Clamp
Clamp
Clamp
1, JANUARY 1996
162
@vm
t
k
6
I. INTRODUCTION
HE flyback topology has long been attractive because
of its relative simplicity when compared with other
topologies used in low power (up to several hundred watts)
applications. The flyback transformer serves the dual purpose of providing energy storage as well as converter isolation,
theoretically minimizing the magnetic component count when
compared with, for example, the forward converter. A drawback to the use of the flyback is the relatively high voltage and
current stress suffered by its switching components. High peak
and RMS current stress is a particular problem for flybacks
when operating in discontinuous conduction mode (DCM) and
is in fact a primary detriment to increasing output power.
An addition, high turn-off voltage (caused by the parasitic
oscillation between the transformer leakage inductance and the
switch capacitance) seen by the primary switch traditionally
requires the use of an RCD clamp to limit the switch voltage
excursion. Unfortunately, in this scheme the energy stored in
the transformer leakage is dissipated in the clamp resistor,
resulting in a difficult design trade-off between clamping
action and clamp circuit power dissipation.
The limitations presented by the RCD clamp can be largely
overcome by replacing the passive clamp with an activeclamp circuit, as shown in Fig. 1. Active-clamp methods have
been explored in detail for forward converters [7], [XI. The
active-clamp circuit provides the benefits of recycling the
transformer leakage energy while minimizing turr-off voltage
Manuscript received June 15, 1994, revised July 12, 1995 This work was
supported by Zytec Corporation
The authors are with the Virginia Power Electronics Center, Bradley
Department of Electrical Engineenng, Virginia Polytechnic Institute and State
University, Blacksburg, VA 24061 USA
Publisher Item Identifier S 0885-8993(96)00603-5.
163
0
0
Vm
0
0
0
T X T A
T4 T5
. ..
T5 -T6
-
T6- T7
Fig. 2. Active-clamp flyback topological states
This paper presents evaluation of a constant-frequency, softswitching, active-clamp flyback converter for dcldc conversion
applications. The basic principle of operation is analyzed and a
design procedure is developed. Experimental results are then
presented, which illustrate converter function and verify the
analysis presented.
11. ACTIVE-CLAMP
FLYBACK
CONVERTER
OVERVIEW
164
For L,
- N
dt
-N
V,n
+ NVO
LT
(3)
EL? 2 Ec,
I ~ ~ 2 t u off
r n
(5)
TO).
Note that the length of the time intervals TI to T3 and T4
to T7 have been greatly exaggerated in Fig. 3 in order to more The last part of the expression in (6) is the value of the
voltage developed across L,. Although an explicit value of
clearly show the transition periods.
the resonant inductor hasn't yet been determined, for the
purposes of estimating the maximum voltage stress on S1
111. SOFT-SWITCHING
FLYBACKDESIGNCONSIDERATIONS and S2 a value of L,/10 can be used as a conservative
For the purposes of ZVS of S1,there must be assurance that design guideline. Also, as long as L , >> L, the converter dut
the switch is turned on during the T5 to T6 time interval. If cycle behavior is approximately the same as for a non-act
not, the resonant inductor current reverses (becoming positive clamp flyback operating in CCM. With the addition of L,, the
again), recharging C,, and ZVS is lost (or at least partially effective duty cycle (as defined by the charge and discharge
lost). Therefore, the delay time between the turn off of S2 and cycle of the flyback inductor) is slightly less than switch S l ' s
the turn on of S1 is critical to ZVS operation. The optimum duty cycle (see Fig. 3)
value of this delay is one-quarter of the resonant period formed
by L, and C,
(4)
Equation (7) is approximate in that it assumes lossless switching. In this application A D M 5% of Dsl so for the purposes
of developing (6) and subsequent design equations it will be
assumed D = Deff = Dsl.
165
=I
ILm,peak
= IL,,peak 2
vCT
(8)
(9)
The difficulty in solving (8) for L, is the fact that the resonant
capacitor voltage (V&) is a function of the value of L,.
However, in a practical design situation, the resonant inductor
voltage at T4 is relatively small (compared to
+ NVo) and
(8) can be solved for an approximate minimum value of L,
necessary to achieve ZVS
v,
51
,peak
(1 - DHL)
,rr2L,F,2
1 DLL
Icclamp,EMS
max
I z t l a m p -/ F .
(11)
and
(16)
Equation (16) was derived using the same simplifying assumption used to derive (12) and (13).
6) Choose Output Rectifier: The maximum theoretical reverse voltage seen by the output rectifier in the active-clamp
flyback is the same as for a standard flyback design. However,
the rectifier current is another matter. Due to the presence of
the clamp circuit, the secondary current is discontinuous in
shape even though the flyback inductor magnetizing current
is always positive and greater than zero. This is illustrated in
Fig. 3. The result is much higher peak secondary currents than
would normally be expected in a non-active-clamp flyback
operating in CCM
Ig?;eak
2pFX
Vo(1 - DLL)
166
Efficiency (h)
0 --
--
Transformer:
core: Toshiba PC40ETD49-Z
primary: 45T of 3 strands of 150/42 Litz wire
secondary: 15T of 5 strands of 150/42 Litz wire
L,: 215 pH, &ak (referred to primary): 2.3 pH
Resonant inductor:
core: MPP 55530
winding: 8T of 10 strands of AWG 26
L,: 7 pH (with no bias)
Cclamp:
2.2 pF, 250 V
S2: IRFF360
* 51: ESP360
* Output stage:
D1 : 2xBYV44-500
CO: 3 x 2200 pF, 100 V Aluminum electrolytic.
167
Efficiency ph)
IRGPC4OU
--c
IRGPC4OU. Cr = 3000 pF
m s o
.-..-.
IRFpcso~~m5o
-..-..
Lleak.
V. CONCLUSION
RES S W 3
kHZ
VBW 3 kHZ
SWP 1.0
sec
This paper has presented the analysis, design, and experimental results for a high-efficiency 500 W flyback dcldc
converter employing active-clamp circuitry. The incorporation
of the active-clamp circuit into the basic flyback topology
provides a mechanism for achieving ZVS for both the primary
and auxiliary switches and soft commutation of the output rectifier while operating with low ripple (unidirectional) current
in the flyback transformer. Low ripple current operation is
desirable as power levels of several hundred watts can realize
higher efficiencies than when compared with a non-activeclamp flyback operating in DCM or an active-clamp flyback
operating with high ripple currents.
APPENDIX
In this appendix are outlines of the derivations of the
equations that appear in the section of the paper describing
the design procedure [(6)-(19)]. Here we make the following
simplifying assumptions:
The clamp current is piecewise linear instead of quasisinusoidal.
Clamp capacitor voltage is constant (i.e., clamp capacitor
is assumed infinite in value).
The duration of the intervals Tl-T2 and T4-T6, shown
in Fig. 3, is short enough compared with the duration of
interval Tl-T6 to allow them to be neglected.
de^
= Dsl.
The peak switch current (which is equal to the peak magnetizing current and peak resonant inductor current) is determined
by applying the principal of power balance to the converter.
168
2l
2 5CrVz7
l@S2turn-off
1QS2 turn-off
(23)
For VC,.M V,, NVo (23) can be solved for Lr to yield (10).
Equation (11): The RMS current flowing in the resonant
inductor can be calculated with the aid of Figs. 3 and 11. If
we assume negligible duty cycle loss, the slope of the resonant
inductor current becomes infinite during the T 4 T 7 interval
shown in Fig. 3 (i.e., the duration of the T4-T7 interval
is zero). The RMS value of the current can then be more
easily approximated using two piecewise linear intervals over
a switching cycle
(b)
Fig. 10. (a) Idealized clamp capacitor current waveform. (b) Simplified
topological state S1 off, S2 on.
S1
s2
Lm
Lr
I S 2 body diode,avg E
ADT -
Lr
- V,,+NVo
(iL,,peak
+ iL,,valley). (21)
The MOSFET
(4 d t
iCclamp
1-D
iSl, peak-
c ~ ~ eisn obtained
t
in a Similar m m ~ r
IS2, MOSFET,RMS E
@=Er---
- iSl,peak
/?*
i&,,,,
( t )d t
(26)
Equations (14)-(16): Equation (14) results from the resonance between L, and C c l a m p . We want half the resonant
period to be greater than S l s maximum off-time
169
<iD1> = Io
Robert Watson (S93) was born in 1958 in Indianapolis, IN. He received the B.S.E.E. degree
in 1982 from Purdue University, West Lafayette,
Indiana and the M.S.E.E. degree in 1986 from
the University of Arizona in Tucson. Currently, he
is a Ph.D. candidate in electrical engineering at
Virginia Polytechnic Insbtute and State University
in Blacksburg.
Prior to joining the Virginia Power Electronics
Center (VPEC) at VPI&SU as a Research Assistant
in 1992, Mr. Watson split 10 years of industrial
employment between Hughes Aircraft Company and Bendix Engine Controls,
a division of Allied-Signal Corporation. Currently, he is a Research Associate at VPEC, specializing in dc and high-frequency ac distributed power
architectures, high-frequency power conversion, and power factor correction
applications.
Mr. Watson is a member of Phi Kappa Phi and was a Hughes Aircraft
Company Fellow in 1985 and 1986.
REFERENCES
Integrating (30) and using the result of (17) yields (18). The
transformer secondary is in series with the output rectifier, so
their RMS currents are identical. Since the dc component of
current is present, we have