Professional Documents
Culture Documents
constant load voltage over the entire battery voltage range, the
converter must be able to work in both buck and boost modes.
The DC-DC converters that meet this operational requirement
{ dx / dt = A2 x + B2 u
y = C2 x + E2 u
(2)
To find the averaged behavior of the converter over one
are Buck-boost, Cuk, and SEPIC converters. However, the switching period, (1) and (2) are weighed average by the duty
Buck-boost and Cuk converters, in their basic form, produce the cycle:
{
output voltage, whose polarity is reversed from the input voltage. d x / dt = As x + Bs u
The problem can be corrected by incorporating an isolation (3)
transformer into the circuits, but this will inevitably lead to the y = Cs x + Es u
increased size and cost of the converters. On the other hands, the where As = A1 d + A2 (1 − d ) , Bs = B1 d + B 2 (1 − d ) ,
SEPIC (Single-Ended Primary Inductor Converter) converter is Cs = C1 d + C2 (1 − d ) , and Es = E1 d + E2 (1 − d ) .
capable of operating in both step-up and step-down modes and Equation (3) is a nonlinear continuous-time equation. It can be
does not suffer from the polarity reversal problem. It is therefore linearized by small-signal perturbation with x = X + x̃, y = Y + ỹ,
attractive for the aforementioned application. u = U + ũ, and d = D + d̃, where the symbol represents a small-
The SEPIC converter consists of an active power switch, a signal value and the capital letter a DC value. It should be noted
diode, two inductors, and two capacitors and is thus a fourth- that X >> x̃, Y >> ỹ, U >> ũ, and D >> d̃. The perturbation
order nonlinear system. Feedback control is usually incorporated yields the steady-state and linear small-signal state-space
into the converter’s circuit to regulate its output voltage, equations in (4) and (5) respectively.
typically by means of Pulse Width Modulation (PWM). To
facilitate the feedback controller design or system stability { dX / dt = AX + BU = 0
Y = CX + EU
(4)
{
analysis, the linear model of the converter is needed. Recently,
modeling of the SEPIC converter has been carried out by some + B d d
dx / dt = Ax + Bu
(5)
researchers. The linear converter models were found by y = Cx + Eu + Ed d
(7) greater than one-half of their ripple components, ΔiL1 and ΔiL2,
⎨
⎩y (s ) = [C( sΙ − A ) B + E C( sI − A ) B d + E d ][u ( s ) d( s )]
−1 −1 T for the circuit to remain in CCM. It can be shown that for CCM
operation L1 and L2 must satisfy the following conditions:
In the DC-DC converters, the input variable u usually contains
⎧ (1− D)2 R 1 1 RE MrC1 rL 2
the input voltage and load current. Hence, u is express as u = [u1 ⎪L1 > 2 Df ( r + + + )
u2]T, the matrix B as B = [Bu1 Bu2], and the matrix E as E = [Eu1 1− D R R R
⎪⎪ C2
+1
Eu2]. Therefore, (7) becomes: ⎨ R (11)
⎧⎪x(s) = ⎡⎣(sI − A) Bu1 (sI − A) Bu2 (sI − A) Bd⎤⎦[u1(s) u2(s) d(s)]
−1 −1 −1 T ⎪L2 > (1− D) R ( 1 + 1 rL 2 + RE + MrC1 − M rC1 + rL 2 )
⎪ 2f rC 2 1− D R R R
⎨ T (8) +1
⎪⎩y(s) = ⎣⎡C(sI − A) Bu1 + Eu1 C(sI − A) Bu2 + Eu2 C(sI − A) Bd + Ed⎤⎦[u1(s) u2(s) d(s)]
−1 −1 −1
⎪⎩ R
For the fourth-order converter, (sI–A)–1Bu1, (sI–A)–1Bu2, and where RE = rC 2 R /(rC 2 + R ) .
(sI–A)–1Bd are the matrices that have four rows and one column.
So, (8) can be expanded into: A. State-Space Description of SEPIC Converter
The state-space equations of the SEPIC converter for the on
⎧ ⎡G vi ( s ) G zi ( s ) G di ( s ) ⎤
1 1 1 and off states of the switch can be written from Figs. 1(b) and (c)
⎪⎪x ( s ) = ⎢G vi ( s ) G zi ( s ) G di ( s ) ⎥[u ( s ) u ( s ) d( s )]T
2 2 2
respectively, which are given by:
⎨ ⎢G vv ( s ) G zv ( s ) G dv ( s ) ⎥ 1 2
(9) diL1 −iL1
=
Ri v R v R I
[r + (1− δ )(rC1 + RE )] − (1− δ )( E L 2 + C1 + E C 2 − E Z ) +
Vg
⎢⎣G vv ( s ) G zv ( s ) G dv ( s )⎥⎦ dt L1 L1 L1 L1 rC 2 L1 L1 L1
1 1 1
⎪ 2 2 2
diL 2 I i v i v
= (1−δ )( Z − L1 − C 2 ) RE − L 2 [rC1δ + rL 2 + (1−δ ) RE ] + C1 δ
⎪⎩y ( s ) = [G vv ( s ) G zv ( s ) G dv ( s )][u1 ( s ) u2 ( s ) d( s )]T dt
dvC1 iL1
L2 L2 rC 2 L2
i
L2 L2
(12)
= (1− δ ) − L 2 δ
where dt C1 C1
−1 −1 −1 dvC 2 1− δ RE R
G vi ( s ) = [( sI − A ) Bu1 ]11 , Gzi ( s ) = [( sI − A ) Bu2 ]11 , Gdi ( s ) = [( sI − A ) Bd ]11 , = (i + i ) R − v − E I
1
−1
1
−1
1
−1 dt C2 rC 2 L1 L 2 E C2 rC 2 R C 2 C2 rC 2 Z
G vi ( s ) = [( sI − A ) Bu1 ]21 , Gzi ( s ) = [( sI − A ) Bu2 ]21 , Gdi ( s ) = [( sI − A ) Bd ]21 , R
2
−1
2
−1
2
−1 vO = (1− δ )(iL1 + iL 2 ) RE. + E vC 2 − RE I Z
G vv ( s ) = [( sI − A ) Bu1 ]31 , Gzv ( s ) = [( sI − A ) Bu2 ]31 , Gdv ( s ) = [( sI − A ) Bd ]31 , rC 2
1 1 1
−1 −1 −1
G vv ( s ) = [( sI − A ) Bu1 ]41 , Gzv ( s ) = [( sI − A ) Bu2 ]41 , Gdv ( s ) = [( sI − A ) Bd ]41 , C1 rC1 iC1
2
−1
2
−1
2 iL1 L1 D
G vv ( s ) = C( sI − A ) Bu1 + Eu1 , Gzv ( s ) = C( sI − A ) Bu2 + Eu2 , and _ iC 2
+ rL 2
−1
.
+
Gdv ( s ) = C( sI − A ) Bd + Ed
Vg Q C2 V R
L2 O
III. MODELING OF SEPIC CONVERTER BY SSA TECHNIQUE rC 2
iL 2
A SEPIC converter is shown in Fig. 1(a). It is comprised of (a). SEPIC converter.
the MOSFET switch (Q), diode (D), two capacitors (C1 and C2), iL1 L1 rL1 C1 rC1 iC1
and two inductors (L1 and L2). The resistor, R, represents a _ iC 2
+ rL 2
+
standing load, and the current source, IZ, models the load current. C2 V R
The resistors, rC1, rC2, rL1, and rL2, are equivalent series Vg L2 O
rC 2
resistances (ESRs) of the capacitors and inductors respectively. iL 2
Their values are usually very small compared to R. In the ideal (b) SEPIC converter during the first state d1T.
converter, these ESRs are zero. In CCM, the converter exhibits iL1 L1 rL1 C1 rC1 iC1
two circuit states. The first state is when Q is turned on (Fig. _ iC 2
+ rL 2
+
Ed = −rC 2 ( MVg + RIZ )/[η R(1− D)] (18) The output impedance transfer function
B. Steady-State Equations G zv(s) = vO (s)/iZ (s) = C(sI − A)−1Bu2 +Eu2
Given the averaged matrices in (13) to (18), the steady-state = R(azvs4 + bzvs3 + czvs2 + dzvs + ezv )/ Δ (23)
solution of converter is obtained from (6): where Δ = as4 + bs3 + cs2 + ds + e
⎧ ⎡rC 2M /(RE R) (rC 2/)RE ⎤ The coefficients in (21) to (23) are listed in TABLE I.
⎪⎡I L1 ⎤ ⎢ ⎥
⎢
⎪ I L2 ⎥ ⎢ rC 2 /( R E R) ( rC 2 /MR E ) ⎥ IV. RESULTS
⎢ ⎥ M
= ⎢ rC 2
⎪ 1 rL2 + R ) −1 (rL2 − MrL1)rC 2 ⎥
⎢V
⎨ C1 ⎥ η (r
⎢ R R C1 M 1− D+ ⎥ (19) To validate the accuracy of the SEPIC converter model, Gdv(s)
⎢ ⎥ MRE
⎪V ⎢ E
⎥ in (21) is plotted and compared with the results from the other
⎢
⎣ C2⎦⎥
⎪ ⎣⎢rC 2 / RE − (rC1 + MrL1 + rL2 / M )rC 2 / RE − rC 2 /(1− D)⎦⎥ two methods, PWM switch model [1, 2] and averaged switch
⎪
⎩VO = [Vg +[rL2rC 2R /[MRE (1− D)]− RE − rC1 − rL1M ]IZ ]MrC 2 /(η RE ) model [3, 4]. They are shown in Fig. 3(a) to Fig. 3(c)
respectively. The following circuit parameters are used to create
where RE = rC 2 R /(rC 2 + R), LT = L1 + L2 , RT = R + rC 2 , M = D /(1− D) = D / D'
these plots: L1 = 100μH, rL1 = 1mΩ, L2 = 100μH, rL2 = 1mΩ, C1
, η =1+ n + r /[(1− D) R], and n = (1+ r / R)(r / R + Mr / R + M 2 r / R).
C2 C2 L2 C1 L1
= 680μF, rC1 = 3mΩ, C2 = 2200μF, rC2 = 1mΩ, Vg = 10V, VO =
Note that if rC1, rC2, rL1, and rL2 are assumed to be zero, the 15V, IZ = 0, and R = 1Ω. The plotted Gdv(s) in Fig 3(a) has four
equation (19) will be reduced to M=VO/Vg=D/(1-D), the same as complex poles at p1,2 = (–0.0158±2.7771i)x103 and p3,4 = (–
the expression for the ideal SEPIC converter in (10). 0.2403±1.1578i)x103; and two real zeros at z1 = –4.5455x105, z2
= 0.0510x105 and two complex zeros at z3,4 =
C. Linear Small-Signal State-Space Equations
(0.0010±0.0277i)x105. It can be seen from Fig. 3(a) to Fig. 3(c)
Given the averaged matrices (13) to (18), the linear small-
that Gdv(s) derived by the SSA technique is exactly the same as
signal state-space equations of the SEPIC converter can be
the other two methods, validating the accuracy of the model
formulated in accordance with (5):
derived by the SSA technique.
TABLE I [3] A. Hren and P. Slibar, "Full Order Dynamic Model of SEPIC Converter,"
COEFFICIENTS OF Gdv(s), Gzv(s), AND Gvv(s). Proc. of the IEEE International Symposium on Industrial Electronics, pp.
553-558, June 2005.
adv =−rC 2 L1C1 L2C2 RT 2 ( DVg + RD'IZ ), [4] R. W. Erickson and D. Maksimović, Fundamentals of Power Electronics,
bdv =[ D'2 rC 2 LT C2 RT − DL2 (rL1rC 2C2 + L1 ) − rC 2C2 DL1 ( DrC1 + rL 2 )]RT2CV 2nd ed., Kluwer Academic Publishers, 2001.
1 g
[5] R. D. Middlebook and S. Cuk, "A General Unified Approach to Modeling
− D'C1 RRT IZ [rC 2C2 LT [ RT ( DrL1 + rC1 ) + rC 2 RD ']
Switching-Converter Power Stages," International Journal of Electronics,
+ RT [ L1 (rC 2 rL 2C2 + L2 ) + rC 2 L2C2 (rL1 − DrC1 )]],
vol. 42, pp. 521-550, June 1977.
cdv = RT2Vg [ D '2 [C1 (rL 2 + R)[C2 ( rL 2 rC 2 + rC 2 rC 1 + rL1rC 2 ) + LT ]
− D2 rC 2C2 L1 ]− C1D ( DrC1 + rL 2 )( L1 + rL1rC 2C2 ) + DL2 rL1 ] −
D 'RT2 RIZ [C1[ D 'rC1 + (1+ D )rL1 ][rC 2C2 (rL 2 + rC1 ) + L2 ]+ L1C1 (rL 2 + rC1 ) +
Magnitude (dB)
DL1 (rL1C1 + rC 2C2 ) + DrL21rC 2C1C2 + D 'C1 RE [ LT + (rL 2 + rC 1 + rL1 )C2 rC 2 ]],
ddv = − D 'RRT I Z [ RT [ D( L1 − rC1rL 2C1 + 2rC 2 rL1C2 ) + C1rL 2 (rL1 + rC1 )]
+ (rC 2 RD'+ rL1 RT D)( rL1 + rL 2 )C1 + (rC1 RT + rC 2 R )(rC1C1 + rC 2C2 ) D ']
+ RT2Vg [ D '2 [C1 R( rL1 + rL 2 ) + rL 2C1 ( rL 2 − rL1 ) + ( rC1C1 + C2 rC 2 )( R + rL 2 )]
− rL1 D2 (rC1C1 + rC 2C2 ) − L1 D2 − DrL 2 rL1C1 ],
Phase (deg)
edv =Vg RT2 [ D'2 ( R + rL 2 ) − D2rL1 ]− D'RT RIZ [2DRT rL1 + D'(rC1RT + rC 2 R)].
avv = rC 2 L2C1C2 RT , bvv = C1 RT [rC 2C2 ( DrC1 + rL 2 ) + L2 ], dvv = DRT ,
cvv = RT [( rC1C1 + rC 2C2 ) D + rL 2C1 ].
azv = − rC 2 L1C1L2C2 RT , ezv =−RT ( D2rL1 − D'2 rL 2 ) − DD'(rC1RT + rC 2 R),
bzv = C1 L1 L2 − rC 2C1C2 [rC 2 RLT DD'+ RT ( L2 rL1 + L1rL 2 + rC1 L1 D + rC1 L2 D')], Frequency (Hz)
czv =−rC 2C1 RDD '[ LT + C2 (rL1rC 2 + rC1rC 2 + rL 2 )]− RT [rC1C1 ( L1D + L2 D ') + (a)
rC 2C2 ( L1 D2 + L2 D '2 )]− C1 RT [rC 2C2 ( DrC1 + rL 2 )(rL1 + rC1D ') + rL1 L2 + rL 2 L1 ],
Magnitude (dB)
dzv =−(C1rC1rC 2 + C1rC 2 rL 2 + C2 rC22 ) RDD '− RT [( D 'rC1 + rL1 )( DrC1 + rL 2 )C1
+ L1D2 + L2 D '2 ] − (rL1C1RD + rC1C2 DD'+ rL1C2 D2 + rL 2C2 D '2 )rC 2 RT .
1 1 L2C2 RT ,
a = LC 2
e = RT (D2rL1 + D'2 rL 2 + DD'rC1 ) + (D'R + rC 2 )D'R,
b =C1RT [C2 RT [L2rL1 + L1 (rL2 + DrC1 )]+C2 D'(rC1L2 RT + rC 2 LT R)+ L1L2 ],
c = C1 RT [ D'rC 2C2 R(rL1 + rL 2 + rC1 ) + rC1D'L2 + L1 ( DrC1 + rL 2 ) + rL1 L2 ]+
RT2C2 [C1 ( DrC1 + rL 2 )(rL1 + rC1 D') + D '2 L2 + L1 D2 ]+ ( RD'+ rC 2 ) D 'RC1 LT ,
Phase (deg)
V. CONCLUSION
Frequency (Hz)
In this paper, modeling of a SEPIC converter in CCM has (b)
been performed with the State-Space Averaging (SSA)
technique. The results yield an insight into the steady-state and
Magnitude (dB)