DC-DC Flyback Converters in the Critical Conduction Mode:
a Re-examination
G. Spiazzi* D. Tagliavia** S. Spampinato**
*
Department of Electronics and Informatics
University of Padova
via Gradenigo 6/a - 35131 Padova - Italy
Phone: +39-049-827.7525 Fax: +39-049-827.7699
e-mail: spiazzi@dei.unipd.it
**
ST Microelectronics
Via Stradale Primosole 50, 95121 Catania - ITALY
Phone: +39-095-740-1111 - Fax: +39-095-740-6006
e-mail: donato.tagliavia@st.com
e-mail: sergio.spampinato@st.com
Abstract - The critical conduction mode for DC-DC flyback
SMPS, in which the converter is forced to operate at the
boundary between continuous and discontinuous conduction
modes, represents an interesting alternative to the classical
constant-frequency PWM technique. In fact, such operating
mode allows for a soft turn off of the freewheeling diode, Zero
Voltage commutations of the switch and reduction of the
generated EMI.
In this paper, this operating mode is re-examined with the aim
of accurately predict switching frequency variation and
component stresses in those applications in which the delay
inserted between the turn off of the freewheeling diode and the
turn on of the switch, used to achieve zero voltage commutations,
cannot be neglected. The analysis presented allows for a correct
prediction of the converter behavior in all operating conditions
as well as for a proper design of the feedback loop through a
suitable small-signal characterization.
The theoretical forecasts are verified by means of a flyback
prototype built using a new smartpower IC developed by ST
Microelectronics in VIPower
M3 technology.
I. I
NTRODUCTION
Standard DC-DC converters (buck, boost or buck-boost) in
the critical conduction mode, i.e. at the boundary between
continuous and discontinuous conduction modes (CCM-
DCM), have the following advantages as compared to normal
constant-frequency PWM operation: soft turn-off of the
freewheeling diode (like constant-frequency DCM operation
but at a reduced current stress), self-protection capability
against short circuit conditions at the output, and reduced turn
on and turn off losses by exploiting the resonance between the
inductance and the switch output capacitance. This feature is
particularly appealing for off-line flyback power supplies in
which the high switch voltage stress increases both switching
losses and EMI.
Analysis of flyback converters in the critical conduction
mode has already been reported in literature [1], but it is
usually done neglecting the resonant intervals that occur at
the beginning and at the end of each switching interval. This
simplification leads to substantial errors in the prediction of
the switching frequency variation and of the component
stresses in converters designed with a low resonance
frequency (for example in order to keep the switch dv/dt at
turn off below a specified limit).
In this paper, a detailed analysis is reported which allows
for an accurate prediction of the converter behavior in all
operating conditions. Moreover, a small signal model is
a)
U
g
R
L
D
C
r
L
µ
S
U
o
N
1
:N
2
+
+
+
u
Cr
i
L
C
L
b)
U
g
C
r
L
µ
S
+
+
u
Cr
i
µ
c)
U
g
L
µ
S
+
i
µ
d)
L
µ
i
µ
U
op
+
Fig. 1 – a) Basic scheme of a flyback converter in the critical
conduction mode; b) equivalent circuit during subintervals T
d
and T
rise
;
c) equivalent circuit during T
on
; d) equivalent circuit during T
off
0-7803-6404-X/00/$10.00 (C) 2000
developed in order to design properly the feedback loop.
A multioutput flyback prototype employing a new
smartpower IC developed by ST Microelectronics in
VIPower
M3 technology was built and tested in order to
verify the theoretical expectations [2].
II. R
EVIEW OF THE
C
ONVERTER
O
PERATION
The basic scheme of the flyback converter in the critical
conduction mode is shown in Fig. 1a in which the resonant
capacitor C
r
accounts for any parasitic capacitance (of the
switch, of the freewheeling diode and of the transformer
windings) as well as added ones. The circuit operation is very
similar to a standard flyback except for the resonant intervals
at the beginning and at the end of each switching period. Two
different situations can occur depending on the value of the
voltage conversion ratio
g
op
U
U
M
=
, where U
op
is the output
voltage reported to the primary side: when M > 1 zero voltage
turn on of the switch is achieved as can be seen from the
converter main waveforms shown in Fig. 2a, while if M < 1
the situation becomes as depicted in Fig. 2b. In both cases the
switching period T
S
= t
4
-t
0
can be divided in four subintervals
which are analyzed in the following, assuming a new time
origin at the beginning of each subinterval.
A. Interval T
d
= t
1
-t
0
.
At instant t
0
the magnetizing current zeroes causing the
turn-off of the freewheeling diode D. The equivalent circuit
during this subinterval is shown in Fig. 1b: the magnetizing
inductance L
µ
resonates with capacitor C
r
bringing its voltage
toward zero. The magnetizing current and the capacitor
voltage are given by the following equations
(u
Cr
(0) = U
g
+U
op
, i
µ
(0) = 0):
() ( )
()
()
tsin
Z
U
ti
tcosUUtu
R
R
op
RopgCr
ω−=
ω+=
µ
(1)
where
r
R
CL
1
µ
=ω
and
r
R
C
L
Z
µ
=
are the angular
frequency and the characteristic impedance of the resonant
circuit respectively.
If M > 1 this interval ends when voltage u
Cr
reaches zero,
while if M < 1 T
d
is chosen to be one half of the resonant
period so as to turn the switch on always at the minimum
value of the voltage across it, as shown in Fig. 2b:
<
ω
π
>
−
ω
=
−
1M if
1M if
M
1
cos
1
T
R
1
R
d
(2)
()
<
>
−−
=
==
µ
µ
1Mif0
1Mif
M
1
1
Z
U
ITi
2
R
op
0d
(3.a)
()
<−
>
=
1MifUU
1Mif0
Tu
opg
dCr
(3.b)
B. Interval T
on
= t
2
-t
1
.
The second interval (T
on
= t
2
-t
1
) corresponds to the normal
charging phase (see Fig. 1c) in which the magnetizing current
i
µ
increases linearly starting from an initial value which is
zero only in the case M < 1 and if the switch is turned on at
the valley point of voltage u
Cr
(this is the case of Fig. 2b). In
the general case we can write:
()
t
L
U
Iti
g
0
µ
µµ
+=
(4)
At the end of the switch on-time, the magnetizing current
reaches its maximum value (indeed, its maximum occurs
during the next resonant period but it differs only slightly
from this value):
()
()
0pk
g
onon
g
0pkon
II
U
L
TT
L
U
IITi
µµ
µ
µ
µµµ
−=⇒+==
(5)
C. Interval T
rise
= t
3
-t
2
.
The third interval (T
rise
= t
3
-t
2
) is the time between the
switch turn off instant and the turn on of the freewheeling
diode at the secondary side which occurs when u
Cr
becomes
greater than U
g
+U
op
. The equivalent circuit during this stage
is, again, that shown in Fig. 1b. Due to the high value of the
energy stored in the magnetizing inductance, it seems
a)
u
Cr
i
µ
t
0
t
1
t
2
t
3
t
4
I
µ
pk
T
d
T
on
T
rise
T
off
t
U
g
+U
op
b)
u
Cr
i
µ
t
0
t
1
t
2
t
3
t
4
I
µ
pk
T
d
T
on
T
rise
T
off
t
U
g
+U
op
Fig. 2 – Magnetizing current i
µ
and resonant capacitor voltage u
Cr
waveforms in a switching period.
a) case M > 1; b) case M < 1
0-7803-6404-X/00/$10.00 (C) 2000
reasonable to consider a linear increase of the resonant
capacitor voltage at a constant charging current, i.e.:
()
(
)
pk
opgr
rise
r
pk
Cr
I
UUC
Tt
C
I
tu
µ
µ
+
=⇒=
(6.a)
()
pk
Iti
µµ
=
(6.b)
However, this approximation can lead to a substantial error
in converters designed with a low resonant frequency (for
example in order to keep the dv/dt across the switch at turn
off below a specified limit), expecially when M becomes
lower than one. In these cases an exact analysis must be
performed, with the following result:
() ( )
() ( )
α+ω=
α+ω+=
µ
tcosI
ˆ
ti
tsinI
ˆ
ZUtu
R
RRgCr
(7)
where
2
R
g
2
Lpk
Z
U
II
ˆ
+=
(8.a)
−=α
µ
−
pkR
g
1
IZ
U
tg (8.b)
The interval duration is given by:
α−
ω
=
−
I
ˆ
Z
U
sin
1
T
R
op
1
R
rise
(9)
and the value of the magnetizing current at the end of this
interval results:
()
−
+==
µµµ
1
M
1
Z
U
ITiI
2
2
R
op
2
pkriseoff
(10)
Note that, this current is lower than I
µ
pk
for M > 1 meaning
a reduction of the energy transfered to the output. This fact is
in agreement with the situation described in Fig. 2a where a
small fraction of the energy store in the magnetizing
inductance is, indeed, returned to the input during the first
fraction of the switch on-time, where the magnetizing current
is negative.
D. Interval T
off
= t
4
-t
3
.
During the fourth interval (T
off
= t
4
-t
3
) the magnetizing
current transfers to the secondary side delivering energy to
the load, until it becomes zero, while voltage u
Cr
remains
clamped to U
g
+U
op
. The magnetizing current is given by (see
Fig. 1d):
()
t
L
U
Iti
op
off
µ
µµ
−=
(11)
while the duration of such interval results:
()
op
off
offoff
U
IL
T0Ti
µµ
µ
=⇒=
(12)
The total switching period is the sum of all these
subintervals, i.e.:
offriseondS
TTTTT
+++=
(13)
Neglecting subintervals T
rise
and T
d
allows for a
straightforward analysis, since the usual relations of the
flyback converter can be used [1]. Unfortunately, when they
are not negligible, substantial errors are introduced in the
prediction of the switching frequency variation as well as of
the component stresses. In the following section, a rigorous
analysis is performed with the aim of precisely forecast the
converter behavior at different output power and voltage
conversion ratios.
III. DC A
NALYSIS
In order to derive the relation between the voltage
conversion ratio and the switching frequency for a given set
of the converter parameters, let’s start with the determination
of the average (in a switching period) current delivered to the
load. From Fig. 2 we can write:
S
off
offDp
T2
T
II
µ
=
(14)
where I
Dp
is the secondary diode current reported to the
primary side. This current equals the average load current,
i.e.:
L
op
Dp
R
U
I
η
=
(15)
where a non unity converter efficiency
η
was assumed. Using
(12) and (15) into (14) we obtain:
L
2
op
2
off
S
R
U
I
T2
L
η
=
µ
µ
(16)
Now, substituting (2), (5), (9) and (12) into (13) the
switching period results:
()
d
op
off
R
op
1
R
op
0pkS
T
U
L
I
I
ˆ
Z
U
sin
1
M
U
L
IIT
++
α−
ω
+
+−=
µ
µ
−
µ
µµ
(17)
Now, equations (16) and (17), together with (2), (3) and (8),
form a system in the two unknowns I
µ
pk
and f
S
, which can be
solved numerically (a simple MatCad sheet was used to this
purpose). The results are shown in Figs. 3a and 3b, which
refer to a converter whose specifications and parameters are
reported in the experimental result section. Fig. 3a, reports
the predicted switching frequency variation as a function of
the output power for three different voltage conversion ratios
corresponding to a nominal peak input voltage of 311V
±
20%
(as shown in the experimental result section, the prototype is
an off line converter whose input voltage is the rectified line
voltage U
IN
= 220V
rms
±
20%). Clearly, as the output power
decreases, the switching frequency increases as a
consequence of a reduced switch on-time. Fig. 3b, instead,
shows the switching frequency variation as a function of the
voltage conversion ratio for four different output power
levels.
For the purpose of comparison, Fig. 4 shows the difference
in the switching frequency prediction as given by three
different approaches: the more accurate one (curve a), by
neglecting only resonant interval T
rise
(curve b) and by
neglecting both intervals T
rise
and T
d
(the values
corresponding to curve c are devided by a factor of 3 in order
to draw the curves at a reasonable scale): as we can see, just
neglecting the short interval T
rise
causes an error in the
0-7803-6404-X/00/$10.00 (C) 2000
switching frequency prediction which can be as high as 35%
at 20W. This result is a consequence of the parameter values
used in the prototype, where attention was payied in reducing
the dv/dt across the switch at turn off by lowering the
resonant frequency
ω
R
.
Lastly, Fig. 5 shows the effect of the approximation usually
done in the calculation of subinterval T
rise
: curve a) reports
the switching frequency prediction given by the approximated
equations (6) as compared to the exact analysis: actually, the
difference is modest for voltage conversion ratios greater than
one and becomes appreciable only when M is lower than one.
In the latter case, at further reduced M values the
approximation gives a completely wrong result since the
corresponding curve bends until it changes slope, which is
clearly an incorrect prediction.
However, it should be mentioned that for converters with
significant resonant intervals T
d
and T
rise
, particular attention
must be payed in order to operate the converter always for
M
≥
1 for any input and load condition, since the lost of the
soft switching condition at the switch turn on, rapidly
increases the switching losses, i.e. from Fig. 2b:
()
s
2
opgreargdisch
fUUC
2
1
P
−=
.
a)
20 80 120 140 16040 60 100
f
S
[kHz]
80
100
120
140
160
P
O
[W]
M
max
M
nom
M
min
180
200
b)
60
100
120
140
160
180
200
80
0.6 1.2 1.6 1.8 2.00.8 1.0 1.4
f
S
[kHz]
M
a)
c)
b)
d)
a) P
onom
b) 0.75P
onom
c) 0.5P
onom
d) 0.25P
onom
Fig. 3 – Predicted switching frequency variation: a) as a function of the
output power for three different voltage conversion ratios corresponding to a
nominal peak input voltage of 311V
±
20%; b) as a function of the voltage
conversion ratio for different output power levels
20 80 120 140 16040 60 100
f
S
[kHz]
80
200
120
240
160
P
O
[W]
a)
b)
c)
280
Fig. 4 – Comparison between predicted switching frequency variation as a
function of the output power for the maximum voltage conversion ratio.
a) accurate analysis; b) by neglecting only resonant interval T
rise
; c) by
neglecting both intervals T
rise
and T
d
(the values corresponding to curve c are
devided by a factor of 3 in order to fit the curve in a reasonable scale)
60
100
120
140
160
180
200
80
0.6 1.2 1.6 1.8 2.00.8 1.0 1.4
f
S
[kHz]
M
a)
b)
Fig. 5 – Effect of the approximation in the interval T
rise
: a) switching frequency
predicted using approximate equations (6); b) switching frequency given by the
exact analysis using eqs. (8-10) (correspond to curve d of Fig. 3).
(P
o
= 0.25P
onom
)
IV. S
MALL
-S
IGNAL
M
ODEL
A proper design of the output voltage control loop requires
the knowledge of the power stage transfer function. As it is
shown in the following section, the IC senses the switch
current and compares it with a reference value provided by
the external control loop, thus realizing a well known peak
current control. Once again, since the converter works at the
boundary between the continuous and discontinuous
conduction modes, we are tempted to use the usual constant
frequency small signal models. However, a dilemma rises: we
should use the model for CCM or for DCM operation?
Fortunately, the difference is not so high except for the phase
behavior at high frequency. However, it is worthy to know the
error magnitude we do when neglecting the resonant
subintervals. Thus, in this section, a more accurate model is
derived based on the real converter waveforms. Since the
interest of this converter is in the soft-switching operation,
which implies the condition M
≥
1, the approximate equations
for the T
rise
subinterval were used. The details of such
analysis are reported in the Appendix. The result is the simple
circuit model shown in Fig. 6 (output section only): from it,
the transfer function G
p
(s) between the control signal I
R
and
the output voltage can be easily found as:
()
()
0LpLp
oLp
Rp
r//RCs1
r//R
hsG
+
=
(18)
0-7803-6404-X/00/$10.00 (C) 2000
where R
Lp
and C
Lp
are the total load resistance and total filter
capacitor reported to the primary side and r
o
is a damping
resistance given by the small-signal model.
Fig. 7 shows the comparison between the bode plots of the
derived G
p
(j
ω
) (curve a)) and those obtained by using the
classical transfer functions of the flyback converter for CCM
operation (curve b)) and DCM operation (curve c)): as can be
seen the latter predict a much higher static gain as compared
with the more accurate one. Note that the error introduced by
the DCM transfer function is mainly due to the high error in
the switching frequency prediction if the resonant subintervals
are neglected. If the correct switching frequency is used in
this transfer function, than the error reduces and curve c)
moves closer to curve a) but still remaining at a higher gain.
g
i
u
g
r
o
C
Lp
R
Lp
h
R
i
R
u
o
+
-
Fig. 6 – Small-signal equivalent model of the flyback converter in the
critical conduction mode (output section only)
10
2
10
4
10
0
Frequency [Hz]
-40
-20
0
20
40
[dB]
|G
p
(j
ω
)|
a)
b)
c)
10
2
10
4
10
0
Frequency [Hz]
-180
-120
-60
[deg]
a)
b)
c)
G
p
(j
ω
)
Fig. 7 - Bode diagram of |G
p
(j
ω
)| and
∠
G
p
(j
ω
) for U
g
= U
gmin
and nominal
power. a) Accurate small signal model reported in the paper; b) standard
small-signal model for CCM operation; c) standard small-signal model for
DCM operation
V. S
MARTPOWER
IC
ST Microelectronics has developed a new smartpower IC
specifically designed for off-line flyback power supplies in
the critical conduction mode [2]. Their new high voltage
VIPower M3 technology, monolithically combining an
Emitter Switching Bipolar-MOSFET Power Stage and a
flexible BCD (Bipolar - CMOS - DMOS) control part, is
ideally suited to develop state-of-the-art off-line SMPS in the
100 to 250 Watts output power range [3-5]. This level of
power covers numerous applications, among which are power
supplies for Monitors, TVCs and Desktop PCs. The block
diagram of such IC is reported in Fig. 8 together with the
scheme showing its application to an off-line multioutput
flyback converter. For a detailed description of the IC internal
blocks refers to [2]; here, it is worthy only to mention the
presence of a high voltage comparator (H.V.Comp) which
directly senses the collector voltage of the power device
(which can reach voltages in excess of 900V) in order to
exactly synchronize its turn on with the zero crossing of
voltage u
Cr
, if the converter is operating with M > 1. An
alternative switch on trigger is provided by the low-voltage
comparator (Comp1) by sensing the auxiliary winding voltage
in the case of M < 1 (in this case the converter behaves as
shown in Fig. 2b). Lastly, note that an inner peak current
control is provided by the internal current sense resistor R
s
and associated comparator (Current Limiter). The internal
Error Amplifier, which is normally disabled in the presence
of an external voltage control loop, allows for an alternative
control loop through the supply voltage when the latter is
provided by an auxiliary transformer winding, as shown in
Fig. 8.
C
r
+
V
CC
Input
Filter
U
o3
TL431
OPT
SOPS
Quasi Resonant
U
o2
U
o1
u
IN
u
g
ZERO
C.
ISO
HVC
GND
COMP
+
-
-
+
+
+
V
REG
Fig. 8 - Block diagram of the smartpower IC and its application to a
multioutput flyback power supply
0-7803-6404-X/00/$10.00 (C) 2000
