You are on page 1of 8

SYSTEM THEORY

IX

FREQUENCY RESPONSE ANALYSIS

1. Bode Diagrams (continued)

1.1. The Quadratic Factor

Statement: The frequency transfer function of the quadratic factor with the natural frequency ω n , and the
dumping ratio ζ is given by the following expression:

G( j ⋅ ω ) =
1
ω  ω 
2
(1.)
1 −   + j ⋅ 2 ⋅ ζ ⋅  
 ωn   ωn 

Proof:

From the differential equation of the integral factor follows:

d 2 y (t ) d y (t )
2
+ 2 ⋅ ζ ⋅ ωn ⋅ + ω n2 ⋅ y (t ) = ωn2 ⋅ u (t ), t < 0, u (t ) = 0 , ⇒
dt dt

Y (s )
ω2
n
ω n2
s ⋅ Y (s ) + 2 ⋅ ζ ⋅ ωn ⋅ s ⋅ Y (s ) + ω ⋅ Y (s ) = ω ⋅U (s ) ⇒ G (s ) =
2 2 2
= ⇒
U (s ) s 2 + 2 ⋅ ζ ⋅ ω ⋅ s + ω 2
n n

G( j ⋅ ω ) =
1
2
.
 ω   ω 
1 + 2 ⋅ ζ ⋅  j ⋅  +  j ⋅ 
 ωn   ωn 

Statement: (a) The expressions of the asymptotes to the log-magnitude curve of the quadratic factor are:

for ω ⋅ T → 0 , G ( j ⋅ ω) dB → 0 , (2.)

ω
For ω ⋅ T → ∞ , G ( j ⋅ ω ) dB = −40 ⋅ log . (3.)
ωn

And, (b) the phase angle curve is given by the expression

 ω 
 2 ⋅ζ ⋅ 
ωn
arg{G ( j ⋅ ω )} = − arctg   (4.)
 ω2 
 1− 2 
 ωn 

Proof:

In the first place, we emphasize the real part and the imaginary part into the expression of the frequency
transfer function, Formula (1):

1
Frequency Response Analysis

Fig. 1: The Bode Diagrams of the Quadratic Fig. 2: The Bode Diagrams of the Reciprocal
Factor. Quadratic Factor.

 ω 2  ω 
G( j ⋅ ω ) =
1 1  
= ⋅ 
1 − 2 
− j ⋅ 2 ⋅ ζ ⋅ ⇒
(i) ω2 ω  ω2  
2
ω   ωn 
2
ωn 
1 − 2 + j ⋅ 2 ⋅ζ ⋅ 1 − 2  +  2 ⋅ ζ ⋅ 
ωn ωn  ω   ωn 
 n  

ω2 ω
1− 2 2 ⋅ζ ⋅
ωn ωn
⇒ G( j ⋅ ω ) = 2 2
− j⋅ 2 2

 ω2   ω   ω2   ω 
1 − 2  +  2 ⋅ ζ ⋅  1 − 2  +  2 ⋅ ζ ⋅ 
 ω   ωn   ω   ωn 
 n    n  

G( j ⋅ ω ) =
1
, and
2 2
 ω  2
ω  (5.)
1 − 2  +  2 ⋅ ζ ⋅ 
 ω   ωn 
 n  

 ω 
 2 ⋅ζ ⋅ 
ωn
(ii) arg{G ( j ⋅ ω )}− arctg  
 ω2 
 1− 2 
 ωn 

The exact expression of the log-magnitude curve may be derived from (ii) as follows,

2 2
 ω2   ω 
(iii) G ( j ⋅ ω ) dB = 20 ⋅ log G ( j ⋅ ω ) = −20 ⋅ log 1 − 2  +  2 ⋅ ζ ⋅  .
 ω   ωn 
 n  

To compute the asymptotes observe in (iv) that:

2
Bode Diagrams

for ω → 0 G ( j ⋅ ω) dB → 0 ,

And for large values of ω,

ω2 ω4 ω2
G ( j ⋅ ω ) dB = −20 ⋅ log 1 − 2 ⋅ + + 4 ⋅ζ ⋅
2
=
ω n2 ω n4 ω n2
ω4 ω2 ω4 ω
= −20 ⋅ log
ωn
4
( 2
)
+ 2 ⋅ 2 ⋅ ζ − 1 ⋅ 2 + 1 ≈ −20 ⋅ log
ωn ωn
4
= −40 ⋅ log
ωn

ω
(iv) And thus G ( j ⋅ ω ) dB = −40 ⋅ log .
ωn ω →∞

The significant values of the phase angle curve are derived from (iii):

(v) for ω → 0 , arg {G ( j ⋅ ω)} → 0

(vi) for ω → ω n , arg{G ( j ⋅ ω )} → 900

(vii) And for ω → ∞ , arg{G ( j ⋅ ω )} → 180 0

Proposition: The intersection of the asymptotes to the log-magnitude diagram of the quadratic factor is
at the natural frequency, i.e.:

ωC = ω n . (6.)

Proof:

The intersection of the two asymptotes is:

y=0

ω ⇒ ωC = ω n .
y = −40 ⋅ log
ωn

Definition: consider a quadratic factor system with the natural frequency ω n and the dumping ratio ζ ,
and a sine wave command with constant magnitude; the resonant frequency, ω r is the angular
frequency of the command that produces a maximum of the response magnitude. The resonant peak
value, M r is the magnitude of the response at the resonant frequency.

Proposition: consider a quadratic factor system with the natural frequency ω n and the dumping ratio ζ ,
then:

a. the resonant frequency is given by the expression,

2
ωr = ωn ⋅ 1 − 2 ⋅ ζ 2 , 0 < ζ ≤ , and (7.)
2

b. the resonant peak value depends only on the dumping ratio and is given by the expression,

3
Frequency Response Analysis

1
Mr = .
(8.)
2 ⋅ζ ⋅ 1 − ζ 2

Proof:

a. We start by searching the minimum of the denominator in the expression of the magnitude of G ( j ⋅ ω ) ,
in the expression (5.). For simplicity let us introduce the variable ν = ω 2 ω n2 ; follows - only for the
denominator:

g (ν ) = (1 −ν )2 + 4 ⋅ ζ 2 ⋅ν = ν 2 − 2 ⋅ν + 1 + 4 ⋅ ζ 2 ⋅ν ⇒
(i)
( )
⇒ g (ν ) = ν 2 − 2 ⋅ 1 − 2 ⋅ ζ 2 ⋅ν + 1

in (i), the value of the variable ν that corresponds to the minimum of the function g (ν ) is

b
(ii) ν min = − = 1− 2 ⋅ζ 2 q.e.d.
2⋅a
2
⇒ ωr = ωn ⋅ 1 − 2 ⋅ ζ 2 ; 0 < ζ ≤
ω r2 2
ν min =
ω n2

b. to compute the value of the peak value of the response we can introduce the expression (7.) into the
expression (5.); follows,

(
G j ⋅ ωr =) 1 (
G j ⋅ ωr = )
2 2
 ω2   
 +  2 ⋅ ζ ⋅ ωr 
(iii) 1
1 − r =
( )
 ω2   ω n  2

 n    ω n2 ⋅ 1 − 2 ⋅ ζ 2 2
ωn ⋅ 1 − 2 ⋅ ζ 2 
  
1 −  +  2 ⋅ζ ⋅ 
 ω n2   ωn 
ωr = ωn ⋅ 1 − 2 ⋅ ζ 2    

(
G j ⋅ ωr =) 1
=
1
=
1
4 ⋅ζ + 4 ⋅ζ ⋅ 1 − 2 ⋅ζ
2 2
( 2
) 4 ⋅ζ ⋅ 1 − ζ
2
( 2
) (
2 ⋅ζ ⋅ 1 − ζ 2 ) q.e.d.

Remarks:

1. The natural frequency is the resonant frequency of an equivalent quadratic factor with zero dumping. In
fact, such a system operates as an oscillator. From the stability point of view, the quadratic factor with
zero dumping is neither stable nor unstable. The magnitude of the response is, theoretically infinite. In
practice, the magnitude is limited by the energy amount provided to the system. Form this point of
view, at resonance, there is an energy exchange between the system and the supply without energy
losses.

2
2. The quadratic factor satisfying the precondition 0 < ζ ≤ resonates at a lower frequency than the
2
corresponding oscillator due to its dumping ratio ζ . In this case, at resonance there are energy
losses. The resonant frequency of a dumped quadratic factor is also referred as the dumped natural
frequency ω d .

4
Bode Diagrams

3. The product σ = ω n ⋅ ζ is called the attenuation of the quadratic factor.

4. The Bode diagrams of the reciprocal quadratic factor are symmetrical to the diagrams of the
corresponding quadratic factor with respect to the horizontal axis as shown in Figures 1 and 2.

5. It is important to know how to control the resonances of a given system especially for machine-tools
and positioning systems.

2. General Procedure for Plotting Bode Diagrams

1. Rewrite the sinusoidal transfer function G ( j ⋅ ω) as a product of basic factors.

2. Identify the parameters such as time constants, corner frequencies, and so associated with these
factors.

3. Draw the asymptotes of magnitude curves with proper slopes between the corner frequencies.

4. Add the magnitude curves of the individual factors to obtain the resulting magnitude curve of the given
system.

5. Draw the phase-angle curve of the individual basic factors.

6. The resulting phase-angle curve can be drawn by adding the phase-angle curves of the individual
factors.

2.1. Example
Sketch the Bode diagrams of the system with the following transfer function:

10 ⋅ (s + 3 )
G (s ) = (9.)
s + 3 ⋅ s3 + 4 ⋅ s2 + 4 ⋅ s
4

Proof:

Step one; derive the sinusoidal transfer function from the expression (9.).

10 ⋅ (s + 3 ) 10 ⋅ (s + 3 )
G (s ) = =
(i) 4 3 2
(
s + 3 ⋅ s + 4 ⋅ s + 4 ⋅ s s ⋅ (s + 2 ) ⋅ s 2 + s + 2

)
10 ⋅ (3 + j ⋅ ω)
⇒ G ( j ⋅ ω) =
[
j ⋅ ω ⋅ (2 + j ⋅ ω) ⋅ ( j ⋅ ω) + j ⋅ ω + 2
2
=
]
 ω  ω
(ii) 10 ⋅ 3 ⋅  1 + j ⋅  7 ,5 ⋅  1 + j ⋅ 
=  3  =  3
 ω  (ω) 2
j ⋅ ω  ω   (ω)
2
j ⋅ ω
j ⋅ ω ⋅ 2 ⋅  1 + j ⋅  ⋅ 2 ⋅ 1 − +  ( j ⋅ ω) ⋅  1 + j ⋅  ⋅ 1 − + 
 2  2 2   2  2 2 

The sinusoidal transfer function is composed of the following basic factors:

 1
G ( j ⋅ ω) = 7 ,5 ⋅
1 1 1
⋅ ⋅ 1 + j ⋅ ω⋅  ⋅
( j ⋅ ω)  1 + j ⋅ ω ⋅ 1   3   ω  2
2 ω
  1 −   + 2⋅ ⋅ j⋅ 
 2 
(ii)   2  4 2 
↑ ↑ ↑ ↑ ↑

(1) (2 ) (3) (4 ) (5 )

5
Frequency Response Analysis

(1) – the gain factor, (2) – the integral factor, (3) – the first-order factor, (4) – the reciprocal first-order
factor, (5) – the quadratic factor.

Note: in (ii) the sinusoidal transfer function of the quadratic factor, (5) has been written in the normalized
form to emphasize its parameters as follows:

ωn = 2
2
1 ⇒ζ= 4
1 1 ⇒
= ≡ 1
(ω) 2
j ⋅ω  ω
2
ω 2⋅ζ⋅ =
1− + 1 −   + j ⋅ 2 ⋅ ζ ⋅ ωn 2
2 2  ωn  ωn

Step two; the parameters of these factors are presented in Table 1.

Table 1: the parameters of the factors (1) – (5).

The Factor Parameter Symbol Value

the gain factor (1) the gain k1 k1 = G1 ( j ⋅ ω) dB = 20 ⋅ log 7 ,5 = 17 ,5

the first-order factor the time constant


= 0 ,5 [s ]
T3 1
(3) T3 =
2

the corner ωC 3 1  rad 


ωC 3 = =2
 s 
frequency
T3

the reciprocal first- the time constant


order factor (4)
T4
T4 =
1
[s ]
3

the corner ωC 4 1  rad 


ωC 4 = =3
 s 
frequency
T4

the quadratic factor the natural ωn 5  rad 


(5) frequency ωn 5 = 2 
 s 

the dumping ratio ζ5 2


ζ5 =
4

the resonant ωr 5 2 3
frequency ωr 5 = 2 ⋅ 1 − 2 ⋅ = 2⋅ =
16 4
6  rad 
=
2  s 

The Bode diagrams of the basic factors and the Bode diagrams of the given system are depicted in
Figure 3.

6
Bode Diagrams

Fig. 3: the Bode Diagrams of the Given System and Its Components.

3. Determination of Error Constants From the Log-Magnitude Curve


The steady-state error of a system to a given input can be determined from the observation of the low-
frequency region on the log-magnitude curve.

3.1. The Static Position Error Constants


Proposition: for a type-zero system, the low-frequency asymptote to the log-magnitude curve is a
horizontal line at 20 ⋅ log K P .

Proof: the static position error constant of a type-zero system is given by:

K P = lim G ( j ⋅ ω) ⋅ H ( j ⋅ ω) = K .⇒
ω→ ∞

7
Frequency Response Analysis

(i) lim G ( j ⋅ ω) ⋅ H ( j ⋅ ω) dB = 20 ⋅ log K P


ω→ ∞

Follows, from (1.) that the low-frequency asymptote is a horizontal line at 20 ⋅ log K P .

3.2. The Static Velocity Error Constants

Proposition: for a type 1 system, the intersection of the initial − 20 dB dec segment with the 0 dB line
has the frequency numerically equal to Kv .

Proof:

For a type 1 system, the frequency transfer function is:

K v ⋅ (1 + Ta ⋅ j ⋅ ω) ⋅ (1 + Tb ⋅ j ⋅ ω) ⋅ K
G ( j ⋅ ω) ⋅ H ( j ⋅ ω) = ,
(i)
j ⋅ ω ⋅ (1 + T1 ⋅ j ⋅ ω) ⋅ (1 + T2 ⋅ j ⋅ ω) ⋅ K ⇒ G ( j ⋅ ω) ⋅ H ( j ⋅ ω) ≈
Kv
.
j ⋅ω
ω << 1

On the log-magnitude curve, the expression (i) is a straight line with the expression:

Kv
(ii) y (ω) = 20 ⋅ log = 20 ⋅ log K v − 20 ⋅ log j ⋅ ω ,
j ⋅ω
⇒ Kv = ωv .

( )
y ωv = 0

3.3. The Static Acceleration Error Constants

Proposition: the frequency at the intersection of the initial − 40 dB dec segment with the 0 dB line
gives the square root of K a , numerically.

Proof:

For a type 2 system, the frequency transfer function is:

K a ⋅ (1 + Ta ⋅ j ⋅ ω) ⋅ (1 + Tb ⋅ j ⋅ ω) ⋅K
G ( j ⋅ ω) ⋅ H ( j ⋅ ω) = ,
j ⋅ ω2 ⋅ (1 + T1 ⋅ j ⋅ ω) ⋅ (1 + T2 ⋅ j ⋅ ω) ⋅K ⇒ G ( j ⋅ ω) ⋅ H ( j ⋅ ω) ≈
Ka
.
(i)
( j ⋅ ω)2
ω << 1

The first asymptote on the log-magnitude curve is the straight line with the expression:

y (ω) = 20 ⋅ log
Ka
(ii) = 20 ⋅ log K a − 40 ⋅ log j ⋅ ω ,
j ⋅ω
⇒ ωa = Ka .

( )
y ωa = 0

You might also like