Static- and dynamic-circuit models of PWM
buck-derived DC-DC convertors
D. Czarkowski, MSc
M.K. Kazimierczuk, MSc, PhD, DSc
Indexing terms: DC-DC convertors, Pulse-width modulation, Buck-derived convertors
Abstract: A systematic derivation of DC-circuit
and small-signal-circuit models and characteristics
for pulse-width-modulated buck-derived conver-
tors operating in the continuous conduction mode
is presented. Starting from voltage- and current-
transfer functions of an idealised switching part of
the convertor and using the linearisation tech-
nique, pure linear-circuit models of lossless con-
vertors are derived. To obtain the models of lossy
convertors, the principle of energy conservation is
used to derive an equivalent averaged resistance of
all parasitic resistances and an equivalent aver-
aged voltage of offset voltage sources of switches.
The advantage of the models is that they can be
directly utilised in standard circuit simulators. The
models are especially convenient in analysing
complicated convertor topologies and for includ-
ing parasitic components. The proposed models
lead to the same characteristics as those obtained
by means of the state-space-averaging method. As
an example, a forward convertor is analysed,
taking into account all parasitic resistances and
threshold voltages of diodes. A complete set of
small-signal characteristics is given for this con-
vertor. Models and expressions for DC and small-
signal characteristics that account for parasitic
resistances are also given for other multiswitch
and transformer buck-derived convertors such as
push-pull, half-bridge, and full-bridge circuits.
1 Introduction
The state-space-averaging method introduced by Mid-
dlebrook and Cuk [l] is widely used to derive expres-
sions for small-signal characteristics of pulse-width-
modulated (PWM) convertors [1-71. This general
method has led to understanding of the dynamic per-
formance of PWM convertors. However, the state-space
method is sometimes tedious, especially when the conver-
tor equivalent circuit contains a large number of ele-
ments. The purpose of this paper is to present a simple,
systematic method of obtaining DC-circuit and small-
signal-circuit models of PWM buck-derived convertors
operating in the continuous-conduction mode (CCM).
The starting point of derivation of the models is the
Paper 8 9 3 4 6 (E10, P6), first received 11th December 1991 and in
revised form 29th April 1992
The authors are with the Department of Electrical Engineering, Wright
State University, Dayton, OH 45435, USA
IEE PROCEEDINGS-G, Vol. 139, N O .6, DECEMBER 1992
voltage- and current-transfer functions of the switching
part of the convertor which are perturbed and linearised
about the steady-state operating point. The resulting
relationships between voltages and currents are used to
develop a linear-circuit model of the convertor-switching
part comprised of dependent sources. To take into
account parasitic resistors and offset-voltage sources of
switches (which are of great importance in low-voltage
high-current applications), the equivalent averaged resist-
ance (EAR) of switched resistors and the equivalent aver-
aged voltage (EAV) of switched voltage sources are
determined using the principle of energy conservation.
The inductances and capacitances are not averaged
because the inductor currents and the capacitor voltages
are continuous functions. The models can be used to find
the convertor-performance static and dynamic character-
istics either analytically or numerically. They offer three
important advantages. First, the switching part of the
convertor may consist of any number of switches, e.g.
transistors and diodes. Secondly, the linear part of the
convertor may consist of any number of elements, e.g.
parasitic components and/or additional input and output
filters. Third, circuit-simulation programs can be used for
analysing the convertor models. Other interesting
approaches to obtaining circuit-convertor models that
are compatible with general-purpose CAD circuit-
simulation packages such as SPICE have been presented
recently by Griffin [8] and Kimhi and Ben-Yaakov [SI.
2 Analysis of forward convertor
2.1 Assumptions
The analysis of the forward convertor of Fig. la is carried
out under the following assumptions:
(i) The transistor output capacitance and diode capa-
citances are neglected; therefore, switching losses are
assumed to be zero.
(ii) The transistor O N resistance rDs is linear and the
transistor OFF resistance is infinite.
(iv) The diodes in the O N state are modelled by a
linear battery VF and a linear forward resistance RF and
in the OFF state by an infinite resistance.
(iv) The transformer-leakage inductances, the mag-
netising inductance, the stray capacitances and the
magnetic-core parallel resistance are neglected.
(v) The diode D, is ideal.
This work was supported by the US National
Science Foundation under Grant ECS-8922695.
669
 (vi) Passive components are linear, time-invariant and where n is the transformer turns ratio, V, is the DC com-
frequency-independent. ponent of the input voltage equal to DC component of
(vii) The inductor current is constant during the entire the ideal switch, and VT is the DC component of the
switching period. voltage across the diode D, if the diodes and the trans-
former are ideal. Assuming that all the parasitic resist-
ances are zero and V, = 0, M v becomes the DC-voltage
transfer function of the entire lossless convertor. Since the
switch S and the transformer in Fig. IC are assumed to be
ideal, the following is true: G I , = V T I L .Therefore, the
DC-current transfer function is

where I, is the DC component of the current through the

p DS T1 inductor L (equal to the DC output current Io) and I, is
the DC component of the input current.
The voltages, current and duty cycle can be decom-
posed into DC and AC components, resulting in the fol-
lowing expressions :
U* = V, + ui (3)
i, = I, + ii (4)
UT = VT + 0, (5)
dToT = D +d (6)
i, = I, + i,
U I I ,
U0 = Vo + U,
C
From eqns. 1 and 2,
Fig. 1 P W M forward convertor
a Circuit
b Equivalent circuit with parasitic components (9)
c Equivalent circuit with rDs and r T , reflected to the secondary side of the trans.
former

(viii) The output impedance of the input-voltage

source is zero for both DC and AC components. Substitution of eqns. 3, 5, and 6 into eqn. 9 yields
2.2 Circuit models of forward convertor for DC and
small-signal operation
A circuit of the forward convertor is shown in Fig. la. It
consists of a MOSFET as a controllable switch S (a BJT, D
an IGBT, or an MCT can also be used), an isolation =-
n
V, +D-n v i +%-
n
d
d +-
n
(1 1)
transformer with a demagnetising winding, diodes D, ,
D,, and D,, an inductor L, a filter capacitor C, and a Likewise, substituting eqns. 4, 7 and 6 into eqn. 10 one
load resistance R. The switch is turned on and off at the obtains
switching frequency f, = 1/T with the ON duty ratio
D = t J T , where to, is the interval when the switch is
ON. Fig. l b depicts an equivalent circuit of the conver- 1, + ii = (:- + -:) ( I , + il)
tor, where rDs is the transistor O N resistance, rT1 is the
D D I , d
winding resistance of the primary of the transformer, rT2 =-I, +- i, + - d + - i,
is the winding resistance of the secondary of the trans- n n n n
former, V, is the threshold voltage of the diode, R, is the
forward resistance of the diode, r, is the equivalent series Now an EAR of the switched components is introduced
resistance (ESR) of the inductor L, and rc is the ESR of using the principle of energy conservation. The switch S
the filter capacitor. It is assumed that the demagnetising and the diode D, are ON for the interval 0 t < DT. -=
winding of the transformer and the diode D, have no The switch current in the circuit of Fig. Ib is approx-
influence on a properly designed forward convertor imately constant and given by
because of their small current. In the equivalent circuit of
Fig, IC, rDs and rT1 are reflected from the primary to the is=L forO<t<DT (13)
secondary side of the transformer.
Steady-state analysis of the circuit of Fig. IC results in
the DC-voltage transfer function. resulting in the RMS value of the switch current

<-In IEE PROCEEDINGS-G, Vol. 139, N o . 6, D E C E M B E R I992

The power dissipated in the resistors rDs and r T 1 on the Hence, one obtains the efficiency of the convertor
primary side of the transformer

1
is equal to the power dissipated in the equivalent aver-
aged resistance r,, of rDs and r T 1 on the secondary side 1 + (r/R)+ (VF/Vo) (25)
of the transformer. Hence Note that the power loss in the ESR of the filter capa-
D(rDS + rT1) citor and switching losses are neglected. Eqns. 24 and 25
rsav = are illustrated in Figs. 3 and 4.
n2
Note that the output resistance of the input voltage
source V, can be added in series with rDs and r T 1 .Similar +
considerations reveal that
IT2.u = DrT2 (17)
The EAR of the forward resistances R,, and R,, for the
diodes D, and D, is obtained analogously:
RFau = D R F l + ( l - D ) R F 2 (18)
a
If the forward resistances of the diodes are identical, the
duty ratio D does not affect their averaged value RFaV
"F 1
which is then equal to R F . Thus, the power dissipated in
the equivalent averaged resistances of the diodes and the
secondary of the transformer is given by
pDT2 = + DrT2) (19)
The EAR of r D S ,r T 1 ,rT2 and RF is obtained from eqns.
16-18: 4-
b
rau = Isau + I T 2 a u + RFau
r L

resulting in the total EAR in series with the inductor L

The EAV of the forward batteries VF, and VF2 of the

diodes D , and D , is C

= + (l - D)VF2 (22) Fig. 2 Model ofthe forward convertor for C C M

a Averaged model for DC and AC operation
Assuming that the voltages of the batteries of the diodes b Model for DC operation
are identical, VFavis equal to VF.Then the power dissip- c Model for AC operation
ated in the equivalent averaged voltage source is
pVF = VF I L
In the above derivations as well as in the state-space-
averaging method, the inductor current is assumed to be
Let us assume that the magnitudes of the AC com- constant, resulting in zero current through the filter capa-
ponents in 3-8 are much lower than the D C components. citor and zero conduction loss in the ESR. In reality, the
Hence, the products (d/n)uiand (d/n)i, in (eqns. 1 1 and 12) current through the filter capacitor is approximately
can be neglected. Using eqns. 16, 17, 18 and 21, a linear- equal to the AC component of the inductor current given
circuit model of the convertor for DC and small-signal by
operation is obtained and shown in Fig. 2a. Since this
model is linear, the principle of superposition can be used Ai, Ai,
for 0 < t < DT
to derive the DC model shown in Fig. 2b and the small-
signal model shown in Fig. 2c.
ji2-T
Ai,
It is apparent from Fig. 2c that uo = 0 if (VI/n)d = ' i ~
(1 - D)T
(t - DT) +-2
for DT < t <T
- (D/n)u,. Accordingly, the control law is d = - (D/V,)u,.
From Fig. 26, the DC-voltage transfer function of the
lossy convertor is found to be expressed as (26)

MvD,~--
- (fl- _ - z) +~

1
1
(r/R)
where Ai, is the peak-to-peak ripple current of the induc-
tor. The RMS value of the capacitor current is

1 E E P R O C E E D I N G S - G . Vol. 139. No. 6. DECEMBER 1992 671

leading to the power-conduction loss in the filter capa- The switching losses in the MOSFET can be approx-
citor imated by

where COis the transistor-output capacitance (assumed to

be linear) and Po is the DC output power. The power
05 loss in the inductor is
r P,, = rL12 (30)
Using eqns. 15, 19, 23, 28, 28, 29 and 30 gives the total
power loss
P D = P S T , + P D T z + PvF + Prc + P s w + Pr, (31)
Hence, one arrives at the efficiency of the convertor
taking also into account the power-conduction loss in the
I I I I
filter capacitor and the switching losses in the MOSFET
neglected in eqn. 25 :
a
P, 1
O
OL-
5r qt=--
Po +PD
-
r
1+-+-+
VF R r d l - 0) f,CoR
00#0 R Vo 12(f,L) +

0 0

The models of Fig. 2 exhibit an important advantage.

They consist of standard circuit elements and can there-
fore be analysed using circuit-simulation programs.
Many previous models of the convertor contain a DC
transformer [ l , 2, 4, 51 which is not a circuit component
and therefore cannot be handled by circuit-oriented
D simulators.
b
The above method of obtaining the linear models of
Fig. 3 DC-voltage transfer function of the forward convertor against D
the forward convertor is particularly convenient for more
n = 2, V, = 0.7 V , rc = 50 mR rL = 40 mR, I D S = 0.5 R rT1 = 50 mR, rT2 = complex convertor topologies, e.g. for more sophisitic-
25 I&, R , = 25 mR
a For a constant value of V, = 40 V ated output filters and/or more complicated loads and/or
b For a constant value of V, = 5 V other parasitic parameters (for instance, the equivalent
series inductance of the capacitor).

2.3 Small-signal characteristics of forward convertor

The small-signal model of Fig. 2c can be used to describe
the convertor performance for frequencies f up to about
one-half the switching frequency f,. Referring to that
model, the control-to-output (or duty ratio-to-output)
transfer function in the s-domain is expressed as

-
- v, Rrc
nL(R rc) +
s +(Wrc)
X

s2 + s
+ Rr + rcr) + L + R + r
C(Rr,
LC(R + rc) LC(R + IC)

-
V, Rw: s + 0,
(33)
no,(R + r) s2 + 25, 0,s + 0,
where
1
0 02 04 06 08 1 w, = - (34)
CrC
D
b C(Rrc + Rr + rcr) + L
5, = 2J{LC(R + rcXR + I)> (35)
Fig. 4 Efficiency of the forward convertor against D
n = 2 , VF=O.7V, rC=50mQ, rL=40mR, r D , = 0 . 5 f k r T , = 5 0 m Q , r T 2 =
25 d, R , = 25 mR
a For a constant value of V, = 40 V
b For a constant value of V, = 5 V
-.=J( LC(RR ++r ) IC)

<,I*
For 5, 3 1, the frequencies of the real poles are The open-loop output impedance of the loaded con-
fpl7 fp2 = f r i t , * JCtf - 1)) (37)
vertor (i.e., including the load resistance R) is

wheref, = 0,/(271). Z,(S) E5 - -

The input-to-output (or line-to-output) voltage-
transfer function, also called an open-loop dynamic line
regulation or an audio susceptibility (which describes the
input-output noise transmission), is
r,rC+L
Mu(s)= -
I d(s) = O
X

s2 + s
sL+s--+-
r,LC
+
r,LC
C(Rr, Rr + rcr) + L R +r
+
-
DRY, LC(R rc) + LC(R r,) +
nL(R I,) + - Rrwf (s + w,Xs + w I )
-

X
s + WCrc) (R + r ) o , wI s2 + 25, ors + wf (45)

s2
+ Rr + rcr) + L + R + r
+ s C(Rr, LC(R The open-loop dynamic load regulation is defined as
+ rc) LC(R + IC)

- DRW; s + w,
- (38)
nw,(R + r) s2 + 25, ors + wf Setting s = 0, one obtains the low-frequency asymptotes
of the above quantities:
The open-loop input impedance is
v ,-
TP(O)= - R
(47)
n R+r
D R
s2 + s
C(Rr, + Rr + r c r ) + L + R +r Mu(0)= - -
nR+r
-_
-
nL LC(R + I,) +
LC(R rc)
n(R + I)
0 2 1 Zi(0)= ~
(49)
D2

(39) Rr
Z,(O) = -
R+r
where
Note that all the above expressions are dependent upon
r. Eqn. 48 is different from eqn. 24 because V, has been
assumed to be a short circuit for a small-signal analysis.
As s -+ 00, one obtains the high-frequency asymptotes
The open-loop output impedance of the unloaded con- of the small-signal characteristics :
vertor (i.e. excluding the load resistance R ) is
Tp(a3)= 0 (52)

ZP(S) --
it(s) I
d(s) = 0,u d s ) = 0 , and R = 00

r,rC+ L
M"(a3) = 0
Zi(a3) = Go
(53)
(54)

s2+s-+-
I Zp(m)= IC (55)
r,LC r,LC
= r' L-
r,+r 1
s2+s-
L
+- LC
3 Plots of small-signal characteristics of forward
convertor

A forward convertor of Fig. la was designed to meet the

where following specifications: V, = 40 to 60 V, V, = 5 V, I , =
2.5 to 20 A, f, = 200 kHz, and V,/V, < 1%, where V, is
I the peak-to-peak output ripple voltage. A design pro-
WI =-
L cedure based on Reference 5 leads to the circuit param-
eters: R = 0.25 to 2 R, n = 2, D = 0.2 to 0.5, L = 5 pH,
C = 1 mF, I , = 50 mR, rL = 40 mR, rDS= 0.5 R, rT1 =
(43) 50 mR, rT2 = 25 m a , R , = 25 ma, and V, = 0.7 V. The
model parameter r calculated from eqn. 21 for D = 0.5 is
0.15 R. The aim of this Section is to show the influence of
parasitic resistances r and I , , and the load resistance
R on control-to-output transfer function Tp, input-to-
If r = I , = J ( L / C ) , then Zp(s)= r. In this case, 2, is inde- output voltage-transfer function M , , open-loop input
pendent of frequency. impedance Z i and open-loop output impedance of the
 loaded convertor Z , . Although the characteristics are 1.06 kHz, and& = 3.18, 5.37 and 7.16 kHz, and the fre-
valid up to one-half of the switching frequency, the plots quency of the zero wasfi = 3.18 kHz.
are drawn to 1 MHz so that they may be useful when Characteristics of T, are plotted in Fig. 7 for V, =
the poles and zeros are shifted to lower frequencies. The 60V, rc=0.05R, r = 0 . 1 5 R and R=0.25,0.5 and 2 0 .
following circuit parameters are held constant in all cal- The effect of R on Tpwas almost negligible. The values of
culations: L = 5 pH, C = l m F and n = 2. Plots were
calculated for fixed values of two of these resistances and
various values of the third one. The fixed values were:
R = 0.25 R, r = 0.15 R and rc = 0.05 R. The minimum
value of R results in the highest sensitivity of the charac-
teristics to variations in r and IC. The value of rc given
above is quite typical for electrolytic capacitors. The
ranges used were: from 0.25 R to 2 Q for R, from 0.05 R
to 0.2 R for r and from 0.01 R to 0.1 R for rc .
Fig. 5 shows the magnitude and the phase of the
control-to-output transfer function T, against frequency -301 I , , \
a

a 01 1 10 100 1000
f,kHz
b
Fig. 6 Control-to-output transferfunction T, 1 T, 1 exp (j4)
V, = 60 V, n = 2, L = 5pH, C = 1 mF,R = 0.25 R, rc = 0.05R
__ r = 0.05 R
_ ~ r =_ 0.1 R~
-~ r = 0.15 R
r = 0.2 R
a I T, 1 against f
P - b $ againstf
01 1 10 100 1000
f,kHz
b
Fig. 5 Control-to-output transfer function T, = I T, I exp (j4)
V, = 6 0 V , n = 2,L = 5fiH, C = 1 mF,R =0.25R,r =0.15R
__ rc = 0.01 R
_ ~ rc ~= 0.03_ R
_ _ rc = 0.05 R
rc = 0.1 R
a 1 T, I againstf
b q5 againstf

for V, = 60 V, R = 0.25 Q, r = 0.15 Q and rc = 0.01, 0.03,

0.05 and 0.1 Q. The maximum input voltage V, = 60 V
was selected because it results in the highest crossover
frequency. As r, was increased, the crossover frequency
increased from about 10 to 70 kHz and the minimum
phase was below -90" only for small values of r c . For
higher values of r c , the circuit behaved like a 1st-order
system. The values of the damping factor were 5, = 1.02,
1.15, 1.28 and 1.56. The corner frequencies weref, = 2.79,
2.69, 2.60, and 2.41 kHz. The frequencies of the poles
weref,, = 2.29, 1.56, 1.26, and 0.87 kHz, andf,, = 3.40,
4.63, 5.37, and 6.63 kHz, and the frequencies of the zero 01 1 10 100 1000
weref, = 15.92, 5.30, 3.18, and 1.59 kHz, respectively.
Fig. 6 depicts plots of T, for V, = 60 V, R = 0.25 R, f, kHz
rc = 0.05 R and r = 0.05, 0.1, 0.15 and 0.2 R. The cross- b
over frequency was independent of r and equal to Fig. 7 Control-to-output transfer function T, = 1 T, 1 exp (j4)
40 kHz. The low-frequency values of 1 T,I decreased with V, = 6 0 V , n = 2 , L = 5 pH, C = 1 mF, rc = 0.05R. r = 0.15 R
increasing r. The values of the damping factor were 5, = ~ R = 0.25 R
_ _ ~ _R = 0.5 R
0.77, 1.04, 1.28 and 1.49. The corner frequencies were _ _ R = 2.0 R
f, = 2.25, 2.43, 2.60 and 2.76 kHz. The frequencies of a I T, I againstf
the poles in the case of 5, > 1 weref,, = 1.86, 1.27 and b $ againstf

lfiv D R O P F F n l N G 9 - G Vnl. 139. N o . 6. D E C E M B E R 1992

the damping factor were 5, = 1.28, 1.33 and 1.39. The
corner frequencies were f, = 2.60, 2.45 and 2.31 kHz. The
frequencies of the poles were fpl = 1.26, 1.11 and
0.98 kHz, and fp2 = 5.37, 5.40 and 5.43 kHz, and the fre-
quency of the zero was f, = 3.18 kHz, equal to that for
the case shown in Fig. 6.
The phase characteristics and the shape of the magni-
tude characteristics of the input-to-output voltage-
quencies and appreciably increased with R at low fre-
quencies. The frequencies of the poles for the case shown
in Fig. 9 weref,, = 0.61, 0.57, 0.53 and 0.45 kHz, and the
frequencies of the zeros were the same as the frequencies
of the poles in the case shown in Fig. 5.
Characteristics of the open-loop output impedance of
the loaded convertor are depicted in Figs. 12-14. IZ,J
increased with increasing r and R at low frequencies and

r
transfer function M u are the same as for T,. To obtain
the exact magnitude characteristic of M , from the magni-
tude characteristic of T,, one should shift it down by a
value of 20 log ( q / D ) .An example plot of the magnitude
of the input-to-output transfer function M O against fre-
quency for V, = 60 V, D = 0.35, R = 0.25 R, r = 0.15 R
and rc = 0.01,0.03,0.05and 0.1 Q is given in Fig. 8.
Figs. 9-11 show plots of the open-loop input imped-
ance Ziagainst frequency. As shown, IZiI was almost
independent of IC, slightly increased with r at low fre-
c
-

N_
.- IooI
10

I I I I I I
a

-45 t
-901 I 1 I
.I
-901
0001
I
001
I
01
I
1
I
10
1
100
I
loo0
01 1 10 100 1000 f,kHz
f,kHz b
Fig. 8 Amplitude of input-to-output transfer function I M , I against f Fig. 10 Open-loop input impedance Zi= I Zi I exp (jdZi)
V, = 6OV, n = 2, L = 5pH, C = 1 mF,R =0.25R,r =0.15R D = 0.35,n = 2,L = 5pH.C = 1 mF,R =0.25R,rc = 0.05R
rc = 0.01 R ~ r = 0.05 R
_ _ _ ~r = 0.1 R
~

~~~- rc = 0.03 R
_- rc = 0.05 R ~~
r = 0.15 R
rc = 0.1 R r = 0.2 R
a I Z , 1 againstf
b $zr againstf

100 i /
/
11 I 1 I I 1 I
a
11 I I I I 1 I

45

a -?
-90 L
0001
I
001
I
01
I
1
I
10
I
100
A
1000
-901 I 1 I I I I
f,kHz 0001 001 01 1 10 100 1000
b f, kHz
Fig. 9 Open-loop input impedance Z , = I Z , I exp (j4z,) b
Fig. 11 Open-loop input impedance Z , = I Z , I exp (j4z,)
D = 0.35, n = 2, L = 5 pH, C = 1 mF, R = 0.25 R r = 0.15 R
~ rc = 0.01 R D =0.35,n = 2,L = 5 pH, C = 1 mF,rc = 0.050,r =0.15R
~-~~ rc = 0.03 R ~ R = 0.25 R
-~ rc = 0.05 R
- - _ ~ R = 0.5 R
~- R = 2.0 R
rc = 0.1 R
a I Z , I againstf a I Z , I againstf
b &, againstf b &, againstf
_------ ---... -,. - ., I ,?n ., , nrr.rnrn ,on1

increased with increasing rc at high frequencies. The fre-
quencies of zerosfi and the frequencies of the polef, were
the same as those shown in Figs. 5-7, respectively. The
frequencies of the zero fi were: 4.77 kHz for the case
,
The ESL affects the small-signal characteristics in the .
I
high-frequency range, especially the control-to-output
transfer function and the open-loop output impedance of
the loaded convertor. Figs. 15 and 16 show magnitudes

I I 1 I I I

1
0001 001 01 1 10 100 1000 0001 001 01 1 10 100 1000
f,kHz
f,kHz
b
b
Fig. 12 Open-loop output impedance of the loaded convertor Z , = Fig. 13 Open-loop output impedance of the loaded convertor Z , =
I z,I exp cidz.) I Z , I exp (idz.)
L = S f l H , C = ImF,R=0.25Qr=O.I5Q L = 5 pH,C = 1 mF, R = 0.25 0,rC = 0.05 Q
__ rc=O.O1 R rc = 0.1 Q ~ r=0.05R r = 0.2 n
-~~~ rc = 0.03 R U IZ,l againstf r = 0.1 Q a 1 Z , I against f
-~ rc = 0.05 Q b bzoagainstf ~- r = 0.15 R b +zo against f

shown in Fig. 12 and Fig. 14, and 1.59, 3.18, 4.77 and and phases of and 2, against frequency for V, = 60 V,
6.36 kHz for the case shown in Fig. 13. D = 0.35, R = 0.25 R, rc = 0.05 R, r = 0.15 Cl, and L , =
1,10,50 and 100 nH.

4 Effect of ESL of filter capacitor on small-signal

characteristics 5 Models of other buck-derived convertors

equivalent series inductance (ESL) L , . Using this equiva- buck convertor shown in Fig. 17 can be derived from the
lent circuit of the filter capicitor in the model of Fig. 2c, model of the forward convertor of Fig. 2 by settng n = 1,
small-signal characteristics of the forward convertor have r T , = rr2 = R F , = 0, and RF2 = RF and replacing V, by
been derived and are

DR s2 + S@C/LC)+ W C C )
Mu(s)= -
( R + r)L, R +r
+ ( R + rc)L + S Rr, + Rr + r c r + ( L / C )+-LL,
s3 + s2 LLC
c
LLC

s3
+ ( R + r,)L
+ s2 ( R + r)L,LLC + S
Rr, + Rr + r c r + ( L / C )+-R + r
nL LLC LL, c
(59)
R+r, 1
s2+s- +-
Lc LCC
rL,+ r,L rr C + L r
s3 + s2 + S L + -
LLC LL,C LL,C
=R (60)
uds) = 0
s3 + s2
( R + r)L, + ( R + r,)L
+ S
Rr, + Rr rcr + ( L / C ) R r+ +- +
LLC LLC LL, c
I E E PROCEEDINGS-G, Vol. 139, N O . 6, D E C E M B E R 1992
(1 - D)V,. Hence, eqns. 16, 18, 20 and 21 become [lo] The DC-voltage transfer function and the efficiency
respectively, are
rsau =

RF,,
D~DS

= (1 - D)R,2 = (1 - D)R,
(61)

(62)
VF(1 - D)
V, I-R
R+r

I,=
, Is,,+ RF,, = DrDS + (1 - D)R, (63)
r = r,, + rL = Dr,, + (1 - D)R, + rL (64)
The small-signal characteristics given by eqns. 33-46
remain the same. The only difference is the value of r.

'. \ 5.2 P u s h p d I convertor

For the push-pull convertor depicted in Fig. 18, the
models from Fig. 2 are still valid [l 13. Assuming identical

004 0 22
a
r /

0021 1 I 1 I
a

0001 001 01 1 lo 100 1000 6or

f,kHz
b

Fig. 14 Open-loop output impedance of the loaded convertor Z ,

I z,I exp ci4Z.l
L = 5 pH, C = 1 mF, r, = 0.05 R, r = 0.15 R
~ R = 0.25 R a I Z , , againstf
_ _ R ~= 0.5 ~R b $zo againstf
~- R = 2.0 R

f,kHz
b
Fig. 16 Open-loop output impedance of the loaded convertor Z , =
I z, I exp ( j 4 Z J
VI = 6OV, n = 2, L = 5 pH, C = 1 mF,R = 0 . 2 5 R , r C = 0.05R,r = 0.15R
__ L , = l n H
L, = 10 nH
~~~~

L, = 50 nH
~~

L, = 100 nH
a I Z , I againstf
b $ - Z , againstf
I I I

a
0-

g:
v
V1
8
-60-

Fig. 17 Buck convertor

01 1 10 100 1000
f,kHz switches, identical diodes and a symmetrical transformer,
b one should only substitute 2 0 instead of D and calculate
Fig. 15 Control-to-output transferfunction T, = I T, I exp ( j 4 ) r from
VI = 6OV, n L = 5pH, C = 1 mF, R =0.25R,rC= 0.05R,r =0.15R
= 2,
___ L , = l n H L, = lOOnH
_ _ L, ~= IO ~nH a 1 T, I againstf
~- L, = 50 nH b $ againstf

I F F P R n r F F n l N C P - C V n l 1 9 9 N n fi n F r F M R F R 1007 677
The DC-voltage transfer function and the efficiency are Also, D must be replaced by 2 0 in expressions for DC
and small-signal characteristics. The DC-voltage transfer
function and the efficiency are the same as for the push-
pull convertor and are given by eqns. 68 and 69, respec-
tively.

5.3 Half-bridge convertor
The models for the half-bridge convertor shown in Fig.
19 are identical to those for the forward convertor.
6 Conclusions
The principle of energy conservation has been used to
derive DC and small-signal linear models of PWM buck-
derived convertors operating in CCM. As an example of
the application of the models, a detailed analysis of the
forward convertor has been presented. The obtained
small-signal characteristics are the same as those derived
using the state-space averaging method. Experimental
results confirming these characteristics can be found in
Reference 1 and are not duplicated in this paper. The
results have been extended to other multiswitch and
transformer buck-derived convertors such as push-pull,
half bridge, and full-bridge. All buck-derived convertors

ri: -
have the same models, but the expressions for the
equivalent average resistance r are somewhat different for
different convertors. The plots of open-loop small-signal
characteristics are useful in designing controllers. It is
shown that the ESR of the filter capacitor makes the
control-to-output transfer function of buck-derived con-
vertors similar to that of a 1st-order system. An impor-
tant advantage of the new models is that they can be
used in standard circuit-simulation programs such as
Fig. 18 Push-pull Convertor
SPICE because they do not contain noncircuit com-
ponents (e.g. DC transformers). The convertors can be
operated in CCM for any load if diodes are replaced with
controllable bidirectional switches such as MOSFETs,
which results also in improved dynamic performance
because of a smaller inductor and a bidirectional power
flow. In this case, the proposed models are still valid with
appropriate modifications in expressions for EAR and
EAV. The presented method may be used for modelling
and analysing other DC-DC convertors and DC-AC
invertors.
Fig. 19 Half-bridge convertor
7 References
1 MIDDLEBROOK, R.D., and CUK, S.: Advances in switched-
mode power conversion (TESLAco, Pasadena, CA, 1981), vols 1
and 2
2 SEVERNS, R.P., and BLOOM, G.: Modern DC-to-DC switch-
mode power converter circuits (Van Nostrand, New York, 1985),
pp. 30-42 and 130-135
3 LEE, Y.S.: A systematic and unified approach to modelling
switches in switch-mode power supplies, IEEE Trans., 1985, IE-32,

-
(4),pp. 445-448
4 MITCHELL, D.M.: Switching regulator analysis (McGraw-Hill,
New York, 1988), pp. 74-76
5 MOHAN, N., UNDELAND, T M., and ROBBINS, W.P.: Power
electronics, converters, applications and design (Wiley, New York,
Fig. 20 Full-bridge convertor 1989), pp. 229-237
6 KISLOVSKI, A., REDL, R., and SOKAL, N.O.: Analysis of
swtiching-mode DC/DC converters (Van Nostrand, New York,
However, r is the same as for the push-pull convertor 1991), pp. 40-48
and is given by eqn. 67. The DC-voltage transfer function 7 VORPERIAN, V.: Simplified analysis of PWM converters using
and the efficiency are given by eqns. 24 and 25, respec- the model of the PWM switch, Part I: Continuous conduction
tively. mode, IEEE Trans., 1990, AES-26, (3), pp. 497-505
8 GRIFFIN, R.E.: Unified converter models for continuous and dis-
continuous conduction mode. IEEE Power Electronics Specialists
Conference Record, Milwaukee, WI, USA, 26-29 June 1989, pp.
5.4 Full-bridge convertor 853-860
The full-bridge convertor depicted in Fig. 20 has the 9 KIMHI, D., and BEN-YAAKOV, S.: A SPICE model for current
same models as the forward convertor. The EAR in series PWM converters operating under continuous inductor current con-
with the inductor is found to be ditions, IEEE Trans., 1991, PE-6, (2), pp. 281-286
10 CZARKOWSKI, D., and KAZIMIERCZUK, M.K.: Circuit
+ IT1 models of PWM DC-DC converters. Proceedings of the IEEE
r =20 2rDS
n2
+ (+ + DXRF + r T 2 )+ rL National Aerospace and Electronic Systems Conference (NAECON
92), Dayton, OH, USA, 18-22 May 1992, pp. 407-413

678 IEE PROCEEDINGS-G, Vol. 139, N o . 6, D E C E M B E R 1992

11 CZARKOWSKI, D., and KAZIMIERCZUK, M.K.: A new and If the condition described by eqn. 79 is satisfied, the
systematic method of modelling PWM D C - D C converters. Pro- peak-to-peak ripple voltage of the buck and forward con-
ceedings of the IEEE Conference on Systems Engineering, Kobe,
Japan, 17-19 September 1992 vertors is

8 Appendix: Ripple voltage in buck-derived

convertors

Consider the buck or forward convertor. The current

through the capacitor is approximately equal to the AC For push-pull, half-bridge, and full-bridge convertors,
component of the inductor current. For the interval
0 < t < DT when the switch is ON and the diode is
OFF, the capacitor current is

Ai,t - -
iAt) = - Ai,
where D,, < 0.5. If the condition described by eqn. 81 is
DT 2 met, the peak-to-peak ripple voltage V, of these conver-
resulting in the AC component of the voltage across the tors is given by
ESR

uEsR(t)= r, i, = rc
Ai, t
(E - $)
and the AC component of the voltage across the filter Waveforms of uESR, uc and U , are depicted in Fig. 21 for
capacitance three values of the filter capacitances. In Fig. Zla, the

where Ai, is the peak-to-peak ripple current of the induc-

tor. Hence, the sum of the two voltages is

u r ( t ) = UESR + uc
-0081 I

0 * 1
The peak-to-peak value of the total voltage U , is equal to
the peak-to-peak value of the ESR voltage uESR if the
voltage drop across the capacitance is sufficiently low.
The derivative of the voltage U, is

(75) -0oel

008
from which the minimum value of U, occurs at 1

DT
tmm. = - - r C
2 ,
The waveforms for the interval DT < t < T can be
obtained in a similar manner. The maximum value of U , -0081 1
occurs at 0 02 04 06 08 1
flT

t,, =-
(1
- rc
+ D)T c (77)
c

2 Fig. 21 Voltage waveforms of v E S Rvc

, and U,
V, = 12 V,f, = 100 kHz, D = 0.35,L = 40 pH, rc 50
For steady-state operation the average value of udt) is ~ 0, a C=33pF
~ ~ _ _ b C = C,,, = 65 pF
zero. Hence ~-
ESR
UC c C = 100pF

peak-to-peak value of U , is higher than the peak-to-peak

The peak-to-peak ripple voltage is independent of the
voltage across the filter capacitor and is determined only
by the ripple voltage of the ESR if tmin< 0 and t,, <
D T . This happens when
value of uESRbecause C < Cmin.Fig. 21b and c show the
waveforms for C = Cminand C > Cmin,respectively. For
both these cases, the peak-to-peak voltages of U, and u E S R
are equal to each other.
Note that the minimum value of inductance L is deter-

= max
C 2 Cmin (icy
- 2) (79)
mined either by the boundary between the continuous
and discontinuous conduction modes or by the ripple
voltage.
I E E PROCEEDINGS-G. Vol. 139. N o . 6. DECEMBER 1992 679