You are on page 1of 72

Modeling approaches for switching

converters

by

Giorgio Spiazzi

University of Padova – ITALY


Dept. of Information Engineering – DEI

Modeling approaches for switching converters 1/72


Summary of the presentation
PWM converters
• Switching cell average model in continuous conduction
mode (CCM)
• Switching cell average model in discontinuous
conduction mode (DCM): first-order model

Modeling approaches for switching converters 2/72


Basic DC-DC Converter topologies: 2°order
is iL Buck
+ S L +
ug iD D
uo Ro
- iL iD Boost
+ L D +
ug iS S uo Ro
is iD -
S -
+ D
uo Ro Buck-Boost
ug iL L
+

Modeling approaches for switching converters 3/72


Basic DC-DC Converter topologies: 4°order

i1 uC i2
+
L1 + C1 - L2 -
+ Cuk
ug Co uo Ro
uS S D uD
- + +

i1 uC D iD
+
+ L1 + C1 -u+ +
D
SEPIC ug uS S i2 L2 Co uo Ro
- -

Modeling approaches for switching converters 4/72


Commutation Cell for 2°order converters
is iD
n
a p
2° order converters can be + D
-
S
described by a unique iL L U
Uon off
commutation cell: - +
c

Buck Boost Buck-boost

Uon Ug-Uo Ug Ug
Uoff Uo Uo-Ug Uo
Uon+ Uoff Ug Uo Ug+Uo
ig iS iL iS
io iL iD iD

Modeling approaches for switching converters 5/72


Averaging
t
1
Moving average: ( )
xt = ∫ x(τ ) dτ
Ts t −T
s

Example: instantaneous and average inductor


current in transient condition
[A]

11 iL
10

9
iL

5
2.8 2.9 3 3.1 [ms]

Modeling approaches for switching converters 6/72


Average model: CCM
• Switching frequency ripples are neglected
• Only low-frequency dynamic is investigated

diL (t )
Example: inductors uL (t ) = L
dt

i (t )
iL (t ) − iL (t − TS )
t
1 L L
uL (t ) = ∫ uL (τ)dτ = ∫ diL = L  
TS t −T TS i (t −T )  T S 
S L S

Modeling approaches for switching converters 7/72


Average model: CCM
d iL (t ) d  1 
t
=  ∫ iL (τ )dτ = ?
dt dt  TS t −T 
 S

β( t ) z
φ(t ) = ∫ f (t, τ)dτ = φ(t, y, z ) = ∫ f (t, τ)dτ with y = α(t ), z = β(t )
α (t ) y
β( t )
dφ(t ) df (t, τ) dα(t ) dβ(t )
= ∫ dτ − f (t, α(t )) + f (t,β(t ))
dt α (t )
dt dt dt

d iL (t ) iL (t ) − iL (t − TS ) d iL (t )
= uL (t ) = L
dt TS dt

Modeling approaches for switching converters 8/72


Averaging approximation

Non steady-state Uon


inductor current iL (dTs ) = iL (0 ) + dTs
L
waveform:
Uoff
uL(t) Uon
iL (Ts ) = iL (dTs ) − (1 − d)Ts
uL L
t
-Uoff
Uoff
− m2 = −
iL(t) L
Uon Uoff
iL(0) m1 =
Uon
L
iL(Ts) iL (Ts ) = iL (0 ) + dTs − (1 − d)Ts
L L
t
uL
dTs (1-d)Ts = iL (0 ) + Ts
L

Modeling approaches for switching converters 9/72


Averaging

• Reactive element voltage-current relations


remain valid also for average quantities;
• for inductors, the current variation in a switching
period can be calculated by integrating their
average voltage;
• for capacitors, the voltage variation in a
switching period can be calculated by integrating
their average current.

Modeling approaches for switching converters 10/72


Continuous conduction mode - CCM
uL(t) Uon
At steady-state: uL = 0
Uon 1 − d
Uon dTS = Uoff (1 − d)TS =
t
-Uoff
Uoff
Uoff d
− m2 = −
iL(t)
M=d
L
iL Buck:
U
m1 = on
L
1
t Boost: M=
is(t) (1-d)Ts 1− d
i S = iLd
d
Buck-Boost: M=
t 1− d
id(t)
Boundary CCM-DCM:
dTs iD = iL (1 − d)
∆iLpp Uon Uoff
Ts
t
iL lim = = d= (1 − d)
2 2LfS 2LfS
Modeling approaches for switching converters 11/72
Switching cell average model: CCM
iS iD iS iD
n
a p
+ S D - + D -
iL L u S uD
uon off uS
- + - +
c
Non linear components

Average switch and diode voltages and currents:

uS = d′(uon + uoff ) d′ iS = d i L



d
iS = iD
 ⇒ uS = uD 
uD = d(uon + uoff ) d iD = d′ i L d′

d’=1-d = complement of duty-cycle

Modeling approaches for switching converters 12/72


Switching cell average model: CCM

uS = d′(uon + uoff ) d′ iS = d i L d


 ⇒ uS = uD  ⇒ iS = iD
uD = d(uon + uoff ) d iD = d′ i L d′

iS iD iS iD
d’:d
+ - iS d′ i D + -
D iS
uS S uD + + d - uS uD
- + uS uD - +
d′
- uD +
d
d’=1-d = complement of duty-cycle

Modeling approaches for switching converters 13/72


Switching cell average model: CCM

• The non-linear components (switch and diode)


are replaced by controlled voltage and current
generators representing the relations between
average voltage and currents;
• These controlled voltage and current generators
can be substituted by an ideal transformer with a
suitable equivalent turn ratio.

Modeling approaches for switching converters 14/72


Buck average model: CCM
iS iL

+ S L +
uug iD D
g uo Ro
-
iL d′
uD = ug − uS = ug − uD
L + d
- +
uS uD = d ug
uD
iS iD Ro
+ - C
iD = (iL − iS )
uo d d
d’:d iS =
+ d′ d′
ug
-
iS = d iL
Modeling approaches for switching converters 15/72
Buck average model (alternative approach): CCM
iS iL

+ S L +
ug iD D
uo Ro
-

ig iL
Switching + S + Independent variables: ug, iL
cell ug uD Dependent variables: uD, ig
D
- -
Modeling approaches for switching converters 16/72
Buck average model (alternative approach): CCM
ig iL ig iL
Averaging d iL +
+ S + + +
ug uD ug uD
D d ug
- - - -

ig iL
1:d
+ +
ug uD
- -

Modeling approaches for switching converters 17/72


Buck average model: CCM
d iL iL
+ + L +
ug d ug C u Ro
o
- -

 d iL
L = uL = d ug − uo
dt
 du u
C C = iC = i L − o
 dt Ro

Modeling approaches for switching converters 18/72


Boost average model: CCM
iL iD

+ L D +
ug iS S uo Ro
-
iL d
uS = uo − uD = uo − uS
L d′
+ -
uS uS = d′ uo
+
ug iS i D uD
- + d′ d′
d’:d + iD = iS = (iL − iD )
d d
C u Ro
o
-
iD = d′ iL
Modeling approaches for switching converters 19/72
Boost average model (alternative approach): CCM
iL iD

+ L D +
ug iS S uo Ro
-

iL iD
Switching + D + Independent variables: uo, iL
cell uS S uo
Dependent variables: uS, iD
- -

Modeling approaches for switching converters 20/72


Boost average model (alternative approach): CCM
iL iD iL
+ d′ i L +
+ D +
uS S uo
uo d′ uo
- - -

iL iD
d’:1
+ +
uS uo
- -
Modeling approaches for switching converters 21/72
Boost average model: CCM
iL d’:1 d′ i L
+ L + +
ug d′ uo C Ro
uo
- -

 d iL
L = uL = ug − d′ uo
dt
 du u
C C = iC = d′ i L − o
 dt Ro

Modeling approaches for switching converters 22/72


Buck-Boost average model: CCM
iS iD
S -
+ D
ug iL L
uo Ro
+
iS 1
d’:d i D ug = uS + uD − uo = uD − uo
+ - - d
uS uD
+ + uD = d(ug + uo ) uS = d′(ug + uo )
ug - Ro
d′ d′
iD = iS = (iL − iD )
C
uo
d d
iL L
+ iD = d′ iL iS = d iL
Modeling approaches for switching converters 23/72
Buck-Boost average model: CCM
uD = d(ug + uo )
iS
d’:d i D
+ - -  d iL
uD L = uL = uD − uo = d ug − d′ uo
uS dt
+  du u
ug - + C C = iC = d′ i L − o
Ro
C
uo  dt Ro
iL L
+ iL
iS iD
+ + L + +
ug d′ iL C uo Ro
d ug d′ uo
d iL -
Modeling approaches for switching converters 24/72
Buck-Boost equivalent average model: CCM
iS iL iD
+ + L + +
ug d′ iL C uo Ro
d ug d′ uo
d iL -

} Buck

iS iL
1:d d’:1 i D
+ + L +
ug
C uo Ro
-
{
Modeling approaches for switching converters
Boost 25/72
Cuk average model: CCM
i1 uC i2
+
L1 + C1 - L2 -
+
ug Co uo Ro
uS S D uD
- + +

Switching cell
iS iD uS = d′ uC d′
 ⇒ uS = uD
+ - uD = d uC d
D
iS = d(i1 + i2 )
uS S uD
d
- +  ⇒ iS = iD
iD = d′(i1 + i2 ) d′

Modeling approaches for switching converters 26/72


Cuk average model: CCM

+ uC
i1 C1 i2
+ L1 + - L2 -
ug uS uD Co u Ro
o
 d i1 - + +
L1 = ug − d′ uC d’:d
 dt
L d i 2 = d u − u
 2 dt C o
 du
C1 C = d′ i1 − d i 2
 dt
 d uo uo
Co = i2 −
 dt Ro
Modeling approaches for switching converters 27/72
Cuk average model: CCM

i1 i2
L1 d′ i1 L2 +
+ + + uC +
ug d uC
C1 Co uo Ro
d′ uC di2 -
Boost
}
i1 d′ i1 d i 2 i2
+ L1 L2 +
+ uC
ug C1 Co uo Ro
-
d’:1 1:d
{
Modeling approaches for switching converters Buck 28/72
SEPIC average model: CCM
i1 uC D iD
+
+ L1 + C1 -u+ +
D
ug
uS S i2 L2 Co uo Ro
- -

Switching cell
is iD uS = d′(uC + uo ) d′
 ⇒ uS = uD
+ - uD = d(uC + uo ) d
D
iS = d(i1 + i2 )
uS S uD
d
- +  ⇒ iS = iD
iD = d′(i1 + i2 ) d′

Modeling approaches for switching converters 29/72


SEPIC average model: CCM
i1 uC iD
+
+ L1 C1 +
ug L uo
i2 2 Co Ro
-
 d i1 + -
L1 = ug − d′(uC + uo ) uD
uS
 dt
L d i 2 = d u − d′ u - +
 2 dt C o d’:d
 du
C1 C = d′ i1 − d i 2
 dt
 d uo
C
 o = d ′(i 1 + i 2 ) −
uo
 dt Ro
Modeling approaches for switching converters 30/72
SEPIC average model: CCM

Alternative approach
i1 i2 iD
d′ i1 di2 d′ i 2 +
L1 + + + L2 + uo
+ d′ uC C1 uC Co Ro
ug + d uC d′ uo d′ i1 -
d′ uo

Modeling approaches for switching converters 31/72


Model perturbation

Generic voltage or current: x = X + x̂


Small-signal approximation: x̂ << X

Product of variables: x ⋅ y ≈ XY + Xŷ + x̂Y

Examples:

d i L = (D + d̂)(IL + îL ) ≈ DIL + D îL + d̂IL

d ug = (D + d̂)(Ug + ûg ) ≈ DUg + Dûg + d̂Ug

Modeling approaches for switching converters 32/72


General switching cell: DC and small-signal model
iS iD
d’:d D(US + ûS ) + d̂US ≈ D′(UD + ûD ) − d̂UD
+ -
uS uD  US + UD  D′
US + ûS + d̂  ≈ (UD + ûD )
- + {  D  D {
uS uD

iD At steady-state:
iS
n
a p uS = US = Uon
+ S D - uL = 0 
iL L u uD = UD = Uoff
uon off

- + US + UD US  UD  US  D  US
c = 1 +  = 1 +  =
D D  US  D  D′  DD′
Modeling approaches for switching converters 33/72
General switching cell: DC and small-signal model
iS iD
D′(IS + îS ) − d̂IS ≈ D(ID + îD ) + d̂ID
d’:d
+ -
D′  IS + ID 
ID + îD ≈ (IS + îS ) − d̂
uS uD

- + { D {  D 
iD iS

US ID
d̂ d̂ i D
iS DD′ DD′
+
+ -
uS uD
- +
D’:D
Modeling approaches for switching converters 34/72
Buck switching cell: DC and small-signal model

ig iL Perturbation and linearization:


1:d
+ +
ig = D i L + IL d̂
ug uD 
- - uD = D ug + Ugd̂

ig Ugd̂ i L
+
+ +
ug uD
- IL d̂ -
1:D

Modeling approaches for switching converters 35/72


Buck DC and small-signal model
ig Ugd̂ i
+ L
+ L +
ug C Ro
uo
IL d̂ -
1:D

Duty-cycle to output voltage transfer function:


Ûo (s ) Ug
Gud (s ) = =
D̂(s ) s s2
1+ + 2
Qωo ωo
1 C
ωo = Q = Ro
LC L

Modeling approaches for switching converters 36/72


Boost switching cell: DC and small-signal model

iL
d’:1
iD Perturbation and linearization:
+ +
uS uo iD = D′ i L − IL d̂

- - uS = D′ uo − Uo d̂

i L Uo d̂ iD
+
+ +
uS uo
- IL d̂ -
D’:1

Modeling approaches for switching converters 37/72


Boost DC and small-signal model
iL Uo d̂ iD
+
+ L +
ug Ro
C uo
IL d̂
-
D’:1

Duty-cycle to output voltage transfer function:


RHP zero
L 2
1− s M
Ro
( )
Gud s = UgM2
L 2
1 + s M + s2LCM2
Ro

Modeling approaches for switching converters 38/72


Boost small-signal model: CCM
Normalized output voltage response to a
duty-cycle step change:
1

0
t
The output voltage initially
moves in the wrong direction
Modeling approaches for switching converters 39/72
Buck-Boost DC and small-signal model

US ID
d̂ d̂ i D
iS DD′ DD′
+
+ - -
uS uD
+ - +
ug D’:D Ro
C
uo
iL L
+

Modeling approaches for switching converters 40/72


Buck-Boost DC and small-signal model

iS Ugd̂ iL Uo d̂ iD
+ +
+ L +
ug
IL d̂ C uo Ro
IL d̂
-
1:D D’:1

Duty-cycle to output voltage transfer function:


RHP zero
L
1− s M(1 + M)
Ro
Gud (s ) = Ug (1 + M)
2
L
1 + s (1 + M) + s LC (1 + M)
2 2 2

Ro

Modeling approaches for switching converters 41/72


Cuk DC and small-signal model
+ uC
i1 C1 i2
+ L1 + - L2 -
ug uS uD Co u Ro
o
- + +
d’:d

uC
i1 + i2
C1
+ L1 US + L2 -
ug d̂ ID
DD′ d̂ Co uo Ro
DD′ +
D’:D
Modeling approaches for switching converters 42/72
Cuk DC and small-signal model

Alternative approach

i1 UC d̂ UC d̂ i2
+ +
+ L1 L2 +
ug +
uC C1 Co uo Ro
I1d̂ I2d̂
-
D’:1 1:D

Modeling approaches for switching converters 43/72


SEPIC DC and small-signal model

i1
uC
+
+ L1 C1 i +
ug 2
L2 Co uo Ro
-
US +
d̂ ID
DD′ d̂
DD′
D’:D

Modeling approaches for switching converters 44/72


Discontinuous conduction mode - DCM
uL(t) At steady-state:
Uon
t uL = 0 uondTS = uoff d′TS
uon d′
-Uoff
Uoff
iL(t)
Uon
− m2 = −
L =
m1 =
L iL uoff d
t
iLpk d2  uon 
is(t) (d’Ts
iL = (d + d ) =
′ uon 1 + 
iS 2 2LfS  uoff 
2
t iLpk d2 uon
iD = d′ =
id(t) 2 2LfS uoff
iD
t iLpk d2
dTs
iS = d= uon
Ts 2 2LfS
Modeling approaches for switching converters 45/72
Discontinuous conduction mode - DCM
Io Ug
uL(t)
Uon
IoN = , IN =
IN 2Lfs
t
-Uoff d2  1  1
U
− m2 = − off Buck: Io = iL = Ug  − 1 M=
iL(t) L
2LfS  M  IoN
m1 =
Uon
L iL 1+ 2
t
d

is(t) (d’Ts d2  1  d2
iS I =
Boost: o D i = Ug  M = 1+
2LfS  M − 1 IoN
t

id(t)
iD
d 2
1 d2
dTs
t Buck- I = i =
o D Ug M=
Boost: 2Lf S M IoN
Ts

Modeling approaches for switching converters 46/72


First order average models - DCM

• The inductor current is always zero at the


beginning of each switching period;
• this loss of the memory effect justifies the
statement that the inductor current is no more a
state variable;
• switch and diode are replaced by non linear
controlled current generators

Modeling approaches for switching converters 47/72


First order average models - DCM
is iD
uL(t) n
Uon a p
t + D
-
S
iL L
-Uoff
U uon uoff
− m2 = − off
iL(t) L
m1 =
Uon - +
L iL c
t

(d’Ts
Inductor average voltage is always
is(t)
iS zero in a switching period! (?)
t
iS iD
id(t) a p
iD
dTs
t iL
Ts
c
Modeling approaches for switching converters 48/72
Buck average model: DCM
uL
+ -
+ S +
L Io
ug D Ro uo
C
- -

io

ug ig
iL C uo

Modeling approaches for switching converters 49/72


Buck small-signal model: DCM

{
d2  ug 
iL = (ug − uo )  = h(ug, uo , d)
Average 2LfS  uo 
quantities: d2
ig = iS = (ug − uo ) = f (ug, uo , d)
2LfS

{
∂f ∂f ∂f
î g = ûg + ûo + d̂ = giûg + gr ûo + k id̂
∂ ug ∂ uo ∂d
Perturbation:
∂h ∂h ∂h
îL = ûg + ûo + d̂ = gf ûg − goûo + k o d̂
∂ ug ∂ uo ∂d
Modeling approaches for switching converters 50/72
Buck small-signal model: DCM
io

ug iS
iL C uo

First order model


îS îL

k id̂ gr ûo
ûg gi go C R ûo
gf ûg k o d̂ o

Modeling approaches for switching converters 51/72


Boost small-signal model: DCM
io d2  uo 
iL = ug  
2LfS  uo − ug 
ug iL
iD C uo
d 2 ug2
iD =
2LfS uo − ug

îL îD

k id̂ gr ûo
ûg gi go C R ûo
gf ûg k o d̂ o

Modeling approaches for switching converters 52/72


Buck-Boost small-signal model: DCM
io
d2 ug2
iD =
ug 2LfS uo
iS uo
iD C
d2
iS = ug
2LfS

îS îD

k id̂ gr ûo
ûg gi go C R ûo
gf ûg k o d̂ o

Modeling approaches for switching converters 53/72


Full order average models: DCM i L (t )
iL + îL

Impulsive perturbation:
+∞
{ } ∫ t̂S (t ) e
t
D̂(s) = L t̂S = − st
dt tS DTs
0
t
1 − e −sd̂Ts
d̂Ts tD D’Ts
− st
= ∫ e dt =
s
≈ d̂Ts
t
0
t S + t̂S
Response to impulsive perturbation: t
tD + t̂D
+∞ D′Ts
{ } ∫ î (t ) e
ÎL (s) = L îL = L
− st
dt = ∆i ∫e
−st
dt t
0 0 t̂S d̂Ts

uon + uoff 1 − e−sD Ts t
= d̂Ts t̂D t=0 d̂′′Ts
L s
t
îL
ÎL (s ) (uon + uoff )D′  1 − e − sD′Ts
 ∆i
Gid (s ) = = 
D̂(s ) Lfs  sD′T
 s

 Ts
t

Modeling approaches for switching converters 54/72


Full order average models: DCM i L (t )
iL + îL

ÎL (s ) ( uon + uoff )D′  1 − e−sD T




Gid (s ) =
s

=  
D̂(s ) Lfs  sD′T
 s

 tS
t
DTs
t
tD D’Ts
 sD′Ts 
 1−  t
First order Padé e − sD′Ts
≈ 2  t S + t̂S
approximation:  1+ s D ′Ts 
  t
 2  tD + t̂D

t
t̂S d̂Ts
 
( uon + uoff )D′  1  t
Gid (s ) ≈   t̂D t=0 d̂′′Ts
Lfs  1+ sD ′Ts 
  t
 2 
îL
2f f ∆i
ωp = s ⇒ fp = s
D′ πD′ t
Ts

Modeling approaches for switching converters 55/72


Full order average models: DCM i L (t )
iL + îL

L∆i  u 
d̂′′Ts = ⇒ d̂′′ = 1 + on d̂
uoff  uoff  t
tS DTs
Overall d’ perturbation: t
tD D’Ts
uon
d̂′ = d̂′′ − d̂ = d̂ WRONG! t
uoff t S + t̂S
t
tD + t̂D
t̂D = −d̂Ts ⋅ δ( t ) + d̂′′Ts ⋅ δ(t − D′Ts )
t
t̂S d̂Ts
Dirac function t
t̂D t=0 d̂′′Ts

t
îL
{ }
D̂′(s ) = L t̂D
 u 
= −D̂(s ) + 1 + on D̂(s) ⋅ e−sD Ts

∆i
 uoff  t
Ts

Modeling approaches for switching converters 56/72


Full order average models: DCM i L (t )
iL + îL

D̂′(s )  u  ′
= −1 + 1 + on e −sD Ts t
D̂(s )  uoff  tS DTs
t
tD D’Ts
Inductor current perturbation:
t
d îL
L = d̂ uon − d̂′ uoff t S + t̂S
dt t
tD + t̂D
sL ÎL (s ) = D̂(s ) uon − D̂′(s )uoff
t
t̂S d̂Ts

t
t̂D t=0 d̂′′Ts

ÎL (s ) ( uon + uoff )D′  1 − e−sD T



 îL
Gid (s ) =
s

=   ∆i
D̂(s ) Lfs  sD′T
 s

 t
Ts

Modeling approaches for switching converters 57/72


Full order average models: DCM
L L n
n c p
c p
iL D iL ip
S
ia
a a

L uD The switch is replaced by a


n +
c p
controlled current generator
iL
while the diode is replaced by a
ia
a controlled voltage generator

Modeling approaches for switching converters 58/72


Full order average models: DCM iL
iLp ia

L uD
n +
c p 0 t
iL
uD
ia uon + uoff
a
uoff
0 dTs (d+d’)Ts Ts t
 1
i L = 2 iLp (d + d )

d
 ia = i L uD = (uon + uoff )d + uoff (1 − d − d′)
1
ia = iLp d d + d′
 2

uon 2Lfs iL
iL = d(d + d′) ⇒ d′ = −d
2Lfs d uon
Modeling approaches for switching converters 59/72
Example: boost in DCM
d iL 1
[ug − uo + uod + (uo − ug )(1 − d − d′)] =
= c
L n
uD
+
p
dt L
iL
 uo  2fs uo d

= 1−  i + ia
 u  d L L
 g a
uD = uo d + (uo − ug )(1 − d − d′)
2
d uo 1  uo  i L d ug u
=  i L − ia −  = − − o
dt C Ro  C 2LCfs CRo 2Lfs iL
d′ = −d
d ug
Duty-cycle perturbation:

 d îL  Uo + ûo  2fs


 =  1 −  I( + î) +
(
(Uo + ûo ) D + d̂ )
 dt  U  D + d̂
L L
L
 g 

 dûo L LI + î D(+ d̂
2
)
Ug Uo + ûo
 dt = C − 2LCf −
CRo
 s

Modeling approaches for switching converters 60/72


Example: boost in DCM
 d îL 2f 2U 2Mfs
 = (1 − M)îL + o d̂ − ûo
 dt D L DRo
Small-signal linear model: 
 dûo = îL − DUg d̂ − ûo
 dt C LCfs CRo

 2fs   s 
 − s 
 1 − 
Û (s) DUg D   ωz 
Gud (s) = o = ≈ KB
D̂(s ) LCfs 2  1 2f (M − 1)  2fs (2M − 1)   
s + s + s  + 1 + s 1 + s 
 R oC D  DRoC  ω  ω 
 pBF  pAF 

2Ug M(M − 1) 2fs 2fs (M − 1) 1  2M − 1


KB = ωz = ωpLF = ωpHF =  
2M − 1 k D D R oC  M − 1 
The same as first order model:

Modeling approaches for switching converters 61/72


Example: boost in DCM

Control-to-output transfer function


[dB] [deg]
Gud ( jω)
-10 -30 ∠Gud ( jω)
Gud (0 ) dB
-20 -60
b)
-30 -90
a)
-40 -120
a)
b)
-50 -150

-60 -180 f 100


10-4 10-3 10-2 10-1 f 100 10-4 10-3 10-2 10-1
fs fs

a) Full order model


b) First order model

Modeling approaches for switching converters 62/72


State-Space averaging (SSA): CCM
 iL  ug  uo 
State, input and output x=  u=  y= 
variable vector: uC   io   ig 
Interval dTs: Interval (1-d)Ts:
x& = A1x + B1u x& = A 2 x + B 2u
 
y = C1x y = C2 x
1 t ∈ t on
Switching function: q( t ) = 
0 t ∈ t off

x& = A 2 x + B 2u + [(A 1 − A 2 ) x + (B1 − B 2 ) u] ⋅ q = A 2 x + B 2u + F ⋅ q



y = C 2 x + (C1 − C 2 ) x ⋅ q = C 2 x + G ⋅ q

Applying moving x& = A 2 x + B 2 u + F ⋅ q


x& = x& 
y = C2 x + G ⋅ q
average operator:

Modeling approaches for switching converters 63/72


State-Space averaging (SSA): CCM
F ⋅ q = ?, G⋅q = ?
Hp: linear ripple approximation Hp: small ripple approximation
iL uC
iL uC

q t q t
q=d q=d
t t
qiL quC
qiL = q i L quC = q u C
0 dTs Ts t 0 dTs Ts t

F⋅q = F ⋅ q = F ⋅d x& = A 2 x + B 2 u + F d = Ax + B u

G⋅q = G ⋅ q = G ⋅d y = C2 x + G d = Cx
A = A 1d + A 2 (1 − d) B = B1d + B 2 (1 − d) C = C1d + C2 (1 − d)
Modeling approaches for switching converters 64/72
State-Space averaging (SSA): CCM
0 = Ax + B u x = − A −1B u
Steady-state solution:  ⇒ 
 y = Cx y = Cx

Small-signal linear model:

x̂& = Ax̂ + Bû + [(A 1 − A 2 )X + (B1 − B 2 )U]d̂ = Ax̂ + Bû + Fd̂



ŷ = Cx̂ + (C1 − C2 )X d̂ = Cx̂ + Gd̂

 X̂(s ) = (sI − A )−1(B Û(s ) + F D̂(s ))



Ŷ (s ) = CX̂(s ) + G D̂(s )

Modeling approaches for switching converters 65/72


Example: boost converter in CCM

 iL  u = ug y = uo
x= 
uC 

Interval dTs: Interval (1-d)Ts:

x& = A1x + B1u x& = A 2 x + B 2u


 
y = C1x y = C2 x
Modeling approaches for switching converters 66/72
Example: boost converter in CCM
 rL   rL + rC // R R 
− L 0  − L

(R + rC )L   1
A1 =  1  A2 =  B1 = B2 = L 
 0 −   R 1   
(R + rC )C  − 0 
  (R + rC )C (R + rC )C 
 R 
C1 = 0   R 
 R + rC  C2 = R // rC
 R + rC 
  IL  Ug  1 
 X = U  = R′ D′R 
Steady-state  C  
x& = 0 ⇒ x = − A −1BUg 
solution: Y = U = UgD′R
 o
R′
R′ = D′2R + rL + DD′(rC // R )
Voltage conversion ratio:
Uo 1 D′2R
M= = ⋅
Ug D′ D′2R + rL + DD′(rC // R )

Modeling approaches for switching converters 67/72


Example: boost converter in CCM

Small-signal linear model:

 rL + D′(rC // R ) D′R  R(D′R + rC )


− −
(R + rC )L   îL    Ug  (R + rC )L 
 &îL   1
L
&  =  ′
   + L ûg +   d̂
ûC  
DR

1  ûC  0  R′  − R 
 (R + rC )C (R + rC )C   (R + rC )C 

 R   îL  rC || R
ŷ = D′(rC || R )    − Ug d̂
 R + rC  ûC  R′

Modeling approaches for switching converters 68/72


State-Space averaging (SSA): DCM
 iL 
State variable vector: x= 
uC 

 3   3  iL iLp Uoff
x& (t ) =  ∑ qk A k  x +  ∑ qkBk u
Uon −
L
L
 k =1   k =1 
d1Ts d2Ts d3Ts t
Applying moving
x& = x& 1
average operator: q1

1
t
q2
&x ≠  ∑ dk A k  x +  ∑ dkBk  u
3 3
t
    1 q3
 k =1   k =1  t

Why?

Modeling approaches for switching converters 69/72


State-Space averaging (SSA): DCM
Hp: linear ripple approximation iL iLp Uoff
Uon −
L
L
iL
Example: iL·q1
d1Ts d2Ts d3Ts t
iLpk
iL = (d1 + d2 ) 1
2 q1
t
iLq1
iLpk d2
q2iL = iD = d2 = iL iL q1 ≠ i L q1
2 d1 + d2
1
= iL q2 Corrective term
d1 + d2
iLpk d1 1
q1iL = iS = d1 = iL = iL q1
2 d1 + d2 d1 + d2
Modeling approaches for switching converters 70/72
State-Space averaging (SSA): DCM
Hp: small ripple approximation

qiuC = qi uC = di uC i = 1,2,3

 3   3  &x =  ∑ dk A k  M x +  ∑ dkBk  u
 3
  3
x& (t ) =  ∑ qk A k  x +  ∑ qkBk u    
 k =1   k =1   k =1   k =1 

M is the correction matrix

 1 
0
M =  d1 + d2
 
 0 1
Modeling approaches for switching converters 71/72
State-Space averaging (SSA): DCM

&x =  ∑ dk A k  M x +  ∑ dkBk  u
  
3 3

   
 k =1   k =1 

 1  2Lfs i L
0 d2 = − d1
M =  d1 + d2 d3 = 1 − d1 − d2
  δ uon
 0 1

If d3 = 0, i.e. in CCM, we have:

1 0 &x =  ∑ dk A k  x +  ∑ dkBk  u
 2
  2
M=      
0 1  k =1   k =1 
Modeling approaches for switching converters 72/72

You might also like