Steady -

State Response
·
Reference Tracking

-want output y to
stay close to reference +

Equivalently error e =r-y should be small

:
-

-steady-state error : how small error

eventually gets

elt)
es =

·
Questions

what is the ideal value steady-state 0

of error
- -

-what might affect design

the steady-state error +

Is
steady-state error
always finite No
->
-

·
Assessing steady-state error

-
Test waveforms For evaluating steady-state errors :

=I = t
=

It

example of ss error :

Ess = O

0
ess
=

2js 0
=

ess ⑳
=
·
key technical tool

knite value theorem :

* signal f with Laplace transform FCS)

* all poles of sFCS) have Recs) < O

,
then

lim f(t)
= 1/M
SFCS)
S- 0
t- x
men
ss value

as long as sFCSI has poles with negative real part .

·
Transfer Function From reference to error

% -

/ I
I
Gre=
1 + PCS) CCS)

(IS) : PCSICCS) =

"loop transfer Funchon"

not the closed -

Loop TF
·
General formula for steady-state tracking error

Reference :

Iaplace transform RCS)

Loop transfer Function : (CS) = CCS) PCS)

If the steady-state error is Finite , then :

lim SRCS)
ess
=

sGreCS) -

RCs) =

sto 0
+ PassCis)
-

Fina value theorem

for ECS)

ECs) =

GreCS) .
RCS)
-

Note :
to use this Formula you must check that sGre' RCS) has

all poles with negative real part

-example
:
cruise control
 * maintain constant
speed despite disturbances

steady-state For disturbances

·
error

Al

Load disturbances can cause

steady-state want control
;

error
- :

to reduce
system

-
V 2

I- I
a

Transfer from to
-

Function : e

Gre
- " cess

· Overall Steady-state Error

-Error : in terms of reference R and disturbance V

PCSS
Else
Res Ves
:

Mess
-

- -

Gre eve
-Final value theorem :

SRCS) lim sPCsVess

ess
-
=

1
+ PCs) Cess 0
PlssVs)
+

1+

reference disturbance
=ess -

See
-

eir From
err From

reference disturbance
System type and Steady-state Response
·
Steady-state Error for a
step input
r(t) Act) T RCS) Ys
Reference
- = :
=
:

zis- cris

steady state error

-

-usi
lim lim
ess
=

sECS)= =

S+ S
kp
+
A

where
kp=1mLCs) is the
position constant

observations :
- a

-
* If Less has a
pole at S
=
0 then
kp =

If kp A then steady state error is zero

=

* .

Ikp
-
0

Steady-state Error Ramp Input

·
For a

Reference
:

r(t) =
t 1A ,
LT : RCS1 =

1/s
-

steady state error :

ess =

M SECs =/Im
S- G
$[re(s) I

. stu

I
- -
-- -
I
Tim lim
sLCS)
"SLCS) Kr
st
S+ 0 S +
0

other
lim
where K= sLcs) is the
velocity constant LIS)
=

/S -

StUFE
S-O
/im
-
observations stoS(s)
=

5 ·

ele - a

If
* LIS) has a
pole of multiplicity, 2 at S =
0 then Kn =
&

*
If kn =

x then steady-state error is zen .

·
Steady state Error For a parabola Input

Reference : rct)
=

+Ye ACA) ,
25
:

RCS1 =

1/53

Steady
:
state error
-

ess =
YosGrels) RCsI- -

Elis
 -= I
lim
S2LCS) Ka
sto

where Ka =

$ S2LLS) is the acceleration constant

-
observations :

* If LCSC has pole of multpliary 3 at s =

0 then Ka =

&

* If Ka =

N then steady-state error is zero

.

system Type
·

Loop transfer Function :

-

25+z , (StZm)
PCS) (CS)
...

LCS) =
=

S"cS+p , ) .
- 25 +PK)
-
E
->

the at
system type is the multiplicity pole s 0
=
-

of
:

* no pole at s 0 type O
=

*
single pole at s 0
type I
=

= . . .

order
-warning system type is different From system
:

steady-state tracking
·
error

#
-
Note : If rit) = < r, Ct) + Brect) then ess
=

Less , i + Bess , 2
m -

unit step unit

ramp

Implication Type s system has zero ss linear

-

error For
any
: .

A parabolas
combination of steps ,
ramp ,
.

2
& +

Bt +
Jt

Example
·
:

A DC & has
motor system with input voltage output angular position

transfer Function PCS) =

Iy) ·
It is connected in a varty Feedback

CONFg · With a
proportional controller of gain K>0 .
 -
-system type type I
:

steady state step Input

-

error af

based on the table above :

ess =

steady-state
-

error
For ramp Input

1)kr = E
t 1 *
ess
=
=
=

kv us
sc
ess
E
: =

Proportional -

Integral Control
·

Integral control

-recall proportional
-

denvative controllers

* For 2nd order

plants
could place closed loop poles in
any desired

vocation but

* could not remove

steady-state entr

aims steady-state
-

Integral control : to
remove error

·
proportional -

Integral(PI) control

-
controller transfer Function : (add a
pole at s =
0 (

Ki
CCS) kp
= +

W
↓

-
r

I
-Time Domain

ult)
=

kpeCt) +
KiSect) de

control &
· Pl steady-state error

-
given :

*
Pl Controller :
CCS) =

kp+ Ki/s
 * plant
:

PCS) -> no tews at s =

loop transfer Function :

-

(15)
=

PCS1 ·

(CS)
=

(kp +

E) PCSS

=Ki
.

pass
S
-

system type
:

type>, I

A steady-state step
error For input is 0 .

-PI controller leads to zero stracking error For step Input .

-
Time Domain Urewpoint
:

controller description

u(t) =

kpeCt) +
KiS, ec2) de

and step reference in put

-
If closed loop system is stable and

* Uss is steady state control

Input

Ha
* ess is
steady-state error

then um
ess>6

See a
Kin Secd1
us Fest
- &

-Limit on the RHS can

only be Inite if ess
=
0

· PI control of 1st order system

/
-plant :

pass -

Controller : a

CCs) =

kp + Ki/s

-
TE From ~ to y

Gryls
= his - is
stke
estki
-

-closed is second
loop system order

order plant
· Closed Loop Response : 1st
 - By changing Ki
Up and ,
can
put poles in any locations

-characteristic polynomial :

S2+ (atkp)s +
Ki =
s +2 Wnst Wi

Implications LIS) PCS) CCS)

*kital
=

-
: ·
=

*
system has type I , so zeross tracking error For step input .

*
Stability For K, >0
kp > a
-

* Introduces a zer at Ki/kp

· Closed Loop Pole locations

S2+ (atkp)s +
Ki =
s +2 Wnst Wi

underdamped case :
0 <I<1 - (Ki > (atkp)/2 > 0

Complex Conjugate pole

*
a Exed
Up 2) Same real
part

T
X
same Is
>
Fixed Ko same we

·
Fixed Up Increase ko+ Proshes
,
pole upwards
Faster oscillations
Im
part controls
. oscollations

Exed Ko Increase kp Pushes pole left

·
to
.

faster Ts
-

Real Part :
-

(at kp)/2
reduced oscillation
Radius :
i

Increasing Ki
* ,
increases oscillations
frequency .

*
Increasing kp ,
increases
settling true -

·
PI control of 2nd order system

-plant it

/
s + a , staz

-controller : (CS) =
Kp+ Kr/s
-

TE From to Y

Engess =

Access" real
I kpS+ Ki
S+ als + (92+kp)s+ ki - closed
loop system
is 3rd order
 · closed loop system :

and order plant

in
By changing up Akr cannot put poles any
location
-

Implications
:

(s) =

( * ) (Estar)
has I
tracking
system type ss error
Input
*

so zero For step

* cannot stabilize every 2 order

system with Pl controller

*
Introduce few at -

Ki/kp .

Proportional -

Integral -

Derivative Controller

·
PIP control

-recall proportional
-

denvative controllers

* For 2nd order

plants
could place closed loop poles in
any desired

vocation but

* could not remove

steady-state entr

recall
proportional Integral controllers
-
-

but
* zew steady-state error fer step imprts

* For end order plants cant put closed loop poles In

any desired

vocahon .

·
Proportional -

Integral -

Derivative Controller (PID)

Controller transfer function :

Ccx =

kp +

b +
kaS

- as
/te
I -
I
-

time Doman :

Uct1=
kpect)
+

KiS ects
an+Kd
·
PID control and steady-state error

-
Given :
 PID controller (CS) kp + Ads
:

* =
+

* plant
:

PCS) With no zeros at 3 = 0

Loop transfer function : LCs) =PCs) -ccs)

=

Pas
(4(+ki)
system type atleast 1
(type 1)
:

zer ss tracking error

for step Input .

·
PIP controller of 2nd order system

-plant :

PCS) = -
S2+ a
, sta2

controller : Casi =

kp+ Ki/s+ Kas

it From
i a he is
-

-
kds2 + kpS + Ki

S + (a , + kd)S2 + (Ac + kp)S + Ki + closed bop system

is 3rd order
·
closed loop Response
-characteristic polynomial : +(a +kd)s2 ,
+
(Az+kp)s
+
Ki

By changing gains put poles in location

-

can
any .

leaista) ( *Kig
+
Implications
- :

*
system has type I So zew Is
tracking error for step input

Introduces
* Zeros at roots of kdS +

kps + Ki

*
If all gains positive , zeros have negative real part .

·
Example
-

given a plant with TF 5tz) And PID controller garns so that

the closed loop characteristic polynomial has

poles at s = -
41s = -

1tj

c acCS) S
=

2 kp + kas =
(s)
= +

bc(s) 32kd K;
skp+
= +

Grycs) = rkps +
Ki =

AcCs ApCS) -

+basIbs
2

2) (S + 4) C(S+1) + 1 (

=
+ 652 +
10s +O =
s +(2 + kd)s + kpS+ Ki
 Less =

<Freetopez system

Zieger -
Nichols PID tuning rules

Eleger Nichols tuning method

·
step response
-
-

step response method

-

I
* :

approximation of delay

*
& :
Steepest slope of the step resp .

* can determine parameters once step

response has reached inflection point

controller parametrisation :

Cs)
=

kp ( 1 +

Is +
TaS) =

kp+ Gi + Ads

P =
ki kp Td .
=

kd

suggested parameter values (based or measured [1al

= T
X X

·
Improved DID tuning Methods (step-responses
-First-order A
time-delay (FOTD) model :

+
- es
Pass -

* 2 :
approx .
or delay

* T =
To - 2 where Tos is time & 63 % K
-
I * K :
zero Frequency gain (ss value (

suggested tuning 0 152 +

.
0 .
35 T 0 . 462 +
0 . 025
-
kp =
,ki
=

For PI control K .

2 k22