Professional Documents
Culture Documents
Buck Boost Converters
Buck Boost Converters
net/publication/3322204
Article in IEEE Transactions on Circuits and Systems I Fundamental Theory and Applications · September 1992
DOI: 10.1109/81.168931 · Source: IEEE Xplore
CITATIONS READS
35 1,617
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Dariusz Czarkowski on 25 February 2015.
nVf
I. INTRODUCTION
�.n:'rn
-
PWM dc/de switch-mode converters are usually modeled D(,ns +Tn)
using the state-space averaging technique and linearization n'(I.D) (rn+R,)
[1]-[5]. This method of modeling is quite laborious, especially '< R Vo
when the equivalent circuit contains a large number of compo • C
nents. To overcome this problem, linear circuit models of con L
vertcrs have been discussed in [6], [7]. Interesting nonlinear L-____________�____� +
obtaining dc and small-signal linear models of PWM dc/de series with the magnetizing inductance of the transformer.
converters, taking into account parasitic resistances of reactive
components and power switches and offsct voltages of power resistance of the secondary of the transformer, VF is the thresh
switches. A comprehensive analysis of the PWM flyback con old voltage ot the diode, RF is the forward resistance of the
verter is given to illustrate the new method. diode, and rc is the equivalent series resistance (ESR) of the
filter capacitor.
II. ANALYSIS OF FLYBACK CONVERTER
A circuit of the flyback converter is shown in Fig. l(a). It is A. Assumptions
comprised of a controllable switch S (e.g., MOSFET), a trans
The analysis of the flyback converter of Fig. Ha) begins with
former with a turns ratio n, a diode DI, a filter capacitor C, and
the following assumptions.
a load resistance R. The switch is turned on and off at the
switching frequency Is = liT with the ON duty ratio D = ton/T, 1) The power MOSFET in the ON state is modeled by a
wherc ton is the interval when the switch is ON. Fig. Hc) depicts constant resistance 'os and in the OFF state by an infinite
an equivalent circuit of the converter, where ros is thc transistor resistance. Its output capacitancc and lead inductance (and
ON resistance, r T1 is the winding resistance of the primary of the thereby switching losses) are zero.
transformer, 'L is the magnetic core series resistance, L is the 2) The diode in the ON state is modeled by a constant-voltage
magnetizing inductance of the transformcr, 'T2 is the winding battery VF and a constant forward resistance RF and in
the OFF state hy an infinite resistance. The chargc-carrier
Manuscript rec eiv ed October 4, 1991; revised May IS, 1992. This work lifetime is zero for p - n ,junction diodes (and therefore
was supported by the National Science Foundation under Grant ECS- switching losses due to thc reverse recovery are zero), the
8922695. This paper was recommended by Associatc Editor F. Y. Chang.
diode junction capacitance and lead inductance are zero.
The authors are with the Department of Elec trical Engineering,
Wright Stat e University, Dayton, OH 45435. 3) The transformer leakage inductances and the stray capaci
IEEE Log Numbe r 9203653. tances are neglectcd.
4) Passive components are linear, time-invariant, and fre An equiva1ent averaged resistance (EAR) of the switched
quency-independent. resistances can be determined using the principle of energy
S) The output impedance of the input voltage source is zero eonsemation. This concept is analogous to the equivalent aver
for both dc and ac components, aged resistance of a switched capacitor in switched-capacitor
6) The ac current ripple on the dc CUTTent I L through thc filters. The dc component of the inductor CUTTent TL is taken as
primary of the transformer is negligibly small. a reference current. The switch S is ON for the interval 0 < t ::;
7) The frequencies of ac components of the input voltage and DT and OFF for DT < t ::; T. The instantaneous switch current
the duty cycle are lower than one-half the switching fre in the circuit of Fig. I(c) is i, '" Iv for 0 < t ::; DT, resulting in
quency, similarly as for the state-space avcraging method the RMS value of the switch current
[2], [3].
1
-fTi; dt = mIL' (IS)
B. de and Small-Signal Circuit Models of Flybaek Converter T 0
ID n(I-D) (17)
M I= - = ---'._-- (2)
I[ D resulting in the EAR of rDS
where Vs is the dc component of the voltage across the ideal (18)
switch, VD is the dc component of the voltage across the ideal
diode and is equal to Va,ID is the dc component of the current Likewise, the EAR of r T1 is
through the diode Dl equal to the dc output current 10, and II (19)
is the dc component of the input current.
The total instantaneous quantities of the converter can be The diode is ON for the interva1 DT < t ::; T during which its
presented as the sum of the dc and ae components current is i 0 '" nlv for DT < t ::; T. The EAR of the diode
forward resistances RF is obtained in an analogous manner and
+ V (3)
v[ = Vi i is given by
il = 1[ + ii (4) (20)
rF = n2(1- D)RF •
V Vs
s= + VS (S)
Similarly, the EAR of rn is
d·mT=D +d (6)
rna" = n2(1 - D)rT2' (21)
iD = ID + id (7)
The same approximation of the EAR's is achieved by determin
V() = Vo + Vo' (8) ing the RMS values of the currents through the resistances using
The magnitudes of the ac components are assumed to be much trapezoidal switch and diode current waveforms.
lower than the dc components. Connecting the EAR's in series with the magnetizing induc
From (1 ) and (2), tance and transferring VF to the primary of the ideal trans
former leads to an equivalent circuit of Fig. l(d).
(9) Since the ac components were assumed to be much smaller
than the dc components, a linear circuit model of the converter
. _ ' _
n (t - dTOT ) .
for dc and small-signal operation can be obtained by neglecting
'D - MI'I - 1[. (10) the products vod and iid in (12) and (14). This model is shown irt
dTOT Fig. 2(a). Since the voltage across the filter capacitor is constant
Substitution of (S), (6), and (8) into (9) yields during the time interva1 (1- D)T, the avcrage inductance cur
rent IL flows through the parallel combination of rc and R
(D + d)(Vs + vs )= n(l - D - d)(Va + vol (11)
during the time interval DT, resulting in the resistance con
which can be reaTTanged using (1) to the form nected in series with the diode
1 n (t - D)
Vs rR = D(rcIIR). (22)
V� + + I5vsd = Va
D
Hence, the EAR of rR is
n(l - D) nVo n
Vo - �-d - -v d. (12)
+
D D2 D U
rRa" = n2D(1 - D) rR' (23)
Likewise, substituting (4), (0), and (7) into (10), one obtains Therefore, the overall EAR in series with the magnetizing
inductancc L is
(D + d)(ID + id) = n (l - D - d)(I[ + i;) (13)
r = rL + rsav + rna" + rF + rna" + rRa"
from which (and (2»
= rL + D( rDS + rn) + nZ(1- D)(RF+ r T2)
. 1. n(l - D) n(1 -D) . nIl n .
ID+ld+I5'dd= 1[+ li- 2d-I5lid.
D D D + n2D(1-D)(rJR). (24)
(14) Bccause the model depicted in Fig. 2(a) is linear, the principle of
690 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS-I: FUNDAMENTAL THEORY AND APPLICATIONS, VOL 39, NO, 8, AUGUST 1992
913
613
10+ io
� 313
t-.�
" ,
13
(a)
-313
.131 .I Ie lee
f(kHz)
(a)
13
-45
R
-913
-135
(b) 's .. ..
-Ise
, .f-I--
-225 ,
-Z7�
.el . I Ie llim
f(kHz)
Y, Y. (b)
Accordingly, the control law is d = (1- DXiJlj). The small-signal dynamic characteristics of the flyback con
The dc model of Fig. 2(b) leads to the dc voltage transfer verter can be derived using the small-signal model of Fig. 2(c).
function They are valid for frequencies I up to about one-half the
switching frequency Is' They will be plotted for the converter
Va nRD(l - D)
( v:)
MVDC"'y =
designed in the Appendix.
v: . The control-to-output (or duty ratio-to-output) transfer function
j rD[n(l- D) + D] + R[n(l- D)]2 1 +
in the s-domain is
I
(25)
v (s)
The efficiency of the converter is found from (2) and (25) as T (s )-
= o
P des) Vi(s)�O [n(l - D) + D](1- D)(R + rc )
Vola VOID n(l - D ) MVDC
T/ '" v:J = v:J = MVDCMj = D
j j j j
(5 + w,c) (s + wll)
R[n(l - D)f (27)
s2 + 2g,w,s + w,2
( �) .
(26)
rD[n(l- D) + D] + R[n(l - D)]2 1 + where
61<!
/1
1 .c::T+
�{ : I
:!;+---i, � 31<! V- t-
If: , �
Q;'f----i V I'--
'�8 Di h"
I<!
j/\ l
;' - V
....�"'.�
.
04D\\ -3�
-�r- \' .0001 .O�l .01 .1 10 100
D J(kHz)
(a) (a)
�
r--
-45
I'--
-91<!
..
-
-135 "" -
� -. -
-lSI<!
-ZZ5
\" :
.
,
. ,
-Z71<!
\.\:
\ j(kHz)
D (b)
(b) Fig. 5. C�ntrol-to-out�ut transfer �nction Tp � I 1J, lej<h for n_� 5, L �
Fig,4, Control-to-output transfer function � n � 5, L �Tp I"lp lei<l> for 1 mH, C - 47 /LF, R - 14 n, TL - 2 U, TDS - _ U.S n, 'n - 50 mn,
Tl2 � 10 mn, RF � 25 mn, 'c � SO mn, and D � 0.91965 (solid line),
1 mH, C � 47 /LF, R � 14 n, 'L � 2 n, 'os � u.5 n, rTl � 50 mn,
0.9197, and 0.91975. (a) IT l against f. (b) '" against J.
'n � lO mn, RF � 25 mn, 'c � 50 mn, and D � 0.1 (solid line), 0.5, p
and 0.9. (a) ITpl
against J. (b) '" against J.
n2(l - D)2Rre } I
{
v
C r(R + rd +
D[n(l D) + DJ
+ L M (s) = o(s)
v
}
_
vieS) d(s)�O
(30)
n2(1 - D)2R
2 LC(R + rd r
{ + [(
D n 1 - D) + DJ
n(l - D) Rrc
0.5, and 0,9, The zero w" is located either in the right half of the (33)
s-plane or the left half of the s-plane, depending upon the value
of D for fixed resistances in the circuit. Therefore, the boost
where
converter is a nonminimum phase system. As D increases from
0.9 to 1, the zero Wt/ is moved from the right half to thc left half
of s-plane. Tp has w" = 0 at D "" 0.9197. The behavior of Tp (34)
when w" crosses zero is illustrated in Fig. 5 for D 0.91965, =
180°. Close examination reveals that, neglecting VF in (25), for D = 0.1, 0.5, and 0.9.
692 IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS-I: FUNDAMENTAL THEORY AND APPLICATIONS. VOL. 39, NO.8, AUGUST 1992
101
1101101101101
-3101
=
11011l1l
iii' s:
::::.
,
� -6101 11011l
-',
,
1101 II ---
-9101
f(kHz) f(kHz)
(a) (a)
'- 0
-9101
" 13 -
-. � I-""
�
-135
-45
f'..
t--t--'
-18101
-913
.1011 .1 1101 IBIl 113 11313
.IH .1
f(kHz) f(kHz)
(b) (b)
Fig. 6. Input-to-output transfer function M" � IM"Ie'<PM for n � 5, Fig. 7. Open-loop input impedance Zi � IZileHzi for n � 5, L � 1
y
L � 1 mH, C � 47 p,F, R � 14 fl, 'L = 2 fl, 'DS = 0.5, fl, 'Tl = 50 mH, C � 47 p,F, R � 14 fl, 'L � 2 fl, 'DS � 0.5 0, 'Tl � 50 mfl,
mfl, 'T2 � 10 mfl, RF = 25 mH, 'c = 50 mfl, and D � 0.1 (solid 'T2 = 10 mO, Rp = 25 mO, 'c = 50 mO, and D = 0.1 (solid line), 0.5,
line), 0.5, and 0.9. (a) 1M) against f. (b) <PMv against f. and 0.9. (a) IZ;lagainst f. (b) <PZi against f.
APPENDIX
�
'I A power stage of a PWM flyback converter shown in Fig. l(a)
�o will be designed to meet the following specifications: continuous
=
2
conduction mode, v;. 240 to 300 V, Va
= = 28 V, 10 0.2 to 2
A, Is =
200 kHz, v,./Vo S; 1%, where v,. is the maximum peak
C " I" to-peak value of the ripple voltage on the dc output voltage.
iii The minimum and maximum values of the load resistance are
• iii1 .1 1 iii Rmin = Vo/ lo max = 1 4 VO/IO min = 140 n, re
=
nand Rm ..
nRmax 2
91i1 Lmin = -- (1 - Dmin) = 809.2 IA-H; let L = 1 mHo (AI)
2fs
0
peak diode current IDJlfmax 10 maJO - Dmax) 3.23 A. As
=
=
iii
0 suming the ESR of the filter capacitor re 50 mn, the maxi
=
"
--s- mum value of thc peak-to-peak voltage across the ESR is
"
-'IS v,.c = rcIDMmax = 50 x 3.23 = 161 mY. Thus, the maximum
permissible value of the peak-to-peak voltage across the filter
capacitance C is Vema x =
v,. - v,.c = 280 - 161 = 119 mY. The
-91i1
minimum value of the filter capacitance is found as
• iii1 .1 1 iii 11i11a
l()max Dmax
!(kHz) Cmm �
= 31.9IA-F; let C = 47 1A- F. (A2)
fsVcmax
(b)