Académique Documents
Professionnel Documents
Culture Documents
Con
ntrol sysstems an
nalysis aand desiign focu
us on thrree specification
ns:
(1) Traansient reesponsee,
(2) Stab
bility, an
nd
the robbustnesss of the design
(3) Steeady-statte errors, taking into account
a
along with
w eco
onomic aand sociial consiideration
ns.
Steeady sta
ate is that parrt of th
he outp
put which rem
mains affter tra
ansients
com
mpletely
y vanish
h from tthe outp
put.
inputts repreesent co
onstant p
position
n. It is used
u
to determ
mine the ability
of the
t conttrol systtem to p
position
n itself with
w resspect to a statio
onary ta
arget.
Ma
ainly thee steady
y state respons
r
he follow
wing specificattion:
e has th
inpu
uts reprresent coonstant--velocity
y inputss to a position control system
and
d can be used to
o track a constan
nt-velocity targeet.
Steeady-sta
ate erro
or is thee differeence bettween th
he inpu
ut and th
he outpu
ut for a
prescribed test input as t .
It iss the diffference betweeen the acctual outtput and
d the dessired outtput.
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 1 of 73
3
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 2 of 73
3
represeent consstant-accceleratio
on inputts to po
osition control
c
s
systems
and
d can be used to
o track an
n acceleerating target.
t
Pag
ge 4 of 73
3
Pag
ge 3 of 73
3
Let us
u examinee the conceept of stead
dy-state errrors. In Fig
gure 7.2(a) a step inpu
ut and two
o possible outputs
o
are
show
wn. Outputt 1 has zero
o steady-sttate error, and outpu
ut 2 has a finite
f
stead
dy-state errror, e2 (). A similar
exam
mple is show
wn in Figurre 7.2(b), where
w
a ram
mp input is compared with outpu
ut 1, which has zero stteady-state
errorr, and outp
put 2, which
h has a finiite steady-sstate error, e2 (), as measured
m
vvertically between thee input and
outp
put 2 after the transieents have d
died down. For the ra
amp input another possibility
p
e
exists.
If th
he outputs
slopee is differen
nt from tha
at of the in
nput, then output
o
3, shown in Fiigure 7.2(b
b), results. Here
H
the stteady-state
errorr is infinitee as measurred verticallly between the input and
a output 3 after thee transients have died down, and
t app
proaches in
nfinity.
FIG
GURE 7.2 Steaady-statee error: (a) step input; ((b) ramp
p input.
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 5 of 73
3
Pag
ge 6 of 73
3
: Fin
nd the ssteady-sttate erro
or for th
he systeem of Figure
F
7
7.3(a)
iff
e input iis a unit step.
T ( s ) = 5 / (s + 7 s + 10
0 ) and the
2
Fro
om Figurre 7.3(a) the errror E(s) write:
Butt
C (s ) = R( s )T (s )
(7.3)
App
plying the
t final value thheorem and lettting t ap
pproach infinity,, we obttain
e() = Lim
L e(t ) = Lim[sE
E(s)]
(7.5
5)
s0
Sub
bstitutin
ng Eq. (7
7.4) intoo Eq. (7.5) yield
ds
e() = Lim
L sE(s)[1 T (s)]
(7.5)
s0
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
G( s) =
Sub
bstitutin
ng Eq. (7
7.3) intoo Eq. (7.2), simp
plifying,, and sollving for E(s) yields
E (s) = R(s)[1 T (s)]
(7.4)
Con
nsider thhe forwaard gainn is given
n by:
Sollution: F
From th
he probleem stateement,
R(ss) = 1/s
(7.2)
E ( s) = R( s) C ( s)
) (
(7..7)
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
(7.8)
E ( s) = R( s) C ( s)
e() = 1/2.
Pag
ge 7 of 73
3
(7.14)
Fro
om Figurre 7.3(b
b) the errror E(s) write:
Sub
bstitutin
ng into Eq.
E (7.4)) yields
E ( s ) = s 2 + 7 s + 100 / s s 2 + 7 s + 10
(s + z1)(s + z2 )..........
) s + p2 )..........
s n (s + p1 )(
(7.9)
Fin
nally, subbstitutin
ng Eq. (7
7.9) into
o Eq. (7..8) and solving
s
for E(s)) yields
E (s) =
R( s )
1 + G( s)
(7.1
10)
plying the
t final value thheorem and lettting t ap
pproach infinity,, we obttain
App
Pag
ge 8 of 73
3
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 9 of 73
3
sR( s)
s 0 1 + G( s)
e( ) = Lim
(7.11)
Equation (7.11) allows us to calculate the steady-state error, e(), given the
input, R(s), and the system, G(s).
nce, to satisfy
s
E (7.13), the denomin
Eq.
nator off G(s) of (7.14) must be
b equal
Hen
to zero
z
as s goes to
t zero. Thus, nn1; thaat is,
s(1 / s)
s 0 1 + G( s)
1
1
=
1 + Lim G( s) 1 + K p
(7.12)
s0
is th
he dc gaain of th
he forwaard transfer fun
nction, siince s, the
t frequ
uency variable,
v
is approach
a
hing zerro.
.
(n1, at least one purre integrration inn forward
d path)
In order
o
to have zeero steaddy-state error,
K p = Liim G(s) =
(7.1
13a)
s0
e( ) = estep ( ) =
Prepared by: Dr. Mohammad Abdul Mannan
Page 10 of 73
1
1
=
=0
1+ K p 1+
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
(n=
=0, no pu
ure integration in forwaard
path
h): If there are no
n integgrations,, then n=
=0. Usin
ng Eq. (77.14), we
w have
z z ...........
K p = Liim G ( s ) = 1 2
p1 p2 ..........
s 0
e() = estep () =
(7.1
14a)
p1 p2 ..........
1
=
1 + K p p1 p2 .......... + z1z2 ..........
(7.14b)
(7.13b)
Pag
ge 11 of 73
7
input to
o a unityy feedbacck systeem,
Forr a
The steady
y-state error w
will be zero
z
if th
here is at leastt one pu
ure integ
gration
in the
t forw
ward pa
ath.
If there
t
arre no inttegratioons, then
n there will be a nonzero finiite errorr.
s 0
K p = Lim G(s)
Thee term
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 12 of 73
7
Hen
nce, to satisfy
s
E (7.17), the denomin
Eq.
nator off G(s) of (7.14) must be
b equal
to zzero as s goes to
t zero. Thus, n2;
n
thatt is,
.
w
R(ss)= 1/s, we
w obtaain
Usiing Eq. (7.11) with,
s(1/ s 2 )
1
1
=
e() = erampp() = Liim
=
s0 1 + G(s) Lim sG(s) Kv
(7.16)
(n=
=1, one pure
p
integration
n in forw
ward
path
h): If there are one
o inteegrationss, then n=1.
n
Usiing Eq. ((7.14), we
w havee
s0
m sG(s) is called
The term Kv = Lim
d
z z ...........
K v = Liim sG ( s ) = 1 2
p1 p2 ..........
s 0
s0
e( ) = eramp ( ) =
1
1
= =0
Kv
e( ) = eramp ( ) =
(7.18a)
1
p p ..........
= 1 2
Kv z1z2...........
(7.18b)
(7.17b)
s 0
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 13 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
e() = eramp () =
1
1
= =
Kv 0
(7.19b)
Pag
ge 14 of 73
7
Forr a
to a un
nity feed
dback sy
ystem,
The steady
y-state error
e
w
will be zero if th
here is at
a least two pu
ure integ
gration
in the
t forw
ward pa
ath.
If there
t
is one inteegration
n, then there will
w be a nonzerro finitee error.
If there
t
arre no inttegratioons, then
n there will be infinitee error.
1
s(1 / s3 )
1
e( ) = e parabola ( ) = Lim
m
=
=
s 0 1 + G( s ) Lim s 2G( s) K a
m s 2G(s) is called
The term Ka = Lim
(7.2
20)
s 0
s0
e( ) = e parabola( ) =
1
1
= =0
Ka
(7.21b)
Hen
nce, to satisfy
s
E (7.2
Eq.
21), the denomin
nator off G(s) of (7.14) must be
b equal
to zero
z
as s goes to
o zero. T
Thus, n
3; that is,
Pag
ge 16 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 15 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 17 of 73
7
(n
n=2, two
o pure in
ntegrations in forward
f
path
h): If there are two
t inteegrationss, then n=2.
n
Usiing Eq. (7.14), we
w havee
z z ..........
K a = Liim s 2G ( s ) = 1 2
p1 p2 ..........
.
s 0
p p ..........
1
e( ) = e parabola( ) =
= 1 2
z1z2 ..........
Ka
(7.222a)
(7.22b)
e( ) = e parabola( ) =
1
1
= =
Ka 0
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
(7.23b)
Pag
ge 18 of 73
7
120( s + 2)
s 0 ( s + 3)( s + 4)
=0
5 5
= =
e( ) = eramp ( ) =
Kv 0
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 20 of 73
7
: Fin
nd the steady-sstate errrors forr inputss of 5u((t), 5tu((t), and
5t u(t)
u to th
he system
m show
wn in Fig
gure 7.6.. The fu
unction uu(t) is th
he unit sttep.
which implies a response similar to output 1 of Figure 7.2(a). Notice that the
integration in the forward path yields zero error for a step input, rather than
the finite error found in Example 7.2.
For the input 5tu(t), whose Laplace transform is 5/s2, the steady-state error
will be five times as large as that given by Eq. (7.16), or
s 0
20( s + 2)
12
K p = Liim G ( s ) = Lim
= 20
s 0
s 0 ( s + 3)( s + 4)
5
5
5
e( ) = estep ( ) =
=
=
1 + K p 1 + 20 21
Pag
ge 19 of 73
7
K v = Lim sG ( s ) = Lim s
F
wee verify that the closed--loop sy
ystem is indeed stable. For
F this
Sollution: First
exaample w
we leavee out thhe detaills. Nextt, for th
he input 5u(t), whose
w
L
Laplace
tran
nsform iis 5/s, th
he steaddy-state error wiill be fiv
ve timess as larg
ge as thaat given
by Eq. (7.1
12), or
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
: Fin
nd the steady-sstate errrors forr inputss of 5u((t), 5tu((t), and
5t u(t)
u to th
he system
m show
wn in Fig
gure 7.5.. The fu
unction uu(t) is th
he unit sttep.
2
For the input 5tu(t), whose Laplace transform is 5=s2, the steady-state error
will be five times as large as that given by Eq. (7.16), or
For the input 5t2u(t), whose Laplace transform is 10/s3, the steady-state error
will be 10 times as large as that given by Eq. (7.20), or
120( s + 2)
=0
( s + 3)( s + 4)
10 10
= =
e() = e parabola( ) =
Ka 0
K a = Lim s 2G ( s ) = Lim s 2
s 0
s 0
Page 21 of 73
100( s + 2)( s + 6)
=0
s ( s + 3)( s + 4)
10 10
= =
e() = e parabola( ) =
Ka 0
K a = Lim s 2G ( s ) = Lim s 2
s 0
s 0
100( s + 2)( s + 6)
= 100
s ( s + 3)( s + 4)
5
5
1
e( ) = eramp ( ) =
=
=
K v 100 20
K v = Lim sG ( s ) = Lim s
s 0
Sollution: For
F the input 5uu(t),whoose Laplace tran
nsform is 5/s, the
t stead
dy-state
erro
or will be
b five tiimes as large ass that giv
ven by Eq.
E (7.12), or
s 0
which implies a response similar to output 2 of Figure 7.2(b). Notice that the
integration in the forward path yields a finite error for a ramp input, rather
than the infinite error found in Example 7.2.
1000( s + 2)( s + 6)
=
s 0
s 0 s ( s + 3)( s + 4)
5
5
e( ) = estep ( ) =
=
=0
1+ K p 1+
K p = Liim G ( s ) = Lim
For the input, 5t2u(t), whose Laplace transform is 10/s3, the steady-state error
will be 10 times as large as that given by Eq. (7.20), or
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 22 of 73
7
Page 23 of 73
find
d the stteady-sttate erroors for inputs of
o 25u((t), 37tuu(t), and
d 47t2u((t). The
fun
nction u((t) is thee unit steep.
Page 24 of 73
e() = estep() =
1
1 + llim G(s)
s 0
s
(7.30)
Forr a rampp input, tu(t):
t
e() = eramp() =
1
lim
m sG(s)
(7.31)
s
0
find
d the stteady-sttate erroors for inputs of
o 15u((t), 25tuu(t), and
d 80t2u((t). The
fun
nction u((t) is thee unit steep.
e() = eparabbola() =
1
lim s 2G(s)
(7.32)
s0
Pag
ge 25 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 26 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 27 of 73
7
: Fo
or eachh system
m of Fiigure 7
7.7, evaaluate th
he static error
con
nstants and fin
nd the expecteed errorr for th
he standdard steep, ram
mp, and
parrabolic in
nputs.
SOLUTION: First verify that all closed-loop systems shown are indeed
stable. For this example we leave out the details. Next, for Figure 7.7(a),
K p = lim
m G( s)
, Kp:
(7.3
33)
s 0
1
1+ K p
G(s)
, Kv: Kv = lim sG
(7.3
34)
s0
e() = estep () =
1
Kv
(7.35)
s0
Pag
ge 28 of 73
7
e() = eramp() =
s 0
s 0
RE 7.7 F
Feedback
k contro
ol system
ms for Example
E
7.4.
FIGUR
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
500(2)(5)(6)
= 31.25
(8)(10)(12)
s 0
Ramp,,
tu(t)
P
Parabol
a,
((1/2)t2u((t)
s 0
e() = eparabola() =
Page 31 of 73
Ty
ype 1
Type 2
Static
S
Staticc
atic
Sta
e
error
Error errorr Erro
or errror Error
E
coonstant
constant
consstant
1
1+ K p
K p = Con.
1
1+ K p
Kp =
Kp =
1
Kv
Kv = 0
K v = Co
on.
1
Kv
Kv =
Ka = 0
Ka = 0
K a = Con.
1
Ka
1
Ka
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Page 30 of 73
s 0
Pag
ge 34 of 73
7
= 875
1
1
=
= 1.14 103
Ka 875
Step, u((t)
1 1
= =0
e() = eramp() =
Kv
Inputt
Pag
ge 29 of 73
7
s 0
s 0
Typee 0
1 1
= =
Kv 0
500(s + 2)(s + 5)
=0
(s + 8)(s + 10)(s + 12)
1 1
= =
e() = eparabola() =
Ka 0
s 0
K p = lim G( s) = lim
1
1
=
= 0.032
e() = eramp() =
Kv 31.25
adyStea
sta
ate
errror
form
mula
=0
1
1
e() = estep () =
=
=0
1+ K p 1+
500(2)(5)
= 5.208
(8)(10)(12)
500(s + 2)(s + 5)
s 0
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
1
1
=
= 0.161
1 + K p 1 + 5.2018
1
Thus, e() = epaarabola() =
Ka
Theese quan
ntities, dependin
d
ng upon
n the form
m of G((s), can assume values of
o zero,
finiite consstant, orr infinitty. Sincce the static error
e
constant appears
a
in the
den
nominatoor of thee steadyy-state error.
e
Eq
qs. (7.30
0) througgh (7.32
2), the value
v
off
the steady--state errror decrreases ass the stattic errorr constannt increaases.
500(s + 2)(s + 5)
s 0
Tab
ble 7.2 ties
t togeether thee concep
pts of steeady-staate errorr, static error
e
con
nstants,
and
d system
m type.
Page 32 of 73
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Staatic erro
or consstants ccan bee used to speecify thhe steaady-statee error
chaaracterisstics of control
c
ssystems..
Thee positio
on consttant, Kp, velocitty constaant, Kv, and accceleratio
on consttant, Ka,
can
n be used
d as specificatio
ons for a control system
ms steaddy-state errors.
Forr examp
ple, if a control
c
ssystem has
h the specificcation Kv = 2000
0, we caan draw
sev
veral con
nclusion
ns:
1. The system is stablee.
2. The system is of Tyype 1, since
s
on
nly Typee 1 systeems hav
ve Kvs that
t
are
finite constantss.
T
0 systems
s
s, whereas Kv = for Type 2 sy
ystems.
Recall that Kv = 0 for Type
ut is the test sig
gnal. Sin
nce Kv iss specifiied as a finite co
onstant,
3. A raamp inpu
and thee steady
y-state errror for a ramp input iss inversely prop
portionaal to Kv,
we kno
ow the teest inputt is a ram
mp.
4. The steady--state errror betw
ween th
he input ramp and
a the output ramp
r
is
1/Kv peer unit of input slope.
s
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 35 of 73
7
Pag
ge 33 of 73
7
: Wh
hat info
ormation
n is con
ntained in the specificcation (ii) Kp =
100
00, (ii) Kv = 100
00, and ((iii) Ka = 1000?
Sollution:
(i) The
T info
ormation
n in the specificcation Kp = 1000
0 are:
Thee system
m is stable.
Thee system
m is of Type
T
0, since on
nly Typee 0 systems hav
ve Kps that are
finiite consttants. Ty
ype 1 an
nd Type 2 system
ms havee Kp=.
A sstep inpu
ut is thee test sig
gnal. Sin
nce Kp iss speciffied as a finite constant
c
for a step in
nput.
Thee steady
y-state errror per unit step is:
e() = esteep () =
1
1
1
=
=
1 + K p 1 + 1000 1001
Pag
ge 36 of 73
7
The system is of Type 1, since only Type 1 systems have Kvs that are
finite constants. Type 0 system has Kv=0 and Type 2 system has Kv=.
A ramp input is the test signal. Since Kv is specified as a finite constant
for a ramp input.
The steady-state error per unit ramp is:
1
1
1
=
=
= 0.001
K a 1000 1000
Prepared by: Dr. Mohammad Abdul Mannan
Page 37 of 73
e( ) = e parabola() =
G (s ) =
Sollution: Since th
he systeem is Type 1, the
t erro
or statedd in the problem
m must
app
ply to a ramp in
nput; onnly a ram
mp yiellds a fin
nite erro
or in a Type
T
1 system.
Thu
us,
1
= 0.1
Kv
Kv = 10
e( ) =
Giveen
th
he
u
unity
s
system
: Giv
ven the uunity feeedback system
s
w G (s ) =
with
K ( s + 2)( s + 4)(s + 6)
s 2 ( s + 5)(s + 7)
K ( s 2 + 3s + 30)
s n ( s + 5)
; find
(7.60)
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 41 of 73
7
(7.60.3)
s 0
s 0
s 0 1 + G1( s)G2 ( s)
sG2 ( s)
D( s )
s 0 1 + G1( s)G2 ( s)
R( s) Lim
Page 43 of 73
(7.61)
(7.61.1)
e() = eR () + eD ()
Where, eR() is the steady-state error due to R(s), which we have given by
s 0
1
is a transfer function relating E(s) to R(s)
1 + G1(s)G2 (s)
G2 (s)
TD ( s) =
is a transfer function relating E(s) to D(s)
1 + G1( s)G2 ( s)
TR (s) =
s
R( s )
s 0 1 + G1( s)G2 ( s)
Where,
Pag
ge 42 of 73
7
s 0
(7.60.2)
(7.59)
C (s ) = R ( s ) E (s )
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
(7.60.1)
(7.58)
To find the steady-state value of the error, we apply the final value theorem to
Eq. (7.60) and obtain
s 0
G2 ( s)
D(s) = TD (s) D(s)
1 + G1(s)G2 ( s)
E ( s ) = ER ( s ) + ED ( s )
Fig
gure 7.1
11 show
ws a feeedback control system
m with a distturbancee, D(s),
injeected beetween th
he contrroller an
nd the pllant. Wee now ree-derivee the exp
pression
for steady-state errror with
h the distturbancee includ
ded.
Butt,
Pag
ge 40 of 73
7
1
ER ( s) =
R( s) = TR (s) R(s)
1 + G1( s)G2 (s)
ED ( s) =
Thee transfo
orm of th
he outpuut is giv
ven by
G2 ( s)
1
R( s )
D( s )
1 + G1( s)G2 ( s)
1 + G1( s)G2 ( s)
Page 39 of 73
with
Substituting Eq. (7.59) into Eq. (7.58) and solving for E(s), we obtain
E ( s) =
Although this gain meets the criteria for steady-state error and stability, it may
not yield a desirable transient response.
find
d the value of K to yieldd a static error constant
c
t of 10,0
000.
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
K (5)
= 10
(6)(7)(8)
Pag
ge 38 of 73
7
the value of
o K and n to yyield a steady-sstate errror consstant of 1/6,000
0 for an
0tu(t).
input of 10
feedbaack
((7.55)
: Giv
ven the unity
u
feeedback system
s
with
w G (s ) =
K ( s + 5)
s 0 s( s + 6)(s + 7)(s + 8)
Applying the Routh-Hurwitz criterion, we see that the system is stable at this
gain.
K ( s + 5)
s( s + 6)( s + 7)( s + 8)
G (s ) =
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Therefore,
s 0
1
1
1
=
=
= 0.001
e( ) = eramp ( ) =
Kv 1000 1000
: Giv
ven the ccontrol system in Figurre 7.10, find thee value of K so
thatt there is 10% error
e
in tthe stead
dy state..
(7.61.2)
And eD() is the steady-state error due to D(s), which we have given by
sG2 (s)
D( s )
s 0 1 + G1( s)G2 ( s)
(7.61.3)
Page 44 of 73
F
FIGURE
E 7.12 Figure
F
7.11 system rearrranged to
t show
w disturbance as input
and
d error as
a outpu
ut, with R(s)=0.
R
Forr step diisturban
nce, D(s)) = 1/s,
eD () =
G2 (s)
s 0 1 + G1( s)G2 ( s)
eD () = Liim
1
1
+L
Lim G1(s)
s0 G2 ( s) s
0
(7.62)
Lim
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 45 of 73
7
This equation shows that the steady-state error produced by a step disturbance
can be reduced by increasing the dc gain of G1(s) or decreasing the dc gain
of G2(s).
If we want to minimize the steady-state value of E(s), we must either increase
the dc gain of G1(s) so that a lower value of E(s) will be fed back to match the
steady-state value of D(s), or decrease the dc value of G2(s), which then yields
a smaller value of e()as predicted by the feedback formula.
: Find tthe
steaady-state
errror
com
mponentt due to
t a sttep
distturbancee for th
he systeem
of Figure
F
7
7.13.
Con
ntrol syystems often do no
ot havee unity feedback beecause of the
com
mpensation used
d to imp
prove peerformance or because
b
of the physical
p
l model
for the systtem.
SO
OLUTIO
ON: Thee system
m is stablle. Using
g Figuree 7.12 annd Eq. (7.62),
(
w find
we
1
eD () =
L
Lim
s 0 G2 ( s)
+ Lim G1(s)
1
1
=
0 + 1000
1000
g
owing th
he inputt transduucer, G1(s), co
ontroller
feedbacck systeem, sho
A general
and
d plant, G2(s), an
nd feedbback, H1(s), is shown
s
in
n Figuree 7.15(a)).
s 0
Page 46 of 73
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 47 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 48 of 73
7
Pusshing th
he input transducer to th
he right past thee summ
ming juncction yieelds the
gen
neral no
onunity feedbacck systeem show
wn in Figure
F
7.15(b), wheree G(s)=
G1(s)G
( 2(s) and H(ss)= H1(ss)/G1(s).
- c((). Th
he first step
s
is tto show
w explicitly E(ss) = R(ss) C(ss) on thee block
diag
gram.
Figure 7.15(d). Finally, combine the feedback system consisting of G(s) and
[H(s) 1], leaving an equivalent forward path and a unity feedback, as shown
in Figure 7.15(e). Notice that the final figure shows E(s) = R(s) C(s)
explicitly.
The steady-state actuating signal for Figure 7.15(a) is
sR(s)G1(s)
s 0 1 + G2 ( s) H1( s)
ea1() = Lim
(7.72)
E 7.15 Fo
orming an equiv
valent unity
u
feeedback ssystem from
f
a general
g
FIIGURE
nonunitty feedb
back systtem.
he errorr is nott
Nottice thatt unlikee a unityy feedbaack system, where H(s)) = 1, th
thee differeence bettween th
he inpu
ut and the
t outp
put. Forr this caase we call
c the
sign
nal at th
he output of thhe summ
ming jun
nction the
t
, Ea(s). Iff
r(t))and c(t)) have th
he same units, we
w can find
f
the steady-sstate errror, e()) = r()
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 49 of 73
7
Tak
ke the nonunity
n
y feedback contrrol systeem show
wn in Fiigure 7.15(b) an
nd form
a unity
u
feeedback system
s
by addiing and subtraccting unnity feed
dback paaths, as
sho
own in Figure
F
7.15(c).
7
This steep requiires thatt input and
a outp
put unitss be the
sam
me. Nex
xt comb
bine H(ss) with the neegative unity
u
feeedback
k, as shown in
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 50 of 73
7
The first step in solving the problem is to convert the system of Figure 7.16
into an equivalent unity feedback system. Using the equivalent forward
transfer function of Figure 7.15(e) along with
G( s ) =
100
1
and H (s) =
s(s + 10)
(s + 5)
Ge (s) =
SO
OLUTIO
ON: Aftter deterrmining
g that th
he systeem is inndeed sttable, on
ne may
imp
pulsivelyy declarre the syystem too be Typ
pe 1. Th
his may
y not be the casee, since
therre is a nonunity
n
y feedback elem
ment, an
nd the plants
p
aactuatin
ng signaal is not
the differennce betw
ween thee input and
a the output.
o
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 52 of 73
7
100(s + 5)
G( s)
=
1 + G(s) H (s) G(s) s3 + 15s 2 50s 400
e() =
1
1
=
= 4
1 + K p 1 (5 / 4)
Page 51 of 73
The negative value for steady-state error implies that the output step is larger
than the input step.
(7.66)
Thus, the system is Type 0, since there are no pure integrations in Eq. (7.66).
The appropriate static error constant is then Kp, whose value is
K p = Lim Ge (s) =
s 0
100 5
5
=
4
400
Page 53 of 73
Page 54 of 73
To continu
ue our discussi
d
on of steady-st
s
tate erro
or for ssystems with no
onunity
feed
dback, let
l us lo
ook at thhe geneeral systtem of Figure
F
77.17, wh
hich hass both a
distturbancee and no
onunity feedbacck.
We will derive a general equation for the steady-state error and then determine
the parameters of the system in order to drive the error to zero for step inputs
and step disturbances.
C(s) = [Ea (s)G1(s) + D(s)]G2 (s) = Ea (s)G1(s)G2 (s) + D(s)G2 (s)
E ( s) =
E ( s ) = R( s ) C ( s )
G1(s)G2 (s)
;
1 + G1(s)G2 (s) H (s)
TD (s) =
ED ( s) = TD ( s) D(s)
e() = 1
Lim{1 + G1( s)G2 ( s) H ( s)} Lim{1 + G1(s)G2 (s) H ( s)}
s 0
s 0
s 0
s 0
Page 58 of 73
=1
and
Lim G2 (s)
s 0
(7.70)
SO
OLUTIO
ON: Usee Eq. (77.72) wiith R(s)) = 1/s, a unit step inp
put, G1(s)
( = 1;
G2(s)
( = 100/[s(s+1
10)], andd H1(s) = 1/(s + 5). Alsso, realiize that ea1() = ea(),
sincce G(s) = 1. Thu
us,
1
s
s
=0
100 1
s 0
1+
s( s + 10) (s + 5)
=0
(7.71)
(7.73)
Now
w use Eq.
E (7.72
2) with R
R(s) = 1//s2, a un
nit ramp input, aand obtaain
1
s
1
s2
=
100 1 2
s 0
1+
s( s + 10) ( s + 5)
e() = Lim
L
Page 59 of 73
(7.74)
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 60 of 73
7
a T
ST :a =
=
T a
a
K
s 2 + as + K
Ks
2
s + as + K
as
=
2
s 2 + as + K
(7.77)
which is, in part, a function of the value of s. For any value of s, however, an
increase in K reduces the sensitivity of the closed-loop transfer function to
changes in the parameter a.
Fraactional cchange in
F / F
n the fun
nction, F
P F
= Lim
= Lim
n the fu
unction, P P 0 P / P P 0 F P
P0 Fractional change in
S F :P = Lim
hich redu
uce to
Wh
P F
F P
e() = Lim
L
s 0
is the ratio
r
of tthe fractional ch
hange in
n the fun
nction to
o the fraactional
chaange in the
t paraameter as
a the frractionall changee of the parameeter apprroaches
zero
o. That is
Page 57 of 73
: Fin
nd the steady-sttate actu
uating siignal foor the sy
ystem off Figure
7.16 for a unit
u step
p input. Repeat for a un
nit ramp input.
s 0
G1( s)G2 ( s)
G2 ( s)
e() = Lim s 1
R( s) Lim s 1 + G ( s)G ( s) H ( s) D( s) (7.69)
s 0 1 + G1( s)G2 ( s) H ( s)
s 0
1
2
S F :P =
Page 56 of 73
Now limiting the discussion to step inputs and step disturbances, where R(s) =
D(s) = 1/s, Eq. (7.69) becomes
Lim G1(s)G2 ( s)
E ( s) = ER ( s) + ED ( s)
G1(s)G2 (s)
G2 (s)
E (s) = 1
R(s) 1 + G (s)G (s) H (s) D(s)
1 + G1(s)G2 (s) H (s)
1
2
C ( s ) = R( s ) E ( s )
Let, TR (s) = 1
E ( s) G1(s)G2 ( s) H ( s) E (s) = R(s) G1( s)G2 (s) H (s) R( s) + G1( s)G2 (s) R( s) + G2 (s) D( s)
[1 + G1(s)G2 (s) H (s)]E (s) = [1 + G1(s)G2 (s) H (s) G1(s)G2 (s)]R(s) G2 (s) D(s)
Pag
ge 55 of 73
7
R( s) E (s) = G1( s)G2 ( s) R(s) G1( s)G2 ( s) H (s) R( s) + G1(s)G2 (s) H (s) E (s) + G2 ( s) D( s)
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
e() =
(7.75
5)
s
w zerro sensittivity (th
with
hat is, ch
hanges in
i the syystem parameteers have
A system
no effect on
o the transfer
t
function
n) is id
deal. Thee greateer the sensitiv
s
vity, the
lesss desira
able the effect oof a parameter changee.
OLUTIO
ON: (i) The
T clossed-loop
p transfeer function is
SO
T ( s) =
K
s 2 + as + K
a
1
=
Kv K
(7.78)
(7.76)
Se:a =
a e
a a
=
=1
e a a / K K 2
(7.79)
Pag
ge 61 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 62 of 73
7
Page 63 of 73
Se:K =
K e
K a
= 1
=
e K a / K K 2
(7.80)
Se:a =
a e
=
e a
a (ab + K )b ab2
K
=
ab
ab + K
(ab + K ) 2
ab + K
(7.82)
OLUTIO
ON: Thee steady--state errror for this
t Typ
pe 0 systtem is
SO
e() =
ab
1
1
b
=
=
1 + K P 1 + ( K / ab
b) ab + K
(7.81)
K e
=
e K
K
K
ab
=
ab (ab + K ) 2 ab + K
ab + K
(7.83)
Equations (7.82) and (7.83) show that the sensitivity to changes in parameter
K and parameter a is less than unity for positive a and b. Thus, feedback in
this case yields reduced sensitivity to variations in both parameters.
e
to changess in paraameter a is
Thee sensitiivity of e()
Page 64 of 73
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 67 of 73
7
Tw
wo meth
hods forr calcullating th
he steaady-statee error will bee coverred: (1)
via inpuut substiitution.
anaalysis viaa final value
v
theeorem and
a (2) analysis
a
5 1 0
A = 0 2 1;
20 10 1
0
B = 0;
1
C = [ 1 1 0]
(7.990)
SO
OLUTIO
ON: Sub
bstitutingg Eqs. (7
7.90) intto (7.89), we ob
btain
s+4
e() = Lim
L sR(s)1
s 0
s3 + 6s 2 + 13s + 20
s 3 + 6s 2 + 12s + 16
e() = Lim
L sR( s)
s 0
s 3 + 6s 2 + 13s + 20
(7.91)
x& = Ax + Br
y = Cx
(7.84a
8 )
(7.84b
8 )
Pag
ge 70 of 73
7
where T(s) is the closed-loop transfer function. Substituting Eq. (7.86) into
(7.85), we obtain
E ( s) = R( s)[1 T (s)]
(7.87)
E ( s) = R( s) 1 C ( sI A) 1 B
(7.88)
s 0
(7.89)
or is
Thee Laplacce transfform of the erro
E ( s ) = R( s ) Y ( s )
(7.85)
(7.86)
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 68 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Page 66 of 73
Pag
ge 65 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 71 of 73
7
Page 69 of 73
9. State how
h
typ
pe of a controll system
m is detterminedd? How
w it affeects the
steady state errror of a system.
: A second
s
o
order
sy
ystem haas unity feedbacck and open-loo
o
op
tran
nsfer funnctions:
(aa) G(s) = s(s500
;
(b) G(s) = s(5000
+ 15)
s + 75)
(ii) Draw
w the blo
ock diaggram for closed--loop system.
(iii) Whatt is the characte
c
eristic eq
quation??
(iiii)Whatt is damp
ping rattio and natural
n
frequenc
f
cy valuees?
(iiv) Calcu
ulate peaak time,, settling
g time, percent
p
o oversshoot forr the sysstem
of
outpu
ut respon
nse wheen exciteed by un
nit step input.
i
(v
v) Sketcch the transient responsse for un
nit step function
f
n.
(v
vi) Calcu
ulate thee static error
e
con
nstants (K
( p, Kv, Ka).
(v
vii) Fin
nd the steeady-staate errorr for the followiing inpu
uts: 70u((t), 70tu((t), and
70t2u(t).
u
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 72 of 73
7
Prepared by
y: Dr. Mo
ohammad Abdul Mannan
M
Pag
ge 73 of 73
7