Upload
hoangkhanh
View
231
Download
3
Embed Size (px)
Citation preview
Frequency Response Methods
• The frequency response
• Nyquist diagram – polar plots
• Bode diagram – magnitude and phase
• Frequency domain specifications
Frequency Response Methods
• The frequency response of a system is defined as the steady-state response of the system to a sinusoidal input signal.
• The sinusoid is a unique input signal, and theresulting output signal for a linear system aswell as signals throughout the system, is sinusoidal in the steady state (the out of thesystem); it is differs from the input waveformonly in amplitude and phase angle.
• In preceding chapters the response and performance of a system have been described in terms of the complex frequency variable s and the location of the poles andzeros on the s-plane.
• A very practical and important alternative approach to the analysis and design of a system is the frequency response method.
Frequency Response• Consider the following
)cos(R)cos()(
ttRtuω
αω=
+=
)cos(B)cos()(
ttBtxss
ωφω
=+=
( )B( ) ( ) ( )( ) ( ) ( )R( ) ( )( )
jj
jj B e Bj G e Gj RR e
φ ωθ
αω ω ω φω ω ω θω ω αω
∠= = = = = ∠
∠G
is known as the gain and is known as the phase of the system.
)(ωG αφωθ −=)(
on
nn
n ajajaj
+++= −
− ....)()(1)( 1
1 ωωωG
For example for a system with dynamics described by the differential equation
uyadtdya
dtyda
dtyda n
n
nn
n
n =++++ −
−
− 011
1
1 ...
The transfer function from u to y is :
on
nn
n asasasG
+++= −
− ....1)( 1
1
This means a frequency dependent system gain (or transfer function):
Frequency Response
The frequency response can be obtained from the transfer function by substituting . Brief derivation:
Write u and y as and substitute into DE:
ωjs =tjtj Beyu ωω == ,Re
tjtjtjtjnn
tjnn
n
n
nn
n
n
BeaBejaBejaBeja
yadtdya
dtyda
dtyda
ωωωωω ωωω Re...)()(
...
011
1
011
1
1
=++++=
=++++
−−
−
−
−
on
nn
ntj
tj
ajajaBej
+++== −
− ....)()(1
Re)( 1
1 ωωω ω
ω
G
This means that the frequency dependent system gain (or transfer function) is :
Hence the substitution rule is justified.ωjs =
Frequency Response
11 1 0[ ( ) ( ) ... ] Ren n j t j t
n na j a j a a Be ω ωω ω −−= + + + + =
Frequency ResponseAlternatively, consider the system
Then in partial fraction form we have
Taking the inverse Laplace transform yields
where pi are assumed to be distinct poles.
where α and β are constants which are problem dependent.
1−l
Frequency Response
1−l
If the system is stable, then all pi are have positive nonzero real parts, (poles are − pi), and
since each exponential term decays to zero as t → ∞.
1−l
• Thus the steady-state output signal depends only on the magnitude and phase of T(jω) at a specific frequency ω.
• Notice that the steady state response as described the above is true only for stable systems, T(s).
Laplace vs. Fourier Transform
Laplace transform:
where
Fourier transform:
Setting s=jω in F(s) yields the Fourier transform of f (t)
(But Fourier Transform is often used for signals that exist for t<0)
Jean Baptiste Joseph Fourier
• Born in 1768 in Auxerre, France
• Died in 1830 in Paris• Was nearly guillotinedin 1794
• Was taught by Laplace,Lagrange and Monge
• Created Cairo Institute• Developed Fourier series
while prefect in Grenoble
Advantage of Frequency response method • The frequency response method is the readyavailability of sinusoid test signals forvarious ranges of frequencies and amplitudes.
• Thus the experimental determination of thefrequency response of a system is easily
accomplished and is the most reliable anduncomplicated method for the experimentalanalysis of a system.
• Furthermore the design of a system in thefrequency domain provides the designer withcontrol of the bandwidth of a system andsome measure of the response of the systemto undesired noise and disturbance.
Transfer Function in Frequency Response Method
• The transfer function representing the sinusoidal steady-state behavior is then a function of the complex
variable jω and is itself a complex function T(jω).
• The frequency response method is that the transfer function describing the sinusoid steady-state behaviorof a system can be obtained by replacing s with jω in the system transfer function T(s).
• Direct correlations between the frequency response and the corresponding transient response characteristics in the time domain are somewhat tenuous (very weak).
Frequency Response Methods
• The sinusoid is a unique input signal, and theresulting output signal for a linear system aswell as signals throughout the system, is sinusoidal in the steady state (the out of thesystem); it is differs from the input waveformonly in amplitude and phase angle.
• The important issue in frequency response methods is how to descript the amplitude and phase angle of the system. We will study different methods to represent amplitude and phase.
Frequency Response Plots
(Review Appendix G in textbook)
)Re()Im(tan
)(Im)(Re|)(|
1
22
ωωφ
ωωω
−=
+=G
Frequency Response Plots
)(tan1/tan
)Re()Im(tan)(
1
1111
ωωωω
ωωωφ −−− −=
−==
1st Ordersystem
Polar Plot or Nyquist Diagram
)1(tantan)Re()Im(tan)( 1
211
ωττωω
ωωωφ −−− =
−−
==KK
2rd system)1()(
+=
τssKsG
Harold Nyquist
• Born in 1889 in Sweden• Died in 1976, USA• Yale PhD, 1917• Career at Bell Labs• 138 patents• Nyquist diagram,
criterion, sampling theorem
• Laid the foundation forinformation theory, datatransmission andnegative feedback theory
Real Axis
Imag
inar
y A
xis
Nyquist Diagrams
-1 -0.5 0 0.5 1-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1G(s)=1/(s3+2s2+2s+1)
Frequency response diagrams
ωj
ωj
1221)( 23 +++
=sss
sG
1)(2)(2)(1)( 23 +++
=ωωω
ωjjj
jG
is plotted for ])0,((),0[ −∞∈∞∈ ωω
The Nyquist diagram of
Bode DiagramPlots of 20log10 | G(ω ) | and of Φ(ω) as a function of log ω
RC filter:
Gain in decibels (dB)
Hendrik Wade Bode
• 1905-1982, USA• PhD from Columbia
in 1935• Entire career at Bell
Labs• Invented magnitude
and phase frequencyplots in 1938
• Many other contributions in electrical engineeringand control
Bode Diagram-3dB
Break or corner frequency
Bode Diagram
• Advantage of logarithmic plot is that multiplicative factors are converted into additive ones
• We can then decompose a high order transfer function into a product of simple standard components to sketch the broad features of the Bode diagram
decade
Form of the Transfer Function
• We treat real and complex poles and zeros separately,and use one of two standard forms.
• We then plot magnitude (in db) and phase on a log scale of frequency
• By using db for magnitude, we can use addition tocombine the effects of each pole or zero since theproduct becomes a sum of log terms.
• The phase effect is already a linear combination• The log frequency scale allows piecewise linear
approximations with reasonable accuracy• Bode plots work best for real poles and zeros. The
complex case is less accurate and requires carefultreatment.
How to draw a Bode plot for a given transfer function?
There are only 8 types of factors in any transfer function - which are these?
Lswyn
j
w
knknkkj
y
m
ii
Lsyn
jj
y
m
ii
Ls
n
m
Lsn
nn
n
mm
mm
esspss
zsk
epss
zsk
epspspszszszsk
easasasabsbsbsbsG
−−−
= =
=
−−
=
=
−
−−
−
−−
∏ ∏
∏
∏
∏
+++
+
=
+
+
=
++++++
=
++++++++
=
2
1 1
22
1
1
1
21
21
011
1
011
1
)2()(
)(
)(
)(
))...()(())....()((
........)(
ωωζ
Ljwyn
j
w
k nknk
kpj
y
m
izi
Ljyn
jpi
y
m
izi
Lj
pnpp
zmzz
Ljn
nn
n
mm
mm
ejjTjj
TjK
eTjj
TjK
eTjTjTjTjTjTjK
eajajaja
bjbjbjbj
ω
ω
ω
ω
ωωω
ωζωω
ω
ωω
ω
ωωωωωωωωωωωωω
−−−
= =
=
−−
=
=
−
−−
−
−−
∏ ∏
∏
∏
∏
⎟⎟⎠
⎞⎜⎜⎝
⎛+++
+=
+
+=
++++++
=
++++++++
=
2
1 12
21
1
1
21
21
011
1
011
1
)(21)1()(
)1(
)1()(
)1(
)1)...(1)(1()1)....(1)(1(
....)()(....)(G
If we knew the look of the Bode plots for each of the 8 types, we could add up the Bode plots from them.
Bode plot analysis techniques
Ljwyn
j
w
k nknk
kpj
y
m
izi
ejjTjj
TjKj ω
ωωω
ωζ
ωω
ωω −
−−
= =
=
∏ ∏
∏
⎟⎟⎠
⎞⎜⎜⎝
⎛+++
+
= 2
1 12
21
)(21)1()(
)1()(G
( )
( )
1
2
1 2
21
( 2 ) 21 1
2
2
Lm ( ) 20 log ( ) 20 log 20 log 1
20 log 1 .... 20 log 1 20 log
20 log 1 20 log 1 ....
220 log 1 20 log 1 ...
220 log 1
z
z zm
p p
p n y wn n
w
nw nw
j j K j T
j T j T y
j T j T
jj T j
jj
ω ω ω
ω ω ω
ω ω
ωζω ωω ω
ωζ ωω ω
− −
= = + + +
+ + + + − −
+ − + −
− + − + + −
− + +
G G
Factorization
Gain in dB :
Bode plot analysis techniques
⎟⎟⎠
⎞⎜⎜⎝
⎛++∠
−⎟⎟⎠
⎞⎜⎜⎝
⎛++∠−+∠
−−+∠−+∠−−+∠+++∠++∠+∠=∠
−−
2
2
21
2
1
1)2(
21
21
)(21
...)(21)1(
....)1()1(10)1(....)1()1()()(
nwnw
w
nnwynp
pp
zmzz
jj
jjTj
TjTjTjTjTjKj
ωωω
ωζ
ωωω
ωζω
ωωπωωωωGPhase:
The laborious procedure of plotting the amplitude and the phase by means of substituting several values of can be avoided when drawing Bode diagrams, because we can use several short cuts. These short cuts are based on simplifying approximations, which allow us to represent the exact, smooth plots with straight-line approximations. The difference between actual curves and these asymptotic approximations is small, and can be
added as a correction.
ω
The 8 types of factors in a transfer function
Transport
delay
Constant
K
Lje ω−
Differentiators
Integrators
yj )(1ω
rj )( ω
The 8 types of factors in a transfer function
First order lag terms(real poles)
First order Lead terms(real
zeros)
rziTj )1( ω+
rpjTj )1(
1ω+
The 8 types of factors in a transfer function
Quadratic lead terms(complex
zeros)
Quadratic lag terms (complex poles
r
nknk
k jj ⎟⎟⎠
⎞⎜⎜⎝
⎛++ 2
2)(21
1
ωωω
ωζ
r
nknk
k jj ⎟⎟⎠
⎞⎜⎜⎝
⎛++ 2
2)(21ωωω
ωζ
The 8 types of factors in a transfer function
Bode DiagramPlots of 20log10 | G(ω ) | and of Φ(ω) as a function of log ω
RC filter:
Gain in decibels (dB)
Bode plot analysis techniques
Ljwyn
j
w
k nknk
kpj
y
m
izi
ejjTjj
TjKj ω
ωωω
ωζ
ωω
ωω −
−−
= =
=
∏ ∏
∏
⎟⎟⎠
⎞⎜⎜⎝
⎛+++
+
= 2
1 12
21
)(21)1()(
)1()(G
( )
( )
1
2
1 2
21
( 2 ) 21 1
2
2
Lm ( ) 20 log ( ) 20 log 20 log 1
20 log 1 .... 20 log 1 20 log
20 log 1 20 log 1 ....
220 log 1 20 log 1 ...
220 log 1
z
z zm
p p
p n y wn n
w
nw nw
j j K j T
j T j T y
j T j T
jj T j
jj
ω ω ω
ω ω ω
ω ω
ωζω ωω ω
ωζ ωω ω
− −
= = + + +
+ + + + − −
+ − + −
− + − + + −
− + +
G G
Factorization
Gain in dB :
Bode plot analysis techniques
⎟⎟⎠
⎞⎜⎜⎝
⎛++∠
−⎟⎟⎠
⎞⎜⎜⎝
⎛++∠−+∠
−−+∠−+∠−−+∠+++∠++∠+∠=∠
−−
2
2
21
2
1
1)2(
21
21
)(21
...)(21)1(
....)1()1(10)1(....)1()1()()(
nwnw
w
nnwynp
pp
zmzz
jj
jjTj
TjTjTjTjTjKj
ωωω
ωζ
ωωω
ωζω
ωωπωωωωGPhase:
The laborious procedure of plotting the amplitude and the phase by means of substituting several values of can be avoided when drawing Bode diagrams, because we can use several short cuts. These short cuts are based on simplifying approximations, which allow us to represent the exact, smooth plots with straight-line approximations. The difference between actual curves and these asymptotic approximations is small, and can be
added as a correction.
ω
Detailed examination of the 8 factors
System type corresponds to integrators (for 0 type there is not integrator factor)Diagram of a constant
πdBlog20Lm KK =
K>0K<0
Ljwyn
j
w
k nknk
kpj
y
m
izi
ejjTjj
TjKj ω
ωωω
ωζ
ωω
ωω −
−−
= =
=
∏ ∏
∏
⎟⎟⎠
⎞⎜⎜⎝
⎛+++
+
= 2
1 12
21
)(21)1()(
)1()(G
Detailed examination of the 8 factors
ωωωω
log20log201log20)(
1log20)(
1Lm yyjj yy −=−==⎟⎟
⎠
⎞⎜⎜⎝
⎛
yjj
yy 90)(1
)(1
−=∠−∠=⎟⎟⎠
⎞⎜⎜⎝
⎛∠ ω
ω
ωj1
Diagram of integrators
Detailed examination of the 8 factors
( ) ωωωω log20log20)(log20)(Lm yyjj yy ===
( ) yj y 90)( =∠ ω
Bode diagram of a differentiator
y=1
Detailed examination of the 8 factors
2)(1log20
1log201log201
1log201
1Lm
T
TjTjTj
ω
ωωω
+−=
+−=+
=⎟⎟⎠
⎞⎜⎜⎝
⎛+ TTj
Tjωω
ω1tan)1(1
11 −−=+∠−∠=⎟⎟
⎠
⎞⎜⎜⎝
⎛+
∠
dB01log201
1Lm =≈⎟⎟⎠
⎞⎜⎜⎝
⎛+ Tjω
.1<<Tω
TTjTj
ωωω
log201log201
1Lm −=≈⎟⎟⎠
⎞⎜⎜⎝
⎛+
.1>>Tω
Bode diagram of a first order lag term
Detailed examination of the 8 factors
frequency error
Corner frequency (bandwidth) -3dB
At half the corner frequency -1dB
At a quarter of the corner frequency
-0.26 dB
Bode diagram of a first order lag term
Detailed examination of the 8 factors( )
2)(1log20
1log201log201Lm
T
TjTjTj
ω
ωωω
+=
+=+=+
( ) TTj ωω 1tan1 −=+∠.1<<Tω
( ) dB01log201Lm =≈+ Tjω
.1>>Tω
( ) TTjTj ωωω log20log201Lm =≈+
First order lead term
Detailed examination of the 8 factors
22 )(121
1
ωω
ωωζ jj
nn
++
1<ζ
22
2
2
22
22
21log20
)(121
1log20)(121
1Lm
⎟⎟⎠
⎞⎜⎜⎝
⎛+⎟⎟
⎠
⎞⎜⎜⎝
⎛−−=
++=
⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
++
nn
nnnn
jjjj
ωζω
ωω
ωω
ωωζω
ωω
ωζ
221
22
/1/2tan
)(121
1
n
n
nn
jj ωωωζω
ωω
ωωζ −
−=⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
++∠ −
Quadratic (second order) Lag
dB01log20)(121
1Lm2
2
=−≈⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
++ ωω
ωωζ jj
nn
For small ω
nn
nn
jj
ωω
ωω
ωω
ωωζ
log40log20
)(121
1Lm
2
2
22
−=−≈
≈
⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
++
For large ω
Detailed examination of the 8 factorsQuadratic (second order) Lag
2121
ζζ −=mM
221 ζωω −= nmFor there is a resonant peak at 707.0<ζ
with peak size
Detailed examination of the 8 factorsQuadratic (second order) Lag
Detailed examination of the 8 factors
Lee LjLj ωωω −=∠= −− ,0Lm
Transport Lag
Non-Minimum Phase System
• A transfer function is called minimumphase if all its zeros lie in the left-handplane.
• It is called non-minimum phase if it haszeros in the right-hand plane.
331)( 2 ++
+=
ssssGH
Apply steps 1-4 and step 5 for breakaway point
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Real Axis
Imag
Axi
s
Root locus examples
331)( 2 ++−
=ss
ssGH
Apply steps 1-4,
Step 4 for crossing and
Step 5 for breakaway point
…
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
Real Axis
Imag
Axi
s
Root locus examples
Non-Minimum Phase SystemNote that |G(jω)| is identical, but the phase is different
Drawing the Bode Diagram
20log5=14
40dB/De
-20dB
-40dB
?
?
Drawing the Bode Diagram
20log5=14
40dB/De
-20dB
-40dB
?
Drawing the Bode Diagram
20log ( ) 20log5 20logG jω ω= −
20log 1 0.5j ω− +
2 0 log 1 0 .1j ω+ +
220log 1 0.6( / 50) ( / 50)j ω ω− + +
(ω <1)
(ω >2 )
(ω >10)
(ω >50)
Drawing the Bode Diagram
Drawing the Bode Diagram
Using Matlab
Example
530)(+
=s
sG
)2.01)(41(10)(
ωωω
jjj
++=G
Example
))125.0(05.01)(21()5.01(4)( 2ωωωω
ωωjjjj
jj+++
+=G
Example
Performance Specifications in the Frequency Domain
• The basic disadvantage of the frequency response method for analysis and design is the indirect link between the frequency and the time domain.
• Then for given a set of time-domain (transient performance) specifications, how do we specify the frequency response?
• Direct correlations between the frequency response and the corresponding transient response characteristics are somewhat tenuous (very weak).
• However, we need to develop a method to evaluate the performance in the frequency response method.
• Like in the time-domain approach, we only consider the performance of a simple second order system to a step input.
1 21( ) 1 sin( cos ), 1tnny t e tζω ω β ζ β ζ
β− −= − + = −
%)2(4== δ
ζωforT
ns
21/100 ζζπ −−= ePO
nrT
ωζ 6.016.2
1+
≅
Settling time
Percentage overshoot
Peak time
Rise time (10% - 90%)
21 ζωπ−
=n
pT
Step response for second ordersystems-in the time domain
Performance Specifications in the Frequency Domain
.2
)( 22
2
nn
nss
sTωζω
ω++
=
The closed-loop transfer function in the frequency domain:
Consider a second order system
The Bode diagram of the frequency response of this feedback system is shown in Fig
dB01log20)(121
1Lm2
2
=−≈⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
++ ωω
ωωζ jj
nn
For small ω
nn
nn
jj
ωω
ωω
ωω
ωωζ
log40log20
)(121
1Lm
2
2
22
−=−≈
≈
⎟⎟⎟⎟
⎠
⎞
⎜⎜⎜⎜
⎝
⎛
++
For large ω
Detailed examination of the 8 factorsQuadratic (second order) Lag
Performance Specifications in the Frequency Domain
rωωpM
• At the resonant frequency, , a maximum value of the frequency response, , is attained.
• The bandwidth, , is a measure of a system’s ability to faithfully reproduce an input signal.
Bω
• The bandwidth is the frequency, , at which thefrequency response has declined 3 dB from its low-frequency value.
Bω
.2
)( 22
2
nn
nss
sTωζω
ω++
=
Performance Specifications in the Frequency Domain
.121 ζζπ −−+= eM tp
ωpMThe resonant peak indicates the relative stability of a system
The resonant frequency and –3dB bandwidth can berelated to the speed of the transient response. Thus as the bandwidth increase, the rise time of the step response of the system will decrease. Furthermorethe overshoot to a step input can be related to, through the damping ratio ζ. by
rω
Bω
ωpM
Performance Specifications in the Frequency Domain
The bandwidth of a system as indicated on the frequency response can be approximately related to the natural frequency of the system. Figure 8.26 shows the normalized bandwidth, versusζ for the second-order system
Bω
,/ nB ωω
Performance Specifications in the Frequency Domain
<ωpM
nζωτ /1=
Thus desirable frequency-domain specifications are as follows:
1. Relatively small resonance magnitude:
2. Relatively large bandwidths so that the system time constant is sufficiently small
1.5, for example.
Performance Specifications in the Frequency Domain
• The usefulness of these frequency response specifications and their relation to the actual transient performance depend upon the approximation of the system by a second-order pair of complex poles, that is the dominant roots.
• If the frequency response is dominated by a pair of complex poles, the relationships between the frequency response and the time response discussed in this section will be valid.
• Fortunately a large proportion of control system satisfied this dominant second-order approximation in practice.
Steady-state error constants
The steady-state error specification can also be related to the frequency response of a closed-loop system.
• As we knew, the steady-state error for a specific test input signal can be related to the gain and number of integrations (poles at the origin) of the open-loop transfer function, i.e.,the type of the system. • In frequency response method, the type of the system
determines the slop of the logarithmic gain curve at low frequency, since steady-state error is defined at
,0→s .0→ωji.e.,
Thus, information concerning the existence and magnitude of the steady-state error of a control system to a given input can be determined from the observation of the low-frequency region of the logarithmic gain curve.
Determine of static position error constants.
For type 0 system (N=0), we have
)(lim)(lim00
ωω
jGsGKjs
P→→
==
Consider the transfer function as follows
.)1()(
)1()(
1
1
∏
∏
=
=
+
+= Q
kk
N
M
ii
jj
jKjG
ωτω
ωτω
For type 0 (N=0) system, at the low frequency, we have
KjG ≈)( ω or KjGKj
P ==→
)(lim0
ωω
∏
∏
∏
∏
=
=
=
=
+
+=
+
+= Q
kk
M
ii
Q
kk
M
ii
j
jK
jj
jKjG
1
1
1
0
1
)1(
)1(
)1()(
)1()(
ωτ
ωτ
ωτω
ωτω
Determine of static position error constants.
KjGKj
P ==→
)(lim0
ωω
Hence, we can determine the steady-state position error by measurethe value from its logarithmic gain curve (let 20logK=c),
KKcpK log)20/)log20()20/( 101010 ===
Determine of static velocity error constant
For type 1 system (N=1), we have
)(lim)(lim00
ωωω
jGjssGKjs
v→→
==
Consider the transfer function as follows
.)1()(
)1()(
1
1
∏ +
∏ +=
=
=Q
kk
N
M
ii
jj
jKjG
ωτω
ωτω
According to the definition, we have
KjGjKj
v ==→
)(lim0
ωωω
(at the low frequency ).
)1()(
)1()(
1
1
1
ωωτω
ωτω
jK
jj
jKjG Q
kk
M
ii
≈+
+=
∏
∏
=
=
Determine of static velocity error constant
||log20log201
vj
v KjK
==ωω
Also, we can find out Kv using the fact that the intersection of the initial –20dB/decade segment (or its extension) with the 0dBline has a frequency numerically equal to Kv
.
1=ωj
Kv or 1ω=vK
At the intersection of the initial –20dB/decade segment (or its extension) with the 0-dB line, the horizontal coordinate, i.e., the frequency isnumerically equal to the. vK
Determine of static acceleration error constant
KjGjKja ==→
)()(lim 2
0ωω
ω
For type 2 system (N=2), we have
Consider the transfer function as follows
)()(lim)(lim 20
20
ωωω
jGjsGsKjs
a→→
==
.)1()(
)1()(
1
1
∏
∏
=
=
+
+= Q
kk
N
M
ii
jj
jKjG
ωτω
ωτω
(at the low frequency ).)()1()(
)1()( 2
1
2
1
ωωτω
ωτω
jK
jj
jKjG Q
kk
M
ii
≈+
+=
∏
∏
=
=
Determine of static acceleration error constant
||log20)(
log201
2 aj
a KjK
==ωω
aω
aK
01log20)(
log20 2 ==ωjKa
aa K=ω .2aaK ω=
The frequency at the intersection of the initial-40db/decade segment (or its
extension) with the 0-dB line gives the square root of numerically.
or which yields
Design Example: Engraving Machine
The goal is to select an appropriate gain K, utilizing frequency response method, so that the time response to step commands is acceptable
Design Example: Engraving Machine
To represent the frequency response of the system,we will first obtain the open-loop and closed-loop Bode diagram.
)2)(1(1)(
++=
sssjG ω
Design Example: Engraving Machine
.223
2)( 23 +++=
ssssT
)8.0(.78.1 == rpMor ωω
Then we use the closed-loop Bode diagram to predict the time response of the system and check the predicted result with the actual result
20log|T|=5 dB at 8.0=rω
.
5log20 =ωpM
ωωωω
ω jsj
jT =−+−
=)2()32(
2)(22
Design Example: Engraving Machine
)8.0(.78.1 == rpMor ωω
5log20 =ωpM
If we assume that the system has dominant second-order roots, we can approximate the system with a second-order frequency response of the form shown in Fig.
29.08.0 == ζωr
Design Example: Engraving Machine
78.1=ωpM
29.08.0 == ζω r
nr ωω / =0.91.
.88.091.08.0==nω
.774.051.0
774.02
)(222
2
++=
++=
sssssT
nn
n
ωζω
ω
Since we are now approximating T(s) as a second-order system, we have
Design Example: Engraving Machine
.sec7.1588.0)29.0(
44 ondsTn
s ===ζω
The overshoot to a step input as 37% for 29.0=ζ
The settling time to within 2% of the final value is estimated as
.774.051.0
774.02
)(222
2
++=
++=
sssssT
nn
n
ωζω
ω
Design Example: Engraving Machine
• The actual overshoot for a step input is 34%, and the actual settling time is 17 seconds.
• We see that the second-order approximation is reasonable in this case and can be used to determine suitable parameters on a system.
• If we require a system with lower overshoot, we would reduce K to 1 and repeat the procedure.
Disk Drive Read System
• The disk drive uses a flexure suspension to hold the reader head mount, as shown in Fig
• we will include the effect of thespringy flexure within the model of the motor-load system.
• We model the flexure with the mounted head as a mass M, a spring k, and a sliding friction b.
Disk Drive Read System
.)/()/2(1
12
)()()(
222
23
nnnn
nssss
sGsUsY
ωωζωζωω
++=
++==
3.0=ζHzfn 3000= 31085.18 ×=nω
The transfer function of a spring-mass-damper was developed in Chapter 2, where
A typical flexure and head has and natural resonance at or
Disk Drive Read System
|,)()()()1(|log20 321 ωωωω jGjGjGjK +
The sketch is a plot of the magnitude characteristics for the open-loop Bode diagram ,or
K=400
Note the resonance at nω
We wish to avoid exciting
This resonance.
Disk Drive Read System
msTn
s 5.24≈=
ζω
8.0≈ζwhere
,400≤KAs long as the resonance is outside the bandwidth of the system.
Plots of the magnitude of the open-loop Bode diagram andthe closed-loop Bode diagram are shown in following
sec/2000radBn =≈ωω
Mpω
Determine the transfer function of the system that has the following frequency response:
15+s
Problems: Experimental determination of transfer function of a system based on its frequency response
Determine the transfer function of the system that has the following frequency response:
)1)(101.0()10(1.0
)1)(101.0(11.0
+++
=++
+ss
sss
s
Problems: Experimental determination of transfer function of a system based on its frequency response
Summary
• In this chapter we have considered the representation of a feedback control system by its frequency response characteristics.
• The frequency response of a system was defined as the steady-state response of the system to a sinusoidal input signal.
• Several alternative forms of frequency plots were considered, including the polar plot of the frequency response of a system G(jω) and logarithmic plots, often called Bode plots, and the value of the logarithmic measure was illustrated.
• The ease of obtaining a Bode plot for the various factors of G(jω) was noted, and an example was considered in detail.
• The asymptotic approximation for sketching the Bode diagram simplifies the computation considerably
Summary of fifteen typical Bode plots
1.
2.
3.
Summary of fifteen typical Bode plots
5.
Summary of fifteen typical Bode plots
7.
9.
Summary of fifteen typical Bode plots
12.
11.
10.
Summary of fifteen typical Bode plots
14.
13.