Chapter 31 Feedback Amplifiers
这一章我们介绍feedback 反馈运放的原理. 负反馈是模拟电路强有力的工具.
1. The Feedback Equation
反馈系统如下图所示
Aol=amp open-loop gain即开环增益. Aol=xo/xi
β
\beta
β 是 feedback factor
β
=
x
f
x
o
\beta=\frac{x_{f}}{x_{o}}
β=xoxf
Closed-loop gain, Acl
A
C
L
=
x
o
x
s
=
A
O
L
1
+
A
O
L
β
A_{CL}=\frac{x_{o}}{x_{s}}=\frac{A_{OL}}{1+A_{OL}\beta}
ACL=xsxo=1+AOLβAOL
Loop gain = T=
A
O
L
β
A_{OL}\beta
AOLβ , 用来判断环路稳定性
2. Properties of Negative Feedback on Amplifier Design
负反馈有几大好处: 降低gain的工艺偏差, 减少非线性失真, 拓展op-amp的有用bandwidth, 控制输入和输出阻抗.
Gain Desensitivity
将上公式取导
d
A
C
L
A
C
L
=
1
1
+
A
O
L
β
d
A
O
L
A
O
L
\frac{dA_{CL}}{A_{CL}}=\frac{1}{1+A_{OL}\beta}\frac{dA_{OL}}{A_{OL}}
ACLdACL=1+AOLβ1AOLdAOL
因为Loop gain T很大, closed-loop gain Acl 的变化对比AOL的变化很小.
Bandwidth Extension
open-loop 传输函数的pole为WH. 经过负反馈后, closed-loop传输函数 的pole为WH=(1+Aol*beta). pole扩展了(1+Loop Gain)倍.
当beta=1, pole最大, 为GBW, 当beta=0, 回到open-loop 场景.
Reduction in Nonlinear Distortion
负反馈可以改善op-amp的非线性行为, 如下图所示:
Input and Output Impedance Control
如下图所示, Ri和Ro是开环输入/输出阻抗, Rinf和Rof是闭环输入/输出阻抗.
负反馈可以增加或者降低闭环输入/输出阻抗 (1 + AOL*beta)倍.
闭环输入/输出阻抗取决于输入/输出类型是电压还是电流. 如下图所示.
输入类型是电压 (串联), 闭环输入阻抗增大 (1 + AOL* beta)倍. 输入类型是电流(并联), 闭环输入阻抗减小 (1 + AOL* beta)倍.
输出类型是电压(并联), 闭环输出阻抗减小 (1+AOL* beta)倍. 输出类型是电流(串联), 闭环输出阻抗增大(1+AOL* beta)倍.
3. Recognizing Feedback Topologies
根据输入/输出是电压还是电流信号, 可以把反馈分为四类:
输入是电压为series串联(voltage mixing), 输入是电流为shunt并联(voltage mixing).
输出是电压为shunt并联(voltage sampling), 输出是电流为series串联(current sampling).
在计算 AOL, 需要注意由于加入负反馈beta网络和其他source对其load的影响.
Input Mixing and Output Sampling
load和feedback network接到同一点的为shunt, 串联在一起的为series.
The Feedback Network
几点rule:
1, forward path是gain最大的通路
2, ac小信号从gate或者source进入, 从drain或者source离开. 从drain到source的增益忽略不计.
3, 每次从gate-to-drain, 小信号符号反向.
上图展示了series-series和shunt-shunt结构.
a)中vi=vs-vf, 因此vf符号为正, vf越大, 输入信号vs越小, 形成负反馈.
b)中VGG 是DC voltage source, ii=is-if,也是相似道理.
Calculating Open-Loop Parameters
In real-life applications, the beta network can cause loading effects on both the input source and the output of the amplifier circuit.
注意在计算open-loop gain时, 需要考虑beta network给输入和输出带来的影响.
计算Rbeta_i和Rbeta_o时, 遵循shunt 就 short(短路)另外一端, series 就sever(开路)另外一端.
例如计算Rbi, 如果输出时shunt, 就短路输出, 再计算Rbi.
Calculating Closed-Loop Parameters
如前面所说, Closed-loop gain:
A
C
L
=
x
o
x
s
=
A
O
L
1
+
A
O
L
⋅
β
A_{CL}=\frac{x_{o}}{x_{s}}=\frac{A_{OL}}{1+A_{OL}\cdot \beta}
ACL=xsxo=1+AOL⋅βAOL
For series input:
R
i
n
f
=
R
i
(
1
+
A
O
L
⋅
β
)
R_{inf}=R_{i}(1+A_{OL}\cdot \beta)
Rinf=Ri(1+AOL⋅β)
For shunt input:
R
i
n
f
=
R
i
(
1
+
A
O
L
⋅
β
)
R_{inf}=\frac{R_{i}}{(1+A_{OL}\cdot \beta)}
Rinf=(1+AOL⋅β)Ri
A C L = x o x s = A O L 1 + A O L ⋅ β A_{CL}=\frac{x_{o}}{x_{s}}=\frac{A_{OL}}{1+A_{OL}\cdot \beta} ACL=xsxo=1+AOL⋅βAOL
For series output:
R
o
f
=
R
o
(
1
+
A
O
L
⋅
β
)
R_{of}=R_{o}(1+A_{OL}\cdot \beta)
Rof=Ro(1+AOL⋅β)
For shunt output:
R
o
f
=
R
o
(
1
+
A
O
L
⋅
β
)
R_{of}=\frac{R_{o}}{(1+A_{OL}\cdot \beta)}
Rof=(1+AOL⋅β)Ro
4. The Voltage Amp (Series-Shunt Feedback)
Voltage-Voltage, Series-Shunt反馈系统如下图所示
transistor level 电路下图所示
open-loop gain:
β
=
v
f
v
2
=
R
1
R
1
+
R
2
\beta=\frac{v_{f}}{v_{2}}=\frac{R_{1}}{R_{1}+R_{2}}
β=v2vf=R1+R2R1
因为是series-shunt
因为是用MOS device, 输入Ri=infinity
R
o
=
R
L
∥
(
R
1
+
R
2
)
R_{o}=R_{L}\parallel (R_{1}+R_{2})
Ro=RL∥(R1+R2)
A C L = A O L 1 + A O L ⋅ β A_{CL}=\frac{A_{OL}}{1+A_{OL}\cdot \beta} ACL=1+AOL⋅βAOL
R i n f = R i ( 1 + A O L ⋅ β ) R_{inf}=R_{i}(1+A_{OL}\cdot \beta) Rinf=Ri(1+AOL⋅β)
R o f = R o ( 1 + A O L ⋅ β ) R_{of}=\frac{R_{o}}{(1+A_{OL}\cdot \beta)} Rof=(1+AOL⋅β)Ro
考虑RG1,RG2和 Rs的影响, 实际Gain和Input Resistance:
考虑下图series-shunt circuit
Open-loop gain和beta:
5. The Transimpedance Amp (Shunt-Shunt Feedback)
shunt-shunt 并联-并联, 电流-电压反馈, 采样输出的Voltage, 在输入端加入反馈电流
A C L = A O L 1 + A O L ⋅ β A_{CL}=\frac{A_{OL}}{1+A_{OL}\cdot \beta} ACL=1+AOL⋅βAOL
R i n f = R i ( 1 + A O L ⋅ β ) R_{inf}=\frac{R_{i}}{(1+A_{OL}\cdot \beta)} Rinf=(1+AOL⋅β)Ri
R o f = R o ( 1 + A O L ⋅ β ) R_{of}=\frac{R_{o}}{(1+A_{OL}\cdot \beta)} Rof=(1+AOL⋅β)Ro
下图是shunt-shunt 电路
Open-loop gain: Aol=v2/is
β
=
i
f
v
2
=
−
1
R
2
mhos
\beta=\frac{i_{f}}{v_{2}}=-\frac{1}{R_{2}}\text{mhos}
β=v2if=−R21mhos
Voltage Gain, v2/v1:
v
2
v
1
=
v
2
i
s
(
R
i
n
f
)
=
A
C
L
1
R
i
n
f
\frac{v_{2}}{v_{1}}=\frac{v_{2}}{i_{s}(R_{inf})}=A_{CL}\frac{1}{R_{inf}}
v1v2=is(Rinf)v2=ACLRinf1
Simple Feedback Using a Gate-Drain Resistor
一个常见的shunt-shunt feedback电路下图所示, 通过很大resistor R2把Gain和Drain连到一起.
分析DC, 因为R2很大, Gate Voltage=Drain Voltage, 没有电流流过R2. 因此确保Device无需其他器件就处于saturation (我真的很好奇 whether it really works).
小信号模型:
A
O
L
=
[
−
g
m
1
(
R
1
∥
R
2
∥
r
o
1
)
]
[
R
2
]
V/A
A_{OL}=[-g_{m1}(R_{1}\parallel R_{2}\parallel r_{o1})][R_{2}] \text{ V/A}
AOL=[−gm1(R1∥R2∥ro1)][R2] V/A
R i = R 2 R_{i}=R_{2} Ri=R2
R o = R 1 ∥ R 2 ∥ r o 1 R_{o}=R_{1}\parallel R_{2}\parallel r_{o1} Ro=R1∥R2∥ro1
β = − 1 R 2 \beta=-\frac{1}{R_{2}} β=−R21
Closed-loop gain:
A
C
L
=
v
2
i
s
=
−
g
m
1
R
o
R
2
1
+
g
m
1
R
o
R
2
1
R
2
A_{CL}=\frac{v_{2}}{i_{s}}=\frac{-g_{m1}R_{o}R_{2}}{1+g_{m1}R_{o}R_{2}\frac{1}{R_{2}}}
ACL=isv2=1+gm1RoR2R21−gm1RoR2
Input, Output Resistance:
R
i
n
f
=
v
1
i
s
=
R
2
1
+
g
m
1
R
o
R
2
1
R
2
R_{inf}=\frac{v_{1}}{i_{s}}=\frac{R_{2}}{1+g_{m1}R_{o}R_{2}\frac{1}{R_{2}}}
Rinf=isv1=1+gm1RoR2R21R2
R o f = R o 1 + g m 1 R o R 2 1 R 2 R_{of}=\frac{R_{o}}{1+g_{m1}R_{o}R_{2}\frac{1}{R_{2}}} Rof=1+gm1RoR2R21Ro
输出输入Voltage Gain:
v
2
v
1
=
v
2
i
s
⋅
i
s
v
1
=
A
C
L
⋅
1
R
i
n
f
=
−
g
m
1
R
o
=
−
g
m
1
(
R
1
∥
R
2
∥
r
o
1
)
\frac{v_{2}}{v_{1}}=\frac{v_{2}}{i_{s}}\cdot \frac{i_{s}}{v_{1}}=A_{CL}\cdot \frac{1}{R_{inf}}=-g_{m1}R_{o}=-g_{m1}(R_{1}\parallel R_{2}\parallel r_{o1})
v1v2=isv2⋅v1is=ACL⋅Rinf1=−gm1Ro=−gm1(R1∥R2∥ro1)
6. The Transconductance Amp (Series-Series Feedback)
Series-Series 串联-串联反馈, 采样输出的电流 反馈到输入的电压.
A C L = i o v i = A O L 1 + A O L ⋅ β A_{CL}=\frac{i_{o}}{v_{i}}=\frac{A_{OL}}{1+A_{OL}\cdot \beta} ACL=viio=1+AOL⋅βAOL
R i n f = R i ( 1 + A O L ⋅ β ) R_{inf}=R_{i}(1+A_{OL}\cdot \beta) Rinf=Ri(1+AOL⋅β)
R o f = R o ( 1 + A O L ⋅ β ) R_{of}=R_{o}(1+A_{OL}\cdot \beta) Rof=Ro(1+AOL⋅β)
transistor level circuit如下所示
Open-loop小信号模型:
Rout:
7. The Current Amplifier (Shunt-Series Feedback)
Shunt-Series 电流-电流负反馈如下图所示,
输入输出阻抗:
transistor电路如下:
Rout和Rin
β
=
i
f
i
o
=
−
g
m
2
r
o
3
A/A
\beta=\frac{i_{f}}{i_{o}}=-g_{m2}r_{o3} \text{ A/A}
β=ioif=−gm2ro3 A/A
8. Stability
Loop gain
T
=
A
O
L
β
T=A_{OL}\beta
T=AOLβ
为了环路稳定性, phase margin至少45 deg, 60 deg更好
phase:
loop gain:
对于pole极点p1: gain在p1处开始以20dB/十倍频 的速度下降. phase在p1/10处开始下降, 降到p1处为-45°, 在10*p1处降到-90°, 结束.
对于zero极点z1: gain在z1处开始以20dB/十倍频 的速度上升. phase在z1/10处开始上升, 升到p1处为-45°, 在10*p1处升到+90°, 结束.
The Return Ratio
Return ratio就是断环, 找loop gain. 断环点一般找MOSFET的gain, 或者impedance高处.
9. Design Examples
9.1 Voltage Amplifiers (series-shunt)
上图展示了series-shunt 即voltage in-out的电路结构.
其中b)为unity gain buffer, 输入阻抗极大, 输出阻抗小, 适合做buffer.
输入电压范围: Vin_min=Vgs+Vdsat, Vin_max=VDD-Vdsat+Vthn.
输出电压范围: Vout swing form Vdd-Vdsat to Vdsat.
比起其他常规unity gain buffer, 这种结构bandwidth更宽.
由于M1的body effect, gain可能不为1, 解决方法就是把M1 local-tie即Source和Body接到一起. 如果工艺不允许这样做, 我们可以用PMOS作为输入管,即下图
Amplifiers with Gain
对于不是unity-gain的应用, 31.48 a) 加入R2,R1 series-shunt有gain. M1的电流为MR3, M2的电流取决与R1和R2的值, 这不好.
31.48 b) 加入source follower解决这个问题, M2 电流为 MRL, 很好控制. MSF可以sink 来自MSL和R1-R2的 current, 因为MSF的gate是free的, 这也是为什么要用PMOS做source follower,而不是NMOS.
Vout的bode plot有一个翘起来的角, 这是因为added source follower. The impedance looking into the output of the source follower (the source of MSF) becomes inductive when it’s driven with a resistive load. 从source-follower的source端看进去会有电感性, 当source接电阻.
9.2 A Transimpedance Amplifier
transimpedance (shunt-shunt) 电压-电流 放大器
Closed-loop gain:
A
C
L
=
1
β
=
−
1
j
ω
C
F
=
v
o
u
t
−
i
d
A_{CL}=\frac{1}{\beta}=\frac{-1}{j\omega C_{F}}=\frac{v_{out}}{-i_{d}}
ACL=β1=jωCF−1=−idvout
1181
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
