ECE 4224 — Lecture 7

Controller Design
(Chapter 9)
Reminders and Announcements
▪Homework #1 solutions and grades are on Canvas
▪Homework #2 due Sept. 16th
▪Refer to Chapter 8.1 in the textbook for a detailed review of Bode plots
▪Instructor office hours: Thursday, 4pm-5pm
▪TA office hours: Tuesday, 7pm-8pm
▪If the office hour times do not work for you, please send me an email
▪Install Matlab Simulink and PLECS (see instructions on Canvas → Files
→ PLECS)
▪Tentative dates:
▪ Project #1 assigned Sept. 20th & due Oct. 4th
▪ Midterm exam on Oct. 11th
2
Controller Design Overview
▪Introduction
▪Negative feedback
▪Loop gain and closed-loop transfer functions
▪Stability
▪Regulator design
▪Measurement of loop gain

3
Controller Design Overview
▪Introduction
▪Negative feedback
▪Loop gain and closed-loop transfer functions
▪Stability
▪Regulator design
▪Measurement of loop gain

4
Small-Signal Block Diagram
𝒊Ƹ𝒍𝒐𝒂𝒅 (𝒔)

ෝ𝒈 (𝒔)
𝒗 𝒁𝒐𝒖𝒕 (𝒔)
𝑮𝒗𝒈 (𝒔)

ෝ𝒓𝒆𝒇 (𝒔)
𝒗 ෝ𝑣ො𝒆𝑒 (𝑠)
𝒗 (𝒔) ෝ𝑣ො𝑒𝒄(𝑠)
𝒗 (𝒔) ෡
𝒅(𝒔) ෝ(𝒔)
𝒗
𝑮𝒄 (𝒔) 𝟏/𝑽𝑴 𝑮𝒗𝒅 (𝒔) +

ෝ(𝒔)
𝑯(𝒔) ∙ 𝒗
𝑯(𝒔)

𝟏 𝟏
ෝ = −ෝ
𝒗 𝒗 ∙ 𝑯 ∙ 𝑮𝒄 ∙ ෝ𝒓𝒆𝒇 ∙ 𝑮𝒄 ∙
∙ 𝑮𝒗𝒅 + 𝒗 ෝ𝒈 − 𝒁𝒐𝒖𝒕 ∙ 𝒊Ƹ𝒍𝒐𝒂𝒅
∙ 𝑮𝒗𝒅 + 𝑮𝒗𝒈 ∙ 𝒗
𝑽𝑴 𝑽𝑴
5
Small-Signal Block Diagram
𝒊Ƹ𝒍𝒐𝒂𝒅 (𝒔)

ෝ𝒈 (𝒔)
𝒗 𝒁𝒐𝒖𝒕 (𝒔)
𝑮𝒗𝒈 (𝒔)

ෝ𝒓𝒆𝒇 (𝒔)
𝒗 ෝ𝑣ො𝒆𝑒 (𝑠)
𝒗 (𝒔) ෝ𝑣ො𝑒𝒄(𝑠)
𝒗 (𝒔) ෡
𝒅(𝒔) ෝ(𝒔)
𝒗
+ 𝑮𝒄 (𝒔) 𝟏/𝑽𝑴 𝑮𝒗𝒅 (𝒔) +

ෝ(𝒔)
𝑯(𝒔) ∙ 𝒗
𝑯(𝒔)

𝟏 𝟏
ෝ = −ෝ
𝒗 𝒗 ∙ 𝑯 ∙ 𝑮𝒄 ∙ ෝ𝒓𝒆𝒇 ∙ 𝑮𝒄 ∙
∙ 𝑮𝒗𝒅 + 𝒗 ∙ 𝑮𝒗𝒅
𝑽𝑴 𝑽𝑴
6
Small-Signal Block Diagram
𝒊Ƹ𝒍𝒐𝒂𝒅 (𝒔)

ෝ𝒈 (𝒔)
𝒗 𝒁𝒐𝒖𝒕 (𝒔)
𝑮𝒗𝒈 (𝒔)

ෝ𝒓𝒆𝒇 (𝒔)
𝒗 ෝ𝑣ො𝒆𝑒 (𝑠)
𝒗 (𝒔) ෝ𝑣ො𝑒𝒄(𝑠)
𝒗 (𝒔) ෡
𝒅(𝒔) + ෝ(𝒔)
𝒗
𝑮𝒄 (𝒔) 𝟏/𝑽𝑴 𝑮𝒗𝒅 (𝒔)

ෝ(𝒔)
𝑯(𝒔) ∙ 𝒗
𝑯(𝒔)

𝟏 𝟏
ෝ = −ෝ
𝒗 𝒗 ∙ 𝑯 ∙ 𝑮𝒄 ∙ ෝ𝒓𝒆𝒇 ∙ 𝑮𝒄 ∙
∙ 𝑮𝒗𝒅 + 𝒗 ෝ𝒈 ∙ 𝑮𝒗𝒈
∙ 𝑮𝒗𝒅 + 𝒗
𝑽𝑴 𝑽𝑴
7
Small-Signal Block Diagram
𝒊Ƹ𝒍𝒐𝒂𝒅 (𝒔)

ෝ𝒈 (𝒔)
𝒗 𝒁𝒐𝒖𝒕 (𝒔)
𝑮𝒗𝒈 (𝒔)

ෝ𝒓𝒆𝒇 (𝒔)
𝒗 ෝ𝑣ො𝒆𝑒 (𝑠)
𝒗 (𝒔) ෝ𝑣ො𝑒𝒄(𝑠)
𝒗 (𝒔) ෡
𝒅(𝒔) ෝ(𝒔)
𝒗
𝑮𝒄 (𝒔) 𝟏/𝑽𝑴 𝑮𝒗𝒅 (𝒔)

ෝ(𝒔)
𝑯(𝒔) ∙ 𝒗
𝑯(𝒔)

𝟏 𝟏
ෝ = −ෝ
𝒗 𝒗 ∙ 𝑯 ∙ 𝑮𝒄 ∙ ෝ𝒓𝒆𝒇 ∙ 𝑮𝒄 ∙
∙ 𝑮𝒗𝒅 + 𝒗 ෝ𝒈 ∙ 𝑮𝒗𝒈 − 𝒊Ƹ𝒍𝒐𝒂𝒅 ∙ 𝒁𝒐𝒖𝒕
∙ 𝑮𝒗𝒅 + 𝒗
𝑽𝑴 𝑽𝑴
8
Small-Signal Block Diagram
𝒊Ƹ𝒍𝒐𝒂𝒅 (𝒔)

ෝ𝒈 (𝒔)
𝒗 𝒁𝒐𝒖𝒕 (𝒔)
𝑮𝒗𝒈 (𝒔)

ෝ𝒓𝒆𝒇 (𝒔)
𝒗 ෝ𝑣ො𝒆𝑒 (𝑠)
𝒗 (𝒔) ෝ𝑣ො𝑒𝒄(𝑠)
𝒗 (𝒔) ෡
𝒅(𝒔) ෝ(𝒔)
𝒗
𝑮𝒄 (𝒔) 𝟏/𝑽𝑴 𝑮𝒗𝒅 (𝒔)

ෝ(𝒔)
𝑯(𝒔) ∙ 𝒗
𝑯(𝒔)

𝟏 𝟏
ෝ = −ෝ
𝒗 𝒗 ∙ 𝑯 ∙ 𝑮𝒄 ∙ ෝ𝒓𝒆𝒇 ∙ 𝑮𝒄 ∙
∙ 𝑮𝒗𝒅 + 𝒗 ෝ𝒈 ∙ 𝑮𝒗𝒈 − 𝒊Ƹ𝒍𝒐𝒂𝒅 ∙ 𝒁𝒐𝒖𝒕
∙ 𝑮𝒗𝒅 + 𝒗
𝑽𝑴 𝑽𝑴
9
Small-Signal Block Diagram Solution
▪Solve for 𝑣:
ො
1 1
ෝ = −ෝ
𝒗 𝒗 ∙ 𝐻 ∙ 𝐺𝑐 ∙ ∙ 𝐺𝑣𝑑 + 𝑣ො𝑟𝑒𝑓 ∙ 𝐺𝑐 ∙ ∙ 𝐺𝑣𝑑 + 𝑣ො𝑔 ∙ 𝐺𝑣𝑔 − 𝑖Ƹ𝑙𝑜𝑎𝑑 ∙ 𝑍𝑜𝑢𝑡
𝑉𝑀 𝑉𝑀

𝟏 1
𝑣ො 𝟏 + 𝑯 ∙ 𝑮𝒄 ∙ ∙ 𝑮𝒗𝒅 = 𝑣ො𝑟𝑒𝑓 ∙ 𝐺𝑐 ∙ ∙ 𝐺𝑣𝑑 + 𝑣ො𝑔 ∙ 𝐺𝑣𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ ∙ 𝑍𝑜𝑢𝑡
𝑽𝑴 𝑉𝑀

1 1
𝑣ො = 𝑣ො𝑟𝑒𝑓 ∙ 𝐺𝑐 ∙ ∙ 𝐺𝑣𝑑 + 𝑣ො𝑔 ∙ 𝐺𝑣𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ ∙ 𝑍𝑜𝑢𝑡 ∙ 𝟏
𝑉𝑀 𝟏+𝑯∙𝑮𝒄 ∙𝑽 ∙𝑮𝒗𝒅
𝑴

1
𝑮𝒗𝒅 ∙ 𝑮𝒄 ∙ 𝑮𝒗𝒈 𝒁𝒐𝒖𝒕
𝑉𝑀
ෝ=𝒗
𝒗 ෝ𝒓𝒆𝒇 ෝ𝒈
+𝒗 − 𝒊Ƹ𝒍𝒐𝒂𝒅
1 1 1
1 + 𝐻 ∙ 𝐺𝑐 ∙ ∙ 𝐺𝑣𝑑 1 + 𝐻 ∙ 𝐺𝑐 ∙ ∙ 𝐺𝑣𝑑 1 + 𝐻 ∙ 𝐺𝑐 ∙ ∙𝐺
𝑉𝑀 𝑉𝑀 𝑉𝑀 𝑣𝑑
10
Small-Signal Block Diagram Solution
▪Solve for 𝑣:
ො
1
𝐺𝑐 ∙ 𝑉 ∙ 𝐺𝑣𝑑 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡
𝑀
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖Ƹ𝑙𝑜𝑎𝑑
1 1 1
1 + 𝐻 ∙ 𝐺𝑐 ∙ ∙𝐺 1 + 𝐻 ∙ 𝐺𝑐 ∙ ∙𝐺 1 + 𝐻 ∙ 𝐺𝑐 ∙ ∙𝐺
𝑉𝑀 𝑣𝑑 𝑉𝑀 𝑣𝑑 𝑉𝑀 𝑣𝑑

1 𝑇 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ
𝐻1+𝑇 1+𝑇 1+𝑇

where 1
𝑇 𝑠 = 𝐻 ∙ 𝐺𝑐 ∙ ∙ 𝐺𝑣𝑑
𝑉𝑀
▪𝑻 𝒔 is the loop gain, which is equal to the products of the gains
around the negative feedback loop
11
Terminology: Open-loop vs. Closed-loop
1 𝑇 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ
𝐻1+𝑇 1+𝑇 1+𝑇
▪Original transfer functions, before ▪After introducing the feedback,
introducing the feedback (“open- these become (“closed-loop transfer
loop transfer functions”): functions”):
1 𝑇
▪ Control-to-output: 𝐺𝑣𝑑
𝐻1+𝑇
𝐺𝑣𝑔
▪ Line-to-output: 𝐺𝑣𝑔
1+𝑇
𝑍𝑜𝑢𝑡
▪ Output impedance: 𝑍𝑜𝑢𝑡
1+𝑇 12
Closed-Loop Line-to-Output Transfer Function 𝐺𝑣𝑔 𝑠
▪Original (open-loop) line-to-output transfer function:
𝑣ො 𝑠
𝐺𝑣𝑔 𝑠 = อ
𝑣ො𝑔 𝑠
෠
𝑑=0
𝑖𝑙𝑜𝑎𝑑
Ƹ =0
▪With the negative feedback (closed-loop):
𝑣ො 𝑠 𝐺𝑣𝑔
อ =
𝑣ො𝑔 𝑠 1+𝑇
𝑣ො 𝑟𝑒𝑓 =0
𝑖𝑙𝑜𝑎𝑑
Ƹ =0
1
▪The feedback reduces 𝐺𝑣𝑔 𝑠 by a factor of
1+𝑇 𝑠
▪If 𝑇 𝑠 is large in magnitude, then 𝐺𝑣𝑔 𝑠 becomes small
13
Closed-Loop Output Impedance 𝑍𝑜𝑢𝑡 𝑠
▪Original (open-loop) output impedance:
𝑣ො 𝑠
𝑍𝑜𝑢𝑡 𝑠 = − ቤ
𝑖𝑙𝑜𝑎𝑑
Ƹ 𝑠 ෠
𝑑=0
𝑣ො𝑔 = 0
▪With the negative feedback (closed-loop):
𝑣ො 𝑠 𝑍𝑜𝑢𝑡
− ቤ =
𝑖𝑙𝑜𝑎𝑑
Ƹ 𝑠 𝑣ො 𝑟𝑒𝑓 =0
1+𝑇
𝑣ො𝑔 = 0
1
▪The feedback reduces 𝑍𝑜𝑢𝑡 𝑠 by a factor of
1+𝑇 𝑠
▪If 𝑇 𝑠 is large in magnitude, then 𝑍𝑜𝑢𝑡 𝑠 becomes small
14
Closed-Loop Reference-to-Output Transfer Function
▪Closed-loop transfer function for 𝑣ො𝑟𝑒𝑓 𝑠 to 𝑣ො 𝑠 :
𝑣ො 𝑠 1 𝑇 𝑠
อ =
𝑣ො𝑟𝑒𝑓 𝑠 𝐻 𝑠 1 + 𝑇(𝑠)
𝑣ො𝑔 =0
𝑖𝑙𝑜𝑎𝑑
Ƹ =0
▪If 𝑇 𝑠 is large in magnitude ( 𝑇 ≫ 1), then:
▪ 1+𝑇 𝑠 ≈𝑇
𝑇 𝑇
▪ ≈ ≈1
1+𝑇 𝑠 𝑇
𝑣ො 𝑠 1
▪ො ≈ , which is independent of the gains in the forward loop path.
𝑣𝑟𝑒𝑓 𝑠 𝐻 𝑠
𝑉 𝑠 1 𝑇 0 1
▪ This result also applies to the DC values: = ≈
𝑉𝑟𝑒𝑓 𝑠 𝐻 0 1+𝑇 0 𝐻 0
15
Example: Construction of 𝑇(𝑠)
LHP Zero at 𝜔𝑧
DC Gain 𝑠
|𝑇| 1+𝜔
𝑧
𝑇 𝑠 = 𝑇0
𝑠 𝑠 2 𝑠
1+ + 1+
𝑄𝜔𝑝1 𝜔𝑝1 𝜔𝑝2

Double Pole Single Pole

at 𝜔𝑝1 at 𝜔𝑝2

▪ At the cross-over
frequency 𝑓𝑐 :
𝑇 =1
𝑓 16
Approximating 𝑇/(1 + 𝑇) and 1/(1 + 𝑇)
1 𝑇 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ
𝐻1+𝑇 1+𝑇 1+𝑇

𝑇 1 𝑓𝑜𝑟 𝑇 ≫1
≈ቊ
1+𝑇 𝑇 𝑓𝑜𝑟 𝑇 ≪1

1
1 𝑓𝑜𝑟 𝑇 ≫1
≈ ൞𝑇
1+𝑇
1 𝑓𝑜𝑟 𝑇 ≪1
17
Example: Construction of 𝑇/(1 + 𝑇)

𝑇 1 𝑓𝑜𝑟 𝑇 ≫1
≈ቊ
1+𝑇 𝑇 𝑓𝑜𝑟 𝑇 ≪1

0 dB

18
Example: Expressions for 𝑣ො𝑟𝑒𝑓 𝑠 to 𝑣ො 𝑠 Transfer Function

▪At 𝑓 < 𝑓𝑐 , 𝑇 > 1, so the transfer function from 𝑣ො𝑟𝑒𝑓 𝑠 to 𝑣ො 𝑠

becomes:
𝑣ො 𝑠 1 𝑇 𝑠 1
อ = ≈
𝑣ො𝑟𝑒𝑓 𝑠 𝐻 𝑠 1 + 𝑇(𝑠) 𝐻 𝑠
𝑣ො𝑔 =0
𝑖𝑙𝑜𝑎𝑑
Ƹ =0

▪This is the desired behavior: 𝑣ො 𝑠 follows 𝑣ො𝑟𝑒𝑓 𝑠 according to

1
the ideal gain
𝐻 𝑠
▪This means the feedback loop works well at frequencies where
the loop gain 𝑇 𝑠 has a large magnitude
19
Example: Expressions for 𝑣ො𝑟𝑒𝑓 𝑠 to 𝑣ො 𝑠 Transfer Function

▪At 𝑓 > 𝑓𝑐 , 𝑇 < 1, so the transfer function from 𝑣ො𝑟𝑒𝑓 𝑠 to 𝑣ො 𝑠

becomes:
𝑣ො 𝑠 1 𝑇 𝑠 𝑇 𝑠 𝐺𝑐 𝑠 𝐺𝑣𝑑 𝑠
อ = ≈ =
𝑣ො𝑟𝑒𝑓 𝑠 𝐻 𝑠 1 + 𝑇(𝑠) 𝐻 𝑠 𝑉𝑀
𝑣ො𝑔 =0
𝑖𝑙𝑜𝑎𝑑
Ƹ =0

▪This coincides with the open-loop 𝑣ො𝑟𝑒𝑓 𝑠 to 𝑣ො 𝑠 transfer

function
▪At frequencies where 𝑇 < 1, the loop has essentially no effect
on the 𝑣ො𝑟𝑒𝑓 𝑠 to 𝑣ො 𝑠 transfer function

20
Example: Construction of 1/(1 + 𝑇)
1
1 𝑓𝑜𝑟 𝑇 ≫ 1
≈ ൞𝑇 𝑠
1+𝑇
1 𝑓𝑜𝑟 𝑇 ≪ 1

21
Interpretation of How the Loop Rejects Disturbances
1 𝑇 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ
𝐻1+𝑇 1+𝑇 1+𝑇

▪At 𝑓 < 𝑓𝑐 , and 𝑇 > 1:

1 1 1
▪ ≈ , and disturbances are reduced in magnitude by
1+𝑇 𝑇 𝑇

▪At 𝑓 > 𝑓𝑐 , and 𝑇 < 1:

1
▪ ≈ 1 , and the feedback loop has essentially no effect on
1+𝑇
disturbances
22
What About the Phase of 𝑇 𝑠 ?
▪What if the phase of 𝑇 𝑠 is equal +180° or -180° at the
crossover frequency 𝜔𝑐 ?
▪ 𝑇 𝑠 = 𝑅 𝜔𝑐 𝑒 𝑗𝜃(𝜔𝑐 )
▪ At 𝜔𝑐 , 𝑅 𝜔𝑐 = 1:
𝑇 𝑠 = (1)𝑒 𝑗𝜃(𝜔𝑐)
▪ If 𝜃 𝜔𝑐 = 180°:
𝑇 𝑠 = 𝑒 𝑗180° = cos 180° + 𝑗𝑠𝑖𝑛 180° = −1 + 0 = −1

1 𝑇 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ
𝐻1+𝑇 1+𝑇 1+𝑇
∞ ∞ ∞
1 −1 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡 We have no control over
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ the output!
𝐻 0 0 0 The system is unstable.
23
For the Feedback to Work Well:
1. The magnitude of 𝑇 𝑠 should be >> 1 over as wide of a
frequency range as possible

2. When 𝑇 = 1 (at the crossover frequency), then the phase of

𝑇 𝑠 must be far away from +180° or -180° to avoid instability

1 𝑇 𝐺𝑣𝑔 𝑍𝑜𝑢𝑡
𝑣ො = 𝑣ො𝑟𝑒𝑓 + 𝑣ො𝑔 − 𝑖𝑙𝑜𝑎𝑑
Ƹ
𝐻1+𝑇 1+𝑇 1+𝑇

24
Controller Design Overview
▪Introduction
▪Negative feedback
▪Loop gain and closed-loop transfer functions
▪Stability
▪Regulator design
▪Measurement of loop gain

25
Stability
▪Adding a feedback loop can cause an otherwise stable system
to become unstable.

▪Even though the open-loop transfer functions and 𝑇(𝑠) do not

contain right half-plane (RHP) poles, it is possible for the closed-
loop transfer functions to contain RHP poles.
▪ The feedback loop then fails to regulate the system at the desired
operating point, and oscillations occur.

▪Even if the system is stable, it is possible for the transient

response to exhibit undesirable ringing and overshoot.
26
Stability
▪When the feedback destabilizes the system, the denominator
(1 + 𝑇 𝑠 ) terms in the closed-loop transfer functions contain
roots in the RHP
▪If 𝑇 𝑠 is a rational fraction of the form 𝑁 𝑠 /𝐷 𝑠 , where 𝑁 𝑠
and 𝐷 𝑠 are polynomials:

𝑁 𝑠
𝑇 𝑠 𝐷 𝑠 𝑁 𝑠 1 1 𝐷 𝑠
= 𝑁 𝑠 = and = 𝑁 𝑠 =
1+𝑇 𝑠 1+ 𝑁 𝑠 +𝐷 𝑠 1+𝑇 𝑠 1+ 𝑁 𝑠 +𝐷 𝑠
𝐷 𝑠 𝐷 𝑠

27
Stability
▪When the feedback destabilizes the system, the denominator
(1 + 𝑇 𝑠 ) terms in the closed-loop transfer functions contain
roots in the RHP
▪If 𝑇 𝑠 is a rational fraction of the form 𝑁 𝑠 /𝐷 𝑠 , where 𝑁 𝑠
and 𝐷 𝑠 are polynomials:

𝑁 𝑠
𝑇 𝑠 𝐷 𝑠 𝑁 𝑠 1 1 𝐷 𝑠
= 𝑁 𝑠 = and = 𝑁 𝑠 =
1+𝑇 𝑠 1+ 𝑁 𝑠 +𝐷 𝑠 1+𝑇 𝑠 1+ 𝑁 𝑠 +𝐷 𝑠
𝐷 𝑠 𝐷 𝑠

▪We could test the stability by evaluating 𝑁 𝑠 + 𝐷 𝑠 , then

factoring to evaluate the roots, but this is a lot of work. 28
Stability
▪Alternatively, we can determine the stability directly from 𝑇 𝑠 .
▪A special case of the Nyquist stability theorem: phase margin
test
▪ Allows determination of the closed-loop stability directly from the
magnitude and phase of 𝑇 𝑠
▪ Provides insight into how 𝑇 𝑠 should be shaped to obtain good
performance

29
Phase Margin Test
1
▪A test on 𝑇 𝑠 to determine whether contains RHP poles.
1+𝑇 𝑠

▪Procedure:
1. Find the crossover frequency 𝑓𝑐 : 𝑇 𝑗2𝜋𝑓𝑐 = 1 → 0𝑑𝐵
2. Find the phase of 𝑇 𝑠 at 𝑓𝑐 : ∠𝑇 𝑗2𝜋𝑓𝑐
3. Calculate the phase margin 𝜑𝑚 : 𝜑𝑚 = 180° + ∠𝑇 𝑗2𝜋𝑓𝑐

▪If there is exactly one 𝑓𝑐 , and if 𝑇 𝑠 contains no RHP poles,

𝑇 𝑠 1
then and contain no RHP poles when 𝝋𝒎 > 𝟎.
1+𝑇 𝑠 1+𝑇 𝑠

30
Example: Loop Gain Leading to a Stable Closed-Loop System

𝒇𝒄

∠𝑻 𝒋𝟐𝝅𝒇𝒄
−𝟏𝟏𝟐°
𝝋𝒎

𝝋𝒎 = 𝟏𝟖𝟎° − 𝟏𝟏𝟐° = +𝟔𝟖°

𝝋𝒎 = +𝟔𝟖° > 𝟎 → 𝒔𝒕𝒂𝒃𝒍𝒆

31
Example: Loop Gain Leading to an Unstable Closed-Loop System

𝒇𝒄

𝝋𝒎 = 𝟏𝟖𝟎° − 𝟐𝟑𝟎° = −𝟓𝟎°

𝝋𝒎 = −𝟓𝟎° < 𝟎 → 𝒖𝒏𝒔𝒕𝒂𝒃𝒍𝒆

𝝋𝒎
−𝟐𝟑𝟎°
∠𝑻 𝒋𝟐𝝅𝒇𝒄

32
Phase Margin and Closed-Loop Damping Factor
▪How much phase margin 𝜑𝑚 is required?
▪For a small, positive phase margin 𝜑𝑚 :
▪ The closed-loop system will be stable
▪ The system will have complex poles near 𝑓𝑐 with high 𝑄.
▪ The transient response will exhibit overshoot and ringing.
▪Increasing 𝝋𝒎 reduces 𝑸.
▪A large phase margin is required to obtain real poles with no
overshoot or ringing.
▪Want to quantify the relation between 𝜑𝑚 and 𝑄.

33
Example: Second-Order System

Pole at Origin Pole at 𝜔2

34
Example: Second-Order System
1
If 𝑇 𝑠 = 𝑠 𝑠 ,
1+
𝜔0 𝜔2

𝑇 𝑠 1 1
then = 1 = 𝑠 𝑠2
1+𝑇 𝑠 1+ 1+ +
𝑇 𝑠 𝜔0 𝜔0 𝜔2

𝑇 𝑠 1
or, = 𝑠 𝑠 2
1+𝑇 𝑠 1+ +
𝑄𝜔𝑐 𝜔𝑐

𝜔0 𝜔0
where, 𝜔𝑐 = 𝜔0 𝜔2 = 2𝜋𝑓𝑐 𝑄= =
𝜔𝑐 𝜔2
35
Example: Second-Order System
1
If 𝑇 𝑠 = 𝑠 𝑠 ,
1+
𝜔0 𝜔2

𝑇 𝑠 1 1
then = 1 = 𝑠 𝑠2
1+𝑇 𝑠 1+ 1+ +
𝑇 𝑠 𝜔0 𝜔0 𝜔2

𝑇 𝑠 1
or, = 𝑠 𝑠 2
1+𝑇 𝑠 1+ +
𝑄𝜔𝑐 𝜔𝑐

𝜔0 𝜔0
where, 𝜔𝑐 = 𝜔0 𝜔2 = 2𝜋𝑓𝑐 𝑄= =
𝜔𝑐 𝜔2
36
Example: Second-Order System
1
If 𝑇 𝑠 = 𝑠 𝑠 ,
1+
𝜔0 𝜔2

𝑇 𝑠 1 1
then = 1 = 𝑠 𝑠2
1+𝑇 𝑠 1+ 1+ +
𝑇 𝑠 𝜔0 𝜔0 𝜔2

1
Standard form: 𝑠 𝑠 2
1+ +
𝑄𝜔𝑐 𝜔𝑐

𝜔0 𝜔0
Equate coefficients: 𝜔𝑐 = 𝜔0 𝜔2 = 2𝜋𝑓𝑐 𝑄= =
𝜔𝑐 𝜔2
37
Example: Second-Order System
1
If 𝑇 𝑠 = 𝑠 𝑠 ,
1+
𝜔0 𝜔2

𝑇 𝑠 1 1
then = 1 = 𝑠 𝑠2
1+𝑇 𝑠 1+ 1+ +
𝑇 𝑠 𝜔0 𝜔0 𝜔2

1
Standard form: 𝑠 𝑠 2
1+ +
𝑄𝜔𝑐 𝜔𝑐

𝜔0 𝜔0
Equate coefficients: 𝜔𝑐 = 𝜔0 𝜔2 = 2𝜋𝑓𝑐 𝑄= =
𝜔𝑐 𝜔2
38
Example: Second-Order System
1
If 𝑇 𝑠 = 𝑠 𝑠 ,
1+
𝜔0 𝜔2

𝑇 𝑠 1 1
then = 1 = 𝑠 𝑠2
1+𝑇 𝑠 1+ 1+ +
𝑇 𝑠 𝜔0 𝜔0 𝜔2

1
Standard form: 𝑠 𝑠 2
1+ +
𝑄𝜔𝑐 𝜔𝑐

𝜔0 𝜔0
Equate coefficients: 𝜔𝑐 = 𝜔0 𝜔2 = 2𝜋𝑓𝑐 𝑄= =
𝜔𝑐 𝜔2
39
𝑄 vs. 𝜑𝑚 𝑄=
𝑐𝑜𝑠𝜑𝑚
𝜑𝑚 = tan−1
1 + 1 + 4𝑄4
𝑠𝑖𝑛𝜑𝑚 2𝑄4

1
𝑇 𝑠 =
𝑠 𝑠
1+𝜔
𝜔0 2

40
𝑄 vs. 𝜑𝑚 𝑄=
𝑐𝑜𝑠𝜑𝑚
𝜑𝑚 = tan−1
1 + 1 + 4𝑄4
𝑠𝑖𝑛𝜑𝑚 2𝑄4

Q very high! 𝑇 𝑠 =
1
𝑠 𝑠
1+𝜔
𝜔0 2

1° 41
𝑄 vs. 𝜑𝑚 𝑄=
𝑐𝑜𝑠𝜑𝑚
𝜑𝑚 = tan−1
1 + 1 + 4𝑄4
𝑠𝑖𝑛𝜑𝑚 2𝑄4

1
𝑇 𝑠 =
𝑠 𝑠
1+𝜔
𝜔0 2

Q ≤ 0.5 (real poles)

≥76° 42
Transient Response vs. Damping Factor

𝑄 > 0.5
Underdamped

𝑄 = 0.5
Critically damped

𝑄 < 0.5
Overdamped

43
Reminders and Announcements
▪Homework #1 solutions and grades are on Canvas
▪Homework #2 due Sept. 16th
▪Refer to Chapter 8.1 in the textbook for a detailed review of Bode plots
▪Instructor office hours: Thursday, 4pm-5pm
▪TA office hours: Tuesday, 7pm-8pm
▪If the office hour times do not work for you, please send me an email
▪Install Matlab Simulink and PLECS (see instructions on Canvas → Files
→ PLECS)
▪Tentative dates:
▪ Project #1 assigned Sept. 20th & due Oct. 4th
▪ Midterm exam on Oct. 11th
44