环路培训

Analog loop：
1、stable state and FFT
2、vector network Analyzer manual
3、self-oscillation conditions
4、phase margin 45°，Gain margin -10dB criterion
5、zero point and pole point characteristic
6、pole-zero point compensation circuit
7、LCR network frequency characteristic
8、system transfer function
9、voltage ripple
Digital loop：
10、continue domain to discrete domain,c2d
11、Tustin bilinear transformation analysis
12、 Matlab calculate difference Equation coefficient
13、 pole-zero point compensation circuit to discrete domain
14、Butterworth low pass filter
15、 incremental PID control algorithm
16、digital flyback power block

数字电源开源社区 1
系统框图

Analog loop Digital loop

Ø Amplifier Ø ADC
Ø Comparator Ø Timer
Ø Lantch Ø PID/loop alogrithm

This picture is provided by iwatt,campbell.

数字电源开源社区 2
矢量网络分析仪

Note：the disturbance signal

should be injected into the feedback
loop by isolation transformer.

Gain
margin：-50dB

Phase
margin:8.9°

数字电源开源社区 3
one pole point(1p)

Z2 Small signal analysis：

1
Z1  R1 Z 2  R2 / /
Z1
S  jw

jwc1
Vin  0 0  V0 Vo  R2 1
 H (s)   
Z1 Z2 Vin R1 1  SR2C1
0V R2 1 R
20 lg H ( s )  20 lg( )  20 lg 2  20 lg wR2C1
R1 1  ( SR2C1 ) 2 R1
f 1
   arctan f0   800Hz
f0 2 R2C1
20 lg wR2C2  3dB

Description：Gain will decrease to infinitely small

with frequency increase.
1
f0 Position： f 0 
2 R2C1
magnitude-frequency characteristic :-20dB/10Wo

Phase-frequency characteristic ：Max phase shift

is 90°,45°@ pole point

数字电源开源社区 4
one zero point(1z)
Small signal analysis：
Z2
Z1 1
Z1  R1 / / Z 2  R2 S  jw
jwc1
Vin  0 0  V0 Vo  R2
0V  H (s)    (1  SR1C1)
Z1 Z2 Vin R1
R2 R
20 lg H ( s)  20 lg( 1  ( SR1C1) 2 )  20 lg 2  20 lg wR1C1
R1 R1
f 1
  arctan f0   1600Hz 20 lg wR1C1  3dB
f0 2 R1C1

Description：Gain will increase to infinitely max

with frequency increase.
1
f0 Position： f 0 
2 R1C1
magnitude-frequency characteristic :20dB/10Wo

Phase-frequency characteristic ：Max phase shift

is 90°,45°@ pole point

数字电源开源社区 5
one pole with one zero point(1p1z)
Z2 Small signal analysis：
Z1 Z1  R1 Z2 R2 
1 S  jw
jwc2
Vin  0 0  V0 Vo (1  SR2C1 )
0V  H (s)  
Z1 Z2 Vin SR1C1
|1  SR2C1 | R
20 lg H ( s)  20 lg( )  20 lg 2
| SR1C1 | R1
f0 1
  arctan f0   800Hz
f 2 R2C1

Pole point position：0Hz

Zero point position：800Hz

f0
magnitude-frequency characteristic :+0dB/10Wo

Phase-frequency characteristic ：Max phase shift

is 0°,45°@ pole point

数字电源开源社区 6
one pole with two zero point(1p2z)
Z2 Small signal analysis：
Z1 Z1  R1 / /
1
Z 2  R2 
1
S  jw
jwC1 jwc2
Vin  0 0  V0 Vo (1  SR1C1 )(1  SR2C2 )
0V  H (s)  
Z1 Z2 Vin SR1C 2
1 1
f zero1   800Hz f zero 2   1600Hz
2 R1C1 2 R2C2

R2
20 lg H ( s)  20 lg ( f zero1  f  f zero 2 )
R1

Pole point position：0Hz

f zero1 f zero 2 Zero point position：800Hz / 1600Hz
magnitude-frequency characteristic :
Phase-frequency characteristic ：

数字电源开源社区 7
two pole point with one zero point(2p1z)
Z2 Small signal analysis：
1 1
Z1  R1 Z 2  ( R2  )// S  jw
jwC2 jwC1
Z1 Vin  0 0  V0

0V Z1 Z2
Vo (1  SR2C2 )
H (s)  
Vin SR (C  C )(1  SR C1C2 )
1 1 2 2
C1  C2

1 1
f zero1   800Hz f pole2   36kHz
2 R2C2 CC
2 R2 1 2
C1 C2
Pole point position：0Hz/36kHz
f zero1 f pole 2 Zero point position：800Hz
magnitude-frequency characteristic :
Phase-frequency characteristic ：

数字电源开源社区 8
two pole with two zero point(2p2z)
Z2 Small signal analysis：
1 1 1
Z1  R1/ / Z 2  ( R2  )// S  jw
jwC1 jwC2 jwC3
Z1 Vin  0 0  V0

0V Z1 Z2
Vo (1  SR2C2 )(1  SR1C1 )
H (s)  
Vin SR (C  C )(1  SR C2C3 )
1 2 3 2
C2  C3
1 1
f zero1  800Hz fzero2   7.2kHz
2 R2C2 2RC1 1

1
f pole1  0 Hz f p o le 2 
C 2C 3
 36 kH z
2 R 2
C2  C3
f zero1 f zero 2 f pole 2 Pole point position：0Hz/36kHz
GBW
Zero point position：800Hz/7.2kHz
magnitude-frequency characteristic :
Phase-frequency characteristic ：

数字电源开源社区 9
three pole with two zero point(3p3z)
Z2 Small signal analysis：
1 1 1
Z1 Z1  R1/ /
jwC1
Z 2  ( R2 
jwC2
)//
jwC3
S  jw

Vin  0 0  V0

Z1 Z2
0V 1 1
(S  )( S  )
Vo R1  R3 R2C2 (R1  R3 )C3
H (s)   
Vin R1R3C1 C  C2 1
S (S  1 )( S  )
R2C1C2 R3C3
1 1
f zero1   340Hz f zero 2   3.6kHz
2 R2C2 2 ( R1  R3 )C3

f pole1 
1
 10kHz 1
CC f pole 2   72kHz
f zero1 f zero 2 f pole1
2 R2 1 2
C1  C2
2 R3C3
f pole 2
Pole point position：10kHz/72kHz

Zero point position：340Hz/3.6kHz

magnitude-frequency characteristic :
Phase-frequency characteristic ：

数字电源开源社区 10
TL431+817

Small signal analysis：

V1
V0    CTR  R6 S  jw
1
R5//(R 7  )
jwc3

V0 R R (1 SC3R7 )
H (s)   CTR 6 5
V1 1 SC3 (R5  R7 )

V0 (1  SR2C2 )
G (s)  
Vin CC
SR1 (C1  C2 )(1  SR2 1 2 )
C1  C2

(1  SR 2 C 2 )
F (s )   CTR
C 1C 2
SR 1 ( C 1  C 2 )(1  SR 2 )
C1  C 2
R 6 R 5 (1  SC 3 R 7 )

1  SC 3 ( R 5  R 7 )

H(S G(S we can add one zero point R7

with c3 to offset the pole
) ) of the optocoupler

数字电源开源社区 11
LCR
Small signal analysis：
1
Z1  jwL  RL Z2  (  RC ) / / R3 S  jw
Z1 jwC
Z2 H (s) 
Vo

Z2

Vin Z1  Z 2
(1  SR2C ) R3
 2
S LC ( R2  R3 )  S [ L  ( R1R2  R1R3  R2 R3 )C ]  R1  R3

(1  SR2C ) R3 1  SR2C
 2
 2
S LCR3  S [L  2R3 )C ]  1 S LC  1

1
f zero   3.4kHz
2 R2C
1
f pole   740Hz
2 LC

Pole point position：740Hz(double pole)

Zero point position：3.4kHz

magnitude-frequency characteristic :
Phase-frequency characteristic ：

数字电源开源社区 12
Open loop

H(S K(S
) )

G(S)

1 1
(S  )( S  )
R1  R3 R2C2 (R1  R3 )C3 1 SR C
F (s)  G (S )  H (S )  K (S )   10 4 2 c S  jw
R1 R3C1 C  C2 1 S LC  1
S (S  1 )( S  )
R2C1C2 R3C3

数字电源开源社区 13
Open loop

Gain
margin：-50dB

Phase
margin:8.9°

数字电源开源社区 14
Open loop

AF>1
Phase=360°

数字电源开源社区 15
C2d,continue domain to discrete domain
 FT 1 
X ( jw)   x(t )e  jwt dt

x( t ) 
2 
X ( jw)e jwt dw Note : s  jw
Fourier transform

FT 1
X (e jw )   x[n]e
n  
 jwn
x[n] 
2   X (e
2
jw
)e jwn dw Note : s  jw

1

X (  jw)   x(t )e t e  jwt dt LS x( t )  

X (  jw)et e jwt dw Note : s    jw Laplace transform
 2 

n  
Z 1 Note : z  e jw
X (e jw )   x[n]e
n  
 jwn
X ( n) 
2  X (e )e dz
jw jw ( n 1)
Z transform
Uniform sampling
x(t ) x(n ) L[ x(t )]  X ( s ) Z [ x(n )]  X ( z )
Tustin transform

How to transform X(s) to X(z), or how to transform X(z) to X(s) ? It's a question.

2 Z 1
Tustin Tustin bilinear transformation: S
T Z 1

X [n  1] Z z 1 X [ z ] 1 unit time shift

z 1

X [ n  2] Z z 2 X [ z ] 2 unit time shift

z 2

X [ n  2] Z z 2 X [ z ] z 2 3 unit time shift

数字电源开源社区 16
one pole point

Z2 Small signal analysis：

Vo  R2 1 2 Z 1
H (s)    S
Z1 Vin R1 1  SR2C1
S  jw

T Z 1

 R2 1
H ( z)   | 2 z 1 T 10uS
0V R1 1  SR2C1 s  T z 1
Y ( n) 1
H ( z)   2
X ( n) 2 z 1
1  0.2 103
T z 1
Y ( n)  0.95Y ( n  1)  0.049 X ( n)  0.049 X ( n  1)
%Matlab code
clear all; %clear old data
s  domain close all; %close old file
Ts=10*10^(-6) %sampling period
Cont_s=tf(-2,[0.2*10^(-3),1]) %S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
z  domain Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v')%Equation coefficient
num_z = -0.0488 -0.0488
den_z =1.0000 -0.9512

If increase the sample frequency ,z domain

will be more closed to s domain.

数字电源开源社区 17
one zero point
Small signal analysis：
Z2
Z1 Vo  R2 2 Z 1
H (s)    (1  SR1C1) S
Vin R1 T Z 1
0V Y (n)  R2
H ( z)    (1  SR1C1 ) | 2 z 1 T 10uS
X ( n) R1 s
T z 1

Y ( n)  Y ( n  1)  42 X ( n)  38 X ( n  1)

%Matlab code
clear all; %clear old data
close all; %close old file
z  domain Ts=10*10^(-6) %sampling period
Cont_s=tf([-0.2*10^(-3),-2],1) %S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
s  domain Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v')%Equation coefficient

num_z = -42.0000 38.0000

den_z = 1 1

数字电源开源社区 18
one pole with one zero point(PI控制器)
Z2 Small signal analysis：
Z1 H (s) 
Vo

(1  SR2C1 ) S
2 Z 1
Vin SR1C1 T Z 1

0V Y ( n) (1  SR2C1 )
H ( z)   | 2 z 1
X ( n) SR1C1 s  T z 1

Y (n)  Y (n  1)  2.05 X (n)  1.95 X (n  1)

%Matlab code
clear all; %clear old data
z  domain close all; %close old file
s  domain Ts=10*10^(-6) %sampling period
Cont_s=tf([-0.2*10^(-3),-1],[0.1*10^(-3),0])%S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v') %Equation coefficient
num_z = -2.0500 1.9500
den_z = 1 -1

数字电源开源社区 19
one pole with one zero point（PI控制器）
Z2 Small signal analysis：
Z1 −�� (1+��2 �1 )
�
�∗ 2+
1
�1 �1 ∗�1
�(�) = =− =−
��� ��1 �1 �
0V 1
�∗�� +
�� �� 2 Z 1
PID(�) = �� + = S
T Z 1
� �

�(�) = �(� − 1) + (�� + 0.5�� ∗ ��)�(�) + (0.5�� ∗ �� − ��)�(� − 1)

%Matlab code ,Kp=Ki=1

clear all; %clear old data
z  domain close all;
Ts=10*10^(-6)
%close old file
%sampling period
s  domain Cont_s=tf([1,1],[1,0])%S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v') %Equation coefficient

num_z =1.0000 -1.0000

den_z = 1 -1

数字电源开源社区 20
one pole with two zero point
Z2 Small signal analysis：
Z1 H (s) 
Vo (1  SR1C1 )(1  SR2C2 )
 S
2 Z 1
Vin SR1C 2 T Z 1
%Matlab code
0V clear all; %clear old data
close all; %close old file
Ts=10*10^(-6) %sampling period
num_c=conv([-0.1*10^(-3),-1],[0.2*10^(-3),1]);
% numerator
den_c=conv([0.1*10^(-3),0],1); %denominator
Cont_s=tf(num_c,den_c) %S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v')%Equation coefficient
z  domain
num_z =-43.0500 79.9000 -37.0500
den_z =1 0 -1
s  domain
Y (n)  Y (n  2)  43.05 X (n)  79 X (n  1)
- 37.05 X (n - 2)

数字电源开源社区 21
two pole with one zero point
Z2 Small signal analysis：
Vo (1  SR2C2 )
H (s)   | 2 Z 1
Vin SR (C  C )(1  SR C1C2 ) S  T Z 1
Z1 1 1 2 2
C1  C2
0V
%Matlab code
clear all; %clear old data
close all; %close old file
Ts=1*10^(-6) %sampling period
num_c=conv([0.2*10^(-3),1],1);
%numerator
den_c=conv([0.1*10^(-3),0],[4.4*10^(-6),1]);
%denominator
Cont_s=tf(num_c,den_c) %S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
z  domain Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
s  domain bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v')%Equation coefficient

num_z =0.2046 0.0010 -0.2036

den_z =1.0000 -1.7959 0.7959

Y (n)  1.8Y (n  1) - 0.8Y（n - 2） 0.205 X (n) - 0.204 X (n  2)

数字电源开源社区 22
two pole with two zero point
Z2 Small signal analysis：
Vo (1  SR2C2 )(1  SR1C1 ) 2 Z 1
H (s)   S
Vin SR (C  C )(1  SR C2C3 ) T Z 1
Z1 1 2 3 2
C2  C3
0V
%Matlab code
clear all; %clear old data
close all; %close old file
Ts=1*10^(-6) %sampling period
num_c=conv([0.2*10^(-3),1],[22*10^(-6),1]);
%numerator
den_c=conv([0.1*10^(-3),0],[4.4*10^(-6),1]);
%denominator
Cont_s=tf(num_c,den_c) %S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
z  domain Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
s  domain bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v')%Equation coefficient

num_z =9.2066 -17.9582 8.7536

den_z =1.0000 -1.7959 0.7959

Y (n)  1.8Y (n 1)-0.8Y（n-2） 9.2X (n)-17.96X (n  2)  8.75X (n  2)

数字电源开源社区 23
three pole with two zero point（Type3补偿器与差分方程）
Z2 Small signal analysis：
Z1 1
(S  )( S 
1
)
Vo R1  R3 R2C2 (R1  R3 )C3
H (s)   
Vin R1R3C1 C  C2 1
S (S  1 )( S  )
0V R2C1C2 R3C3
2 Z 1
S
T Z 1
%Matlab code
clear all; %clear old data
close all; %close old file
Ts=1*10^(-6) %sampling period
num_c=conv([1,1/(470*10^(-6))],[1,1/(44*10^(-6))]);
z  domain %numerator
den_c=conv([0.33*10^(-6),0],conv([1,1/(15.51*10^(-
s  domain 6))],[1,1/(2.2*10^(-6))]));
%denominator
Cont_s=tf(num_c,den_c) %S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v')%Equation coefficient

Y ( n )  2 .5 6 7 Y ( n  1)  2 .1 5 7 Y ( n  2 )
 0 .5 9 Y ( n  3)  1 .2 1 0 9 X ( n )  1 .1 8 1 X ( n  1)
 1 .2 1 X ( n  2 )  1 .1 8 X ( n  3)

数字电源开源社区 24
Butterworth low pass filter
1
H (s)  2
| 2 Z 1
2( SRC )  2 SRC  1 S  T Z 1
R  1K C  22nF
%Matlab code
clear all; %clear old data
close all; %close old file
Ts=10*10^(-6) %sampling period
num_c=1;
%numerator
den_c=conv([0.968*10^(-9),44*10^(-6),1],1);
%denominator
Cont_s=tf(num_c,den_c) %S domain function
bode(Cont_s,'g');grid on;hold on; %S domain bode
Cont_discret_s=c2d(Cont_s,Ts,'tustin') %Z domain function
bode(Cont_discret_s,'r');grid on;hold on; %Z domain bode
[num_z,den_z]=tfdata(Cont_discret_s,'v')%Equation coefficient

Y (n)  1.95Y (n  1) - 0.955Y（n - 2） [0.252 X (n)

 0.505 X (n - 1)  0.252 X (n  2)] 10 3

数字电源开源社区 25
PID

PID control algorithm: Proportion Integration Differentiation

continue domain B
1 t de( t)
0
A
u ( x)  k p [e(t )  e (t ) dt +Td ]
Ti dt

discrete domain
T k
Td
u ( k )  k p [e ( k ) 
Ti
 e(n)+
n 0 T
(e(k )  e(k  1))]

incremental PID algorithm

u (k )  u (k )  u (k  1)  K p[e(k )  e(k  1)]  k ie(k )  k d (e(k )  2e(k  1)  e(k  2))

Y (n)  x (n)  x (n  1)  K p[e(n)  e(n  1)]  k ie(n)  k d (e(n)  2e(n  1)  e(n  2))

https://blog.csdn.net/qq229596421/article/details/51419813

数字电源开源社区 26
Incremental PID c-code
#define max(a, b) (a>b? a:b)
#define min(a, b) (a<b? a:b) continue domain
#define range(x, a, b) (min(max(x, a), b))
#define exchange(a, b, tmp) (tmp=a, a=b, b=tmp)
#define myabs(x) ((x<0)? -x:x)
typedef struct
{ float limit; //max limit
float target; //target voltage
float feedback; //pid output to control
float Kp;
float Ki;
float Kd;
float e_0; //current error voltage
float e_1; //last error voltage
float e_2; //before last error voltage
}PID; discrete domain
float pid_calc(PID *pid)
{
float out;
float ep, ei, ed;
pid->e_0 = pid->target - pid->feedback;
ep = pid->e_0 - pid->e_1;
ei = pid->e_0;
ed = pid->e_0 - 2*pid->e_1 + pid->e_2;
out = pid->Kp*ep + pid->Ki*ei + pid->Kd*ed;
out = range(out, -pid->limit, pid->limit);
pid->e_2 = pid->e_1;
pid->e_1 = pid->e_0;
return out;
}

数字电源开源社区 27