Small-Signal Characterization of the

Forward-Flyback Converters with Active Clamp

Ionel Dan Jitaru Serban Birc5-G515teanu,Senior Member, IEEE

Rompower Inc., Phone (520) 326-8401, Fax. (520) 326-8366 LR 2EP - GE 44, IRESTE, Fax. +33 240 68 30 66
4400E Broadway, Suite 414, Tucson, AZ 85711, U.S.A. La Chantrerie, BP 60601,44306Nantes cedex 3, FRANCE

Abstract - Linear equivalent models of the The for-fly with hybridge rectifier has a single transformer
forward-flyback converters with active clamp are secondary which always conducts half of the output current.
deduced, using the Vorpbrian model of the PWM Transformerratio is then the same for the forward and for the
switch. Control-to-output transfer functions are boost-buck sections.
plotted against frequency, for a 36-72 V to -
5 V - The for-fly with tapped secondary has separate windings for
converter, for different load resistances, thus
allowing the design of the error amplifier as to the two sections, thus allowing for different transformer
ensure circuit stability. Experimental results ratios, but the two sections are connected towards the output
confirm models validity. at the same filter inductance. This leads to dynamic behavior
different from the preceding topology when the converter is
I. INTRODUCTION controlled by a feedback loop.

The trend towards higher power density and higher III.THE LINEARIZED MODEL
efficiency imposed converter topologies which can use the
entire range of quasi-linearity of the transformer core: push- To deduce the small-signalbehavior of the converters, the
pull, bridge, half-bridge and active-clamp forward. In these method of the PWM model [31] will be used, rather than the
converters, in steady state, the transformer magnetizing state-space average method [30]. The PWM model is
current has the average value zero for any value of the load versatile and may be applied to the complex, non-standard
current. In the active-clamp flyback and forward-flyback topologies, as the converters with active clamp are.
converters, the average value of the magnetizing current is The for-fly converter with hybridge rectifier (Fig. 1.a) may
proportional to the load current. The magnetizing current be considered as being the association of two dc-dc converters
may even not change the polarity - at medium and heavy connected at the same output (Fig. 1.c) :
loads. Transformer core is used less well than in push-pull or - a forward converter with the duty factor D , and
active-clamp forward converters, but these topologies still. - a boost converter with the duty factor D ,the transformer
have high efficiency and other advantages. The forward- and a buck converter with the duty factor (1 - D ) . Indeed, the
flyback converter (hereafter called "for-fly") has lower losses operation of the circuit L2-D2-D1 is that of a buck converter
in the secondary than the push-pull converter. where D2 is the freewheeling diode and D , is the control-
This paper aims to deduce small-signal equivalent circuits led switch (called the "active" switch in [31]), because it
for the for-fly converters, in order to make it possible to conducts when energy is stored in inductance L2 . The input
calculate the error amplifier phase correction as to ensure voltage for this one is ideally E D / (n (1 - D)) .
circuit stability when controlled by a feedback loop. The If the clamping circuit is connected across the transformer
circuits were deduced using the versatile model of the PWM primary, as in [33] and others, than the clamping circuit is a
switch [3 11. Experimental verification demonstrated the flyback [27]. For this reason we maintained in this paper the
validity of the models. same name offorward-flybackconverter, though the operation
is slightly different.
II. CIRCUITS OPERATION Consequently, the linear equivalent circuit for duty cycle
variations only is that of Fig.2, with the parameters' values:
Two active-clamp converters are studied: the for-fly with
hybridge (current-doubler)rectifier (Fig.1) and the for-fly with I
tapped transformer secondary (Fig.5). In both converters, Ic1= -
when the main switch Q conducts, energy is transferred from 2
the dc source E to the filter inductor (L1 or L ) . At the same E
VDI= Vapl - ICI rEl (1 - D) = - - I,, rEl (1 - D)
time the magnetizing current increases, hence energy is stored n
in the transformer magnetic field. When Q switches off, RI = D (1 - D)rE1+ r,,
energy stored in the transformer is eliminated: partly through The transformer ratio n = np / ns .
QA in C A ,partly towards the output as in a boost or flyback
rE = 'CA R A =rcA 1 v
converter. Circuit operation is mixed: forward and boost-buck IC= --- - 0
[I1 - WI. rCA RA 1-D RA

0-7803-4340-9/98/$10.000 1998 IEEE. 626

rt 5 ’
------l

At w ’

tf ‘ f

U *

w
 ~

1 E parameters: E = 36 -72 V , U = 5 V , R L = 0.25 - 1 Q , L , =

4 pH , L 1 = L z = 2 pH , L p = 64 pH , n = 4 , rL1= rL2 = 3
m a , r , = 3 mQ ,rp = 50 m a , Rg = 0.15 a , C = 390 pF ,
rc = 60 m a ,CA= 10 nF and 47 nF ,r,-A = 10 mQ .
The calculated and measured gain and phase curves are close
to each other for different E and load values, at frequencies
V is the voltage across capacitor CA . R A is the load lower than half the switching frequency. The load resistance
resistance of the boost section, that is the resistor connected values seems not to have a dramatic impact on these
in parallel with CA. In the for-fly converters R A is infinite, functions. As it was always noticed, the gain and the phase
hence relationships for R E , I c and V are simpler. rL1 shift are larger for higher values of the load resistance. On
and rLz are the series resistances of inductors L1 and Lz , the contrary, variations of the duty factor lead to significant
respectively. variations of the curves (Figs.3 and 4). The curves are
relatively smooth, because of the very low value of the ratio
I L1IC , which leads to a very low value of the critical series
la=- damping resistance (= 0.14 a).This is an advantage of the
2
switching power supplies with very low output voltage.
E D
VD2= Vap2- IC, rm D = - -- IC2 m D Phase lead at high frequencies is due to the transfer function
n 1-D zero produced by the equivalent series resistance rc of the
output capacitor C .
Capacitor CA value is a key parameter, because the reso-
A transformer must be inserted before, between or after the nance of C A and transformer primary inductance creates a
equivalent circuits of the boost and the buck converters (even right-half-plane zero with a strong impact on the gain and
with unity tums ratio), in order to avoid a short-circuit due to phase curves. The position of this zero strongly depends on
the crossed-connection of the ground nodes of the two the input voltage and its damping - on the load. CA value is
converters. First, the transformer was inserted in front of the to be chosen low, in order this zero to have a frequency as
equivalent circuit of the forward converter and between the high as possible. Also an RD C D damping circuit may be
boost and the buck converters as to be equally taken into added in parallel with CA to reduce phase shift. We used C ,
account. Inserting the transformer limits the domain of = (8 - 10) CA and RD =180 .
model validity towards low frequencies, because in the real The same method was used to deduce the model of the for-
circuit the duty-factor variations are transmitted by PWM and fly converter (Fig.5) with transformer tap (previously deduced
rectification, not directly as slow signals through the by the averaging method [331). The small-signal equivalent
transformer. This leads to completely false results at circuit (Fig.6) is very similar to the previous one, with two
relatively low frequencies. Consequently, the isolation task remarks. First, the transformer ratio in the forward section
was left to the ideal transformer from the equivalent circuit of ( 1 ) may be different from the ratio for the boost-buck section
the buck converter. The real transformer was moved in front (2); thus the weights of the two sections may be different.
of the boost converter and the other component values of the Second, the circuit contains a single output inductor L and
boost equivalent circuit converted according to the turns ratio the two sections are connected together in front of this
(voltages divided by n ,currents multiplied by n and impe- inductor.
dances divided by n2 ). Now, the two transformer icons may For this topology, the curves are very sensitive to the
be put together in a single one, with the converters connected transformerratios. The higher the contribution of the forward
at the secondary. The real transformer initially appears in the section, the higher the high-frequency phase shift. The phase
boost equivalent circuit with opposite primary/secondary shift is larger than for the for-fly converter with current-
polarities. In order to make it possible to unify the two icons doubler rectifier, hence the dynamic stability is more difficult
of the real transformer, it must appear with the same winding to obtain.
polarities. Therefore, the relative windings' polarity of the
ideal transformer from the buck converter equivalent circuit CONCLUSION
was inverted and that of the current and voltage generators of
the equivalent circuit, too. Finally, the real transformer could
even be eliminated from the equivalent circuit and only the Small-signal equivalent circuits of the forward-flyback
generator and winding resistances be taken into account. converters were deduced, in order to make it possible to
The model is deduced for all inductances operating in calculate the error amplifier phase correction as to ensure
continuous mode (with never-zero current), for two reasons. circuit stability when controlled by a feedback loop. The
First, it was noticed that the stability conditions are more circuits were deduced using the model of the PWM switch.
easily satisfied in the discontinuous than in the continuous Experimental verification demonstrated models' validity. The
mode [32]. Second, the model is simpler for the continuous curves are relatively smooth, because of the very low value of
mode. Anyway, the boost section always operates in the the ratio L I / C (which leads to a very low value of the
continuous mode, because the auxiliary switch QA conducts critical series damping resistance). The phase shift is larger
in both directions, therefore the current in the transformer for the converter with tapped secondary than for the converter
with current-doubler rectifier, hence the current-doubler
primary never remains zero for an interval during each cycle.
We have consbvcted for-fly converters with the following rectifier is to be preferred.

628
 lk 10k

frequency (HZ)

Fig.3. Calculated and measured gain and phase curves of the for-fly converter with hybridge rectifier delivering 15 A and 5 A ,

?/d-7 2v

lk 10k lOOk ’
iI
frequency (Hz)

Fig.4. Gain and phase curves of the for-fly converter with hybridge rectifier delivering l!j A , at 3 different supply voltages.

629
 Fig.5. Forward-flyback converter with secondary tap.

I-------------i--c--i----c--l
a

Fig.6.Linear equivalent circuit of the converter with secondary tap

630
 __
1 measured 1

1I I E=4av I

//(j-bA

I a

j 0 -.

I"/d-lSA

~ -10-1

1
I
I
I
v/d_5A calculated

measured
2
v/d_lSA

lk 1Gr

- frequency ( H z )

Fig.7. Gain and phase curves of the for-fly converter with tapped secondary delivering 15 A and 5 A.

REFERENCES [ 121 I.D. Jitaru, "A new high-frequency zero-voltage-switched

PWM converter,'' Proc. of APEC'92, pp.657-664.
[l] J.A. Bassett, "New topology constant frequency ZVS [I31 I.D. Jitaru, "Zero voltage PWM, double ended converter,"
converter," Proc. of Int. Conf. on Power Conversion Proc. of HFPC'92, May 1992, pp.394-405.
PCIMPO, June 1990, pp.249-258. [ 141 LD. Jitaru, "High-frequency soft-transition converter," Proc.
[2] C. Peng, M. Hannigan and 0. Seiersen, "A new efficient high- ofAPEC'93, March 7-1 1, 1993, pp.880-887.
frequency rectifier circuit," Proc. of HFPC91, pp.236-243. [15] I.D. Jitaru and G. CocinB, "High-efficiency DC-DC conver-
[3] K. O'Meara, "A new output rectifier configuration optimized ter," Proc. of APEC'94, February 13-17, 1994, pp.638-644.
for high-frequency operation," Proc. of HFPCYI, pp.219- [16] I.D. Jitaru, "The impact of low output voltage requirements
225. on power converters," Proc. of HFPCPS, May 6-12, 1995,
[4] J.A. Bassett, "Constant-frequency ZVS converter with pp. 1- 10.
integrated magnetics," Proc. of APEC'92, pp.709-716. [ 171 I.D. Jitaru, "DC-DC converter technologies," seminar notes,
[5] P. Vinciarelli, "Optimal resetting of the transformer's core in APEC'96, March 3-7, 1996, PAC 3.
single-ended forward converter,'' LIS. Patent 4,441,146, [I81 I.D. Jitar:~and A. Ivascu, "Quasi-integrated magnetics - an
Apr. 3, 1984. avenue for high power density and efficiency in power
[6] O.S. Seiersen, "Power supply circuit," US.Pafent 4,899,271, converters," APEC'97, Feb. 23-27, 1996, Atlanta, GA,
Feb. 6, 1990. pp.395-402.
[7] B. Carsten, "Design techniques for transformer active reset [19] G. Hua, C.S. Leu and F.C. Lee, "Novel zero-volt-transitions
circuits at high frequencies and power level," Proc. of PWM converter," Proc of Virginia Power Electronics Center
HFPC'90, pp.235-246. Seminar, 1991, pp.81-88.
[8] I.D. Jitaru, "Fixed-frequency single-ended forward converter [20] C.S. Leu, G. Hua and F.C. Lee, "Comparison of forward
switching at zero voltage," US.Patent 5,126,931, June 30, topologies with various reset schemes," Proc of Virginia
1992. Power Electronics Center Seminar, 1991, pp.101-109.
[9] I.D. Jitaru, "Square-wave converter having an improved zero- [21] C.S. Leu, G. Hua, F.C. Lee and C. Zhou, "Analysis and
voltage switching," US.Patent 5,231,563, July 27, 1993. design of R - C - D clamp forward converter,'' Proc. of
[lo] I.D. Jitaru and V. Beldianu, "Fixed-frequency converter HFPC'92, May 1992, pp.198-208.
switching at zero voltage," U.S. Patent 5,434,768, May 18, [22] R. Watson, G.C. Hua and F.C.Lee, "Characterization of an
1995. active clamp flyback toplogy for power factor correction
[ 111 I.D. Jitaru, "Constant-frequency forward converter with reso- applications," Proc. of APEC'94, Feb. 13-17, pp.412-418.
nant transition," Proc. of HFPCPI, pp.282-292.

631
[23] L. Huber, D. Sable, G . Hua, F.C. Lee and N. LaPrade, "Design [29] L. Huber and M.M. Jovanovic, "Forward converter with
of a high-efficiency power converter for a satellite solid- current-doubler rectifier: analysis, design and evaluation
state power amplifier," 1994, pp.645-651. results," APEC'97, Feb. 23-27, 1996, Atlanta, GA, pp.605-
[24] W.A. Peterson, R.Kolbet, and R. Saint-Pierre, "A clamped 611.
transformer power supply producing zero voltage resonant [30] G.W. Wester and R.D. Middlebrook, "Low-Frequency
transitions over the full load range," May 1994, pp.287-292 Characterization of Switched dc-to-dc Converters,'' ZEEE
[25] W. Andreycak, "Active clamp and reset technique enhances Power Processing and Electronics Specialists Conference,
forward converter performance," Power Supply Design Atlantic City, NY, May 22-23, 1972.
Seminar SEM-1000,Unitrode Integrated Circ. Corp., 1994. [31] V. Voqkian, "Simplified analysis of PWM converters using
[26] L. Balogh, "The current-doubler rectifier: an alternative model of PWM switch. Part I: continuous conduction mode.
rectification technique for push-pull and bridge converters," Part 11: discontinuous conduction mode," IEEE Tr. on Aero.
Design Note DN-63, Unitrode Integrated Circuits Corp., DG- and Elec. Syst., vo1.26, no.3/May 1990, pp.490-505.
300, Dec. 1994, pp.12-14. [32] S . Birc%-GBlSiteanu,"Parameters' calculus for high voltage
[27] D. Dalal and L. Wofford, "Novel control IC for single-ended ratio switching power supplies," Ph. D. dissertation, Univ.
active-clamp converters," Proc. of HFPC'9.5, May 1995, of Nantes, France, Jan. 11, 1993.
pp. 136- 146. [33] G. Stojcic, F.C. Lee and S. Hiti, Small-signal characteriza-
[28] L. Balogh, "The performance of the current doubler rectifier tion of active-clamp PWM converters," Power Systems
with synchronous rectification," Proc. of HFPC'9.5, May World'96, Sept. 7-13, 1996, Las Vegas, NV.
1995, pp.216-225.

632