06/03/2017

UWE Bristol Today’s Lecture

Industrial Control • Frequency Response
UFMF6W-20-2

Control Systems Engineering • Amplitude Ratio and Phase

UFMEUY-20-3

• System Analysis using

Lecture 6: Frequency Response
– Bode Plot
Bode Plots and Nyquist Diagrams – Nyquist Diagram
• System Identification

Frequency Response Frequency Response

• Up to now – Time response • System behaviour determined from the
– How the system behaves to a specific input steady state response to a sinusoidal input
with respect to time in the form:
• Now – Frequency Response R = A sin wt
– System response to a sinusoidal input • Sine wave used:
– Range of frequencies used – easy to analyse
– Used for system identification – easy to generate
– Used for stability analysis (Next week) – easy to measure experimentally

Frequency Response Frequency Response

• Sinusoidal applied to LINEAR system: Input
G(s)
Output

– output will be sinusoidal A sin wt

– output amplitude proportional to input
amplitude
– harmonic input produces harmonic output at
same frequency
• Variation in Amplitude and Phase
• Function of Frequency

© 2017 University of the West of

England, Bristol. 1
 06/03/2017

Frequency Response Frequency Response

Input Output
G(s)
Input Output A sin wt Bsin(wt + f )
G(s)
A sin wt Bsin(wt + f )

A
B

B
• Amplitude Ratio (AR) AR = [dimensionless]
A

Frequency Response Frequency Response

Input Output Input Output
G(s) G(s)
A sin wt Bsin(wt + f ) A sin wt Bsin(wt + f )
One Period = 360! f

• Phase Shift (angle) f [degrees]

• If output follows input à LAG à negative

Frequency Response Frequency Response

• Data gained experimentally • First order system:
• Can be gained from OL transfer function
1 1
– Input: R = A sin wt G (s ) = ® G ( jw ) =
1 + ts 1 + jwt
– R can be represented as: R = Ae jwt (j = -1 )
– Take first derivative:
dR • Second order system:
R = A sin wt = Rs = jwA jwt
dt wn2 wn2
– Replace s terms in TF with jω G (s ) = ® G ( jw ) =
s + 2zw n s + wn
2 2
- w + 2zw n jw + wn2
2
• Moving from s-domain to frequency domain
Note: ω is driving frequency, ωn is natural frequency

© 2017 University of the West of

England, Bristol. 2
 06/03/2017

Frequency Response Frequency Response

• Consider second order transfer function: • Complex quantity that depends on frequency:
5 5
G (s) = 2 G ( jw ) =
(1 - 2w )+ j(3w )
2
2 s + 3s + 1
• Multiply by complex conjugate to separate real
• Substitute for s and imaginary parts:
5 (
ù é 1 - 2w 2 - j (3w ) ù )
G ( jw ) = é 5
2( jw ) + 3( jw ) + 1
G ( jw ) = ê
2
( ) úê (
ë 1 - 2w + j (3w ) û ë 1 - 2w - j (3w ) û
2 2 ú
)
5
G ( jw ) = [( ) ]
5
(1 - 2w )+ j(3w )
2 G ( jw ) =
(1 - 2w )
2 2
(
5 1 - 2w 2 - j (3w )
) ( )
+ j (3w ) 1 - 2w 2 - j (3w ) 1 - 2w 2 + ( j (3w ))(- j (3w ))
G ( jw ) =
- 2w 2 + 3 jw + 1 [( )
5 1 - 2w 2 - j (3w ) ] 5 - 10w 2 15w
G ( jw ) = G ( jw ) = -j 4
1 - 4w 2 + 4w 4 + 9w 2 4w 4 + 5w 2 + 1 4w + 5w 2 + 1

Frequency Response Frequency Response

• Complex number in Re + j Im form: • From complex number theory:
5 - 10w 2
15w • Amplitude ratio:
G ( jw ) = -j 4
4w 4 + 5w 2 + 1 4w + 5w 2 + 1
AR = Re 2 + Im 2
5 - 10w 2

• Real part: Re = • Phase angle:

4w 4 + 5w 2 + 1 æ Im ö
f = tan -1 ç ÷
è Re ø
15w
• Imaginary part: Im = - • Used to generate a table of results
4w 4 + 5w 2 + 1

Frequency Response Representing Data

Frequency (Rad/s) Real Imaginary AR Phase • Some methods
0.01 4.9965 –0.1499 4.9988 –1.7187
0.02 4.9860 –0.2994 4.9950 –3.4364
– Nyquist Diagram
0.03 4.9686 –0.4480 4.9888 –5.1520
– Bode Plot
0.05 4.9135 –0.7407 4.9690 –8.5730
0.1 4.6649 –1.4280 4.8786 –17.0205
0.2 3.8130 –2.4867 4.5522 –33.1113
0.3 2.7658 –3.0356 4.1066 –47.6630
0.5 1.0000 –3.0000 3.1623 –71.5651
1 -0.5000 –1.5000 1.5811 –108.4349
2 -0.4118 –0.3529 0.5423 –139.3987
3 -0.2297 –0.1216 0.2599 –152.1027
5 -0.0933 –0.0286 0.0976 –162.9795
10 -0.0246 –0.0037 0.0248 –171.4270

© 2017 University of the West of

England, Bristol. 3
 06/03/2017

Nyquist Diagram Nyquist Diagram

5 - 10w 2 15w
• Polar plot on Argand diagram of open loop • Example G( jw ) = -j 4
4w 4 + 5w 2 + 1 4w + 5w 2 + 1
frequency locus (i.e. G(jω)H(jω)) Frequency
(Rad/s) AR Phase
• Amplitude ratio is radial coordinate. 0.01 4.9988 –1.7187
0.02 4.9950 –3.4364
• Phase angle is the angular coordinate. 0.03 4.9888 –5.1520
0.05 4.9690 –8.5730
0.1 4.8786 –17.0205
0.2 4.5522 –33.1113
0.3 4.1066 –47.6630
0.5 3.1623 –71.5651
1 1.5811 –108.4349
2 0.5423 –139.3987
3 0.2599 –152.1027
5 0.0976 –162.9795
10 0.0248 –171.4270

Nyquist Diagram Nyquist Diagram

5 - 10w 2 15w 5 - 10w 2 15w
• Example G( jw ) = -j 4 • Example G( jw ) = -j 4
4w 4 + 5w 2 + 1 4w + 5w 2 + 1 4w 4 + 5w 2 + 1 4w + 5w 2 + 1
Frequenc Frequenc
y (Rad/s) AR Phase y (Rad/s) AR Phase
0.01 4.9988 –1.7187 0.01 4.9988 –1.7187
0.02 4.9950 –3.4364 0.02 4.9950 –3.4364
0.03 4.9888 –5.1520 0.03 4.9888 –5.1520
0.05 4.9690 –8.5730 0.05 4.9690 –8.5730
0.1 4.8786 –17.0205 0.1 4.8786 –17.0205
0.2 4.5522 –33.1113 0.2 4.5522 –33.1113
0.3 4.1066 –47.6630 0.3 4.1066 –47.6630
0.5 3.1623 –71.5651 0.5 3.1623 –71.5651
1 1.5811 –108.4349 1 1.5811 –108.4349
2 0.5423 –139.3987 2 0.5423 –139.3987
3 0.2599 –152.1027 3 0.2599 –152.1027
5 0.0976 –162.9795 5 0.0976 –162.9795
10 0.0248 –171.4270 10 0.0248 –171.4270

Bode Plots G( jw ) =
5 - 10w 2
-j 4
15w Bode Plots
4w 4 + 5w 2 + 1 4w + 5w 2 + 1
• Two logarithmic plots • Amplitude measured in dB, converted by:
– Amplitude ratio vs frequency Frequency AR(dB) = 20 log( AR )
(Rad/s) AR AR(dB)
– Phase angle vs frequency 0.01 4.9988 13.98
0.02 4.9950 13.97
0.03 4.9888 13.96
0.05 4.9690 13.93
0.1 4.8786 13.77
0.2 4.5522 13.16
0.3 4.1066 12.27
0.5 3.1623 10.00
1 1.5811 3.98
2 0.5423 –5.32
3 0.2599 –11.70
5 0.0976 –20.21
10 0.0248 –32.11

© 2017 University of the West of

England, Bristol. 4
 06/03/2017

Bode Plots Bode Plots

• Phase measured in degrees: • Constructing Bode plots
Frequency
– Sections of TF can be represented as straight
(Rad/s)
0.01
Phase
–1.7187
lines = asymptotic approximations
0.02 –3.4364 – Example:
0.03 –5.1520
0.05 –8.5730
0.1 –17.0205
0.2 –33.1113
0.3 –47.6630
0.5 –71.5651
1 –108.4349
2 –139.3987
3 –152.1027
5 –162.9795
10 –171.4270

Bode Plots Bode Plots

• Each part of TF has certain type of Gain:
dB
0 dB/decade
phase

frequency response: G(s) = K

AR(db) = 20 log K
0
frequency
0
Phase = 0°
• Building blocks: dB phase
+20 dB/decade
Differentiator:
– Gain K G(s) = s
+90
0 0
+20 dB/decade rise frequency
– Differentiator
1 Rad/s
s Phase = +90°

1 dB phase
– Integrator Integrator:
s 1 G(s) = 1/s
0
1 Rad/s
0
– First order lead/lag 1 + ts 1 + ts
–20 dB/decade fall
Phase = –90° dB -20 dB/decade
0 dB/decade
frequency
phase
-90

– Second order lead/lag dB0 +20 dB/decade

phase0
frequency +90
1 SEE HANDOUT ON BLACKBOARD
2z
2
s 0 dB/decade
0 0
+ s +1 s
2
2z frequency
wn2 wn + s +1 dB 1/t phase 0.1/t 1/t 1/10t
wn2 wn +20 dB/decade
+90
dB0 phase0
1 Rad/s frequency
1/t 1/t
0 0
0 dB/decade frequency
dB phase
-90
-20 dB/decade
dB phase 1 Rad/s
0 dB/decade frequency
0 0
+40 dB/decade frequency
0 0 dB -20 dB/decade phase-90
+180
frequency
wn +90
0 dB/decade
dB0 +20 dB/decade
phase0
frequency wn
dB phase +90
+20 dB/decade 0 dB/decade
+90 0 0
1/t frequency 0.1/t
0 0 1/t 1/10t
1 Rad/s frequency dB phase
dB wn phase
wn
0 dB/decade

Bode Plots Bode Plots

dB phase 0 1/t 0 1/t
frequency -90
1 Rad/s 0 0
0 dB/decade frequency -180
-40 dB/decade -90
0 0 -20 dB/decade
-20 dB/decade frequency
-90
frequency

dB phase 2st order lead: dB +40 dB/decade

phase
+20 dB/decade
2z
1st order lead: 2
s +180
G(s) = 1 + ts 0 dB/decade
+90
G(s) = + s +1 0 dB/decade
wn +90
Corner frequency: 1/t 0 0 wn2 wn 0 0
1/t frequency 0.1/t frequency wn
True curve 3dB above break 1/t 1/10t Corner frequency: wn
point True curve at break point
dB phase depends on z
1st order lag:
1/t
dB phase
1/t
1 2st order lag:
G(s) = 0 0 wn wn
0 dB/decade frequency 0 dB/decade
1 + ts -20 dB/decade
-90
1 0 0
Corner frequency: 1/t frequency -90
2z
G(s) = 2
True curve 3dB below break s
frequency
+ s +1 -40 dB/decade -180
point
dB +40 dB/decade
phase wn2 wn
wn
+180 Corner frequency: wn
+90
0 dB/decade True curve at break point
0 0 depends on z
frequency wn

SEE HANDOUT ON BLACKBOARD SEE HANDOUT ON BLACKBOARD

dB phase
wn wn
0 dB/decade
0 0
frequency -90

-40 dB/decade -180

© 2017 University of the West of

England, Bristol. 5
 06/03/2017

Bode Plot Bode Plot Example

• Example: • Take each term separately:
• Draw the Bode diagram for the following
(a) R1 = 20 log 5 = 14dB (b) R2 = -20 log w
transfer function:
B 5 B 5 R2
(s ) = ® ( jw ) =
14 w=1
s(1 + 0.2 s )(1 + 0.5s ) jw (1 + 0.2 jw )(1 + 0.05 jw )
0
E E R1

slope = -20 dB/dec

AR(dB) = 20 log 5 - 20 log w - 20 log 1 + (0.2w ) - 20 log 1 + (0.05w )

2 2

f = -90! - tan -1 (0.2w ) - tan -1 (0.05w )

Bode Plot Example Bode Plot Example

(c) R3 = -20 log 1 + (0.2w ) 2 (d) R4 = -20 log 1 + (0.05w ) 2
1 1 1
break point at w = = = 5 rad/s break point at w = = 20 rad/s
t 0.2 0.05
for w < 5 rad/s slope = 0 dB/dec for w < 20 rad/s slope = 0 dB/dec
for w > 5 rad/s slope = -20 dB/dec for w > 20 rad/s slope = -20 dB/dec

w = 20
w=5 0
0
R3 R4 slope = -20
slope = -20 dB/dec
dB/dec

Bode Plot Example Bode Plot Example

(e) combining plots: for w < 5 rad/s

R1 slope = -20 dB/dec

R1
1 5 20
0
1 5 20
R 0
slopes = -20 dB/dec
R

R2 R3 R4
R2

© 2017 University of the West of

England, Bristol. 6
 06/03/2017

Bode Plot Example Bode Plot Example

for w < 20 rad/s for w > 20 rad/s

slope = -20 dB/dec slope = - 20 dB/dec

1 5 20 1 5 20
0 0
R R slope = - 40 dB/dec
slope = -40 dB/dec

slope = - 60 dB/dec

Bode Plot Example Bode Plot Example

40

Phase lag given by: 20

f = -90! - tan -1 (0.2w ) - tan -1 (0.05w )

R - dB

f = f1 + f2 + f3 -20

-40 R
ω – rad/s Φ1 - deg Φ2 - deg Φ3 - deg Φ - deg -100°
-60
1 -90 -11 -3 -104 -140°

phase
5 -90 -45 -7 -149
-180°
10 -90 -63 -27 -180
20 -90 -76 -45 -211 Φ -220°

50 -90 -84 -68 -242

.1 1 5 10 20 100
Frequency – rad/s

System Identification System Identification

• Can use Bode Plot to estimate transfer • Step 1: Plot amplitude and phase data on
function Bode plot
• Use same rules as before • Step 2: Fit asymptotic approximations to
• Example amplitude data
– slopes must be 0 dB/dec, ±20 dB/dec,
±40 dB/dec etc.
– approximations must pass through the data
points at the lowest and highest frequencies
• Step 3: Use relationships to determine
values for transfer function

© 2017 University of the West of

England, Bristol. 7
 06/03/2017

System Identification System Identification

10
0 dB/dec
Following experimental data obtained 0

from a frequency response test -10

6 dB

R - dB
undertaken on control system -20 -40 dB/dec
-20 dB/dec

-30
Freq - (rad/s) R - dB Phase - deg
2 8 -112 -40 -60 dB/dec -100°
-40 dB/dec
6 -2.5 -212
-50 -140°
12 10 -152

phase
30 26 -202 -60 -20 dB/dec -180°
90 52 -245
-70 -220°
200 72 -260
-80 -260°
1 10 20 100 1000
Frequency – rad/s

System Identification System Identification

• Amplitude: Asymptotic slopes • High frequency amplitude and phase
– low frequency = -20dB/decade – -60 dB/decade slope and phase à 270°
Indicates a 3rd order system
– high frequency = -60dB/decade
• Break point
– no intermediate asymptotes
– Only one internal break point at intersection of low and
• Low frequency amplitude and phase high frequency asymptotes
– -20dB/decade and phase à -90° • slope change = -40dB/dec
• break point frequency = 20 rad/s
1 • difference between point of intersection and true curve = -6dB
Indicates a integrator term
𝑠
Indicates two coincident 1st order lags

System Identification System Identification

• System transfer function: • Transfer function terms
– integrator and two coincident first order lags: – Integral term supposed to cross zero at 1
𝐾 rad/s
TF =
𝑠 1 + 𝜏𝑠 ) – In this case, integral term crosses at 5 rad/s
à integrator is 14 dB higher
• Transfer function terms
– Therefore, gain: 20 log(K ) = 14dB
– Break point frequency:
log(K ) = 0.7
- - -
𝜔𝑏𝑝 = →𝜏= = = 0.05 sec
. 012 )3 K = 100.7 = 5.01
5.01
TF = )
𝑠 1 + 0.05𝑠

© 2017 University of the West of

England, Bristol. 8
 06/03/2017

Today’s Lecture
• Introduction to Frequency response
• Determining Amplitude Ratio and Phase
• Representing diagrammatically
– Nyquist
– Bode
• Plotting frequency response
• System Identification

© 2017 University of the West of

England, Bristol. 9