Professional Documents
Culture Documents
CH9-2
CH9-3
CH9-4
variations in element values
Vo/Vo=2%
Gc(s) H( s )
Snubber
PWM IC 3842
CH9-8
Direct duty cycle (voltage mode)
control
carrier
CH9-9
Direct duty cycle (voltage mode) control
CH9-10
Current Mode control
Inner loop
Outer loop
CH9-12
H(s)
vc ( s )
Tc (s)
ve ( s )
ve ( s )
H(s)
CH9-13
ve vref Hv
v fb (s) R1
H(s)
v(s) R1 R2 Ch2, 3, 5, 7 SC model
vc (s) Gc 0 1 s / z
Gc (s) ( e.g. type II )
ve ( s )
s 1 s / p
1 Gvg(s)
Gc(s) VM
Gvd(s)
T(s)
vfb(s)=H(s)v(s)
H(s)
Loop gain T(s): product of forward & feedback path
gain 1
T(s)=Gc(s) Gvd(s) H(s)
VM CH9-15
If loop gain T(s) is large in magnitude, the
disturbance influence on v is small ve vref Hv
Large loop gain also cause ve0, then v vref(s)/H(s)
but nonetheless small, adding Gc(s) can obtain
small error
Simple phase margin (PM) criterion for
assessing stability is used. When the PM of loop
gain T(s) is positive, then the feedback system is
stable
Increasing PM causes system transient response
to be better behaved, with less overshoot &
ringing
CH9-16
Compensator is designed to attain adequate PM
& good rejection of expected disturbances
Lead compensator (PD controller) are used to
improve PM & extend feedback loop bandwidth.
Better rejection of high-frequency disturbance
Lag compensator (PI controller) are used to
increase low-frequency loop gain. Better
rejection of low-frequency disturbance and very
small steady-state error
CH9-17
Power stage
CH9-18
(9.1)
CH9-19
CH9-20
(9.1)
v (s) Gvd (s)d (s) Gvg (s)v g (s) Zout (s)iload (s)
CH9-21
Y ( s) forward path gain G
TF
R( s) 1 loop gain 1 GH
0dB
CH9-26
f << fc
f >> fc
f << fc
f >> fc
CH9-27
asymptote
CH9-28
H (s)Gc (s)Gvd (s)
T (s)
VM f << fc
f >> fc
feedback CH9-29
Disturbances: vg and iload
Gvg (s)
1 T ( s)
Zout (s)
1 T ( s)
CH9-30
At low frequency, follow 1/ T
CH9-31
At low frequency, T is large, disturbance transfer
function from vg and ilaod to v are
CH9-33
CH9-34
N ( s)
T ( s)
D(s)
CH9-35
CH9-36
Phase Margin (PM) Test
PM : m 180 T ( j 2 fc )
o
CH9-37
Phase & Gain Margin
CH9-38
PM:
c 02
CH9-45
f2 f0
CH9-46
(9.25)
CH9-47
Homework#16 Derivation of Eq. (9.25)
Hint
1
T (s)
s s
1
0 2
Let T ( j ) 1 find cross over frequency c [c f (0 , 2 )]
find m 180o T ( j 2 f c ) [m f (0 , 2 ) f (Q)]
o f0 f0
Q Q
2
2 fc f2
1 1 4Q4 sinm
tanm
2Q 4
cosm
CH9-48
1
T (s)
s s
1
0 2
obtaining real poles (Q < 0.5) requires a phase margin of at least 76. To
obtain Q = 1, a phase margin of 52 is needed
The system with a phase margin of 1 exhibits a closed-loop response
with very high Q
With a small phase margin, T ( j) is very nearly equal to 1 in the vicinity of
the crossover frequency. The denominator (1 + T) then becomes very
small, causing the closed-loop transfer functions to exhibit a peaked
response at frequencies near the crossover frequency fc
CH9-49
1
T (s)
s s 1 T ( s)
1 Output response Inverse Laplace
0 2 s 1 T ( s)
v ( s) Gvg ( s)
v g ( s) 1 T ( s) Line regulation
Low Q-factor
CH9-53
Effect of load current variations on v regulation
v (s) Zout (s)
i ( s ) 1 T ( s )
load
CH9-54
Transient response time: when large disturbance
occurs, the v may undergo a transient and it will
deviate from specified allowable range. Eventually,
feedback loop should operate to return v within
tolerance. Increasing feedback loop crossover
frequency can shorten the transient time
Overshoot & ringing: allowable overshoot &
ringing should be limited, such requirement
implies the PM must be sufficiently large
Each of these requirements imposes constraints on
the loop gain T(s). Therefore, the design of the
control system involves modifying the loop gain, a
compensator network is added for this purpose
CH9-55
fp
Gc (s jf max f z f p ) Gc 0
fz
fz f p
fz sufficiently far
below the fc
Often application
in systems contain
two-pole response
Feedback loop bandwidth can be extended while
maintaining an acceptable PM CH9-56
s
1
z
PD type compensator Gc (s ) Gc 0
s
1
p
2
1 j (1 ) j( )
z zp z p
Gc (s j ) Gc 0 Gc 0
2
1 ( )
1 j p
p
CH9-57
2
1 j (1 ) j( )
z zp z p
Gc (s j ) Gc 0 Gc 0
2
1 ( )
1 j p
p
max
2
f2max
1 1
zp fz fp
fp fz
1
f max ( f p f z ) 1
fz fp ( fp fz ) 1
fz fp
tan tan tan (9.34)
2 fz fp 2 fz fp 2
CH9-58
Derivation of (9.35)
fp fz
fz fp fp fz
Gc ( f max ) tan 1
2 tan
2 fz fp
fpfz fp f 1
4 tan 2 , Let x z
2
fz fp fz fp x
x 2 ( 2 4 tan 2 )x 1 0,
( 2 4 tan 2 ) ( 2 4 tan 2 )2 4 ( 2 4 tan 2 ) 4 tan 1 tan 2
x
2 2
( 2 4 tan 2 ) 4 tan sec sin 2 sin
2 1 2
2 cos 2 cos 2
2 sin 2 2 sin cos 2 (1 sin )2 (1 sin )2
cos
2
cos 2
1 sin 2
1 sin
(9.35)
1 sin
CH9-59
Derivation of (9.36)
fp 1 sin
fc fz fp , x ( 9.35)
f z 1 sin
fz 1 1 sin
f z fc fc fc
fp x 1 sin
fp 1 sin
f p fc fc x fc ( 9.36)
fz 1 sin
Derivation of (9.37) s
fc 2 1
1 ( ) z
fz Gc ( s) Gc 0
Let Gc ( f f c ) 1 Gc 0 s
fc 2 1
1 ( )
fp p
fz
Instituting fc f z f p Gc 0 (9.37)
fp
CH9-60
s
1
Gc (s ) Gc 0 z
s
fz f p 1
p
CH9-61
s
1
Gc (s) Gc 0 z
s
1
p
CH9-62
fp/10 10fz
L
Gc (s j ) Gc 1 j
1 L
90 o
f fL
Gc tan o
0 f f L
Inverted zero is added
At low f, the inverted zero causes compensator to integrate
error signal CH9-64
Uncompensated loop gain Tu 0
Tu (s)
s Single pole
PI type compensator 1
0
L
Gc (s) Gc 1
s
Compensated loop gain L
1
T (s) Tu (s)Gc (s) Tu 0Gc s
s
1
0
L L L
1 j (1 ) j ( )
T G 0 0
T (s j ) Tu 0Gc
u 0 c
2
1 j 1 ( )
0 0
L 2 L
1 ( )
1 0
T Tu 0Gc , T tan
2 L
1 ( ) 1
0 0 CH9-65
L 2 L
1 ( )
T Tu 0Gc , T tan 1 0
2 1 L
1 ( )
At high f 0 0
Tu 0Gc Tu 0Gc
T , at f f c , T 1 f c Tu 0Gc f 0
( f / f0 ) ( fc / f 0 )
fc
for desired f c Gc
Tu 0 f 0
CH9-66
PI Gc makes low freq. loop gain
increase, but PM unchanged
Gc
Gc
f0
10 -20db/dec
fL
10
L
1
T( s) Tu ( s)Gc ( s) Tu0Gc s
s
1
0 CH9-67
1
1 , Low f , T large
T
1 T
1, High f , T small
CH9-68
Zout (s) Gvg (s)
(9.30), (9.31)
1 T ( s) 1 T ( s)
CH9-70
fL f z f p1 f p 2
-45o/dec.
+45o/dec. -45o/dec.
CH9-71
Inverted zero at fL functions in the same manner
as PI compensator
Zero at fz adds phase lead in vicinity of cross-over
frequency, as in PD compensator
High frequency poles fp1 & fp2 must cause the gain
to roll off at high frequencies and to prevent
switching frequency ripple from disrupting the
operation of PWM
fc of loop gain is choose to be greater than fL & fz,
but less than fp1 & fp2
CH9-72
Output: 15V/5A
CH9-73
CCM
VM4V
D 1
Vc DVM
Vc VM
Vref
ve vref Hv 0 H
V CH9-74
CH9-75
Gd0=15/0.536
L=50H, C=500F
R=3
CH9-76
CH9-77
CH9-78
Gc ( s) H ( s) V 1
T ( s)
VM D 1 s s
( )2
Q00 0
-20.26
0.097 Tu (w 2 5kHz) 0.097
+20.26dB=10.3
(9.36)
HW#17
CH9-81
1 HV
Tu ( s) Tu0 , Tu 0 2.33, Gc 1
s s 2 DVM
1 ( )
Q00 0
1 2.33
Tu Tu0 0.097 20.3dB
f fc
c 2 2 c 2 24
[1 ( ) ] ( )
0 Q00
Q0 9.5, 0 2 1k , c 2 5k c 2
1 ( )
z
Gc ( f f c ) 10.3( 20.3dB) Gc 0
c 2
1 ( )
p
c 2 5k , z 2 1.7k , p 2 14.5k
10.3
Gc 0 3.51 10.91dB 11dB
2.935
CH9-82
52o
fz 1.7kHz
fp 14.5kHz
CH9-83
2.333.51=8.178=18.25 dB fz 1.7kHz
fp 14.5kHz
dc gain
fz
10
fp
10 10 f z
-128o
fp
please verify T ( jc ) 128
o
10
f 10 f z PM 52o
CH9-84
1
1 , low f , T large
T
1 T
1, high f , T small
1 1
v ( Dvg ) (0.536 1V ) 0.062V
T 8.6
500Hz
CH9-86
1
1 , low f , T large
T
1 T
1, high f , T small
CH9-87
D=0.536
CH9-88
Supplementary
Materials for
Compensator
Design
CH9-89
Compensated Error Amplifier
TOL ( s ) T1 ( s )Tc ( s )
vo ( s ) vo ( s ) d ( s )
T1 ( s ) Tp ( s )Tm ( s )
vc ( s ) d ( s ) vc ( s )
Tc ( s ) T.F. of compensated EA
For a given T1(s), Tc(s) must be properly tailored so that
TOL(s) meets performance requirements expected of power
supply CH9-90
Desired Characteristics of Open-
loop TF
1. Low-freq. Gain should be high to minimize
steady-state error in PS output
2. Crossover freq. should be high but an order of
magnitude below switching freq. to allow quick
response for transients
3. Phase margin (PM): PM=OL+1800 (PM should be
positive quantity, determines transient response
of Vo in response to sudden change in load & Vd)
OL = phase angle of TOL at crossover freq. (negative)
Desired phase margin range: > 450
CH9-91
General Compensated EA
A>0, z < p
CH9-92
General Compensated EA
Due to zero
If C2 C
1 then
1 1
S S
vc ( s) R2C1 R2C1
vo ( s) C1 C2 1
R1C2S( S ) R1C2S( S )
R2C1C2 R2C2
C1 C2 1
poles 0 p
R2C1C2 R2C2
1
zero z
R2C1 CH9-95
Type 2 CEA
Middle-
frequency gain
Gc = R1/R2
CH9-96
Type 2 CEA
Control Loop TF Frequency Response
(Type II)
z p
cross
1
20 log
Power stage & Filter VP
20log Gvd
vc ( j) j 1 R2
Gain :
vo ( j) R1C2 j p R1C2 (
1
) R1
R2C2
1 1
Phase : CEA 180 tan ( ) 90 tan ( )
0 0
z p
1 1
= 270 tan ( ) tan ( )
0
z p
CH9-98
Type 2 CEA Design Procedure
1. Choose desired crossover freq. of total open-loop
T.F., usually around an order of magnitude less
than converter switching freq.
2. Except for CEA, determine all units TF & freq.
response that in control circuit
3. Determine mid-frequency gain of CEA required
to achieve overall desired crossover freq., this
obtain R2/R1 ratio
4. Choose desired phase margin needed to assure
stability, typically > 450. From middle-freq. gain
obtain R1, R2; C1, C2 are determined by pole & zero
CH9-99
K-Factor Method
11
Phase : CEA = 270 tan ( ) tan ( )
0
z p
Phase angle of CEA at crossover freq. is
cross
1 1 cross
CEAcross 270 tan (
0
) tan ( )
z p
cross p
Let K
z cross
1 1 1
CEAcross 270 tan K tan ( )
0
K
Pole & zero freq. selectionK-factor method (D. Venable, The K factor: A
New Mathematical Tool for Stability Analysis and Synthesis, Proceedings
Powercon, 1983) CH9-100
K-Factor Method
Using trigonometric identity
1 1 1 1
tan ( x ) tan ( ) 90 tan ( ) 90o tan 1 ( K )
1 o
x K
1 1
CEAcross 270 tan K tan ( ) 2 tan 1 ( K ) 360 o
0 1
K
2 tan 1 ( K ) (1)
CEAcross
K tan (2)
2
from (1), 0o CEAcross 180o for 0 K
CEAcross PM converter (3)
The K can be calculated from (2) if the CEA-cross is determined
from (3). Where PM is the desired phase margin
CH9-101
K-Factor Method
Accordingto desired phase margin (PM ) and control
loop phase shift (converter ), the needed CEA-cross can
be obtained form (3), then K is determined by (2)
cross & K are known, then z & p can be obtained,
If
C1, C2 are then found
cross p
K
z cross
1 cross K
z C1
R2C1 K cross R2
1 1
p K cross C 2
R2C 2 cross R2 K
CH9-102
Design Example: Type 2 CEA
A Buck converter: Vs=10V, Vo=5V, fs=100 kHz,
L=100 H with SR=0.1, C=100F with ESR = 0.5
, R = 5 , VP = 3V in PWM circuit. Design a type
2 CEA that results in a stable control system (PM
450).
Solution
1. Let fco=10 kHz < fs
2. Pspice simulation shows that (at 10 kHz): power
stage & filter gain: 2.24 dB, phase angle: -1010 ,
PWM gain=1/VP=1/3=-9.54 dB. The combined
gain= 2.24 dB+(-9.54 dB)=-11.78 dB (= G1) CH9-103
Design Example: Type 2 CEA
CH9-104
Design Example: Type 2 CEA
4. PM=450, converter= -1010 , then
CEAcross 146 0
K tan tan 3.27
2 2
K 3.27
C1 13.4nF
2f co R2 2(10k )(3.88k )
1 1
C2 1.25nF
2f co R2 K 2(10k )(3.88k )3.27
CH9-105
Implemented Type 2 CEA circuit
C2 = 1.25 nF
R2 = 3.88 k
R1=1 k
C1 = 13.4 nF
V0
VC
+
V0,ref
CH9-107
Type 3 CEA Design
CH9-108
Small-signal Transfer Function for
Type 3 CEA
1 1 1 1
R
2
/ / s s
vc ( s) Zf sC sC R R3
R C
2 1 R R C
1 2 1 3
( )
GCEA ( s) 1 3
vo ( s) Zi 1 R1R3C2 C1 C2 1
R1 / / R3 s s s
sC 3 R C C
2 1 2 R C
3 3
(1)
CH9-109
Vref is dc and has no effect on small-signal TF.
Assuming C2 << C1 & R3 << R1, then
1 1 1 1
s s s s
R1 R3 R2C1 ( R1 R3 )C3 1 R C
2 1 R C
1 3
GCEA ( s) ( 2)
R1R3C2 C1 C2 1 R3C2 1 1
s s s s s s
R C C
2 1 2 R C
3 3 R C
2 2 R C
3 3
CH9-110
There are 2 zeros and 3 poles in the G(s)
1 1
s s
For C2 << C1 & R3 << R1 GCEA ( s)
1 R2C1 R1C3
R3C2 1 1
s s s
j z1 j z 2 R2C2 R3C3
1
GCEA ( s j) (3)
R3C2 jp1 j p 2 j p 3
1 1
1 1 1 s s
z1 , z 2 G ( s)
v ( s)
R R
c 1 3
R2C1 ( R1 R3 )C3
R2C1 ( R1 R3 )C1 R1C3
CEA
v ( s) RRC C C2 1
o 1 3 2
s s 1 s
R2C1C2 R3C3
C1 C2 1 1
p1 0 , p 2 , p 3
R2C1C2 R2C2 R3C3
from (3), phase angle of the CEA
1 1 1
1
CEA 180 tan
0
tan 90 tan
0
tan
z1 z 2 p 2 p 3
1 1 1
1
270 tan
0
tan tan tan ( 4)
z1 z 2 p 2 p 3
CH9-111
Type 3 CEA Design
CH9-112
Type 3 CEA Design
The double zeros & poles are placed at frequencies
GCEA ( s j)
1 j z1 j z 2
co
(3)
R3C2 jp1 j p 2 j p 3
z , p co K (5)
K type 2
j z
2 1
z cross
1
From (3), GCEA ( j)
R2C1 K
R3C2 j j
2 1
p K cross
p R2C 2
jco z
2
1
At crossover freq. GCEA ( jco )
R3C2 j j
2
co co p
co co
1 1 1
270 2 tan K tan
0
K
1 1 1 1 1
tan x tan 90 tan
0
90 0 tan1 K
x K
CEA -co 900 4tan1 K
2
CEA -co 90
0
K tan (7)
4
CEA -co 900 ~ 270 0 for 0 K CH9-114
Type 3 CEA Design
The maximum angle of the type3 CEA is 2700, but
the type2 is 1800 1 1 1
z1 , z 2
R2C1 ( R1 R3 )C1 R1C3
C C2 1 1
p1 0, p 2 1 , p 3
CEA -co PM converter (8) R2C1C2 R2C2 R3C3
K 1 1 1
From (5), co K z , p 2p
R1C3 R2C2 R3C3 R2C2 R3C3
K
j
1 jco 1 R1C3 R2
GCEA ( jco ) ( j K ) (10)
R3C2 2
p
R3C2 1
R2C2 R3C3
R1
CH9-115
Type 3 CEA Design
1 jco j KR2 R2 GCEA ( jco )
GCEA ( jco ) (11)
R3C2 2
p
R1 R1 K
Solution
1. Select fco=10 kHz < fs
2. Pspice simulation shows that (at 10 kHz): power
stage & filter gain: 10.5 dB, phase angle: -1440 ,
PWM gain=1/VP=1/3=-9.5 dB. The combined gain
= 10.5dB+(-9.5 dB)=-20 dB (= G1)
CH9-117
Design Example: type 3 CEA
2
CEAcross 90
0 2
K tan tan 69.75 7.35
0
CH9-118
Design Example: type 3 CEA
R2 GCEA ( jco ) 10
3.6885
R1 K 7.35
R2
3.7 Let R1 1k , R2 3.7 k
R1
CH9-119
Design Example: type 3 CEA
5. Calculating the other component values
K K 7.35
C1 11.6nF
co R2 2fco R2 2(10k )(3.7k )
1 1 1
C2 1.58nF
co R2 K 2fco R2 K 2(10k )(3.7k ) 7.35
K K 7.35
C3 43.1nF
co R1 2fco R1 2(10k )(1k )
1 1 1
R3 136
coC3 K 2fcoC3 K 2(10k )( 43.1n ) 7.35
CH9-120
Design Example: type 3 CEA
6. Implemented type 3 CEA
C2=1.58nF
C3=43.1nF R =3.7k
R3=136 2
C1=11.6nF
R1=1k
CH9-121
Design Example: type 3 CEA
C2
R3 C3
R2 C1
R1
V0
VC
+
type III CEA
V0,ref
CH9-125
Manual Placement of Poles &
Zeros in Type 3 CEA
In addition to the K-factor method, the poles &
zeros assignment at specified frequencies can be
used to design the type 3 CEA
A frequency of particular interest is the resonant
frequency of the LC filter in the converter (neglect
any resistance in the L & C)
1 1
LC fLC
LC 2 LC
CH9-126
Zero or Pole Expression Placement
First zero z1 = 1/R2C1 50% to 100% of LC
Second zero z2 = 1/(R1+R3)C3 1/R1C3 At LC
First pole p1 = 0 --
Second pole p2 = (C1+C2)/R2C1C2 At the ESR zero = 1/rCC
1/R2C2
Third pole p3 = 1/R3C3 At one-half the switching
frequency, 2(fsw/2)
CH9-128
G1 G2
CH9-129
CH9-130
CH9-131
CH9-132
CH9-133
CH9-134
CH9-135
CH9-136
CH9-137
CH9-138
Neglet the injection source impedance. The current source is converted into
voltage source using Thevenin-equivalent circuit
CH9-139
i ( s) v z ( s) G1v e ( s) , i ( s) v z ( s)
y x
Z1 Z2
Z
Z1
G
v z ( s) Z2ix ( s) iy ( s) 2 ix ( s) 1 HG2 Z2ix ( s)
Z1
iy ( s) Z2 Z2 Z1 Z2
Ti ( s) (1 G1HG2 ) (1 T ( s) )
i ( s) Z1 Z1 Z2
x
Z2 Z2
T ( s) 1
Z1 Z1 CH9-140
CH9-141
CH9-142
CH9-143
CH9-144
CH9-145