Ch.

5 Feedback Control System

Objectives
Covered: Math. Modeling & Transfer Function
New: include control system characteristics
1. Sensitivity to model uncertainties
2. Steady-state errors (t): Y() , E()
3. Transient response characteristics to input test
signals
4. Disturbance rejection
Find TF of a feedback control system
R

Closed-loop control
Has ability to reduce system sensitivity
If G(s)H(s)>>1 for all complex freq. of interest,
then:
G
G
1
R= R
R

Y (s) =
GH
H
1+ G H
If H = 1, Y(s) = R(s) the desired result
R

By increasing the gain of G(s)H(s) it reduces the

effect of G(s) on the input -> variation of the
parameters of the process, G(s), is reduced
(advantage of a feedback system)
But, making G(s)H(s)>>1 can lead to highly
oscillatory & even unstable response

When Process, G(s), is changed

<Open loop> Y ( s) = G( s) R( s)
G(s)

R(s)

Y(s)

<Closed loop>
Y ( s) + Y ( s) =

G(s) + G( s)
R( s )
1 + (G( s) + G( s) )H ( s)

Y (s) =

Then the change in the output is

Y ( s) =

G
R
1+ G H

G( s)
R( s).
(1 + GH (s) + GH (s))(1 + GH (s))

When GH>> GH(s), as is often the case, we have

Original T. F.
G
1+ G H

G ( s )
Y ( s ) =
R ( s ).
2
[1 + GH ( s)]

Change of the output is reduced by [1 +GH]

<System Sensitivity>

S =
T
G

Ratio of % change in sys T.F.

Ratio of % change in Process T.F.

T = ln T = T G
G
ln G G T
G
Open-loop

Y (s) = G(s)R(s),

T ( s) =

Y ( s)
= G ( s )
R( s)

Closed-loop
1
T (1 + GH ) G ( H )
G
=
=
T ( s) =
(1 + GH ) 2
(1 + GH ) 2
1 + GH , G
S GT =

T G
1

=
G T
(1 + GH

)2

G
G
1 + GH

1
(1 + GH

T
Reduced S G below that of the open-loop sys by
increasing G*H (>>1.0).
T H
GH
T

SH =

* if GH >>1.0 S GT = 1

H T

(1 + GH )

Feedback components should not be varied with

environmental changes change in H(s) directly
effect output response

< Disturbance in a system >

Disturbance

D(s)
Control

M(s)

G1(s)

+
+

G2(s)

C(s)

H(s)
Sensor

D(s)

M(s)

Gd (s)

By superposition

Gp (s)

K G1G2
C(s) = 1 + K G G H M ( s)
C(s)
1 2
G2
D ( s )
+
1 + K G1G2 H

= G (s)M(s) + Gd(s)D (s)

p

State Variable Model

X = Ax + Bu
y = Cx

m(t)
u = d(t)

Then T.F. G(s) = C[sI-A]-1B

Gd(s)]
= [Gp (s)
T.F. from the control
Input M(s) to the output

T.F. from D(s) to the

output

Disturbance in a closed-loop system

Where

C(s) =

Gc G p

1 + G c G p H

Gd
R
(
s
)
+

D(s)

1 + G c G p H

= T(s) R(s) + Td(s) D(s)

The loop gain Gc Gp H must be made large to reduce
the system sensitivity.

Gd
Gd
=
Td = 1 + G G H G G H
c p
c p

Reducing Disturbance
1. Reduce the gain Gd(s)
2. Increase the loop gain Gc Gp H (Choice of Gc)
3. Reduce the disturbance d(t)
4. Feed forward method if the disturbance can be
measured

Example: Feed forward method

T =

T.F of the disturbance

Td =

D(s)

Td

Gd
Gcd Gc GP
+
1 + Gc GP H 1 + Gc GP H

if Gcd is selected to make Td = 0

Gd Gcd Gc GP = 0
Gd
Gcd = Gc
GP

P
k

C(s)

Steady State Error: Error after the transient

response has decayed (t)
Open Loop

R(s)

G(s)

Y(s)

Error Eo(s) = R(s) - Y(s) = R(s) - G(s)R(s)

= R(s) [1- G(s)]
Closed Loop
Error

1
EC (s) =
R(s)
1+G(s)
The final value theorem

lim e(t ) = lim sE ( s )

t 0

Error for unit step input

1
u (t )
s
1) Open Loop case

t0

e0 () = lim s[E ( s)]

s 0

2) Closed Loop Case

s 0

1
= lim[1 G( s)] = 1 - G(0)
s s 0

1 1
1
ec () = lim s
=

s 0
1 + G ( s ) s 1 + G ( 0 )

G(0) is the dc gain and usually greater than 1. If

G(0) >> 1, closed-loop error ec () is very small

<Example> Consider T.F. G ( s) =

Open Loop case

K
1
Input R(s) =
s +1
s

Ec ( s ) = [1 + G ( s )]R ( s ) = [1 G ( s )]

1
s
e0 ( ) = lim Ec ( s ) = 1 G (0) = 1 K
s 0

Closed Loop Case

1
1
1
R(s) =
1 + G (s)
1 + G (s) s
1
ec ( ) = lim sEc ( s ) =
s 0
1+ K
Ec ( s ) =

1. For open loop if K = 1; e0 ( ) = 0 . However, during

the operation the parameter of G(s) will change due
to environment changes.
2. Closed loop error ec () can be reduced by selecting
1

high gain of K (if K = 100, ec () = 101 )

3. In case of the gain setting drifts or changes

k
( k = 0.1 ), open loop e0 () = 0.1 (10%) while

closed loop: K = 100 90

ec ( ) =

1
1
= 0.0011
101 91

(0.11%)