Professional Documents
Culture Documents
م
َ ُْ ََ
Suez Canal University
Faculty of Engineering, Ismailia
Electrical Engineering Department
FEEDBACK CONCEPTS
Feedforward
Vin
+ Verr
Vout
_+
(Actuator)
AOL
Vfb Feedback
(Sensor)
β
2
Closed-loop Gain
𝑉𝑜𝑢𝑡 = 𝐴𝑂𝐿 𝑉𝑖𝑛 − 𝑉𝑓𝑏 = 𝐴𝑂𝐿 𝑉𝑖𝑛 − 𝛽𝑉𝑜𝑢𝑡
𝑉𝑜𝑢𝑡 𝐴𝑂𝐿 𝐴𝑂𝐿
𝐴𝐶𝐿 = = =
𝑉𝑖𝑛 1 + 𝛽𝐴𝑂𝐿 1 + 𝐿𝐺
❑ Loop gain = 𝐿𝐺 = 𝛽𝐴𝑂𝐿 ≫ 1
1
𝐴𝐶𝐿 ≈
𝛽
❑ Closed-loop gain is independent of open-loop gain!
Feedforward
Vin
+ Verr
Vout
_+
(Actuator)
AOL
Vfb Feedback
(Sensor)
β
3
Error Signal
𝑉𝑒𝑟𝑟 = 𝑉𝑖𝑛 − 𝑉𝑓𝑏 = 𝑉𝑖𝑛 − 𝛽𝑉𝑜𝑢𝑡 = 𝑉𝑖𝑛 − 𝛽𝐴𝑂𝐿 𝑉𝑒𝑟𝑟
𝑉𝑖𝑛 𝑉𝑖𝑛
𝑉𝑒𝑟𝑟 = = →0
1 + 𝛽𝐴𝑂𝐿 1 + 𝐿𝐺
❑ Loop gain = 𝐿𝐺 = 𝛽𝐴𝑂𝐿 ≫ 1
𝑉𝑖𝑛
𝑉𝑒𝑟𝑟 = →0
1 + 𝐿𝐺
❑ Negative feedback loop works to minimize the error signal
Feedforward
Vin
+ Verr
Vout
_+
(Actuator)
AOL
Vfb Feedback
(Sensor)
β
4
Feedback Example
❑ Op-amp functions: (1) subtraction and (2) amplification
❑ The network 𝑅1 and 𝑅2 functions: (1) sensing the output voltage and (2) providing a
𝑅2
feedback factor 𝛽 =
𝑅1 +𝑅2
Vin
+ Verr
Vout Vin
_+
AOL
Verr AOL Vout
Vfb R1
β Vfb
β
R2
5
Loop Gain
❑ Deactivate the input → Break the loop → Apply a test source → Calculate the gain around
the loop
𝑉𝑜𝑢𝑡
❑ Loop gain = 𝐿𝐺 = = 𝛽𝐴𝑂𝐿
𝑉𝑡
❑ But biasing/loading changes when we break the loop!
▪ Make sure dc biasing is properly set
▪ Add a dummy load
Vin = 0 + Verr
+ AOL Vout
_
Vfb
β Vt
6
Loop Gain
𝑉𝑜𝑢𝑡
❑ Loop gain = 𝐿𝐺 = = 𝛽𝐴𝑂𝐿
𝑉𝑡
❑ Circuit simulators can compute the loop gain without explicitly breaking the loop
▪ Insert a 0V dc voltage source or iprobe in the loop
▪ Use stability (STB) analysis
+ Verr
Vin + AOL Vout
_
V=0
Vfb
β
7
Why Negative Feedback?
❑ We use a very high gain amplifier 𝐴𝑂𝐿 , but end up with a much smaller gain
𝐴𝑂𝐿 1
𝐴𝐶𝐿 = ≈
1+𝛽𝐴𝑂𝐿 𝛽
❑ We can design high gain amplifiers, but we really do not need all that gain
❑ High gain is the balance that we use to buy other useful properties
❑ Negative feedback useful properties
1. Gain Desensitization (Gain de-sensitivity )→ Accurate, stable, and linear
gain
2. Bandwidth Extension
3. Modification of I/O Impedances
8
Gain Desensitization
❑ In IC design, we cannot control absolute values due to PVT, load, and input signal variations
❑ But we can precisely control ratios of MATCHED components
𝑌 𝐴𝑂𝐿 𝐴𝑂𝐿 𝑅1 + 𝑅2 𝑅1
𝐴𝐶𝐿 = = = ≈ =1+
𝑋 1 + 𝛽𝐴𝑂𝐿 1 + 𝑅2 𝑅2 𝑅2
⋅ 𝐴𝑂𝐿
𝑅1 + 𝑅2
❑ 𝑅1 = 3𝑅 𝑎𝑛𝑑 𝑅2 = 𝑅 ⇒ 𝐴𝐶𝐿 = 4 ⇒ Accurate, stable, and linear
Vin
Verr AOL Vout
R1
Vfb
β
R2
[Razavi, 2014] 9
Bandwidth Extension
❑ Assume the op-amp (OL) is a first order system
𝐴𝑂𝐿𝑜
𝐴𝑂𝐿 𝑠 = 𝑠
1+
𝜔𝑝,𝑂𝐿
❑ The CL transfer function is also a first order system
𝐴𝑂𝐿𝑜 𝐴𝑂𝐿𝑜
𝐴𝑂𝐿 𝑠 1 + 𝛽𝐴𝑂𝐿𝑜 1 + 𝐿𝐺𝑜
𝐴𝐶𝐿 𝑠 = = 𝑠 = 𝑠
1 + 𝛽𝐴𝑂𝐿 𝑠 1+ 1+
1 + 𝛽𝐴𝑂𝐿𝑜 𝜔𝑝,𝑂𝐿 1 + 𝐿𝐺𝑜 𝜔𝑝,𝑂𝐿
▪ But the pole is at a much higher frequency
𝜔𝑝,𝐶𝐿 = 1 + 𝐿𝐺𝑜 𝜔𝑃,𝑂𝐿
10
Bandwidth Extension
❑ Cascade of feedback amplifiers provides the same gain and a much faster response
▪ But power consumption and area doubled
[Razavi, 2017] 11
The Price We Pay
❑ Negative feedback useful properties
1. Gain Desensitization → Accurate, stable, and linear gain
2. Bandwidth Extension
3. Modification of I/O Impedances
❑ The price we pay to buy these useful properties
1. Gain reduction
2. The risk of instability
12
Stability of Feedback System
𝑉𝑜𝑢𝑡 𝑠 𝐴𝑂𝐿 𝑠
𝐴𝐶𝐿 𝑠 = =
𝑉𝑖𝑛 𝑠 1 + 𝛽𝐴𝑂𝐿 𝑠
Vin
+ Verr
Vout
_ + AOL
Vfb
β
13
Types of Amplifiers
14
How to Connect the Feedback
15
Effect of Feedback on I/O Impedances
16
Conclusion
17
Thank you!
18
20
21
22
23
24
25
26
27
28
29