Second Order Systems

Reading Assignment: Chapter 5

 Origin of 2 nd order systems
• Two 1st order systems in series
• Two 1st order systems in parallel
• Two interacting 1st order systems
• “Inherently” second order systems that
do not come from two 1st order systems.
Physically, they arise as mechanical or
electrical systems.
Physical origin of inherently 2nd order systems
Prototype second-order system: the mass-spring system
dy
y = (displacement ), = v = ( velocity )
dt
Forces acting on the mass:

Newton’s law:
d  linear 
  =  ( forces )
dt  momentum 

dy
Equations =v
dt
of motion: dv
m = −k y − Cv + f
dt
Physical origin of inherently 2nd order systems
Prototype second-order system: the mass-spring system
dy
y = (displacement ), = v = ( velocity )
2 dt
d y
2
= ( acceleration )
dt
In some applications, y is the
Newton’s law: only variable of interest…
d  linear 
  =  ( forces ) ( mass )  ( acceleration) = ( forces )
dt  momentum  d2y dy
m 2 = −k y − C + f
dy dt dt
Equations =v
dt
of motion: dv d2y dy
m = −k y − Cv + f m 2 +C +k y = f
dt dt dt
Physical origin of inherently 2nd order systems
Prototype second-order system: the mass-spring system
dy
y = (displacement ), = v = ( velocity )
2 dt
d y
2
= ( acceleration )
dt
Input = f
Input = f Output = y
States = y, v input-output model
state-space model d2y dy
m 2 = −k y − C + f
dy dt dt
Equations =v
dt
of motion: dv d2y dy
m = −k y − Cv + f m 2 +C +k y = f
dt dt dt
 Mass - spring system

dy
y = (displacement ), = v = ( velocity )
2 dt
d y
2
= ( acceleration )
dt
Define: u = f
k
 dy Natural frequency : n =
 dt = v
m
 1 m
m dv = −k y − Cv + f = =
n k
 dt
C
or Damping factor :  =
2 mk
d2y dy
m 2 +C +k y = f 1
dt dt Steady state gain : k =
k
 Mass - spring system
2
d y dy  dy
 + 2  + y = ku  dt = v
2
2 or
dt dt

 2 dv = − y − 2 v + ku
 dt
u= f
k
 dy Natural frequency : n =
 dt = v
m
 1 m
m dv = −k y − Cv + f = =
n k
 dt
C
or Damping factor :  =
2 mk
d2y dy
m 2 +C +k y = f 1
dt dt Steady state gain : k =
k
A 2nd order system may be represented
either as two 1st order differential equations
or as one 2nd order differential equation
What about two 1st order systems in series?

Yes, we can eliminate the states and

get a 2nd order differential equation …
The result is:

1 + 2 But if we convert to 2nd order

It has  = 12 ,  = , k = k 2 k1 differential equation, we may
2 12
lose physical meaning …
 2
d y dy
2 nd order system:  2
2
+ 2  + y = ku
dt dt
• k ≠ 0 is the steady-state gain
• τ > 0 is the characteristic time
------ also called natural period
• ζ ≥ 0 is the damping factor
How can we calculate its response?
The most convenient method is the Laplace transform:
 
2  s 2Y ( s ) − sy ( 0 ) − ( 0 )  + 2 ( sY ( s ) − y ( 0 ) ) + Y ( s ) = kU ( s )
dy

 dt 
2 s + 2 2 dy k
Y (s) = 2 2 y ( 0) + 2 2 ( ) 22
0 + U (s)
 s + 2s + 1  s + 2s + 1 dt  s + 2s + 1
Inverting the Laplace transform, we find y(t)
 2
d y dy
2 nd order system:  2
2
+ 2  + y = ku
dt dt
Laplace transforming,
2 s + 2 2 dy k
Y (s) = 2 2 y ( 0) + 2 2 ( 0) + 2 2 U (s)
 s + 2s + 1  s + 2s + 1 dt  s + 2s + 1

k
Under zero initial conditions, Y (s) = 2 2 U (s)
 s + 2s + 1

k
Transfer Function: G( s) = 2 2
 s + 2s + 1
 Second Order System:

Roots of  s + 2  s + 1
2 2

  2
−1
  1  simple real roots −   OVERDAMPED
 
1
 = 1  double real root −  CRITICALLY DAMPED

 1−  2
0    1  complex roots −  i  UNDERDAMPED
 
1
 = 0  pure imaginary roots  i  UNDAMPED

 Second Order System:

How can we calculate the step response?

M
When U ( s ) = ,
s
kM
Y ( s) =
s(2 s 2 + 2 s + 1)

Must expand in partial fractions …

But the form of the expansion depends on the kind of roots …
 Second Order System:

How can we calculate the step response?

  
  
C
 kM  1 + C C  , if   1
2
+ 3
 s   2 −1   2 −1 
  s+ − s+ + 
      
  
  
 C
 kM  +  , if  = 1
C C
1 2
+ 3
Y (s) =  s s+1  1 
2

   s+  
    
  
  
 kM  C1 + C2
+
C3  , if 0    1
  s 2 
  1 −  2
 1 − 
 s + −i s+ +i 
      
Second Order System: Step Response

Three alternative expansions → three alternative formulas …

After quite a bit of algebra, the final answer is:
  1   −  2 −1    +  2 −1 
 kM 1 −  1 +   e 
− t 1 − t
 − 1 − e   , if   1
  2   − 1 
2
2
  − 1 
2

  
   t  − t 
 kM 1 − 1 +  e  , if  = 1
y (t ) =     

  − t 
  1−  2    1 −  2  
 kM  1 − e  cos  t+ sin  t    , if 0    1
  
    −  2    
  1   


Qualitative
behavior: ζ≥1

0<ζ<1

ζ=0
Effect of  on Overdamped Response
Effect of  on Underdamped Response
Effect of  on Response
 Characteristics of an
Underdamped Response (0 < ζ < 1)
• Rise time tr
• Peak time tp1
• Period (T)
• Overshoot (A/kM)
• Decay ratio (B/A)
 − t 
  1−  2    1−  2  
y ( t ) = kM  1 − e  cos  t+ sin  t  

    
 1−  2  

 
 
  ( − cos −1  ) 
Rise Time: tr = Peak Time: t p1 =
1−  2 1−  2

2
Period of Oscillation: T =
1−  2

A   
Maximum Overshoot : MOS = = exp  − 
kM  1−  2 
 
B  2 
Decay Ratio: DR = = exp  − 
A  1−  2 
 
 Second Order System:
Sinusoidal Response when ζ > 0

M
U ( s) =
s2 +  2

kM 
Y ( s) =
( s +  )( s + 2  s + 1)
2 2 2 2

Again, must expand in partial fractions …

C1 C2  Terms from the roots 
Y (s) = + +  
s − i s + i  of  s + 2  s + 1 
2 2

The terms from the roots of τ2s2+2ζτs+1 die out as t →∞

 Second Order System:
Sinusoidal Response when ζ > 0

The end result is:

y ( t )  kM
( )
1 −  2 2 sin (t ) − ( 2 ) cos (t )
, for large t
(1 −   ) 2 2 2
+ ( 2 )
2

Combining sine and cosine,

kM
y (t )  sin (t +  ) , for large t
(1 −  
2 2 2
) + ( 2 )
2

 
 1−   2 2

where  = − cos −1  

 (1 −   ) + ( 2 )
2 2 2 2


• Same frequency ω
• Different amplitudes
k
Amplitude ratio: AR =
(1 −   )
2 2 2
+ ( 2 )
2

• Not in phase  
 1−   2 2

Phase shift:  = − cos −1  
(1 −  2 2 ) + ( 2 )
2
 
2
 
Bode Diagrams: k
AR =
(1 −   ) + ( 2 )
2 2 2 2

Amplitude Ratio
in log-log
 Bode Diagrams: k
AR =
(1 −   ) + ( 2 )
2 2 2 2

Amplitude Ratio
in log-log
1
The frequency c =

is called corner frequency
 Bode Diagrams: k
AR =
(1 −   ) + ( 2 )
2 2 2 2

Amplitude Ratio
in log-log

Resonance:
For , there is
a maximum at the
frequency

Peak value of AR:

k
ARmax =
2 1 −  2
Bode Diagrams:  
 1−   2 2

 = − cos −1  

 (1 −   ) + ( 2 )2
2 2 2


Phase Angle
in lin-log
 General Form of a
Linear Second Order System
dx1
= a11 x1 + a12 x2 + b1u
dt
Input = u
dx2
= a21 x1 + a22 x2 + b2u States = x1 , x 2
dt
Output = y
y = c1 x1 + c2 x2 + du
where a11, a12, a21, a22, , b1, b2, c1, c2, d
are constant parameters
 General Form of a
Linear Second Order System
In matrix form,
d  x1   a11 a12   x1   b1  Input = u
 x  = a    +  u
dt  2   21 a22   x2  b2   x1 
State vector =  
 x1  x2 
y =  c1 c2    + du
Output = y
 x2 
 General Form of a
Linear Second Order System
Laplace transformation, under zero initial conditions:
dx1
= a11 x1 + a12 x2 + b1u  sX1 ( s ) = a11 X 1 ( s ) + a12 X 2 ( s ) + b1U ( s )
dt
dx2
= a21 x1 + a22 x2 + b2u  sX 2 ( s ) = a21 X 1 ( s ) + a22 X 2 ( s ) + b2U ( s )
dt
y = c1 x1 + c2 x2 + du  Y ( s ) = c1 X 1 ( s ) + c2 X 2 ( s ) + dU ( s )

b1s + ( a12b2 − a22b1 )

Solving for X1(s) X1 (s) = U (s)
s − ( a11 + a22 ) s + ( a11a22 − a12 a21 )
2
and X2(s):
b2 s + ( a21b1 − a11b2 )
X 2 (s) = U (s)
s − ( a11 + a22 ) s + ( a11a22 − a12 a21 )
2

From the above, we can calculate the output …

 General Form of a
Linear Second Order System
Laplace transformation, under zero initial conditions:
dx1
= a11 x1 + a12 x2 + b1u  sX1 ( s ) = a11 X 1 ( s ) + a12 X 2 ( s ) + b1U ( s )
dt
dx2
= a21 x1 + a22 x2 + b2u  sX 2 ( s ) = a21 X 1 ( s ) + a22 X 2 ( s ) + b2U ( s )
dt
y = c1 x1 + c2 x2 + du  Y ( s ) = c1 X 1 ( s ) + c2 X 2 ( s ) + dU ( s )

The previous results on X1(s) and X2(s) give:

 (c1b1 + c2b2 ) s + (c1a12b2 + c2 a21b1 − c1a22b1 − c2 a11b2 ) 
Y ( s) =  + d U ( s)
 s − (a11 + a22 ) s + (a11a22 − a12 a21 )
2

Transfer function