck and boost converters. IEEE
Transactions on Power Electronics. 4, 36-52.

-1

~
'0;- --7-

746

..-"
...

-.-

'~.,

You might also like