Professional Documents
Culture Documents
118836
118836
C CONTROL
SY TEMS
DES.
GN
Fe
ngy
u Yang
LANZHOU
UNI
VERSI
TY
PRESS
PREFACE
HYDRAULI C
CONTROL SYSTEM DESI GN fil
ls
t
he educ
ati
on ga
p bet
we en
pra
cti
cal
t
exts
t
hat
empha
size hard
ware
and a
ppic
l ati
ons
at
the
expense o
f t
heoy,
r a
nd t
heoret
ica
l lu
f i
d d
ynamics
text
s
tha
t h
ave l
it
tle
r
ela
ti
onship
to
th
e r
eal
wo r
ld
of f
un c
tio
ning s
yst
ems.
Bl
ack
bu n,
r Lee and Shear
er'
s mag n
ifi
cent b
ook Fl
uid Power Cont
rol g
ives an al
mo s
t
ine
xhaust
ible s
ourc
e of f
undamenta
l i
n f
ormation.
The best
book i
s tha
t of Merr
it
t,
Hydrauli
c
Contr
ol
Systems,
Wile
y,
Ne w
York,
1967.
Th i
s
boo k
of
3 5
0 pages
i
s di
rect
ed more
spec
ifi
cal
ly
to
hydra
uli
c contr
ol
pr
oblems.
Except
f
or
the
last
chapte
rs,
i
t
is
excel
len
t.
Howe ver
,
it
doe
s no
t co
ver
desi
gn p
roblems.
Thi
s t
ext
book'
s mai
n s
our
ce
of
cont
ent
s comes
rom;
f
・An al
y s
is,
Synthesis
and Desig
n of Hydraul
ic Serv
osy
ste
ms an d
Pipeli
nes
, by
Taco J.
Vi
er
sma ,
1980.
.Co
ntrol
Syst
em Pr i
nci
ples
and
De si
gn,勿Erne st
O.
Doebein,
l 1985.
・Mo
de r
n Control
Syste
ms ,
勿 Ri ch
ard C.
Dorf
and
RobetH
r .
Bisho
p,2002.
An
aly s
is,
S ynt
hesi
s and Desig
n of
Hydrauli
c
Serv
osy s
tems
and
Pipel
ines
di
scuses
many
desi
g n
t
opi
cs and ex p
e r
imental
resul
ts.
There a
re var
ious
ca s
e st
udi
es and
specia
l t
opic
s in
Co n
trol
S
yste
m Pr i
nciples
and Design
wh i
ch make t
he s
tudy
become en
joyabl
e.
Mo der
n Contr
ol
Syst
ems
s
uppl
ies update i
nfor
ma ti
on, f
or ex
amp l
e to
pics
in
robust
cont
rol
sys
tems and
in
digi
tal
c
ontrol
s
yst
ems.
Combi
nin
g the
se s
ple
ndi
d con
tent
s,
wit
h a
n e
mphsi
a s
on
th
e a
nal
yti
cal
t
echni
que
a
nd
t
he
d
esi
gni
ng
metho
ds,
is
t
he
mai
n a
im of
HYDRAULIC
CONTROL
SYST
EMS DESIGN.
La
nzh
ou
Un
ive
rsi
ty
o
f Te
chno
log
y
Ma
rch,
2004
CONTENTS
1.
TOOLS
FOR
THE
ANALYSIS
OF
HYDRAULIC
SERVO'
S
1
.1
Contr
ol Syst
em Desig
n Pro
cedur
e s
,
、) 一
1
.2
Int
egra
tor wi
th Propor
i
t onal
Fe
edback
勺
1
.3
Dominant Pr
ope r
tie
s of a
Si
mple Serv
osyst
em
9 10 12 16 17 23
1
.4
Veloc
ity Gai
n K,,
and Valve
Config
urati
on
1
.5
Mul
ti
ple
-Ho
le
Va
lve
1.
6 Tur
bul
ent
Por
t Fl
ow
1
.7
Compr
ess
ibi
it
l y
o
f Oi
l
1.
8 For
ces
and
Cont
inui
ty
1.
9 Perf
ormance
Indi
ces
2.
ANALYSI
S OF
HYDRAULIC
SERVOSYSTEMS
32 33 39 43 科 46 48 51
2.
1 I
ntr
oduct
ion
2.
2 Asymme t
ri
c Motor
s Cont
roll
ed
by
3-
Wa y
Val
ve
2.
3 Load-
Co mpens
ati
on
2.
4 Symme t
ric
Motor
s Cont
roll
ed
by
a
Symme t
ri
c 4-Way
Valve
2.
5 Asymme t
ri
c Moto
rs Co
ntrol
led
by
a
Symmetr
ic 4-
Way Cri
ti
cal
-Ce
nte
r Va
lve
2.
6 Asymme t
ri
c Moto
rs Cont
roll
ed
by
an
Asymmetric
4-
Wa y
Valv
e
2.
7 Pump-Contr
oll
ed Hydr
aulics
2.
8 Fr
om
M at
hem at
ical
M odel
t
o Tr
ansf
er
funct
ion
3.
DYNAMIC
ANALYSI
S AND
SYNTHESIS
55 56 58 60 61 65 69 71 74 75 6 78 80
3.
1 I
ntr
oduct
ion
3
.2
Pro
por
ti
ona
l Co
ntr
ol
3
.3
Dynamic
Load
3
.4
Dynamic
Dead Zone
Due
to
Co
ulo
mb
Fr
ict
ion
3
.5
Damping
Methods
3
.6
Ele
ctr
o-Hyd
rauli
c Se
rvo
val
ve
3.
7 De
rivat
ive
Cont
rol
Modes
3
.8
Bl
ock
Di
agr
am
f
or
Ha
rdwa
re
Comp
one
nts
3.
9 PI
-Cont
rol
3.
10
Fi
rst
Or
d e
r Sy
ste
m in
Se
rie
s
3.
11
Compli
ant St
ruc
tur
es
3.
12
LF-
Con
tro
l o
f Comp
lex
Hyd
rau
lic
Se
rvo'
s
3.
13
Loa
d Ac
tua
tor
s
80 82 料 85
3.
13
Load
Act
uat
ors
3.
14
Cou
lomb Fri
cti
on a
nd St
abi
i
lty
3.
15
Ana
log
Simulat
ion
of
Coulo
mb Fri
ct
ion
3.
16
Des
cri
bin
g Funct
ion
of
Coul
omb Fr
ict
io
n
4.
PERFORMANCE
DIAGRAM
AND
DESI
GN
OF
SERVOMTOR
9
4
4.1 I
ntr
oduct
ion
0了 9
4
4
.2
Rel
at
ion
Be
twe
en
Max.
St
rok
e S=2
夕二a
nd
Ba
ndwi
dt
h
co
b
l
lj l
4
.3
Rel
at
ion
Be
twe
en
Va
lve
Capa
cit
y
Q,.a
n
d Max
imum
Speed
y.
9
勺
4
.4
Rel
at
ion
Be
twe
en
Ca
vit
at
ion
Bound
ay
r a
nd Ma
x.
Acc
ele
rat
ion夕
二
9
户
0
4.
5 Pr
eimi
l n
ary
Pe
rfo
rmanc
e Diag
ram
0产 g
气了 OU
4.
6 Desi
gn
of
Symmetr
ic
Mo t
ors
4.
7 Fi
nal
Pe
rfor
mance
Dia
gra
m
Q产 0
O了 ntj} ,
4.
8 Desi
gn of
Asy
mme tri
c Se
rvomoto
rs
一
.
4.
9 Desi
gn Example
1 扣
1
11} (
.‘ 八口
.
4.
10
In
vers
e Desi
gn Example
工 J
U
.
4.
11
Case
St
udies
and
Special
Topic
s
1
5.
ROBUST
CONTROL
SYSTEM S
2
飞〕
. .J
5.
1
Int
roduc
tion
2
4
, 1
5.
2
Robust
Cont
rol
Syste
ms and
Syst
em Se
nsi
t
ivi
ty
.
勺自 八
}吕
1
. 门
5.
3
Analys
is
of
Ro bust
ness
1
J
.
盆 .
5.
4
Sys
tems wit
h Un c
ert
ain
Paramet
ers
3
2
.
二
5.
5
The Desi
gn
of Robust
Cont
rol
Syst
ems
J
乡
9
.
几 ‘
5.
6
PI
D Cont
rol
ler
s
却
0
. 几
5.
7
The
De
sig
n o
f Ro
bus
t PI
D-Co
ntr
oll
ed
Sy
ste
ms
4
1
‘
勺
. 孟
5.
8
Des
ign
Exampl
e:
Ai
rcr
aft
Aut
opi
lot
4
了0
.. 且
J
5
.9
The
De
sig
n of
a
Space
Te
les
cope
Cont
rol
Sy
ste
m
1
勺
0了
}
且
5
.10
The
De
sig
n of
Robu
st
Bobbi
n Dr
ive
 ̄
11
勺
刁
.且
5
.11
The
Ro
bust
In
tena
r l
Model
Cont
rol
Sy
ste
m
‘) ‘〕 口
内
..昌
,) 门
5
.12
Th
e De
sig
n of
a
n Ul
tr
a-Pr
eci
si
on
Di
amo
nd
Tu
ni
r ng
Ma
chi
ne
‘
了
.1
5.
13
The
Ps
eudo-
Quant
it
ati
ve
Feedbac
k Sys
tem
勺
0了
,.1
5
.14
Robus
t
Co n
trol
Sys
tems
Usi
ng
MATLAB
6
2
..占
5
.15
Seq
uenti
al
De s
ign
Exampl
e:
Di
sk
Dr
ive
Re
ad
Sy
ste
m
6.
DI
GITAL
CONTROL
SYSTEM S
‘U
,了
‘
.且
6.
1 I
ntr
oduc
tion
}
OO
瓜U
J
.1
6
.2
Di
gi
ta
l Co
mput
er
Co
ntr
ol
Sy
ste
m Ap
pli
cat
io
ns
工
C己
‘U
召.1
6.
3 Sa
mpl
ed-
Dat
a Sy
ste
ms
,1
7
‘U
6
.5
Ope
n- Loop
Dig
ita
l Con
tro
l
06
1
工
.1
、J
6
.6
Cl
osed-Loop
Feed
back
Sampl
ed-
Dat
a Sy
ste
ms
1
C6
n龙
.1
6.
7 S
tab
il
it
y Ana
lys
is
i
n t
he
z-
P
lan
e
U
C6
.
O矛
.
I
6.
8 Perf
ormanc e
of
a Sampled-Data,
Second-Order
Syste
m
J
9
,山
.l
6.
9 Clos
ed-Loo p S
ystems wit
h Digit
a l
Comput e
r Compensa
ti
on
l
0了 0 了
 ̄
、) , J
6.
10
Th e
Design of
a Wo r
ktable
Mo tio
n Co n
trol
Syste
m
J
.1
6.
11
Th e
Roo t
Locus
of Digi
tal
Con t
rol
Syste
ms
今白
0
,1
6.
12
Impleme nta
tio
n of Digi
tal
Co nt
rolle
rs
2
0
曰
.三 月
6.
13
Digit
al Control
Syste
ms Usi
ng MATLAB
,‘
八曰}
了
6.
14 S
equentia
l Desi
gn Examp le
:
Di s
k Drive
Re ad S
yst
em
2
月
0
.人
REFERENCES
2
,且
4
APPENDIX
LHYDRAULIC
SYM BOLS
内‘
2
,t
2.
CONTROL
TERM S
AND
CONCEPTS
CHAPTER
1 TOOLS
FOR
THE
ANALYSI
S OF
HYDRAULIC
SERVO'
S
1.
1 CONTROL
SYSTEM
DESI
GN
PROCEDURES
J
ust
s
a f
or an
y oth
er
engi
nee
red pr
oduct
or
s
erv
ice
, th
e o
ver
all
de
sig
n of
fe
edbac
k cont
rol
s
yst
ems
proce
eds i
n s
yst
emat
ic
fas
hion t
hrou
gh
as
equence
of
s
tep
s s
a i
n Fi
g.l
.l.
Rat
her
th
an
di
sc
uss
ing
t
hi
s
ge
ner
al
p
roc
ess
we
wi
l
l
co
nsi
der
t
hos
e
ppe
cts
p
ecu
li
ar
t
o
fe
edb
ack
c
ont
rol
s,
whi
ch
a
ppl
y t
o t
he
p
ort
ion
s o
f t
he
o
ver
all
de
sig
n p
roc
ess
s
o de
sign
ate
d o
n Fi
g.
l
.1.
Due
t
o t
he
c
los
ed-
loo
p
nat
ure of t
he sys
tem confi
g ur
ati
on, t
he input of
e a
ch compon ent
is
afected by al
l t
he othe
r
compon e
n t
s,
r
equir
ing
a desi
gn/anal
ysis
approach t
hat
consi
der
s the ent
ir
e s
yst
e m s
imulta
n e
o us
ly.
Fundamen t
all
y,
the
syst
e m is
d e
scri
bed by diferent
ial
equat
ions ,
t
he most
general
and highl
y
devel
oped the
oy
r bei
ng based
o n o
rdina
ry lin
e ar
equat
ions
wit
h const
ant
coefici
ents,
i
n wh i
ch
ti
me i
s t
he i
ndependent
var
iable.
Be
fore
one c
an wri
te
a n
d solve
the
syst
em difer
e nt
ia
l e
quat
ion s
to
obtai
n s
y s
tem perf
orma nc
e
dat
a, one must f
irs
t concei
ve a s
yst
em configur
a t
ion a
nd ope ra
ti
ng pri
ncipl
e. New sy
s t
ems
gener
ally
uti
li
ze
comb i
nati
ons and/o
r va
riat
ions of
well-
known pri
nc i
ple
s,
thus
a
designe
r mu s
t
be
fa
mi l
ia
r with
thi
s
background.
In addi
ti
on to f
amili
a r
it
y wit
h bsi
a c
pri
ncipl
es,
pra
ctica
l
de s
igner
s
must
al
so
maint
ain
u
p-t
o-da
te
knowl
edge
of
avai
la
ble
har
dwar
e i
n t
hei
r
fi
eld
of
a
ppli
cat
ion.
Trad
e
j
ouna
r ls,
wher
e man
ufa
ctur
ers
pr
esen
t a
dver
ti
sement
s an
d t
echni
cal
pa
pers
exp
lai
nin
g the
lat
est
dev
elo
pme nt
s,
ar
e a
major
so
urc
e of
such
i
nformat
ion.
Al
thou
gh new h
ardwar
e i
s co
n t
inu
all
y
b
ein
g d
eve
lop
ed,
th
ere
a
re
ma
ny
c
lss
a e
s o
f e
qui
pme
nt
t
hat
h
ave
b
een
i
n
wi
de
us
e
fo
r d
eca
des
a
nd
t
hat
s
how n
o si
gns
of
d
isa
ppe
ari
ng.
Si
nce
th
e phi
los
op hy
of
t
his
t
ext
a
tta
ches
gr
eat
i
mporta
nce t
o
t
he b
ene
fit
s o
btai
ned
fr
om a c
onc
urrent
t
rea
tment of
the
oy
r a
nd pr
acti
ce,
h
ardwar
e of s
uch
e
ndur
ing
na
tur
e wi
ll
al
so
be
di
scus
sed
at
s
ome
l
engt
h.
Once
a t
ent
at
ive s
yst
em concept
ha
s been
deve
lop
ed th
rough c
omb i
na t
ion
of
bas
ic
pri
nciple
s
a
nd h
ardware
famil
iar
it
y,
comp onent
s must
be "s
ized" t
o meet the needs
of
the spe
c i
fi
c
a
ppl
icati
on.
For
e
x a
mp l
e,
i
n t
he t
racer
la
the,
use
of
a
ser
voval
ve
a n
d hydrauli
c
act
uat
or
rep
resent
s
t
he de
sig
n e
r's
c
ho i
ce
(ro
f m s
e v
er a
l a
vail
abl
e a
lt
ena
r ti
ves)
of
a par
tic
ular・
f
orm o
f d
riv
e sys
tem;
however
, f
urt
her
deci
sions
mus
t be
made
s
a t
o t
he
s
ize
of
val
ve
and
act
uat
or.
Thes
e ar
e
d
ete
rmined
f
rom cons
ide
rat
ion of
speci
fi
c a
tio
ns s
uch
as
the
des
ire
d maxima of
ac
tuat
or
f o
rce
and
s
peed
re
quir
ed
in
the
machini
ng operat
ion.
Tent
ati
ve
choi
ces
of
th
e fo
rm and "
siz
e" of
all
sy
ste
m
c
omponent
s can be
made i
n this
wa y,
and once t
his
phase
is c
omp le
te,
the fo
rm of t
he
c
omponent
s'
difer
ent
ial
equat
ions (
and th
e numeri
cal
val
ues
of
coef
fici
ent
s i
n the
se equat
ion
s)
b
ecome ava
il
able
, t
he si
multaneous s
et of al
l t
hese
equat
i
ons becoming t
he over
all sy
ste
m
desc
rip
tio
n.
Ac t
ual
ly
, at
t
his
st
age,
one
nu mer
ica
l val
ue,
the
loop
gai
n,
is
le
ft o
pen t
o cho
ice.
Usual
ly,
l
oop
gain
i
s c
omposed
of
sever
al
"bit
s
and
piece
" of
gai
n,
on
e f
or
each
component
,
so
it
s
ad
just
me nt
t
o s
ome
desir
ed
val
ue p
rese
nts
no pr
act
ica
l di
f
ficul
ty.
The
"prop
er"
valu
e f
or
lo
op gai
n
i
s th
e one t
hat
meets
the
sys
tem ac
curacy s
peci
fi
cati
ons whil
e ma
int
aini
ng ade
quate
st
abil
it
y
e h
t y
e
sse
tm
e
nc
olv
a
y s
r
ub
l
t
d
s
i
te
d
a
/
e
d
p
n
o
e
mma
e t r
口0亡心一 J
OdXa
d o
e
d
c
b
一
P 1
份 怪
i
ln
e
t
l
r
s
u
u
si
七
月 曰}
』0
o
wo
de
.
。U.后侣 咎
e
e
的u}
拐a卜
s
胜言 芝乙
乞
e
oc
u
h h
5 奋巴n‘』
况
l
t
r
a
gn
s
瓦
5 沉。P
p
v
u
灿筋 a
P 5品认公
P卜。 0
口 1
之
口 们 J
O
t
』
应
忍 -
与
0
wi
.。轰任各七胜 分 尘一
侣 eJ巴工e 护
。n艺Jd 爹。u
hi
份 0
U
二0 扫云 E.U‘之
e
2
留一
脚生』
.居
i
ng
(
UO云 召 2几 缪1书一
巴 竺憩舒P 卜O
‘
句
0卜Pa中U
t
a二
扫
芍 PU
e
s
节 uol
倒工
u兮苗匕
t
一
‘.uJ;。』
d‘一 ‘。.-
乞:uoo
de
妇
e.三公-
罗占
l
』
O UO-
1月n一
绍囚
b
di
a
仍易 .-
的aP
i
c
r
u
n
a
巴 n尸
韶 助二‘目任 a,书.uJ .一一
.
i
d
o
d
f
v
』
卫
o
节
1
Pr
脚
‘ 的占
h
艺里-
.二V
t
l
ole
g.
.
c
r
l
a
t
o
on
r
Fi
e
p
c
san
b ̄
u二兰 Pazl
l
.一。践5
" ap
r
h
t
乍
乡
1村
e
‘J
O
ce
o
的
老 二.
。一的
全d}
。叻u
lPu
.下J
.Pu
匀
n
r
f
e o
,
巴
,卫月 ‘瞥 一
拐【xa乞 价aua招宝之
f
o
命云 ̄』e。 ̄
^l一
.‘日』
胜〕.。 u.一
七江鱿a妇5.
d
i
t
t
u
a
r
qu
b
l
A "
e
.
nta
i
s
n
e
i
r
g
fe
mar
di
i
nputs
ap pl
ied)
for a
rang
e of
gain val
ues and sel
ect t
he best
. Compute
r-ai
ded-
desi
gn (CAD)
t
echniqu
e s ca
n ma ke
such at
r
il-
a and-e
rror t
ec h
n i
que quit
e cost
efect
ive
. However,
cla
ssi
cl
a
c
on t
rol
theoy
r provide
s more
dire
ct (
and usua
lly one-
step)
me t
h o
ds f
or
solvi
ng t
his
most bsi
a c
s
ystem de s
ign
problem of
gai
n set
tin
g. The two mos t
succes
sful
pra
cti
cl
a me t
hods,
re
f quency
r
esponse an
d roo
t lo
cus,
wil
l
be
explai
ned l
at
e r
i
n thi
s t
ext.
Al
t
hou gh
evey
r f
eed
back s
yst
em must
have i
ts
gain
set,
t
here
is
no g
uarant
ee
that
t
he o
ptimum
g
ain for
the i
nit
il
a choi
ce of
component
s wi l
l
me et
all
syst
em speci
ic
f at
ions.
Wh e
n t
his f
ail
ure
o
cc ur
s,
changes
other
t
han
loop
gain
must b
e tr
ied.
These i
ncl
ude:
a
.
I
mpr
ove
d
co
mpo
nen
t
dy
nami
cs
(
2's
,城s
,
et
c.
)
b.
Add
iti
on
o
f d
yna
mic
c
ompe
nsa
ti
on
e
leme
nts
.
c.
Change
o
f
co
ntr
oll
er
t
ype
(
on-
o氏pro
poti
ronl,
a i
nte
grl,
a et
c.)
d.
Tot
all
y
new
s
yst
em co
nce
pt
(c
han
ge
ro
f m
el
ect
ri
c
dri
ve
to
hydr
auic
l dr
iv
e,
f
or
e
xamp
le)
Fo rt
unat
ely
,
the
re
f q
uency-r
esponse
and r
oot
locus
desi
gn
me t
hods
incl
ude t
echnique s
t
hat
provi
d e
guidance
on t
he
type
and a
mou nt
of
change
ne e
ded t
o h
elp
sat
isf
y a
pati
r cu
lar
unme t s
pecif
icat
ion.
The modi
ie
f d s
yst
e m must
the
n again
have
it
s gai
n se
t,
and
it
i
s ho
p e
d t
hat su
fic
f i
en t
i
te
r a
ti
ons of
this modi
fy set
gai
n/eva
luate p
erformance
p roc
edure l
ead t
o an acc
eptabl
e desig
n. Both t
he
re
f quenc
y-r
esponse
and r
oot
locus
approa
ches wee
r ori
gin
all
y i
mp l
eme n
ted
s
a "ma nua
l" gra
phical
pro c
edur
es,
but
they
are
now availa
ble
s
a s
of
twa re
package
s f
or
rapid
comp ut
er -
ai
de d
d e
sign
o n
ma ny co
mpu t
er s
y s
tems.
In f
act, t
he co
ntr
ol syst
ems ie
f ld
was t
h e
lead
er in develop
ing CAD
procedur
es,
which
are
now spr
eading r
api
dly
to ot
her
ie
f ld
s.
1.
2 I
NTEGRATOR
W ITH
PROPORTIONAL
FEEDBACK
Complex e
lect
ro-hy d
ra u
ic
l se
rvomechanisms in
volve s
o many f
a ce
ts,
eleme nts
and
compo nent
s
and
so
ma ny
dis
ciplines
that
a
syst
emati
c approach
rom
f t
he b
eginning may take many s
pecial
iz
ed
chapt
ers
bef
ore
some underst
andin
g of
t
he esse
nt i
lc
a haracter
is
ti
c s
of
a hydraulic
serv
o i
s ea
r ched.
I
n
an att
empt to discover
dir
ectl
y some i
mp ort
ant c
ontr
ol featu
res, we sha
ll der
ive i
n t
his
i
n t
rod
uc t
ion
a v
ey
r simpl e
mo d
el
of a
hyd r
auic
l copying syst
em, showing tha t
a
hydrauic
l s
ervo
sys
tem behaves
i
lke a n i
nteg
rator
wit
h proport
ionl
a po si
tion
feedback.
To sho w
the
relat
ions
hip
bet
ween mechani
c ld
a e
sign a
n d
cont
rol
behavior
the i
nfluence
of mo t
o r
and v lve
a c
onig
f ur
ati
ons
on
the
loop
gain
i
s discussed
ro
f m the
beginnin
g.
I
n
Fig.
1.
2 a)
vey
r simple
asymme t
ri
c "mo t
or" with a s
o-cll
a ed t
hree-way vlv
a e i
s s
ho wn.
Fig
1.2
a) ca
n be si
mp liie
f d to gi
ve b)
, while c
) gives the nor
ma i
lzed presentati
on in which
the
mechanic
lv
a lv
a e conf i
gurat
ion
is
not
i
ncluded.
Ass
umin
g a
n a
re
a r
at
io
of
2 :
1,
a
t
one
s
id
e o
f
the
pi
st
on
th
e c
ons
ta
n t
supp
ly
pr
ess
ur
e
PS
pr
eva
il
s,
wh
il
e a
t
th
e o
the
r s
id
e o
f t
he
pis
ton
t
he
pre
ssu
re
is
P,=0.5P, when
t
hee
r i
s
no
ext
ena
r l
or
f
ri
ct
io
nll
a
oa
d.
Than
ks
to
Newt
on's
l
aw,
P,=0.5P,
no
t
onl
y at
a
p i
st
on
spe
ed
y=0 b
ut
ls
a o
at
夕=co
nstant.
Ev
ide
ntl
y,
t
he
magn
it
u de
of
s
pee
d夕of t
he
pi
st
on (
wit
hout
l
ode
)
is
d
ete
rmine
d b
y t
he
i
nl
et
o
penin g
a,
and
the
ou
tl
et
op
en i
ng
a2,whi
ch
i
n t
urn
ar
e c
ontr
oll
ed
by
t
he
di
spla
cement
xo
f
t
he
s
pool
valve.
Assumi
ng
an
in
let
low
f Q,
and
an
o
utl
et
QZ,we
fi
nd
y
A=Q,
一Q2=a,
v,一R
2V2
wh
ere
v,
a
nd
v
2 a
e
r t
he
s
pee
d o
f t
he
oi
l t
hr
oug
h a,
a
nd
a2,r
espe
cti
vel
y.
Th
e lo
fws
Q,
an
d
Q2
de
pen
d o
n t
he
ma
gni
t
ude
o
f
th
e c
orr
es
pon
din
g o
pen
ing
s
as
we
ll
a
s
on
t
he
p
res
sur
e
dr
op
a
cros
s
th
ese
o
pen
ing
s.
‘
峨
,
a
.
_
:
,}立
.门.
- .呀. ̄
山
Q
2
Q1 P
5.
b
)
Fi
g.1
.2
Th
ree
-wa
y-v
alv
e-c
ont
rol
l
ed
a
symme
tri
c
mo
tor
As
suming
t
urbu
len
t lo
f w,
t
he (
ave
rage
) s
pee
d vo f
t
he o
il
thr
ough a
n o
peni
ng i
s d
ete
rmin
ed
e
xc l
usiv
ely
by
the
pr
essu
re
drop
acr
oss
t
hat
open
ing.
Si
nce t
he
pres
sure
d
rops
ac
ross
in
let
a
nd
o
utl
et
o
pen
ing
s a
re
e
qua
l(P,一P
,=0.
5P,
a
nd
P
,=0
.5P
, r
es
pec
ti
ve
ly)
,
th
e s
pee
ds
i
n t
he
n
o-l
ode
s
it
uat
i
on
a
re
a
ls
o e
qua
l:
v
,=v
2=v
o,・T
hus
y
A=v
o(a
,一a2
),,
an
d f
or
t
he
v
alv
e
c
onig
f u
rat
io
n o
f
Fi
g.
1
.2
妙一
vo -A
a(
a,一a2
)
击
-
一
Km
=c
ons
ta
nt.
击
Wi
thi
n a
limi
ted
val
ve t nK,
ravel
the v
elo
cit
y ga
i of
o
ur
m
oto
r
re
甲l
y
is“c
on
st
an
t
s
o
K.
,=0
-
)
}/a
x二}
/X,and
t
he
i
npu
t t
o o
utp
ut
t
ran
sfe
r f
unc
ti
on
of
t
he
o
pen
l
oop
s
yst
em
becomes
y_Km
x
s
4
知卜 山 尸
bed
Fi
g.
1
.3
Co
pyi
ng
l
at
he
wh
ere
sis
t
he
d
ife
rent
i
al
o
r La
pla
ce
o
per
ato
r.
F
rom
t
he
c
ont
ro
l
po
int
o
f
vi
ew
t
he
s
yst
em
i
s
an
i
nt
egr
ato
r,
me
ani
ng
t
hat
t
he
o
utp
ut
s
igna
l
y
c
an
be
d
eri
ved
b
y i
nte
gra
ti
ng
t
he
i
npu
t s
igna
l
x.
The
p
hys
ica
l
imp
lic
ati
on
of
t
his
i
s
tha
t
th
e mo
tor
i
s
pu
she
d b
y lo
f w
a
nd
no
t
by
f
orc
e.
Thi
s
is
no
t
a
cont
radi
cti
on
of
Newt
on'
s l
aw,
s
ince
e
xter
nal
, ri
f ct
ional
a
nd
i
ner
ti
al
f
orces
s
houl
d be
s
een
s
a
di
st
urb
anc
e s
ign
als
i
nlue
f nci
ng
t
he
i
npu
t o
f t
he
i
nte
gra
tor
.
Th
is
wi
ll
b
e c
lar
iie
f d
l
at
er.
Gi
vin
g t
he
i
nte
gra
tor
a
p
ropo
rti
ona
l p
osi
t
ion
f
eed
bac
k,
t
he
s
yst
em
des
cri
bed
a
bov
e b
eco
mes
a
r
eal
s
erv
osy
ste
m a
ble
t
o c
ont
rol
d
ispl
ace
ment
s.
By
way
o
f i
ll
ust
rat
ion,
F
ig.
1.
3 s
hows
v
ery
s
che
mat
ica
ll
y
th
e h
ydr
aul
ic
c
ont
rol
sy
ste
m o
f a
c
opy
ing
l
at
he.
He
re
t
he
pi
st
on
i
s
st
at
io
nay,
r t
he
c
yli
nde
r
is
mo
vab
le,
whi
le
t
he
s
poo
l
va
lve
c
an
mo
ve
r
el
at
iv
;,
t
o
t
he
c
yli
nde
r.
The
i
npu
t s
ign
al
ro
f
the
c
ont
rol
s
yst
em
is
g
ive
n by
t
he
mas
ter
, t
he
(
rot
ay)
r
wo
rkpi
ece
t
o b
e c
opi
ed.
In
pe
rfe
ct
c
opy
ing
t
he
t
ool
di
spl
ace
men
t y(
sha
pin
g t
he
wo
rkp
iec
e b
ein
g
ma
nuf
act
ure
d)
e
qua
ls
t
he
ma
ste
r d
ime
nsi
on
r.An
y d
ife
ren
t£=r一y
wi
ll
r
esu
lt
i
n a
c
orr
espo
ndi
ng
s
poo
l v
alv
e d
isp
lac
eme
nt
x=£c
aus
ing
c
yli
nde
r mo
veme
nts
wh
ich
t
y
r t
o ma
ke
x=£=r一Y二Oo
ry=r.
Her
e
th
e f
eedb
ack
ha
sb
een
o
bta
ine
d i
n
a
pu
rel
y
me
cha
nic
al
wa
y.
Th
e c
ompl
et
e c
ros
s-s
li
de
t
oge
the
r wi
t
h t
he
hy
dra
ul
ic
s
erv
osys
tem,
i
s t
rav
eli
ng
wi
th
a
co
nst
ant
s
pee
d a
lon
g t
he
b
ed
o
f t
he
l
at
he
(
per
pend
icul
ar
t
o t
he
p
lan
e o
f d
rawi
ng)
.
Th
e s
yl
t u
s wi
l
l f
oll
ow
t
he
va
ri
abl
e
co
ntou
rs
o
f
th
e ma
st
er
a
nd
c
ont
rol
le
d by
t
he
v
alv
e-t
he
c
yli
nde
r
wi
l
l
fo
ll
ow
a
cco
rdi
ng
t
o
夕=Km
x,
where
x 二£ = r一 Y-
Of
c
our
se
i
t
is
n
ot
n
ece
ssa
y
r t
hat
x=£,
s
a t
her
e c
oul
d b
e a
l
eve
r b
etwe
en
s
tyl
us
a
nd
s
poo
l o
r e
ven
a
n
el
ec
tr
o-
hyd
rau
li
c
se
rvo
val
ve,
ma
kin
g
x=K,
,
E
an
d
”
了-X
Y , KS
Km=兰
-
-
一 = 几s
E
S
Bu
t t
his
d
oes
no
t c
han
ge由e
e
sse
nti
alchar ours
erof
act erv
o,
wh
ich
i
s t
hat
o
f a
n i
nte
gra
tor
wi
t
h
pr
opo
rt
io
nal
p
osi
ti
on
f
eed
bac
k.
1.
3 DOMI
NANT
PROPERTI
ES
OF
A
SI
MPLE
SERVOSYSTEM
Di
sr
ega
rdi
ng
compl
i
cat
ion
s c
onc
eni
r n
g s
tab
ili
ty
and
l
oad,o
ne
cou
ld
cha
rac
ter
ize
a
ny
s
erv
osy
st
em
a
s
an
i
nt
egr
at
or
wi
t
h f
eed
bac
k..
灿Mo
st
i
mpo
rt
a4p
rop
ert
i
es
k
i
f
ave
b
een
s
umma
i
rze
d
i
n
Fi
g.
1
.4.
igu
F r
e 1
.4
a
) s
hows
t
he
op
en
l
oop
t
ran
sfe
r f
unc
tio
n of
t
he
s
erv
o
夕=K,
e
or
s)=Y
H( =K,
,
E E
,
一K,
f
e
dt
The
c
los
ed
l
oop
t
ran
sfe
r f
unc
ti
on
i
s
th
at
of
a
ir
fst
o
rde
r s
yst
em
Kv
-
Y
:二二
H
s
0
K
r
1+H — + t 二 +l
Kv
Kv
is
c
all
ed
t
he
v
elo
cit
y g
ain
o
f t
hei
nt
egr
at
or
.
An
i
mpo
rta
nt
i
mpl
ic
ati
on
i
s
th
e s
o-c
all
ed
v
eloc
it
y
error:i
n or
der
t
o mai
ntai
n ac
onst
ant s
pee
d夕 of
t
he
s
erv
o,
f
und
ame
nta
ll
y
a
c
ont
ro
l
er
or£ha
r s
t
o bep
res
en
t.
Th
e
co
nt
ro
l。 “=yI
K,
i
s
ca
ll
ed
t
he
v
el
oc
it
y
er
ro
r.
Th
e we
ll-
kno
wn
c
los
ed
l
oop
s
tep
es
r p
ons
e a
nd
re
f qu
enc
y es
r p
ons
e a
re
s
how
i
n Fi
gs.
1.
4
c)
a
nd
d)
.
Fi
gs
1
.4
b
),
c
)
and
d)
ma
ke
c
lea
r
th
at
a
l
ar
ge
v
elo
ciyg
t
ai
n
Kv
i
s
de
si
ra
ble
t
o
ob
tai
n
a
.s
mal
l
vel
oci
y
t e
rror
・f
ast
s
tep
r
esp
ons
e
.l
arge
bandwi
dth.
I
n
pr
act
i
ce,
a
d
esi
ra
ble
i
nc
ea
r se
i
n
Kt
i
,
is
i
l
mit
ed
b
y c
ons
id
era
ti
ons
re
gar
di
ng
s
ta
bi
i
lya
t
nd
d
amp
ing
.
Th
e e
fec
t
of
a
gi
ven
l
oad
h
as
b
een
n
egl
ect
ed
s
o f
ar
.
In
t
he
s
impl
es
t
ca
se
o
ne
ma
y s
ta
te
t
hat
a
l
oad
F
(
oppo
sin
g t
he
mo
ti
on
y)g
ive
s
th
e o
pen
l
oop
s
erv
o a
p
ropo
rt
io
nal
o
fse
t
in
t
he
s
pee
d夕.
In
t
he
mo
to
r
of
Fi
g.
1.
2
a)
th
is
c
an
b
e
ex
pl
ai
ne
d
esi
al
y.
A
l
et
f
war
d
fo
rc
e
F
w
il
l
es
rul
t
i
n
an
i
nc
re
se
a d
p
res
sur
e君,c
aus
ing
t
he
p
res
sur
e d
rop
a
cro
ss
t
he
i
nle
t o
pen
ing
t
o d
ecr
ese
a a
nd
t
hat
a
cro
ss
t
he
o
utl
et
o
pen
ing
t
o
in
cre
se
a .
Q1
an
d Qz
wi
l
l
ch
ang
e a
cco
rdi
ngl
y
an
d a
s
pee
d (
i
n t
he
d
ir
ec
ti
on
o
f
t
he
l
oad
b
ein
g a
ppl
i
ed)
wi
ll
r
esu
lt.
Thi
s
ha
s b
een
s
umma
riz
ed
i
n Fi
g.
1
.5
a
),
s
howi
ng
t
hat
a
n e
xtr
a
6
e
rro
r s
ign
al
i
s
ne
ede
d i
n t
he
c
los
ed
l
oop
t
o c
omp
ens
ate
t
he
l
oad.
It
s
co
nse
que
nce
s a
e
r s
hown
i
n b)
a
nd
d)
.
Fr
om
a)
we
de
rive
r.
。。_}
凡1 _
-甲闷气入护一 p .一 卜翩一 -尸一  ̄,, ,
宁
,r Is
I
i
一
产
0
-0代声.
/
1
r
厂
l
彭 00
.
八了一
书
二
_
__
一
{
-一
r
、 ;
么
.J
-
-
--一丫 -  ̄ 朴 r 一  ̄
〔\
{ ! }
\一
一oP
户e
que
nel
`
re
spo
nse{
率盆?州
1.
.
6
一
\
-一一 一 十一 ̄— 一
勿o
p
C
los
ed
l
oop一
:
/
0,as 0.4 肠
,!
卜t落 1
a
k
y
mp
t
oe
s
-
}
l j 卜 |!
一
一
一  ̄份一 山 _ 一
0.
.1
Q.心 口
Fi
g.
1
.4
I
nte
gra
tor
wi
th
p
ropo
rti
ona
l
fe
edba
ck
£
_二+
丝 _二+F
Q十二.
Kv
c
K,
c
c
He
re
c
i
s
t
he
l
oa
d
st
i
fne
ss
a
nd
I
F
t
he
"
to
ta
l"
l
oa
d
in
cl
udi
ng
e
xt
ena
r l
fo
rc
es凡(
pa
y-
lo
ad)
an
d
t
he
(
al
way
s p
res
ent
)
Co
ul
omb
ri
fct
i
on,
s
ee
F
ig.
1.
5
c)
.
I
Fl
c i
s
ca
ll
ed
t
he
l
oa
d-e
rr
or
,
wh
ic
h
mu
st
b
e
ad
ded
t
o
th
e
ve
lo
ci
ty
e
rr
or对K t
o
obt
ai
n
t
he
c
omp
le
te
c
ont
r
ol
e
rr
or
8.
The
e
fec
t
of
t
he
Co
ulomb
f
ri
ct
ion
o
n a
s
erv
osys
tem
wit
h l
oad
s
ti
fne
ss
ci
s
s
hown
i
n
Fi
g.
1
.5
d
).
Co
ulo
mb
f
ri
ct
ion
, r
eve
rsi
ng
i
ts
s
ign
wi
th
t
he
s
pee
d夕,i
nt
rod
uce
s a
d
ead
z
one
,
wh
ich
p
oses
a
s
eri
ous
t
hre
at
t
o t
he
a
ccur
acy
o
f a
s
erv
osy
ste
m.
Unp
red
ict
abl
e v
ari
at
ions
i
n Co
ulo
mb
ri
fct
ion
wi
ll
c
aus
e c
orr
espo
ndi
ng
v
ari
ati
ons
i
n t
he
c
onro
t l
e
rro
r a
nd
c
oul
d h
ave
f
ata
l c
ons
equ
enc
es
f
or
t
he
"
smo
oth
nes
s"
o
f o
per
ati
on,
p
art
i
cul
arl
y wi
th
r
ega
rd
t
o t
he
s
pee
d夕a
nd
t
he
a
cce
ler
ati
on夕o
f t
he
s
erv
osy
ste
m.
r
o
a
d
c
F二F一F
e 十
;
EI
i
S
-
,
1
.干
一
)
a
t
f
ea过-
Ott
e
I
d
)
Fi
g.
1
.5
I
nf
lue
nce
o
f e
xte
na
r land
f
rict
ional
l
oad
To
mea
sur
e t
he
p
erf
orma
nce
of
t
he
s
erv
osys
tem
one
i
s i
ncl
ine
d i
n ny
ma cas
es
t
o mea
sur
e t
he
r
esu
lt
,
t
he
ou
tpu
t
si
gna
l
y.
But
,
as
sumi
ng
t
he
i
npu
t
si
gna
l
r
t
o
be
pe
rf
ecttsometimes
,
i make
mor
e
s
ens
e t
o me
su
a r
e t
he
e
rro
r s
igna
lE
d
uri
ng
o
per
ati
on
o
r t
est
ing.
Thus
ma
chi
ne
t
ool
co
p外ng
s
yst
ems
Fi
g.1
.6
Me
su
a r
ing
s
et-
up
s
hou
ld
n
ot
b
e t
es
ted
p
rima
ri
ly
b
y me
asu
rin
g t
he
wo
rkpi
ece
s t
hat
h
ave
b
een
ma
nuf
act
ure
d,
b
ut
by
me
asu
rin
g t
he
e
rro
r s
ign
al
E
di
rec
tl
y,
s
a i
s s
hown
i
n Fi
g.
1
.6,
whe
re
a
mi
crome
ter
(
att
ach
b
y me
ans
o
f a
ma
gne
ti
c
cl
amp
) i
ndi
cat
es
t
he
s
pool
d
ispl
ace
men
t wi
t
h r
esp
ect
t
o i
ts
ho
usi
ng.
I
n t
he
c
los
ed
l
oop
s
it
uat
ion
s
eve
rl
a p
oint
s o
f t
he
c
har
act
eri
st
ic
of
Fi
g.1
.5
d)
c
oul
d b
e me
su
a r
ed
e
si
a l
y,
d
ete
rmi
ni
ng
t
he
v
elo
cit
y
ga
in
Kv
an
d l
oa
d s
t
ifn
ess
c,
sw
a
ell
st
a
h
e d
ead
z
one
.
Th
e v
lu
a e
o
f
su
ch
si
mp l
e tes
ts, i
n which
el
r evant
syst
em para
me te
rs ar
e measued
r (which i
s the ai
m.),
can be
il
lus
tra
ted by t
he f
act
th
at s
ome fi
rms are us
ing mainl
y t
he mesur
a eme nt
of t
he dead
zone
s
a a
cr
ite
rion i
n sel
ect
ing copying machines to be inst
all
ed i
n t
her
e worksho ps. Wit
hout any
pepa
r rat
ion,
the
mesu
a re
me n
t o
f t
he
dea d
zone o
f a
copyin
g s
ystem c
o u
ld一a cco rdi
ng t
o Fi
g.
1.
6一
be c
arie
r d out
in
a fe
w mi nute
s.
It
is
t
h er
efore
surpri
si
ng t
o obse
rve a
t machine t
ool
exhi
bit
ions
th
at
neit
her s
le
a s
manager
s nor
ma ch
ine t
ool
operator
s u
nders
tand
the
ess
e n
tia
ls of t
he e
xperi
me nt
.
1
.4
VELOCI
TY
GAI
N Kv
AND
VALVE
CONFI
GURATI
ON
I
n t
he
pr
evi
ous
s
ect
ions
we
hav
e f
ound
t
hat
Kv gain
a
y 0
v.
a(
a,
.
-a2
)=K
_
_
-
vo
.
}a
(a
,
-a
2.
aE A
aE
3
A
ax
Th
us
t
he
v
eloc
it
y Kvi
s p
rop
ort
ion
al
t
o t
he
s
lope
o
f t
he c
harac
teis
r t
ic,
i
n whic
h t
he
-di
fer
ence
bet
ween i
nlet
and
out
letopen
ing (
a,一a2)is
pl
ot
te
d s
a a
funct
ion
of
the v
a l
ve
di
spl
ace
ment
x.That cha
ract
eri
st
icdependscompl et
ely on
t
he vlv
a e config
urati
on a
nd i
ts
di
mensi
ons
.
In
Fi
g.
1
.7-
some e
xamp l
esar
e gi
ven.
Fi
g 1
.7 a)
s
hows
anopen-
center
3-way
va
lve
a
nd
b
) a
cr
iti
cal
-ce
n t
er
valv
e, both
wit
h cy
lin
d r
ical
por
t
openi
ngs.
The c
orre
匆on di
ng c
har
act
eri
st
ics
A
re
giv
en al
so.
It
a
ppears
that
t
he
open-
cente
r val
ve
has
wit
hin
t
he under
lap-
regi
on
a
gai
n t
wice
as
o
uts
ide,
whi
le
the
cri
ti
cal
- ce
nter
v
alve
has
a
const
ant
gai
n.
Fu
rt
heri
in
ore
, i
t
shoul
d be no
ted
th
at
t
he open-c
enter
v
alv
e i
mp l
i
es an
oi
l
suppl
y (
rom
f t
he
pump)
t
hat
i
s pa
rti
cul
ar
i
nde
pen
dent
of
t
he
s
poo
l p
osi
ti
on
i
n t
he
u
nde
rla
p-r
egi
on.
it
W h
a
cr
it
ica
l-c
ent
er
v
alv
e t
her
e i
s
no
o
il
c
bns
um沁询 a
t x
=0
.
Th
is
a
bse
nce
o
f "
lea
kag
e f
low"
(
at
l
eas
t
th
eor
eti
cal
ly)
c
oul
d be
a
n a
dva
nta
ge,
e
spe
cia
ll
y i
n s
yst
ems
(
suc
h a
s a
irc
rat)
f wi
t
h al
arg
e n
umb
er
o
f mo
tor
s
conne
ct
t
o one
pump
s
et.
A
!
泊r
 ̄
.
-
-
.
,
卜
卜
,
,一
毖1 口
11、
』、
产 口
伪乙
JJ
性.I P
s.
崇结凳 !
}一 「
I S 4)
Fi
g.
1
.7
Op
en-
cen
ter
a
nd
c
ri
少al
-c
ent
er
3
-wa
y v
alv
e
1.
5 M ULTIPLE-
HOLE
VALVE
I
n F
ig.
1.
8
a
v
ery
s
pec
ial
3
-wa
y v
alv
e i
s
sho
wn,
a
"
mul
t
ipl
e-
hol
e"
v
alv
e,
whe
re
t
he
i
nl
et
a,
and
t
he
ou
tl
et
a2
c
ons
is
t o
f a
l
ar
ge
nu
mbe
r of
c
ir
cul
ar
h
ole
s a
rr
ang
ed
i
none,
t
wo
or
more
rows.
Di
sp
lac
ing
t
he
s
poo
l,
e
it
he
r
th
e
in
let
ho
les(
a,)o
r
th
e o
utl
et
h
ole
s(a2
)wi
l
l倪o
pen
ed.
砒ei
A}
"
und
erl
ap"
U
is
p
rop
erl
y c
hos
en,
s
ee
Fi
g.1
.8
b)
, t
he
c
har
act
er
is
ti
c o
f a,
-a2
v
ers
us、
x
(
det
er
min
ing
K,
,.
)is
r
at
he
r s
tr
ai
ght
,
se
e c
).
Di
sc
los
in
g o
ne
s
in
gle
h
ole
wi
t
(n
d
iame
ter
d o
ver
,
a
1
0
di
st
anc
e h,
one
in
f ds
a
n o
peni
ng
口1 口2
上
粼鑫赵土上窦羞么 二
、 蠢卜.‘面画‘ 肠白砰测 一
时
)
FS
井
a
)
Fi
g.
1
.8
Mul
ti
ple
-hol
e
va
lve
2
h.__2
h.)
不-万
尤一
ar
ccos
(1一了 )一2(
卜 --),
I一
万一下
d
d
d Vd
d“
a
-
一
4
兀
a
s
a
h
_
2了
一
d
h-
k
h2_
2d,压二
Vd
可.
d`
u
nde
rla
p U
a
cco
rdi
ng
t
o
Fi
g.
1
.8
(
b)
we
f
nd
i i
nsi
de
t
he
u
nde
rla
p-r
egi
on
a
t
a
(a
,一a
2)
ax
a
(
a
,
a-
x
a
2)
一rd
-
U
a2
= Maxi
mum,
Ou
tsi
de
t
he
u
nde
rla
p r
egi
on
we
f
in
a
(
a
,
a
d
t
h
-
x
a
2)
-
e
2
2
ma
x
U
d
-
i
4
U
d)
2
mu
m g
ain
a
t
=M i
nimum .
d 一2
a
(a,
一a2
)
U
_a
a,=d
.
X
一
一
一
ax
ax
T
he
p
at
te
no
r
f
a(
a,一a
2
Ya
x i
s
r
ep
et
it
i
ve,
si
a
s
s
how
n i
n
Fi
g.
1
.
8 c
)
an
d
d)
.
Tr
yi
ng
t
o
ma
ke
v
ar
ia
ti
on
s
in
a(
a,一a
,
Va
x
as
s
mal
l
a
s
po
ss
ib
l
e,
we
e
qua
i
lz
e
th
e
ma
xi
ma
i
ns
i
de
a
nd
o
ut
si
de
t
he
u
nde
rla
p r
egi
on,
re
sul
ti
ng
i
n
U
1立_
0.
06
7.
d 2 4
The maxi
mum and mi
nim um val
ues o
f a
(a,
一a.
,l
a
x t
hen
、
b
eco
me
da
nd
0
.68
1 d
1 ‘ 1 1
r
espe
cti
vel
y.
Le
aka
ge
f
low
i
n a
mu
lti
ple
-ho
le
v
alv
e d
epe
nds
o
n t
he
nu
mbe
r o
f r
ows
wi
t
h h
ole
s.
I
n t
he
n
eut
ral
v
al
ve
p
osi
t
ion(
x =
O),
wi
t
h.
U/
d=0
.06
7,t
he
i
nl
et
o
pen
ing
i
s,
in
p
erc
ent
ag
e
of
t
he
ma
xi
mum
o
pen
ing
, 3
% 1
工%a
nd
1
%
wh
enthere
a
re,
r
espec
tivel
y,
1,
2
a
nd
3
rows
wi
th
hol
es,
Thus
t
he
2
c
hoi
ce
o
f n
umb
er
of
rows
d
epe
nds
o
n t
he
a
mou
nt
o
f l
eak
age
lo
f w
l
eaka
ge
low
f be
ing
a
ll
owe
d.
1.
6 TURBULENT
PORT
FLOW
To
a
nal
yze
hyd
rauli
c
s e
rvosys
tems
. Oi
l
lo
f w t
hrough va
ria
ble
p
oto
r
pen
ing
s(
v
alv
es)
,
. Co
mpr
ess
ibi
li
ty
o
f o
il,
an
d
. For
ces
a
nd
c
ont
inui
ty
mus
t be
t
aken
i
nto
account.
12
Th
e s
pee
d v
o
f oi
l
th
roug
h a
n
arr
ow
o
rif
ice
i
s
de
ter
min
ed
b
y
v
_C,
_2
t}
P. (
1.1
)
“1
1 P
I
n
191
7
Von
Mi
ses f
ound
a
c
ons
tan
t
"
dis
cha
rge
"
c
oefi
ci
entCd.The
ore
ti
cal
ly
,
C
d=二
/伽+2
)
=
0
.61
1.
Thi
s
ty
pe
of
lo
f w
i
s
ca
ll
ed
t
urbu
lent
b
eca
uset
he
p
res
sur
e dr
op
OP
‘
--
2
acr
oss
t
he
openi
ng
i
s propornonai
t
o尸v
The
low
f Q
ca
n no
w be
f
oun
d a
s t
he
p
rod
uct
o
f po
rt
o
pen
ing
a
an
d t
he
s
pee
d v。
。
=。
一。
hC
d
2
0
Pp
・ (
1.2)
e h
Her (
the
va
riabl
e)
is
t
he s
mall
est
d
imens
ion
of t
he
(r
ectang
ula
r)
ori
fi
ce;
s
ee
Fig.
1.
9 a
),
a
nd
b
t co
he ns
tant
wi
dth.
Tur
bul
ence
is
e
nsur
ed
onl
y at
"
lar
ge
enough"
Reynol
ds
numbe
rs.
万
v
/ P
4
A
r N-
P
7 1
717又
P h入 、 _
._ _ r
.,.
一
节 ‘
一“‘
V半
毋
0.
公 )
I
V
「一 不一一 尸一 一 一 ’
I
l'! /i
I
句poi
het
ica I }
a5
l
r 匕, ! L--
I
/
}
奋
/
{
a
dge
s / 一
0.
4 p门
} I 夕
}
}
I
」
a3 : -一 洲
一7 一’I
} /・
l : }
!
,/ } I
}
a1 t -, .节
I
困 {
/ ! I
}
门 .
’ 一
}
:
,
} }
!
/ 一
一
!
}
}
/
l
{
}
一
0.
1 / 」 !
一
} ! I
{
{ } }
」I !
} I
V
lO 场
,、
俪 vi
ie
DJ
Fi
g.
1
.9
Di
s gec
char oefi
ci
ent
Cd
。__2
八c= —
v
h
2
= —
pv
h
一 . (
1.3)
V
P
He
reV
i
s
th
e k
ine
mat
ic
v
isc
osi
ty
, ,
t
l t
he
d
yna
mic
v
is
cos
it
y
an
d p
t
he
s
pec
if
ic
ma
ss.
E
xpe
ri
men
ts
p
rov
ed
t
hat
at
v
ery
s
har
p
ed
ges
i
n
na
rr
ow
o
if
r i
ces
t
he
c
ri
ti
cal
va
lu
e
Re
,,
i
s
s
a l
ow
s2
a
0,
whi
l
e s
li
ght
ly
r
oun
ded
o
ff
e
dge
s i
ncr
eas
ed
Re,
,‘8
0 o
r
hi
ghe
r.
Th
us
a
t
ve
ry
s
har
p e
dge
s
we
ma
y ss
aume
Cd
t
o
be
c
ons
ta
nt
a
t
Re
>
Re
cr“2
0.
At
s
mal
le
r v
alue
s of
Re
a
b
oun
day
r l
aye
r wi
ll
bu
il
d u
p.
I
n 1
954
Wue
st
d
eiv
r e
d,
f
or
e
xtr
eme
l
ow
va
lues
of
Re,
C才=S
Re. (
1.4)
He
re
S
i
s
a
c
ons
ta
nt,
de
pen
ding
o
n g
eome
tr
y.
Co
mbi
ni
ng
Eq
ns.
(I
.
l)
,(1
.3)
a
nd
(
1.
4)尹e
lds
,
一。J
,2
0p
一。R
e
1
li
p一
V
:
、2
pv
h
2
AP
P
V P
V P
P
r
esul
ti
ng
i
n
v= —4
82h__
AP
p
and
。二
。,一
。,,
=44
5
P
2
bh
2
AP.
(
1.5)
Ex
tra
pol
at
ing
l
ami
nar
low
f t
o l
arg
er
v
alu
es
o
f Re
a
nd
a
ssu
min
g a
s
udd
en
t
ran
sit
ion
rom
f l
ami
nar
low
f t
o t
urbul
ent
low,
f we
f
ind
S Cd
t
ur
b
0.
611 0.
136.
:二 一 一 二昌 一 牛之
(
1.6)
R
e
,万 2
0
Th
is
lo
f w
mod
el
o
f l
ami
nar
a
nd
t
urbu
lent
lo
f w
i
s
pr
ese
nte
d i
n Fi
g.
1
.9
b
) c
an
be
us
ed
s
a t
he
ba
sis
f
or
c
alc
ula
tion
of
lows
f .
App
ie
l d
t
o a
c
rit
ica
l-
cen
ter
v
alv
e of
Fi
g.
1
.10
a
) a
l
ami
nar
low
f a
t s
mal
l
po
rt
o
pen
ings
wi
ll
o
ccu
r
wi
t
h a
p
ara
bol
i
c r
el
at
io
n b
etwe
en
Q
a
nd
x
a
cco
rdi
ng
t
o
Eg
n.(
1.
5)
.
At
x二hT
th
e c
ri
ti
cal
Re
yno
lds
n
umb
er
Re,
,
wi
ll
b
e p
ss
a e
d,
r
esu
lt
in
g i
n a
c
ons
tan
t
low
f g
ain
a
cco
rdi
ng
t
o
Eg
n.(
1.2
).
Fo
tu
r n
ate
ly,
t
he
p
oor
lo
f w
g
ain
a
rou
nd
x=0
i
s
re
str
ict
ed
t
o a
s
mal
l
ar
ea.
Equa
ti
ons
(
1.2
)一(
1.6)
如el
d
V
Rec
r
hT= (
1.7)
2
气rb
担Op
Ip
L
et
ti
ng
,
fo
r e
xa
mpl
e,、
,=3
0x1
06m2
- /S,
Re
cr=2
0,
Cd=0
.61
1,
AP=5
0x1
05N/
m2
a
nd
p=9
00k
g /
m
3
,we
f
i
nd
h
T=4
.66
x1护m=4
.66
,
um.
Th
e l
ow
f
low
ga
in
a
rea
c
oul
d b
e r
est
ri
ct
ed
i
n p
ropo
rti
on
t
o t
he
wh
ole
a
rea
b
y o
rif
ice
s wi
th
a
s
mal
l
er
wi
dt
h
b,
as
s
hown
i
n
Fi
g.
1.
1t
)
b)
.
He
re
h
T
is
t
he
s
ame
a
s
be
for
e,
b
ut
o
win
g t
o
th
e s
mal
l
wi
dt
h b<MI
i
t
ta
kes
pl
ace
a
t
a
l
owe
r v
alu
e o
f t
he
low.
f Th
us
a
s
a
p
erc
ent
age
o
f t
he
ma
ximum
fo
l w
o
r v
alve
s
tr
oke
the
l
ow g
ain
ar
ea
i
s
re
str
ict
ed
con
sid
era
bly
.
I
n
add
iti
on
it
i
s p
ossi
ble
t
o i
ntr
odu
ce
av er
y s
mall
und
erl
ap
U
in
ou
r "
cri
ti
cal
-ce
nte
r"
v
alv
e.
As
su
min
g
in
le
t
an
d
ou
tl
et
op
eni
ng
s a
,=b
h,=b
(U+x
j a
2=b
h2=b
(U一x
),
a
nd
p
re
ss
ur
e
14
t
nrhul
eat
fl
ow
\
‘
‘ .‘. .
r
ect
angu
lar
hol
es
Q,一Q
.
1
句 1
paw
Fi
g.
1
.10
Tr
ans
it
ion
rom
f l
ami
nar
t
o
tu
rbul
ent
f
low
d
rop
s△只 =△凡 =AP,
we
in
fd i
n
th
e u
nde
rl
ap-
reg
ion
Q
:一
Q2一
432
b(
h
i
2
A
P,
一。
2
2
P2
A)-
16
4
52
-
b
APU
x
So
u
井
)
刁
一82.一__ = 16- bAP U = Const
ant
.
a
公,一Q2
、〕
户、f
“. 卜.
ax
井
Re
qui
ri
ng
t
hi
s
lo
fw ga
in
t
o
be
e
qua
l
to
t
he
t
ur
bul
ent
lo
f w
ga
in
a
t
x>h,
.
acc
ord
ing
t
o
Ec
l
n.
(
1.
2)
,
e
『
F
15
fi
nd
v
Re__ h}
U 二— 详 兰二二 =一
.竺‘. (1
.9)
8
Cdt
u
rb
V
2
A /
P
4
Th
us,
i
n t
he
e
xamp
le
a
lre
ady
g
ive
n,
e
ven
a
t a
n e
xtr
eme
ly
s
mal
l
un
der
lap
U=1
.16
ynt
,a
v
ery
go
od
c
ons
tant
lo
f w
gai
n i
s o
bta
ine
d t
hro
ugh
out
t
he
wh
ole
wor
king
r
ang
e o
f t
he
v
alv
e.
Thi
s i
s
c
onf
ir
med
b
y ma
nuf
act
ure
r's
p
rac
tic
e,
wh
ere
a
n u
nde
rla
p o
f a
f
ew
mi
cro
ns
i
s u
sua
l t
o p
rov
ide
a
c
ons
tan
t lo
f w
gai
n.
Th
is
i
n s
pit
e o
f t
he
f
act
t
hat
t
he
a
ssu
mpt
ion
s l
ead
ing
t
o Eg
n.(
1.8
) a
re
n
ot
f
ulf
il
led
a
utoma
tica
lly.
Co
ncl
udi
ng
t
his
s
ect
ion,
we
st
at
e t
hat
i
t
is
pe
rmi
t
ted
i
n n
orma
l
pr
act
ic
e t
o us
e t
he
t
urb
ule
nt
lo
f w
e
qua
ti
on
(
1.2
) r
ega
rdl
ess
o
f t
he
Re
yno
lds
n
umbe
r i
f we
b
ear
i
n mi
nd
t
hat
a
t v
ery
s
mal
l
po
rt
openi
ngs
cons
ider
abl
e de
via
tions
i
n t
he
ca
lcul
ate
d low
f gai
n may
oc
cur
.
1.
7 COMPRESSIBI
LITY
OF
OIL
I
t i
s quit
e s
urpri
si
ng t
o outsi
ders
th
at
oil
compress
ibi
li
ty i
nlu
f ence
s t
he dy
namics
of hy
draul
ic
se
rvo'
s s
ignif
ica
ntl
y.
Hydraulic
lui
f ds s
uch a
s oil
,
even th
ough f
ar l
ess
compres
sib
le t
han gs,
a
when comp r
esse
d in
ac
yin
l der
compart
men t
act
l
ike
a spri
ng a
nd i
ntr
oduce t
her
efor
e a
second
or
der mas s
-sp
ring
sys
tem wh ose
natur
al
re
f quency i
l
mits t
he ban
dwidth
of
any hyd
rauic
l s
ervo
ab
ruptl
y .
The compres
si
bilit
y of a hyd
rauic
l lu
f i
d, t
rapped i
n a vol
ume v
and
ap res
sure
P
is
def
ined
by
AV △尸
(1
.10)
v E
E
i
s
ca
ll
ed
t
he
b
ulk
mo
dul
us,
of
t
en
i
ndi
ca
te
d b
y刀.
The
b
ulk
mo
dul
us
h
st
a
he
s
ame
d
ime
nsi
on
(
NI
m
z
)sY
a
oun
g'
s
modu
lu
s
of
e
la
st
ic
iyf
t
o
r
met
al
s.
For
r
ea
son
s
of
a
nal
ogy
i
t
ha
s
be
en
gi
ve
n t
he
s
ame
s
ymb
ol
E.Fo
r mi
ne
ra
l
oi
ls
,£ma y
be
s
a hi
gh
as
15
,0
00
bar
(1
5
X 1
0¥
N
/
m2).I
ts
e
fe ct
iv
e val
ue
in
pract
ic
e,
however,
i
s
muc h
lower,
main l
y du
e to
a s
ma l
l perc
entag
e of
ent
rai
ned
a
ir,
which
in
fac
t c
annot
be av
oided.
Howe v
e r
,
bad practi
ce
in r
eservo
irs (r
etun
r l
ine
splsh
a in
g,
s
ma l
l s
urfa
ce ar
ea,
cl
ogged su
ction
fi
lt
ers,
etc
. e
tc.)
s
a well
s
a bl
ind holes
and pocke
ts i
n t
he
h
yd r
aui
lc sys
tem it
sel
f coul
d reduce
E to
less t
han 1
0% of i
ts t
heo re
tic
al val
ue. I
n good
e
ng i
neer
ing
pra
cti
ce,
val
ues
bet
we en
6,
000 an
d 12,
00 0
b ars
can
be
expected.
El
st
a ici
ty
of
stee
l c
yli
nders
and t
ubi
ng has
some efe ct
on
the
efecti
ve E, but
could
normal
ly
b
e negl
ecte
d.
Howe ve
r,
t
he e
lst
a ici
ty
of
(h
igh
pressure)
hoses
coul
d have
a disst
a r
ous e
fect
on t
he
e
fec
tiv
e bul
k modul
us
E.
Th
e o
il
s
pri
ng
st
ifne
ss
co
of
v
a r
iou
s s
ervo
syst
ems d
e p
ict
ed
in
Fi
g.
1
.11
hs
a be
en d
eri
ved
o
n t
he
b
si
a s
of
a
det
er
mined
bul
k modul
us
E.I n
Fig.
1.
1 1
a),
i
n t
er
ms of
Egn.
(1.
10)
, AV=y A and
V二AL+叭,wh e r
e AL is
the e
ff
ecti
ve v
olume of
the
cyl
in
der co
mpart
ment a
nd玫 the
i
nefe
cti
ve
o
ne.
The
p
ress
urei
ncr
eas
e AP
i
s
cau
sed
by
t
he
loa
d F=AAP.Th
us
we
f
ind
Y F
A F
A2
E
a
nd
co二一一 二二 — . (1
.11
)
L十从 AF
,
A Y AL
+VL
16
The
min
us
si
gn i
s
du
e t
o c
onvent
i
onal
(
opposi
ng)
p
osi
ti
ve
di
re
cti
ons
of
y
a
nd
F.In
Fi
g.
1.
11
b)
an
d c
)
the
v
alu
es
of
co
hav
e be
en
fou
nd
by add
ing
up
t
he
oi
l
sp
rin
g s
ti
fnes
s
on
bo
th
s
ide
s o
f t
he
p
ist
on.
Th
e i
l
nea
r
mo
tor
s
a)
,
b)
a
nd
c
)
al
l
sho
w s
ti
fne
ss
c
. he
avi
ly
d
epe
ndi
ng
o
n t
he
pi
st
on
po
sit
io
n.
^
,、l
o
wes
t
po
ss
i
bl
e
na
t
ura
l
r
fe
qu
en
cy
0
c
)
0司动M
i
s
mo
st
i
mp
or
ta
nt
,
co
rmh
as
b
ee
n
c
alcu
lat
ed
for
ev
ey
r moto
r i
n F
ig.
1.11.
A
spec
ial
c
ase
is
th
e r
ota
y
r mot
or of
Fi
g.
1
.11
d).
I
ts
r
ota
tiona
l o
il
s
pri
ng
st
if
fness
(
in
Nm/
rad)
c
ould
be
deri
ved
by a
nal
ogy wit
h t
he s
ymme t
ri
c in
l e
ar
motor,
r
epl
aci
ng
the
p
isto
n ar
ea
A
by
th
e
mot
or d
ispl
ace
ment
Vr
per
rad
ian
and
by re
placi
ng
the s
tr
oke
S b
y 7
r r
adi
ans
(
hal
f a
r
evol
ut
ion).
Of
cou
rse
c
o
ca
n b
e de
ri
ved
di
re
ctl
y
also
by
l
et
ti
ng:
7
r__ V, _ __. S .
V,
ComDart
ment
vol
umes:
V ,二 一 V + 一
。 i
nst
ead
一of
‘V ,二 一 A + 一 .
一“
甲一“
’一’
“’一
‘“
’‘
一‘”1
,
2一2’
r’2, 一 下 1
,
2一2“’2’
F
lows
:
Q1
一Q2
=
t
V
Vr
,
in
st
ead
o
f Q,
一Q2=y
A
L
oa
d:
T二V
r・(
PI
一P2
),
in
st
ea
d
of
F=A・
(
P,一P
2)
Re
le
van
t
in
ma
ny
a
ppl
i
cat
i
ons
i
s
th
e f
act
t
hat
t
he
"
ine
fec
ti
ve"
l
i
ne
v
olu
me
0.
5VL
c
ann
ot
b
e
n
egl
ec
te
d wi
t
h r
es
pec
t
to
t
he
e
fec
ti
ve
v
olu
me
t
o
0.
5
7
rvr
.Th
us
o
ne
h
st
a
o
be
e
xt
re
mel
y
ca
ref
ul
i
n
c
alc
ula
ti
ng
o
r e
sti
mat
i
ng
co
f
or
r
ota
y
r mo
tor
s.
A s
econ
d c
ompli
cat
ion
in
rot
ay
r motor
s wi
th
t
hei
r
us
ual
tra
nsmiss
ion
is
t
he
iner
ti
a "
fel
t
by
the
mo
tor
".
The
i
ner
ti
a
of
gear
t
rai
ns
et
c.
re
duced
to
t
he
mot
or
a
xis
is
s
umma r
ize
d a
lso
i
n Fi
g.
1
.11
d
).
Us
ual
ly
t
he
"
addi
t
iona
l
in
ert
ia
" i
s
5-
10
t
ime
s l
ar
ger
t
han
t
he
i
ner
ti
a
JM
of
t
he
r
ota
ym
r o
tor
i
ts
el
f.
I
t
is
t
her
efo
re
a
lmo
st
a
n un
der
sta
teme
nt
t
o s
ay
t
hat
c
are
ful
d
esi
gn
of
g
ear
t
rai
ns,
et
c.
i
s
wo
rthwhi
l
e
i
n
o
rd
er
to
ob
t
ai
n
a
cc
ep
ta
bl
e
v
al
ue
s
of
th
e
na
t
ure
r
fe
qu
en
cy
c
o
o司瓦7
天万.
1.
8 FORCES
AND
CONTINUITY
The
c
omp
lex
s
it
uat
ion
r
ega
rdi
ng
f
orc
es
a
nd
c
ont
inu
it
y wi
lb
e t
rea
ted
he
re
f
or
t
he
mos
t
ge
ner
al
c
ase
of
a
n a
symme
tr
ic
l
ine
ar
mot
or(A,
#A2)a
nd
an
as
ymme
tr
ic
4-
way
val
ve
(
a,
#a3
,a2
#
a4)
,se
e Fi
g.1
.12
. Ot
her
c
ase
s a
re
f
ar
s
imp
ler
a
nd
t
hei
r
re
sul
ts
c
oul
d b
e f
oun
d
ro
f m t
he
ge
nera
l c
se
a qui
t
e e
asi
ly.
Th
e mos
t
import
ant
for
ces
ar
e l
is
ted
b
elo
w a
cco
rdi
ng
t
o o
rigi
n
an
d b
rie
ly
f di
scus
sed.
(
1)
Oi
l
pes
rsur
es
P
a
nd凡c
aus
ing
o
ppo
sin
g f
or
ces
,只A,
a
nd凡A2
(
se
e
Fi
g.
1
.1
2)
(
2)
E
xte
rna
l
fo
rce
F
e
co
min
g ro
fm o
uts
i
de
t
he
h
ydr
aul
i
c mo
tor
,
fo
r i
ns
ta
nce
c
utt
i
ng
f
or
ces
o
f a
ma
chi
ne
t
ool
.
(
3)
Fr
ic
ti
ona
l f
orc
es,
op
pos
ing
t
he
s
pee
d夕.On
e mus
t d
ist
in
gui
sh
b
etwe
en
(
li
nea
r)
v
isc
ous
f
ri
cti
on
a
nd hi
ghl
y n
on-
li
n e
ar
Coul
omb f
ri
cti
on
or
dr
y f
ri
ct
ion.
(
4) Cou
lomb ri
fct
ion
i
s
the
most
mag
ic
non-l
ine
ari
zi
ng
fa
ctor
i
n h
ydr
aul
i
c s
erv
o's
a
nd
i
s
al
way
s
17
」
、
犷 ‘1饮 产 1
尺
<
,
l
v
一
, 卜卜1
卜、
官}
r卜
}I
V
N
引公
111
?
,
丁 万
右
.0
』
护‘
七 。。 味鑫
之
补
a
p
岛叫、
卜,)户.
,J)
勺 屯
矛二。件
户
1
、
,、
1
1
1"
阳
t』
户
7
的
、.丁飞 .不叫L
闪」气闲,十一
S
IC
昨才州
闷二.
外
、 ̄・5吧
瞥・
险曰 ,咨
w
闷
娜 ,
甲
“卜
犷
,护
,0
贾 V
毛 V
』
L
`
N
子
,J上+J上 /
a
门 ̄+才宁・ ,卜』、、
。
(
闷 r丫
尸一
飞,气芯囚扣
{
十尸二
5、户
哟
『
}
思
的
常〉
 ̄
叹,
带叭〉带冷
\ 川扩
11二
创二
飞1
』
.
乞。
、
。‘
.心
乏才
勺
冷
1
夕 1
冉
之七、1
。
冷 夕 .
奢
、 』
二t
、
之1
,7
盯,
‘
峨\
仑
比川
时卜瞬.・、
目
脚
卜 。、、
零扩
卜
SvI
闷 甲
闷
阵。
甲
切
l
r
代
l
〉v『
‘
 ̄了
7
l
』
L
』岛
Fi
g.
1
.1]Co
mpr
ess
ibi
li
ty
o
f o
il,
oi
l
sp
ring
,
st
ifne
ss
a
nd
n
atu
ral
re
f qu
enc
y
s
us twh
pec e
n s
omet
hi
ng
i
rr
egu
lar
i
s h
app
eni
ng.
f
n n
orma
l p
rac
tic
e,
Co
ulo
mb
f
ic
r t
ion
c
an
b
e
5-15% o
f t
he
ma
ximu
m l
oad
-ca
pac
iy.
t Tl
a
us
i
t
ca
nno
t b
e n
egl
ect
ed
wi
tho
ut
s
eri
ous
c
ons
equ
enc
es,
is
s
a i
s
don
e i
n ma
ny
t
ext
book
s a
nd
ma
nuf
act
ure
rs'
ca
tal
ogs
.
In
t
his
bo
ok
t
he
s
imp
lif
yin
g ss
a u
mpt
ion
i
s
ma
de
t
hat
i
n a
mov
ing
mo
tor
Co
ulo
mb
f
ri
ct
ion
i
s
co
nst
ant
,
re
ver
si
ng
i
ts
s
ign
wi
t
h t
he
s
pee
d夕
(
see
F
ig
1
.13
).
I
n s
pit
e o
f s
ome
mi
sl
ead
ing
p
ubl
ica
ti
ons
i
t
sho
uld
be
s
tre
sse
d h
ere
(
and
wi
ll
b
e
pr
ove
d l
at
er
o
n)
t
hat
Coul
omb
f
ri
cti
on
i
ts
elf
c
anno
t
pr
ovi
de
a
deq
uat
e
st
abi
li
yi
t
n
pr
act
ic
e.
(
5)
Vi
sc
ous
f
ri
ct
io
n i
s
pr
opo
rti
ona
l t
o t
he
s
pee
d夕a
nd
c
ont
ri
but
es
t
o s
ta
bil
i
ty
a
nd
d
amp
ing
i
n
ma
ny
h
ydr
auic
l s
yst
ems
. No
rma
ll
y,
i
t c
anno
t b
e p
red
ict
ed
s
ufi
ci
ent
ly,
i
ts
a
mou
nt
c
anno
t b
e
a
ccu
rat
el
y d
esi
gne
d a
nd
d
ire
ct
me
su
a r
eme
nt
i
s h
ardl
y po
ssi
ble
. Ne
ver
the
les
s i
ts
p
res
enc
e c
an
b
e
pr
ove
d i
ndi
rec
tl
y.
Nu
mer
ous
e
xpe
rime
nts
s
how
a d
egr
ee
of
s
yst
em
damp
ing
f
or
whi
ch
t
he
ss
a ump
tio
n o
f v
isc
ous
f
ri
ct
ion
i
s t
he
on
ly
e
xpl
ana
tio
n.
Thu
s i
t s
houl
d n
eve
r b
e n
egl
ect
ed
i
n
dy
nami
c
an
aly
sis
.
.门卜一 .
F
,,
一F,
_”了
Fi
g.
1.
12
For
ces
t
otal
Fi
g.
1
.13
Cou
lomb
f
ri
ct
ion
a
nd
v
isc
ous
ri
fct
ion
Our
exp
eri
ments
s h
owe d
too,
i
n an
indir
ect
wa
y,
a st
ron
g corr
ela
ti
on bet
wee
n vis
cous
and
Cou
lomb
ri
f ct
ion.
Hydros
tat
ic
be
ari
ngs
i
n hydr
aul
i
c c
yli
nder
s f
or
ins
tanc
e e
li
mina
te
not
onl
y t
he
Cou
lomb
fr
ict
ion
(whic
h is
t
he
purp
ose
.) b
ut
al
so
t
he
vis
cous
ic
f ti
on,
maki
ng a
ddi
ti
onal
da
mping
mea
s ures
necess
ary.
Su
mma
riz
ing
s
o f
ar,
we
a
ssu
me
a
s
impl
e (
but
c
omp
ic
l a
ted
e
nou
gh)
f
ri
ct
ion
mod
el
d
epi
ct
ed
i
n
认
Fi
g.
1.
13
『
Fc
阿
Ff
nc
t
io
n二Fc十吟 二 nl 兀
+ wy.
.
Co
mbi
nin
g p
rev
iou
s f
orc
es
o
f Fi
g.
1
.12
wi
th
Ne
wto
n's
l
aw,
we
g
et
P
,
A,
一凡A:一凡一凡一w夕=My
or
P
,
A,
一凡A2=凡+凡+砂 +
My・
Now,
de
fni
i ng
t
he
t
ota
l l
oad艺F
t
o be
艺F二F
e+F
c+w
y+My
,
we
ge
t f
ina
ll
y
P,
A,
一P2
A:二IF=FQ+Fc+砂十搏 ・
S
o t
he
r
el
at
io
n b
et
wee
n t
he
o
il
p
res
sur
e
P
a
nd几d
epe
nds
o
n t
he
t
ot
al
l
oad
I
F.T
he
p
res
sur
es只
a
nd凡 i
n
tu
n
r d
ete
rmi
ne
t
he
pr
ess
ure
d
rop
s a
cro
ss
t
he
p
ort
ope
nin
gs
i
n
th
e v
alv
e.
Us
ing
。_。
,_a
CJ_EP
一 “v
P
t
he
lo
f ws
t
hro
ugh
t
he
o
pen
ings
a,
,
a2
,
a3
a
nd
a
4 i
n
Fi
g.1
.14
be
come
LI 丈
?"
LI一L
2
=
.S
-Fc一,
r}
-一场・
Fi
g.1
.14
F
orc
es,
lo
f ws
a
nd
c
ont
inu
it
y
。
,
一。
I
Cd
一2
P
S
_
P
I Q
2
一。
2
2
。
C
、
d
P
2
P,
'
Q
3
=a
3C
d
俘 Q
4一
。
4
C
d
2
平一
Co
nsi
der
ing
t
he
c
omp
res
si
bil
it
y o
f o
il
(
acc
ordi
ng
t
o F
ig.
1.
11)
a
s we
ll
a
s c
ont
inu
ity
we
g
et
t
wo
e
qua
ti
ons
f
or
t
he
s
pee
d夕of
t
he
mo
tor
.
y=
Q
l
-Q
2一A
,
L,
+1
鱼户,
入
AI
E
20
‘・
?“兮少个一
瞥・+
1。
)
・内?
乙叱口“.:.牙 右 勺、‘
之下‘
之』‘之勺乍户
勺魂勺q、
‘,、白』
飞 心・
,I’」树、
。亡
夕+
-叫
夕、
・夕!
已陈万?
“,.
 ̄犷} 1
.
竺逻令
・
匕 、・
一
咚,・
‘
7
二 叫‘
-z 菩:
:
、.兮半中一瞥。1陋
『长乙叱J;、 ・
。 (
州・ ‘
.。
.二
、、
、、
,,
t
.!
J代、
。一心
}!
、, 宁
.
豹.匀
I
",
心}臀
}・留公}
+‘}
・叼・
"公・。
;,.
‘宁・
夕矛・
‘补 』
、勺油
、二洁 勿
・4`I
-1
a` cP
Q
L
Ci
。
. ̄・
心・如+
、。
』。
4
; 味“ ̄
‘
1
+7
勃!、
臼心 ; 叭・
:1
l
兰
#
-
J
W
}
4I
4
Oi
C
弓
、!
、
,气
”兮报・
公少
”
k
!心.
夕・公; 了
'1
、
;瞬+
公少
s
r
; 叭。鑫旬
or
I
4
vomot
}
;
a
』
、1
匆 ̄"
‘勺.
岛七之。
,01气
『
肛件口
寸1比J
拿・享 比亡
。!陈味J
"11
价犷
b
`
J夕.、
白
er
4
ic
s
叱J
夕! 比 ?1
陈味。・
夕分
比 J
。
・夕
aul
口1
勺. 卜
. 1
.
勺 八 、
.队髯泞宁.矿
hydr
of
21
i闷
. 目伪
心
'
_ 1嗽
I
G !
I
y
护
淤 勺卜
、
ve
W卜
‘叫泊卜,
公。
冰・??
..
`
沙・公・
叩
卜明IN
衅
心
.
0
Sur
斗
C
盯 “
比“
.资
}+、
。丫穿・、
下舀
5
.
片 +净十罕十蛤.瑞 ・切即』,吟
g. 1
心翔 十心二十心十,1”卜『。盆(公・荤、
1.
.
渝 +,2.+幼十
睁11』闪.,(
卯。分
二
-肾・
冷
,犷 枷
丫,吟
Fi
粗 与‘为』
1
叫 ,
、
.一
二 r
. 1
5气 C
; 1 !
勺
t之 一- .  ̄
卜之,
之、。户・日.
.1 ‘二俄 侣 、、邃‘
合沂 勃
之,,・
户二』・‘
欲傲 与 ̄
Jl补
‘注称劳  ̄
七
,尸 匀
二 。
J二
户、
口
,
1
宕 户 』
・、 、‘
之
J
o,
0、之・
L,oloJ 勺、
尧‘抽尧攀一坷 O
J 、
勺 ‘
之‘
。‘t
之 七‘
11",
之・沂号 ‘
之。.
"‘n
、‘之,三匆
‘之、
二・
I’t,
.右 ‘0,01
1
1‘舒『1之勺1
‘、、J
,I
工tl.
爵,仓
仑
夕= Q
3
-
Q
,十A
兰过V
,2户
2
A 2
A,
E
Combi
ni
ng
p
rev
iou
s
equ
ati
ons
a
nd
n
egl
ect
in
g玫,an
d从:we
ge
t
fi
nal
l
y
夕
=
Cd
PP
S
}
a
A-
'
,
2
1
-
P
P
S
)
一a
A
2
,
2
P
S
,
L, 三
一一
-二尸 I (
1.1
2)
E
w 「
a.
2P7。月厅,P }
)1
L, ;
夕“Cd
飞}
言}AJ
11
P
`一A飞
}“
}’一P }+一二 尸2.
}E (
1.13)
,r
L‘-
3
l
'‘S
.
.2
V\ ‘S
l
j
wh
ere
P,
A,
一P2
A2
=Y
,F二F,
+凡+砂 十脚 . (
1.
14
)
Fr
om
t
hes
e e
qua
ti
ons
,只 a
nd几 a
s we
ll
a
s夕ha
ve
t
o be
s
olv
ed.
I
n g
ene
ral
, a
nal
yt
ic
l
a
s
olu
tio
ns
are
i
mpos
si
ble
.
A sp
eci
al
sol
uti
on
could
be
found
o
nly
f
or
ver
y s
peci
al
c
ondit
ions
.
Any
so
lut
ion
mus
t
fu
lfi
l
l t
he
req
uir
eme n
t t
hat
th
e sp
eed夕acc
ord
ing
to Eg
ns.
(1.
12)
and
(1
.13
)
a
re
t
he
s
ame
.
A
s
pec
ial
s
olu
tio
n c
an
e
asi
ly
b
e f
oun
d b
y e
quli
a z
ing
a
ll
c
orr
esp
ondi
ng
f
act
ors
i
n
Eg
ns.
(1
.12
)
an
d (
1.1
3).
Th
e "
con
dit
io
ns"
a
re
a,
a3
a
nd丝二生 (
1.1
5)
A,
A2 A,
A2
S
L.二L,二一 , (1
.16)
2
君十PZ二P
S , (1
.17)
Condi
ti
on
(1.
15)
conc
er
ns
val
ve
and
mot
or
di
mens
ions
a
nd
cou
ld
be
f
ulf
il
le
d t
heo
ret
ica
ll
y.
Con
dit
ion
(
1.1
6)
r
est
ri
ct
s
th
e s
olu
tion
t
o o
nly
on
e po
sit
iono
f t
he
p
ist
on.
Cond
iti
on
(
1.1
7)
t
oge
the
r
wi
th
Eqn.
(
1.1
4)勇el
ds
P, _A
Z+艺F
l
Ps and 2_A
P ,一艺Fl
Ps (1
.18)
尸: A,+AZ
PS A,+AZ
Subs
ti
tut
ing
Eqns.
(
1.1
5),
(1
.16)
and
(
1.1
8)
i
nto
Eqn.
(
1.1
2)
gi
ves
心= r
2
P
I
}
-
l
a,
I
A,
"
一 ‘_iw
-
S'
.
F/P
,。
e止‘
i
,I
A,十Y.
FI
Pc飞y户
_一二一一止竺巴二-二一一二 一 一立 {,一二一‘竺竺二一二一二二 1
一 rr
J
- d、】 _ f , 11 , .A ‘ 1t , . , 1 (1
.19)
u
U in , u
H :十 八 , ti. u 八 ,十 H .
,
I
C
L ”
“ ‘’ 且 ‘ 」
wh
er
e c
o=2
E认+A,
)/
s.
As
i
s
ma
de
c
lea
r,
e
gn.
(1
.19
)
re
pre
sen
ts
o
nly
o
ne
s
pec
ial
sol
ut
ion
,
be
ing
v
aid
l i
f
co
ndi
ti
ons
(
1.1
5)
a
nd
(
1.1
6)
a
re
met
.
Th
e qu
est
ion
i
s wh
ethe
r o
r n
ot
t
his
s
olu
tio
n i
s a
go
od
a
ppr
oxi
mat
ion
i
f t
he
c
ond
iti
ons
a
re
n
ot
f
ulf
il
led
. I
n t
he
l
it
era
tur
e t
his
q
ues
ti
on
i
s a
voi
ded
s
yst
emat
i
cal
ly
, t
aci
tl
y
a
ssu
min
g t
hat
it
i
s
no
t
impo
ta
r n
t whe
the
r t
he
c
ond
iti
ons
a
re
me
t o
r
no
t.
I
t wi
ll
b
e pr
ove
d i
n Ch
apt
er
2
t
hat
Eg
ns.
(1.
18)
a
nd
(
1.1
9)
do
n
ot
e
ven
s
how
t
he
s
li
ght
es
t
r
ese
mbl
anc
e t
o ph
ysi
cal
r
eal
it
y i
f
co
ndi
ti
on
(
1.1
5)
i
s n
ot
f
ulf
il
le
d,
s
ugge
sti
ng
t
hat
a
n a
symme
tri
c
22
mo
tor
n
eeds
a
n a
symme
tri
c v
alv
e.
I
t
wi
l
l b
e pr
ove
d t
hat
t
his
s
ugg
est
i
on
i
s c
orr
ect
b
ut
t
his
i
s
no
t
t
oo
h
elp
ful
s
inc
e a
symme
tri
c v
alv
es
n
orma
lly
a
re
n
ot
a
vai
lab
le
c
omme
rci
all
y.
So
s
pec
ial
a
tt
enio
t n
i
s
ne
ede
d t
o
sol
ve
t
his
p
rob
lem.
Sy
mme
tri
c
mo
tor
s
wi
t
h s
ymme
tri
c
va
lve
s d
o n
ot
c
aus
e c
omp
lic
ati
ons
.
Re
lat
ing
t
he
o
pen
ings
a'
an
d a2
to
t
he
s
poo
l
di
spl
ace
men
t
x
l
in
ear
iz
ati
on
a
nd
a
ppr
oxi
mat
io
n wi
l
l
le
ad
t
o
a
h
ands
ome
di
ffe
rent
ial
equat
ion
t
hat
wil
l
permit
d
ynamic
anal
ysi
s,
s
ynthe
sis
an
d des
ign.
Thi s
i
s
done
in
t
he
fo
llo
wing c
hapt
ers
f
or
sev
eral
t
ypes
of
hydr
aul
ic
ser
vorn
otor
s.
Fig.
1.1
5 gi
ves
a
fundament
al
sur
vey
of
hydr
aul
ic
ser
vomotor
s as
a
"su
mma r
y i
n a
dvance
" i
n whi
ch t
he
abov
e tool
s f
or
a
naly
sis
ha
ve
bee
n appl
ied
and
t
he
l
inear
izat
ion
ca
rri
ed
out
.
1.
9 PERFORMANCE
INDICES
In
c rea
sing emp h
si
a s on the ma th
ema tic for
mu l
ai
ton and measur
eme nt o
f control
syst
em
perf
orma nc
e c
an be found i
n t
he r
ecent it
l erat
ure
on aut
oma t
i
c cont
rol.
Mod e
r n
contro
l t
heory
ass
ume s t
hat t
he syste
ms eng
ineer can
s pecif
y quant
it
ati
vel
y the
equ
r i
red s
ystem perfo
rmance.
Then ap
erf
o r
ma nce i
nd e
x can b
e c
alculated or
measur
ed and us
ed to eval
uate
the sys
tem's
perf
orma nc
e. A q
uantit
ativ
e mea su
re of
t
h e perf
ormance
of
a
syst
em is
necess
ay
r f
or
the
operat
ion
of
mo dern
adapti
ve cont
rol
syst
e ms f
or a
u toma t
ic
par
ame t
er
opt
imizat
ion
of a
cont
rol s
yst
em, and
for
t
h e
desi
gn of
opti
mu m syst
em.
Wh
ethe
r the ai
m is
to improve t
he de si
gn of a s
yst
em or
to des
ign ac ont
rol sys
tem, a
per
forma nc
e inde
x mu st
be
chosen and
me a su
red.
A
perf
orma nce
index
is
a quanti
taiv
t e
me asur
e of
t
he
perfo
rmance o
f a
syst
em a
nd is
chos
en so
empha s
is
is
given
to
the
imp or
tant
syst
em.
A
syst
em is
consider
ed an o
ptimu m control
syst
em when t
he
syst
em paramete
rs ar
e adj
ust
ed so
t
hat
t
he
i
nde
x r
eac
hes
a
n e
xtr
emum
v
alu
e,
c
ommo
nly
a
mi
ni
mum
v
alue
.
A
p
erf
or
man
ce
i
nde
x,
t
o
b
e u
sefl
_
mu
tt
h
e。n
umb
er
t
ha
t
is
a
lwa
ys
p
osi
t
ive
o
r
ze
ro.
Th
en
t
he
b
est
s
ys
te
m i
s
de
fi
ne
d
as
s
yst
em
t
hat
mi
nimi
zes
t
his
i
ndex.
A
s
uit
abl
e p
erf
orma
nce
i
nde
xis
t
he
i
nte
gra
l of
t
he
s
qua
re
o
f t
he
e
rro
r,
I
SE
whi
ch
i
s
d,
in
f e
d
as
二=
f
0
,e
2
(
t
)d
t. (1
.20)
Th
e up
per
l
i
mit
T
i
s
a
f
ini
te
t
ime
c
hos
en
s
ome
wha
t
ar
bi
tr
ari
l
y s
o t
hat
t
he
i
nt
egr
al
a
ppr
oac
hes
a
s
te
ady
-s
ta
te
v
alu
e.
I
t
is
u
sua
ll
y c
onv
eni
ent
t
o c
hoo
se
T a
s t
he
s
et
tl
i
ng
t
ime
,兀.Th
e s
te
p
r
espo
nse
f
or
a
s
pec
ifi
c
fe
edb
ack
c
ont
rol
s
yst
em
i
s s
hown
i
n Fi
g.
1
.16
b)
,
an
d t
he
e
rro
r i
n Fi
g.
1
.16
c)
.
The
er
ror
s
qua
re
i
s s
hown
in
Fig.
1.
16
d),
a
ci
d t
he i
nt
egr
al
o
f t
he
er
ror
s
quar
e i
n Fi
g.
l
.
16
e
).
Thi
s
cr
it
er
ion
wil
l
di
scr
imi
nat
e b
etwee
n e
xce
ssi
vel
y ove
rda
mped
and
exc
ess
ive
ly
under
damped
sy
ste
m.
The
mini
mu m va
lue
of
t
he
int
egr
al
occ
urs
for
a
compro
mise
val
ue
of
t
he da
mpi
ng.
Th e
pe
rfo
rman
ce
in
dex Fi
gn.(
1.2
0)
is
ea
sil
y a
dap
ted
`f
or p
rac
ti
cal
mesu
a re
ments
be
cause
a
s
qua
rin
g
c
ir
cui
t
is
r
ead
il
y
ob
tai
ned.
Fu
rt
her
mor
e,
t
he
s
`
qua
red
e
rro
r i
s'
mat
her
nat
ic
all
y
co
nve
nie
nt
a
nal
yt
ic
al
a
nd
c
omput
ati
onal
p
urpose.
An
oth
er
r
eadil
y
inst
rument
ed
pe
rfo
rma
nce
c
rit
eri
on
i
s t
he
i
nt
egr
al
o
f t
he
a
bsol
ute
ma
gni
tud
e o
f
23
t
he
e
rr
or,
IAE,
whi
ch
i
s
wr
it
te
n a
s
一
曰
1
-(0
一
曰
‘
」
11
八
0
a) r
- -1
卜
00
0
e(
t)」
e=(
t)
41
0
,
f
e
=(t
)d
t
e)
Fi
g.
1
.16
The
c
alc
ula
tio
n o
f t
he
i
nte
gra
l
squ
are
d e
rro
r
I
A:一
fl
e
(
t)
dt
l
. (
1.21)
Th
is
i
ndex
i
s
part
i
cular
ly
usef
ul
f
or
c
omp u
ter
s
imul
ati
on
st
udi
es.
To
red
uce
the
con
tri
buti
on
of
t
he
lar
ge
ini
ti
al
er
ror
t
o t
he
val
ue
o
f t
he
pe
rfo
rma
nce
i
nte
gra
l,
a
s
we
ll
a
s
to
e
mph
si
a z
e
er
ror
s
oc
cur
ri
ng
l
at
er
i
n
th
e r
espo
nse
,
th
e f
ol
lowi
ng
i
nde
x h
s
a b
een
p
rop
ose
d:
二二一
f
0
'
Tt
l
e(
t
)l
d
t. (
1.22)
This i
nde
x i
s d
esi
gna
ted
t
he
i
nte
gra
l o
f t
ime
mu
lt
ip
li
ed
by
a
bsol
ut
e e
rro
r,
I
TAE.
24
Ano
the
r
si
mil
ar
i
nde
x i
s
th
e i
nt
egr
al
o
f
ti
me
mul
t
ipl
i
ed
b
y t
he
s
qua
red
e
rro
r,
I
TSE:
I
T
SE・
f
0
,
te
2
d
t (1
.23)
The
perf
orma
nce
in
dex I
TAE
pro
vid
es
the
bes
t
sel
ecti
vit
y o
f t
he
per
for
man
ce
ind
ices
;
t
hat
is
,
the
mini
mum val
ue
of
t
he
in
teg
ral
i
s
re
adil
y di
sce
ni
r bl
e a
s t
he
s
yst
em
par
amete
rs
a
re
va
rie
d.
The
gen
era
l
form
of
t
he
per
for
mance
int
egr
al
is
,
一ff
(e
,
(t
),
;
(‘
),
y(
t
),
t
)d
t, (1
.24)
whe
re
fi
s
a
f
unc
tio
n of
t
he
e
rro
r,
i
npu
t,
o
utp
ut,
a
nd
t
ime
. One
c
an
o
bta
in
n
ume
rou
s i
ndi
ces
b
ase
d o
n va
ri
ous
c
omb
ina
tio
ns
o
f t
he
s
yst
em
var
iab
les
a
nd
t
ime
. I
t i
s wo
th
r n
oti
ng
t
hat
t
he
mi
ni
miz
at
io
n o
f I
AE
o
r I
SE
i
s
of
t
en
o
f
pr
act
i
cal
si
gni
ic
f a
nce
.
Fo
r
ex
amp
le,
th
e mi
ni
miz
at
io
n
o
f a
pe
rfo
rma
nce
i
nde
x c
an
be
d
ire
ctl
y
re
lat
ed
t
o t
he
mi
nimi
zat
ion
o
f f
uel
con
sump
tion
f
or
a
ir
cra
t
f
a
nd
s
pace
ve
hicl
e.
Pe
rf
ormance
i
ndic
es a
re
use
ful
fo
r the
a
nal
ysi
s a
nd
d
esi
gn
o
f c
ont
rol
s
yst
ems
.
Two
e
xamp
les
wi
ll
i
ll
ust
rat
e
the
u
til
iyo
t
f
th
is
app
roach.
EXAMF生E1.
1 PERFORMANCE
CRITERI
A
A
s
ing
le-
loo
p f
eed
bac
k c
ont
rol
s
yst
em
i
s s
hown
i
n Fi
g.
1
.17,
wh
ee
r t
he
na
tur
al
re
f qu
enc
y i
s
th
e
no
rma
li
zed
v
alu
e,
0
)二1.
The
c
los
ed-
loo
p t
ra
nsf
er
f
unc
ti
on
i
s
th
en
1
T(
s)= (1
.25)
S2+2
}
S+1
2 -5
I
(
s+J‘)
Fi
g.
1
.17
Si
ngl
e-l
oop
f
eed
bac
k c
ont
rol
sy
ste
m
1
7
.
.
V
】
.
\ /
[
IT.1
E 10 /
ITSE/
.
I
I
I入 / ( ,
/
\ \ / 11/
洲
尸尸
卜‘_ 一‘
叮 曰曰 尸尸口
WW'-'
、
议 声
目侧.口‘翻口
口口户
尸 声 IS石
叹二二二二a
0
.0
0.
2 0.
4
0.
6 0
.8
1.
0
1.
2
1.
4 1.
6 1
.8 2.
o
杏
Fi
g.
1
.18
Th
ree
p
erf
orma
nce
c
rit
eri
a
fo
r a
s
eco
nd-
ord
ersyst
em
,5
Thr
ee
p
erf
orma
nce
i
ndi
ces
叫ISE,
ITSE,
and
I7
AE
-
cal
cul
at
ed
f
or
v
ari
ous
v
alue
s
of
t
he
d
amp
ing
r
at
io否a
nd f
or
a
st
ep
i
npu
t a
re
sh
own i
n F
ig.
1.
18.
T
hese
cur
ves
s
how
th
e s
ele
ct
iv
it
y o
f t
he
IT
AE i
nde
x i
n
co
mpar
i
son
wit
h
th
e I
SE
ind
ex.
Th
e v
alu
e o
f
the
da
mpi
ng
rat
i
o杏s el
ec
te
d on
t
he
b
asi
s
of
I
TAE
i
s
0.
7,
wh
ich,
fo
r a
s
eco
nd-
ord
er
s
yst
em,
r
esu
lts
i
n a
s
wif
t
re
spo
nse
t
o a
s
tep
wi
th
a 4.
6%
over
shoot
.
EXAMPLE
1.
2 SPACE
TELESCOPE
CONTROL
SYSTEM
Th
e s
igna
l-Mow
g
rap
h o
f a
s
pac
e t
el
esc
ope
p
oint
i
ng
c
ont
rol
s
yst
em
i
s s
hown
i
n Fi
g.1
.19
. We
d
esi
r
e t
o
se
le
ct
t
he
ma
gni
t
ude
o
f
th
e g
ain
K3
t
o
mi
ni
miz
e t
he
e
ff
ec
t
of
th
e d
ist
ur
ban
ce
D(
s)
.
Th
e
d
ist
urba
nce
i
n t
his
c
se
a i
s
equ
iva
len
t t
o a
n i
nit
il
a a
tit
ude
e
rro
r.
Th
e c
los
ed-
loo
p t
ran
sfe
r f
unc
ti
on
f
or
t
he
di
stur
bance
i
s obt
aine
d by
us
ing
Mas
on'
s
si
gna
l-low
f gai
n f
ormul
a a
s f
oll
ows
:
S)_只(
Y( S)0,
(s)_ 1
・(1
+K,凡S一
,) S(
S+凡凡S 一
,) (
1.26)
D(s) 0(
s) 1+K,
K3S-
,+K,
K2KP
S-
2
S
2+K,K3
S_
1+K,
K2KP
脚 ‘t
nrb4l
tPe
In
V
R/x
/
At
titude
R(
S)
Pos
it
ion
fe
edba
ck
b)
Fi
g.
1
.19
A
spa
ce
t
el
esc
ope
po
int
ing
c
ont
rol
sys
tem.
a)
Bl
ock
d
iag
ram.
b)
Fl
ow
g
raph
Ty
pic
al
v
alue
s f
or
t
he
c
ons
tant
s
aeK
r ,=0
.5
a
ndK,
K2K,二2
.5・
The
n
th
e
na
tu
ra
l
r
fe
que
ncy
o
f
t
he
v
eh
ic
le
i
s
f
=了
2
.5
/
27
r=0.
25
c
ycl
e/
s.
Fo
r a
u
nit
s
te
p d
ist
ur
ban
ce,
t
he
mi
ni
mum
I
SE
c
an
be
a
nly
a t
ic
all
y
ca
lcul
at
ed.
The
a
tt
it
ude
y(t
)
is
,
(
;)
一迎[
。
。My
si
n
(粤,
十、
)
P L
{
,
Z 」
(
1.27)
26
wher
e 刀=K3
寸(K孑/
8)一5
. S
qua
i
rng
y
(t
)
a
nd
i
nt
e
gra
ti
ng
t
he
re
su
lt
,
we
ha
ve
e
-
o.
5
K3
t
si
ne
(
粤 t、、)、
,一厂 奖。 一
。
,‘
。
,(
喜一 喜Co
s(
刀,
十:
、)
)J
,
1
0
I二 「一兰竺
J
o n
月z L
"
U P一 L
L
(1
.28)
1
二(
,二,,+U.
1K,)
.
K3
Di
fer
ent
ia
ti
ng
I
and
e
qua
ti
ng
t
he
r
esu
lt
t
o z
ero,
we
obt
ai
n
d
l二一
Kz
2
+0.
1=0
. (1
.29)
d尤3
T
her
e
fo
re
t
h
e
mi
n
imu
m
IS
E
is
ob
t
ai
ne
d
wh
en
K3
=俪 =3
.2
.
Th
i
s
va
l
ue
o
f
K3
co
rr
es
po
nd
s
t
o a
da
mping
r
at
io杏of
0.
50.
The
va
lue
o
f IS
E a
nd IAE for
thi
s
sy
ste
m ar
e
plo
tt
ed
i
n
Fig.
1.
20.
The
mini
mum f
or
the
IAE pe
rf
or
mance
in
dex
is
obt
ai
ne
d when K3=4 .
2
and否=0. 6
65.
Whil
e
the
I
SE c
ri
te
i
ron
is
n
o t
as
s
el
ect
iv
e a
s
the
IAE cit
reri
on,
i
t
is
c
lea
r
tha
t
it
i
s
pos
si
bl
e t
o
so
lve
ana
lyt
i
cal
ly
for
t
he mini
mum val
ue
of
ISE.The mi
nimum of
IAE is
obt
ai
ned
by
meas
uri
ng
t
he
act
ual
val
ue
of
IAE f
or
se
ver
al
va
lue
s
of
t
he
par
amet
er
of
i
nt
ere
st
.
I了
.5
了.了
\
1.
1 \““ /
1
7
\ \ 踌
a
0.
夕
E、、、火
尸一
I
S /
叹
,/
-9
0
7-
1
"
} .阮
}"
.  ̄
劝‘召.口巨.
}
1 ̄一‘川p,
...月..网一
仪j
4 5 ‘ 夕 s 9 10
K;
Fi
g.
1
.20
The
p
erf
orma
nce
i
ndi
ces
o
f
the
t
el
esc
ope
c
ont
ro
l
sy
ste
m v
ers
us
K3
A c
ont
rol
s
yste
m is
o
pti
mum wh e
n t
he
sel
ect
ed
per
for
mance i
ndex
is
mini
mize
d.
Howe v
er,
t
he
o
ptimum va
lue of
t
he
para
me t
ers
dep
ends
dir
ect
ly
on t
he def
ni
i ti
on of
opt
imum,
that
is,
t
he
p
erfor
mance
in
dex.
The
ref
ore,
in
Examples
1
.1
and
1.
2,
we f
ound t
hat
t
he
opti
mum s
etti
ng
vari
ed
f
or
di
ff
erent
pe
rfor
mance
i
ndi
ces
.
Ta
ble
1
.1
Th
e Opt
i
mum
Co
eff
ic
ien
ts
o
f T(
s)
Ba
sed
o
n t
he
I
TAE
Cri
t
eri
on
f
or
a
St
ep
I
npu
t
S+叭
S
2+1
.
4o)
,
,
s+。万
s+
5
31
.7
500S2
+2.
15
嵘s +嵘
5
s+
42.
1o,
,
s
'+3.
A
4
t)
n
2S
2+2.7
_J
)
3
nS+。‘
S+2
.
8a
)s
4
+5.
0可S3+5.5
叫S 2+3.
4心S+嵘
5
S+
63.25
0),
s5+6.
6
0
0)
n2
S4+8.
60武S3+7.
45
叫5S+
23.9
5嵘S+。分
27
The
c
oefic
f i
ent
s
th
at
wil
l
mini
mize
t
he
ITAE p
erfor
man
ce
c
rit
eri
on
f
or
a
s
tep
i
npu
t ha
ve
b
een
d
eter
mined
fo
r t
he
ge
ner
al
cl
ose
d-lo
op
tr
ansf
er
f
uncti
on:
Y(
s)
bo
T
(
s)=
R
(s
)一,
‘+b
n
-,
sn
-
'
-
+
-
.
.
・+
、、+
bo (
1.30)
Thi
s t
ra
nsf
er
f
unct
io
n has
a
st
eady-
st
at
e er
ror
equa
l
to
zer
o f
or
a
st
ep
i
nput
.
No
te
th
at
th
e t
ra
nsf
er
fun
cti
on
hs
a n
pol
es an
d no
ze
r o
s.
The
opti
mu m
coef
fi
ci
ents
f
or
t
he IT
AE cri
t
eri
on
aeg
r i
ven
i
n Ta
ble
1
.l.
Th
e r
esp
ons
es
u
sin
g o
pti
mum
coe
fic
f i
ent
s f
or
a
s
tep
i
npu
t a
re
g
ive
n i
n Fi
g.1
.21
f
or
I
SE,
IAE
a
nd
I
TAE.
The
r
es
pon
ses
a
re
p
rov
ide
d f
or
n
orma
li
ze
d t
i
me,
0t
)
nt.
Ot
he
r
st
an
dar
d
f
orms bas
ed
on di
feen
r t
p
erf
ormanc
e indi
ces
ar
e a
vai
lable
an
d can
be
us
eful
i
n ai
din
g th
e
d
esig
ne r
t
o d
ete
rmi
ne t
he
ra
nge o
f coe
fic
f i
ent
s f
or
a s
pe c
iic
f pro
ble
m. A
fi
nal
exa
mple wi
ll
i
ll
ust
ra
te
t
he
u
ti
li
t
y o
f
th
e s
ta
nda
rd
f
orms
f
or
I
TAE.
EXAMPLE
1.
3 TWO-
CAMERA
CONTROL
A vey
r ac
cura
te
and
ra
pidl
y
resp
onding
co
ntr
ol
syst
em i
s
requ
ire
d fo
r a
s
ystem
that
al
lows iv
l e
a
cto
rs
seemi
ngly
to
pe
rfo
rm i
nsi
de
of c
omple
x min
iat
ure
set
s.
Th
e t
wo -
camera
sys
tem
is
shown i
n
F
ig.
1.2
2, wher
e one
camera
is
tr
aine
d on
the
acto
r an
d the
other
on t
he mini
atu
re s
e t
.
Th e
c
hal
len
ge i
s t
o o
btain
rap
id a
nd
acc
ura
te
coo
rdi
nati
o n
of
the
two
camer
as b
y usi
ng s
enso
r
i
nf
ormat
ion
ro
f m
the
for
egro
und
ca
mera
t
o c
ont
rol
t
he move
me n
t
of
t
he
back
grou
nd c
amera
.
The
b
loc
k di
agr
am
of
the
bac
kgrou
nd
ca
mera
s
yst
em
is
shown i
n
Fig.
1.
22
fo
r o
ne
axi
s
of
movement
o
f
t
he
b
ack
gro
und
c
ame
ra.
Th
e c
los
ed-
loo
p t
rans
fer
f
unc
tio
n i
s
、的uo冬之 P、芝.』
1.夕
10
协 认人甲.了吧补妙
炸
互\ ,,
二之一
.
产二、
X、.
二《、 云嘴扮:
{
1
1`
1
i
}
y
£心代
,
口召
n6
I
1, 2
了
,
L刁f
卜’
n
=
01 {
r
'
+
i
}
}-
,
一’
「”
}
}
02
1
,
t {
幼
15 20
川
Nonnahz
ed
0me
28
、六二0身之 P、,1
月
仰
尸 代 声翻.月沁之.
、、
乞1
.... 口.. 曰口. _ .....‘‘. _
吧洲巴巴巴日甩护脚-
啼二留,,
一
、
八
『
17-
‘匀\’
/I
拐
I
产
口
傀 i。
,:
、 尸
,昌J
La万
刀湃厂n=
}
6
1
1
}/
匆 b)
15
10
Cor
mnl
red
t
ime
a认‘
门名
井 .
』0冬七飞 “,I。攀屯护一
芝
才甲群份会"
r。一,
0
.夕
对r
人 ,,
I
I
{
)1
一/
{
l
刀 叫1石 !
}
I
1
/
I
1
(
i
1
户
了j
:
}
- n
a
-
L
a
-
}
c
’厂/ }
i 厂
一
一”一
“
I
{
却
1了 2口
d
八or
rna石“ d打”r
心
Fi
g.
1.
21
Cont
inued
T(
s)=
KQ
K.
(0
0 (1
.31
)
S
3+2
甄SZ+叫S+K,
K,
n
耐‘
Th
e s
tan
dar
d f
orm
f
or
a
t
hir
d-o
rde
r s
yst
em
gi
ven
i
n Ta
ble
1
.1
r
equi
res
t
hat
2
甄 =1
.
75(
),嘴=2
.15
可,K.
K.
耐==嵘.
Exami
ning
Fig.
1.21 c
) f
or
n=3,
we es
ti
mat
e t
hat
t
he
s
eti
ng
t
ime
i
s a
ppr
oxi
mat
el
y 1
4 s
eco
nds
(no
rmaiz
l e
d t
ime).
There
for
e we
e
sti
mate
).
0 T
s二8
.
Bec
aus
e a
ra
pid
res
pon
se
is
r
equ
ire
d,
a
l
arg
e叭 will
be
s
ele
cted
so
t
hat
t
he
se
tt
ing
t
i
me wi
ll
be
l
ess
t
han 1s
ec
ond.
Thus
,叭 will
be
s
et
equ
al
t
o 10
ra
d/s
. The
n f
or
an I
TAE s
yst
em
it i
s
ne
ces
say
r t
hat
t
he
pa
rame
ter
s o
f t
he
c
ame
ra
d
yna
mic
s b
e
C
OO=1
4.67r
ad/s
29
and
杏=0.
597
.
Backrot
utd
矛飞
,J
egJ
’ 刀“,
才d
n
u
g
e
r
o
F c口即sera
ca月2era
POSl
ttol
posi
tron
Fi
g.1.
22 The foregr
ound c
ame r
a,
wh i
ch may be ei
ther a
fi
lm or v
ide
o camera,
i
s t
rai
ned on
t
he b
lue
cyclor
a ma s
tag
e.
The e
lect
roni
c
servoc
ontrol
in
stal
lat
ion
per
mits
t
he
slav
ing,
by means
of
el
ect
roni
c ser
vo de
vices
,
of
t
he t
wo camera
s.
The back
g r
ou n
d ca
me r
a r
eac
hes
int
o t
he
mini
atur
e se
t
wi
th a
peris
cope l
ens a
nd i
nst
anta
neousl
y r
eproduces
all
movements
of
t
he f
oregr
ound
camera i
n
t
he sca
le of t
he miniat
ure
. The vi
deo con
trol i
nst
alatio
n all
ows t
he co
mp os
ite
image t
o be
moni
tor
ed
and
r
ecor
ded
ive.
l
Th
e a
mpl
if
ier
a
nd
mo
tor
g
ain
a
re
r
equi
re
d t
o b
e
。, _ n_
之、 _才、_ = 一-井 = —
n_
_(0n_
_,‘
一 —
二 一 , 。uJ.
叫 2
.
15心 2.
15
Then
t
he
cl
osed-
loo
p t
rans
fer
f
unc
tion
i
s
1000
T
(
s)
S
3+1
7.5,
2+21
5s+1
000
(
1.32)
1000
(
s+7.08)(
s+5.21+j10.68)(
s+5.21一j10.
680)
The
l
oca
ti
ons
o
f
th
e c
los
ed-l
oop
r
oot
s d
ict
at
ed by
t
he
I
TAE s
yst
em a
re
s
hown
i
n
Fi
g.
1
.23.
Th
e
d
amp
ing
r
at
io
o
f
th
e c
omp
le
x r
oot
s
is杏=0.
44.
Ho
wev
er,
th
e c
omp
lex
r
oot
s
do
n
ot
d
omi
nat
e.
Th
e a
ctua
l
re
spo
nse
t
o a
s
te
p i
npu
t
us
ing
a
c
ompu
ter
s
imu
lat
ion
s
howe
d t
he
ov
ers
hoo
t
to
b
e o
nly
2
% a
nd
t
he
s
ett
ing
t
ime
t
o wi
thi
n 2
% of
t
he
f
ina
l
va
lue
e
qua
l
to
0
.75
s
eco
nd.
Fo
r a
r
amp
i
npu
t,
t
he
c
oefi
ci
ent
s
ha
ve
b
een
d
ete
rmi
ned
t
hat
mi
ni
miz
e t
he
I
TAE
cri
te
ri
on
f
or
t
he
g
ene
rlc
a
los
ed-
loo
p t
ran
sfe
r
fu
nct
ion:
b
,
s+bo (1
.33)
T
(
s)=一
i+方 。n
-1+…十b
S
,
t ,
s十b
o
30
了口
,
12-
’
\ l ,  ̄
O
0 -
;
一{\ 9.
.
4夕
N
I
一 a一
入
自 \ .
口
一
f
-
I
-
]o 一
Fi
g.
1
.23
Th
e c
los
ed-
loo
p r
oot
s
of
a
mi
nimum
ITAE
s
yst
em
Thi
s t
ran
sfer
fu
nct
ion h
s
a a
st
eady-
st
ate
er
ror
e
qual
to
z
ero
fo
ra r
amp i
npu
t.
The
opt
imum
co
efic
ient
s f
or
thi
s t
ran
sfe
r f
unct
ion
ar
e g
iven
i
n Ta
ble
1.
2.
Th
etr
ansf
er
f
unct
ion
r
equ
ire
d t
o
pr
ovi
de
z
ero
s
teady-
sta
te
Tabl
e 1.
2 TheOpt
imu
m Co
efic
fien
ts
of
T
(
s)
Ba
sed
o
n t
he
I
TAE
Cri
te
ri
on
f
or
a
Ra
mp
I
npu
t
s
2
+3
.2o
),s
+。厂
S
3+1
.75
o)
,,
S2+3
.25
(1
)
n+
2S。灵
S
4+2
.41
(t
).
S3+4
.39
叫s 2
+5
.14
o刃
( S
+W4
S
S+2
.1
9(
t)
"
s
4+6
.5
0耐s
3
+6
30}n
s2
+5
.2
4叫s
+。刃
.
31
CHAPTER
2 ANALYSI
S OF
HYDRAULIC
SERVOSYSTEM S
2.
1 I
NTRODUCTION
T he
re
is
a
l
arg
e v
ari
et
y o
f h
ydr
aul
ic
mo
tor
s
an
d va
lve
s t
o b
e c
ombi
ned.
We
di
st
in
gui
sh
(
see
a
lso
Fi
g.
1.
15)
:
(
1)
As
ymme
tr
ic
l
i
nea
r
mo
tor(
A,”2
A2)
con
tr
ol
le
d
by
a
3
-wa
y v
al
ve
.
(
2)
As
ymme
tr
ic
i
l
nea
r
mo
tor(
A,
#A2
)co
ntr
ol
le
d
by
a
s
ymme
tr
ic
4
-wa
y v
al
ve.
(
3)
As
ymme
tr
ic
l
ine
ar
mot
or(A,
#A2)c
ont
rol
l
ed
by
an
as
ymme
tr
ic
4-
way
val
ve
(
a,
¥
a3
,a2
:
#a
4)・
(
4)
S
ymme
tr
ic
i
ln
ear
mot
or(A,=A2)c
ont
r
oll
ed
b
y a
s
ymme
tr
ic
4-
way
va
lve
(
a,=a3
,a2=a4)
・
(
5)
Symme t
ric
ro
tay
r mot
or
con
trol
led
by
a s
ymme tic
r 4-way
valve
.
I
n
each o
f t
hese
cse
a s
one
can
di
sti
nguis
h open-
c e
nte
r a
s wel
l a
s c
rit
ica
l-c
ent
er va
lve
s h
aving
a
p
osi
ti
ve under
lap
U and
zer
o l
ap
U=0 r
espect
ive
ly.
So
thi
s an
alysi
s
could
be e
xtend
ed t
en t
imes,
b
ut
thi
s would o
f c
ours
e be
ra
ther
la
borio
us.
Therefor
e t
he a
pproach
in
thi
s cha
pter
wil
l
be t
he
f
oll
owi
ng.
Firs
t t
he
mo s
t s
imple
cas
e of
an
sy
a mmetr
ic
li
near
mot
or co
ntr
oll
ed
by a
3-way
val
ve wi
ll
be
t
rea
ted,
bot
h for
an ope
n-cen
ter
val
ve
and
for a
cr
it
ica
l-c
ent
er va
lve.
Aft
er l
i
near
iza
ti
on and
a
pproxi
ma t
ion
a
univ
ersa
l
mo del
a
cco
rdi
ng
t
o th
e bl
ock
di
agram s
how i
n Fi
g.
2.
1 wi
ll
be
der
ive
d,
d
et
er
min
ing
t
he
h
ydr
aul
i
c
lo
ad-
st
ifn
ess
C
h
an
d t
he
v
el
oc
it
y
ga
in
K.=o
-
!
}Ia
x
of
th
e
mo
to
r-
va
lv
e
c
ombi
nat
ion
i
tsel
f.
么F
J
.
r+ 凡,
, +
Fi
g.
2.
1
Si
mpl
e
bl
ock
d
iag
ram
f
or
t
he
o
pen
l
oop
o
f a
h
ydr
aul
ic
s
erv
o
To
av
oid
re
peat
ing
t
his
p
roc
ess
e
ndle
ssl
y,
some s
hot-
rcut
s
ar
e mad
e.
As 3-
way-
val
ve-
cont
rol
led
a
symme
tri
c
li
nea
r mot
ors
a
nd
4-wa
y-val
ve-
contr
oll
ed
symmet
ri
c mo
tor
s (
li
near
a
nd
rot
ary)
d
ifer
f
rom
e
ach
ot
her
on
ly
by
a
f
act
or
2
i
n t
he
l
oad
i
nlu
f e
nce(
Ch
i
s
dou
ble
d i
n s
ymme
tr
ic
mo
tor
s,
s
a
t
her
e a
re
2
compatme
r nt
s
wher
e t
he
pres
sur
e c
an
rea
ch on
t
he
loa
d),
open-
cen
ter
v
alv
es
ar
e t
rea
ted
o
nly i
n c
onnect
ion
wit
h 3
-way-va
lve
-cont
rol
le
d asy
mme t
ri
c mot
ors
. I
n all
o
ther
c
ases
only
32
c
ri
ti
cal
-c
ent
er
v
alv
es
(
hav
ing
a,=a2=a3=a4
a
t
x=0)a
re
t
rea
ted.
Pr
agma
ti
cal
ly
s
pea
kin
g,
e
lec
tro
hyd
raul
i
c se
rvo
val
ves
commerc
ial
ly
avai
la
ble
no
rmall
y h
ave
ze
ro
la
p,
i
.e.
c
rit
i
cal
-c
e n
ter
v
alve.
A (
somet
imes
e
xag
gera
ted)
r
eson
a f
or
th
is
is
t
he
abs
ence
of
o
il
c
ons
umpt
ion
i
n t
he
neutr
al
s
ta
te(x=0,
夕=0)
,wh
ich
i
s
imp
ort
ant
i
n
ma
ny
a
ppl
ic
ati
ons
.
Howe
ver,
und
erl
ap-
val
ves
al
so
hav
e a
dvanta
ges
(mai
nly
bec
ause
sys
tems
be
havio
r i
s
far mo
re
i
l
nea
r)
res
ult
ing i
n g
ood c
onst
ant
val
ues o
r sy
stem pa
ramet
ers
. Advan
ced c
ombinat
ions
of
f
eedf
orwa
rd
a
nd
f
eed
bac
k c
oul
d r
esul
t
in
s
upe
ri
or
s
yst
em
be
hav
ior
,
es
pec
ial
l
y a
rou
nd
x=0
a
nd夕二0.
Af
ter
t
his
i
nt
rod
uct
ion,
we
b
egi
n wi
th
t
he
a
nal
ysi
s o
f v
ari
ous
s
erv
omo
tor
s,
a
imi
ng
a
t t
he
h
ydr
aui
l
c s
t
ifn
ess
C
h a
nd
t
he
v
el
oc
it
y
ga
in
K,
.=妙/
ax.
Int
ro
duc
in
g t
he
o
il
s
pr
in
g s
t
ifn
ess
c,
ac
cor
din
g t
o
Fi
g.
1
.11
,
we
g
et
丫F V 户
夕=Km
(
x一匹州‘一)一盛月 ̄
一.. (
2.1)
Ch
CO
The
math
emati
cal
mode
l wil
l
b e
co
mplet
ed
in
the
l
st
a sec
ti
on of
t
his
c
hap
ter
,
de
riv
ing
t
he
s
yst
em
t
rans
fer
f
unc
tio
n n
eed
ed f
o r
dynami
cal
a
naly
sis
a
nd
syn
thes
is.
2.
2 ASYMMETRIC M OTORS CONTROLLED BY 3-
WAY VALVE
Th
e
mos
t
si
mpl
e
se
rvomo
tor
s
ar
e
sy
a mme
tr
ic
mo
tor
s
co
ntr
oll
ed
b
y 3
-way
va
lve
s (
see
Fi
g.
2.
2)
.
They
ar
e re
f q
uen
tly
use
d i
n mac
hine
t
ool
copy
ing
syst
ems wi
th
mechan
ical
f
eed
bac
k.
As
sumi
ng
tu
rbu
len
t
lo
f w
and
us
ing
t
he
bsi
a c
f
ormul
s
a o
f Cha
pte
r 1
f
or
th
e mo
tor
b
eha
vior
,
we
fi
nd
P
,
、一
。、=
:;,
s
o
P
s二
AZ
, 丫F
PS
A,-
。
,
=・
a
,
,
c
Cd
d
‘
2(
P
S
P一
-
P
,
)c
・
rP
S) )
2
(A,
一AZ
)2艺F
A, PS
人
L
奋酬‘具
Q,
一Q2
Qr
、.
曰洲
r一 、
交) ar
映二 p =口 P,
Fi
g.
2.
2
As
ymme
tri
c
mo
tor
c
ont
rol
led
by
a
3
-wa
y v
alv
e
33
2
又F
Q
2
一C
几d
2
P,
一C
0
F
}
_{
2
A,
2 十 一一二二二二一一
Ps
A,
夕
一Q,
_Q2一JF一
A, CO
Q
,
_Q
2
Al
一李y
LA,一
,
F
I
t
c
an*se
en
t
h
at
f
or
A
,/
A2
=2。。l
oa
d-
ca
pa
ci
t
y
is
s
ymme
t
ri
c,
mea
ni
ng
艺F .==一艺F;・
S
hou
ld艺
F
co
nt
ai
n
a
"
dc-
c
omp
one
nt
"
Fs
r
(
t
o
be
c
ons
i
der
e
d a
s
a
c
ons
t
ant
pr
e
-l
oad
),
an
a
r
ea
r
at
io
A2
`
A,=0.5一Fs
t
I
P,
A,
co
ul
d
compen
sat
e
for
i
t
.
Si
nc
e t
he
e
fec
t
of
a
cons
ta
nt
p
re
-l
oa
d
Fs
,
wi
ll
be
t
re
at
ed
ag
ai
n
la
te
r
on
,
we
no
w as
su
me Fs,=0 a
nd
ch
oos
e
th
ere
fo
re
A,=2A2
=
A・
乏一
At
t
he
s
ame
t
ime
we
i
ntroduce
t
he
s
o-cal
led
l
oad-
pres
sur
e
F 一。5
1 一2
1 凡
AP
s
o二
+
一
一
二 一 +
PS
2 PS
Now
the
lows
f become
Q1一
,。
J
伶2P
,,
01
_。。
C
一 ‘“
J
_F
f
PS
1
U户N
+
2
PL
Ps
and
一
C
A
d
a
l
f
F
I
_
2
:
Ps
L 一
。,I
I
+
,
2P
,]
‘
V 只」 E
一一 尸L
L 生
(
2.2)
He
re
t
he
s
pee
d夕o
f t
he
mo
tor
h
as
b
een
s
olv
ed
a
s a
f
unc
ti
on
o
f t
he
v
ari
abl
es
a,
,a2
a
nd
PL=I
FI
A.Eq
n
(2
.2
)
ho
ld
s
fo
r
al
l
s
or
ts
of
val
ve
s,
o
pen
-c
ent
er
as
we
ll
as
c
ri
ti
ca
l-
ce
nt
er
va
lv
es.
Th
e on
ly
ss
a u
mpt
i
on
ma
de
s
o f
ar
i
s
tu
rbu
len
t p
ort
lo
f w.
Fo
r t
he
o
pen-
cent
er
v
alv
e of
F
ig.
2.
3
a)
we
c
an
wr
it
e
.
x<-U,a
,=
0,
a2=7
r
d (
U一x)
,
一U<x<十U,a,=i
r
d (
U+x)
,a2=i
r
d (
U一x
),
x>+U,a
,=耐(
U+*
),a
2=0・
S
ubs
ti
tu
ti
on
i
nt
o
Eq
n.
(
3.
2)
y
iel
ds
wi
th
a.=rd
U
一夕
二A
Cd
P
P-
S
-
(
1 于)
U
,1
厂2
P
l十一
,一
I L。
V Ps」 七
卜 _ PL
一U <x< +U,
a
F
v二 一=c,_
A “
I
P
-
_-
FJ_
= 1
VpL’
0+ 于)
一 U 1
-
2P
S
4一
(
,
v V YS」
一x
),
1+
2P
L}
一冬、 乙
L一
・
几
一,
一
aA
o
调(1
+
U
一
E
Th
e v
elo
cit
y g
ain
c
har
act
eri
st
ic
i
s d
rawn
u
p i
n Fi
g.2.
3
c)
,
sh
owi
ng
a
r
ema
rka
ble
l
i
nea
r c
har
act
er
wi
thi
n t
he
u
nde
rla
p r
egi
on一U<x<+U.
{ /
=
F
B
石
n
o
c
l
)
0
-
(
PY
a
{
谁
川目 卜一 一
}
_}
C
2
+
1
0
r
,
l"
+
1
l
o
}
,
N 芳
V
/r
-
2
a
P
.
r(
I
P
S
=
-1
蒸
/ o
I
o
"
;
一
/
矛 -
2
}i
!
+nde
ll+
p
o
pen
-ce
nte
r-3
-
n qr
r
alr
e
川
形
+
:
/
+
t
a 夕
尸z
/
s产
+了
 ̄叫., 戈 _3
月... .勺. .日.口 .
・乙
+I
b)
Fi
g.
2.
3 As
ymme
tr
ic
mot
or
c
ont
rol
le
d by
a
n o
pen
-ce
nte
r 3
-wa
y v
alv
e.
Th
e s
ame
v
elo
cit
y-
gai
n
c
ha
ra
ct
er
i
st
ic
h
ol
ds
fo
r
sy
mme
t
ri
c
mo
to
rs
wi
t
h。4
-wa
y
va
l
ve
i
f
2P
L
I
P
,
is
re
pl
ac
ed
b
y之/
PS
Th
e l
oad
i
s i
l
mit
ed
by
p
res
sur
e b
oun
dar
ie
s 0<P,<P,
f
or
o
bvi
ous
r
eas
ons
.
To
ha
ve
s
ome
s
afe
ty
ma
rgi
n,
we
r
equi
re
1 一飞以
1、P
,
_
1
+PL
<
Ss
o
I
PLI
一
<
(
2.4)
6
P,
2 P,
6 PS
N
ow
,
a
pp
r
ox
im
at
i
ng
1
.
1
2
P
L
I
PS
。,1
1鱼w
PS
e
ge
t
,
i
n
si
d
e
t
he
un
d
er
l
ap
re
g
io
n,
夕
一ao
C
H
d,
P
S干
(,
、
VPL
委
U
)
(
,一
P
共
L)
一(
,
一弃
U
)
(
1十
PL
))一会汽
共 」 C
_2
ao
。
C,
a_
-:
PS
I一竺川一
L
P,
AU “VPL 凡 一
」E-
35
I
ntr
odu
cing
the
por
t wi
dth b
=ao
/U=n
d
U/
U=r
d a
nd
s
ub
st
i
tu
ti
n
g艺F二A
PL
a
nd
C
o=E
A/L,
we
f
in
all
y
co
me
t
o
ou
r
"u
niv
ers
al
"
eq
uat
i
on
Y.
F
Y户
夕 =K,
(x一 ̄ )
一三竺
巴一
Ch Co
wher
e
2
b_ 阿 AP
EA
K
m二AC
‘V节,
“”一
才’“
“一了‘
F
or
t
he
c
rit
ica
l-
cen
ter
v
alv
e o
f Fi
g.2
.4
a
),
we
c
an
wr
it
e
・
‘
0,
a
,一
0
,
a
2
一黔二一
一1a
x_
,
m
x
=
、,
a
2一
Su
bst
i
tut
ion
i
nto
Egn.
(
2.2
) yi
el
ds
。_}
,,「1
,
・AU
d
V
P x一l
L x
12P
L
」 x Ps
Th
e v
elo
cit
y ga
in
c
har
act
eri
st
ic
i
s
dr
awn
u
p i
n F
ig.
2.
4
c)
,
sho
win
g a
f
ar
l
ess
l
i
nea
r c
har
act
er
F
i
g
.2
.
3・
)
・T
h
er
e
is
a
j・
二一 0
i
n
t
h
e
v
e
l
oc
i
ty
g
ai
n
瓮de
p
en
d
in
g
on
th
e
lo
a
d.
Th
us
a
ppr
oxi
mat
ing
mo
re
r
oug
hly
t
han
b
efo
re,
we
ge
t
Y=二 C'
b _
只「 ,
I_飞乙二
X
d 一 !x一二丁 ̄rL!
一_ PL
A
PL 芍 」七
Tr
ans
for
mat
ion
a
gai
n i
nto
Egn
.(2.
1)
y
iel
ds
b_ {
万一 AP
E
A
K
。一Ac
eV奋“
”“亩’C
O“L’
I
t
sh
oul
d b
e n
ote
d t
hat
i
n s
pit
e o
f t
he
a
ppr
oxi
mat
ion
t
he
c
ri
ti
cal
-ce
nte
r-v
alv
e-c
ont
rol
le
d mo
tor
i
s
s
ti
ll
a
no
n-l
i
nea
r
sy
ste
m,
a
s
th
e pa
rame
ter
Ch
de
pend
s o
n t
he
o
pen
l
oop
i
npu
t
si
gna
l
x・
36
火一
J训-I/.,
一
代
f / J产
岁
, 』
/
护
c
a
n
e
t
eC
.
-
1
a
P
S
/
P!
卢
万F
=:1 +1召
门叫卜 ̄一
十2/_
?
C
心 目卜 一 一 卜 ̄州..
}
I
X
C.
1
。公i
cal
-cent
er
3-
rrpp
v
alv
e ・2月 /
/
一川 一
b
) ・1召
-1
a价“、
+1/_
?
aJ
//ma
t{ 十1乙? '
P.一月
P
s
-
Iama
%
I [amax
+ i
Fi
g.2.
4 As
ymmetri
c
motor
con
tro
led
by
a
cr
it
ic
al-
cen
ter
3-
way
val
v
The s
ame
vel
oci
ty-
gai
n cha
rac
ter
is
ti
c
hol
ds
f
or
symmetr
ic
mot
orwith
a
4-
way
v
al
ve
i
f
2P
L/
P
,,
is
re
pl
ac
e
by
P
L
I
P
,,
.
Compar
e d wit
h hy
draul
ic
te
xtb
ook s
' a
nd manufact
urer
s'
cat
al
ogs,
our
pr
esenta
ti
on b
y means
of
vel
oci
ty
ga i
n ch
arac
ter
ist
ics
i
s r
ather u
nusual.
Normall
y "va
lve
chara
cter
is
tics"
ar
e pres
ent
ed,
sho
wing
the lo
f w as
a
funct
ion
of
t
he l
oad, t
he
valve
di
splac
ement
bei
ng a
par
a met
er.
Us i
ng
Eqns.
(2.
3)
an
d (2.
7 ),
t
he v
alve
char
act
eri
sti
cs of
open
-and c
rit
ica
l-
cent
er-
valve
-contr
oll
ed
mo t
ors
ha
ve
bee
n work
e d out
and a
re
shown i
n Fig.
2.5.
These
chara
cte
ris
ti
cs
ma k
e sens
e ,
as
t
hey s
how t
h e
c
apa
cit
y
of
t
he
v
alv
e wi
th
r
esp
ect
t
o
mo
tor
lo
f w
y
A a
nd
t
he
l
oad
p
res
sur
e凡.I
n
pr
inc
ipl
e,
t
her
e
is
n
o diferen
ce bet
ween t
he two t
y p
es o
f cha
ract
eri
st
ics
, becau
se bot
h cover
exact
ly
the
same
b
ehavior
.
But,
rom
f t
he c
ontr
ol
point
of
v
iew t
he
vel
ocit
y gai
n ch
arac
teri
st
ic
i
s f
ar mor
e i
mport
ant
,
a
s i
t
dire
ctl
y i
ndic
ate
s par
ame t
ers
th
at
gove
rn t
he
contr
ol e
rro
r and
t
h e
dyna
mic pro
peti
r es
o
f t
he
s
yst
em.
To
i
nve
st
ig
ate
a
h
ydr
aul
ic
s
erv
o i
n
pr
act
i
ce,
on
e s
hou
ld
me
asu
re
f
r
ist
o
f
al
l
t
he
v
elo
cit
y
ga
in
K.,
whic
h is
prop
ort
ional
t
o t
he
slope o
f t
he
str
aight
i
lne
s i
n t
he v
e l
oci
ty
gai
n c
harac
ter
ist
i
c.
The
curv
es
of
t
he valve
cha
ract
eri
st
ic
s o
f Fig.2
.5
do not
indi
cat
e d
ire
ctl
y par
amet
ers
si
gniic
f ant
f
or
the
co
ntr
ol
er
ror
and
the
dynami
c s
yst
em be h
avior
.
Ir
regula
rit
i
es
in
t
he ve
loci
yg
t ai
n
charac
ter
ist
i
c,
suc
h a
s
・
dec
re
as
e
of
Km=妙/
ax
a
ro
und
x
=0,
du
e
to
i
mpe
rf
ec
t
t
ur
bul
en
t
po
tf
r l
ow
(
ca
use
d
by
we
ar
o
f
37
,
护一
见片
p.
夕;俘弓‘
1、
勺勺飞、‘又』分 奋 :气七‘
、,.,勺‘
右冬、常
a
门 1
l
下
吸
r
一
//0//
de
了 一
l
lun
/
ma
/
a
/奋今!?/ 」
s
}
o
/一
e t
爪
du
Km
/
of
ics
哟0,{ 咬0・
。
t
e
今乡
「
is
一r
! 叶
er
act
as
— 一—
心 ・
t’q 一
trl
今i
e
I、1
char
r
、,之冬』趁,
0之、
//
n nc
ve
i
和,
38
y
n
下
0
下
吸0 1
a
Val
or
b
5
工//
d
2.
f
斗八
g.
e
Fi
t
含
1仁Ik门以犷而
nsa
丁一
“
’
二
mpe
曰
红
s
o
/
c
尸勃 、‘
、
卜仆 1
 ̄ 一
s
一
一
r
补
e e
l
/
了/产
一
gs)
e mo
, ro
一
1 , 一 一
a
ln d
d
e
.dead
zone,
due
t
o Coul
omb
f
rict
ion,
・l
oca
l
ju
mp
i
n
K,
d
ue
t
o
ba
ckl
as
h i
n
th
e s
erv
ova
lve
f
eedb
ack
pa
th
(
see
s
ect
io
n3.
6)
s
hou
ld
b
e me
asu
red
c
are
ful
ly
, be
caus
e i
mpe
rfe
cti
ons
c
oul
d d
ete
rio
rat
e t
he
c
ompl
et
e s
yst
em,
i
ncl
udi
ng
i
ts
dyna
mic
behavi
or.
Fi
nal
l
y,
we
s
umma
riz
e s
ome
i
mpo
rta
nt
d
ife
ren
ces
b
etwe
en
o
pen
- a
nd
c
rit
ica
l-
c
ent
er
v
alv
e
c
ont
rol
led
a
symmet
ri
c mot
ors:
(
1)
Th
e o
il
c
ons
ump
tio
n of
a
c
rit
ic
al-
cent
er
v
alv
e i
s p
ropo
rti
ona
l t
o t
he
mod
ulus
o
f t
he
mot
or
s
pe
ed.
QS
P
p
ly=0
.5
AJy
J.S
o
at
st
an
d-
st
i
ll
no
o
il
su
ppl
y
i
s
ne
ce
ss
ay.
rAt
an
o
pe
n-
ce
nt
er
va
l
ve
t
h
e
oi
l
co
ns
um
pt
io
n
i
s
a
l
mos
t
i
n
de
pe
nd
en
t、t
h
e
mo
t
or
s
p
eed
j
a
s
l
on
g。、
}
习<U
.
I
n
a
pp
l
ic
at
i
on
wit
h o
ne s
ing
le
se
rvo-
mo t
or
fe
d by
a
c
onsta
nt
deiv
l e
y
r pump,
t
his
i
s h
ard
ly a
d
rawba
ck
bec
aus
e
t
he
energ
y wil
l
be
dis
sip
ated
any
way,
ei
the
r i
n t
he
ser
vova
lve
or
i
n t
he pr
ess
ure
co
ntr
ol
v
alv
e
k
eep
ing
P5
=c
ons
ta
nt.
(
2)
Th
e v
elo
ci
ty
g
ain
Km=a
yla
x o
f a
n o
pen
-ce
nte
r v
alv
e o
per
at
in
g i
n t
he
u
nde
rl
ap
r
egi
on
(
1
x1<U)
is
tw
ic
e
as
la
rg
e
as
Km
o
f
a
c
ri
t
ic
al
-c
en
te
r
va
lv
e.
(
3)
Wi
th
r
ega
rd
t
o t
he
l
oad
s
ens
it
iv
it
y t
he
c
ri
ti
cal
-ce
nte
r v
alv
e i
s s
upe
ri
or
a
s t
he
hy
dra
uli
c
s
t
ifn
es
s
c.=P
S
A1
J
x},e
sp
ec
ia
ll
y
ar
ou
nd
x=0,i
s
fa
r b
et
er
t
ha
n i
t
s c
ou
nt
er
pa
rt
,
t
he
o
pen
-c
ent
er
v
al
ve
h
avi
ng
c
h=P
S
A/
U.Ho
wev
er
,
t
he
o
pen
-ce
nt
er
s
ys
te
m b
eha
ves
p
erf
ec
tl
y
l
in
ear
ly
a
roun
d x=0.F
or
t
hi
s
re
aso
n t
he
l
oad
e
rr
or
c
oul
d
be
c
ompe
nsa
ted
by
f
eed
for
war
d,
a
s
i
s shown
i
n t
he
next
sect
ion.
2.
3 LOAD-
COM PENSATION
Th
e l
oad
e
rro
r。=艺F/
ch
i
n a
n o
pen
-ce
nte
r-v
alv
e-c
ont
rol
l
ed
mot
or
c
an
be
exp
lai
ned
p
hys
ic
al
ly
.
In
tr
odu
ci
ng
a
l
oad
YF
i
n F
ig.
2.
6,
t
he
p
res
sur
e只wi
l
l r
is
e i
n o
rde
r t
o ma
int
ai
n
e
qui
l
ibr
iu
m o
f
fo
rce
s.
Co
nse
que
ntl
y
th
e p
res
sur
e
dr
op
PS一Pl
a
cro
ss
t
he
i
nl
et
p
ort
wi
l
l
de
cea
r s
e
a
nd
s
o d
oes
t
he
i
nl
et
low
f QI
.On
t
he
o
the
r h
and,
Q2
wi
ll
r
is
e owi
ng
t
o i
ncr
eas
ed
p
res
sur
e
dr
op
P,
a
cro
ss
t
he
o
utl
et
po
t.
r S
o wi
th
Q,<Q2,
the
s
yst
em
wi
ll
ge
t
a
s
pee
d,
un
les
s
a
c
han
ge
i
n
al
an
d
a2
c
omp
ens
at
es
f
or
t
he
lo
f w
du
e t
o
the
l
oad.
As
al一a2
u
nti
l
no
w h
as
b
een
pr
opo
rti
ona
l
to
x
a
nd
£,t
he
i
mpl
ic
aio
t n
i
s a
c
ont
rol
e
rro
r.
F
rom
t
his
d
iag
nos
is
a
t
her
apy
c
oul
d b
e f
oun
d.
A
c
han
ge
i
n
t
he
p
res
sur
e只 (
re
pre
sen
ti
ng
t
he
l
oad
) mus
t
ca
use
a
c
han
ge
a,
an
d /
or
a2
i
nde
pend
ent
o
f t
he
c
ont
rol
e
rro
r s
ign
al
E.
Fro
m Eq
n.
(
2.2
) be
ing
de
riv
ed
wi
tho
ut
r
est
ri
ct
io
ns
39
m em brane
a
1
2
( el
ear
2-
tr
.
a
p h
}
n
l
hl
#F
穷碑次欢欢长长州
..目口. “ .一
Fi
g.
2.
6
As
ymme
tr
ic
l
oad
-comp
ens
at
io
n
Th
e i
nle
t
ope
ning
s o
f t
he
s
yst
ems
a
) a
nd
b
) a
re
i
nf
lue
nce
d by
t
he
c
yli
nde
r p
es
r s
ure
P
,=0
.5P
,+尺,c
aus
in
g
c,
,=0
0
40
P
S
a
1
-
2
P
P
S
L一1
+
2P
S
L
_._
C,
“
y = 一下一
A
c
onc
eni
rng
a
,
or
a2
,it
ca
n b
e
se
en
t
ha
t
by
ma
kin
g a
,
so
lel
y
de
pen
den
t
on几a
nd
a
2
onx ,as
s
hown
i
n Fi
g.2.
6,
t
he
f
irs
t or
der
e
fect
of
t
he
l
oad
coul
d be
c
ancel
led.
Li
nea
ri
zi
ng
a
rou
n
y
一
CA
d
丹。(
1
・
2P
S
L
)
1
-
d几 =0
a
2
P
S
L
n
一(
1
一
U
d
)
1
+
P5
L
x=0
r
e
・
sul
t
s i
n
c,=0
0.Ex
amp
les
o
f l
oad
-co
mpe
nsa
ted
(
2.8)
sys
tems t
hat
hav
e been o
per
ati
ng s
ucces
sfu
lly a
re s
hown i
n Fi
g.2.
6.
Fig.
2.
6 a) i
s part
of
a
numeri
cal
ly
cont
rol
led
machi
ne,
b)
of
a copyi
ng l
at
he (o
ne
pot
r onl
y) a
nd t
her
efore not
f
ull
y
eff
ect
ive
at
la
rge l
oads
, a
s c
an be
en s
een ro
f m the
"loa
d c
hara
cter
is
ti
c" of
Fi
g.2.
6 c )
wher
e
2P
L
/Pr
is
p
lo
tt
ed
a
s
a
f
unc
ti
on
o
f x
l
U
a
t y=0.
Sub
st
it
ut
io
n
of
y
=0
i
n
Eq
n.
(
2.
8)
y
ie
ld
s
(
2
PT
L
)
2・
‘,
一
音,
’一
‘
・
On
ly
a
t s
mal
l
lo
ads
i
s t
he
s
lo
pe
o
f t
he
c
har
act
er
is
ti
c i
nfi
ni
te
, c
aus
ing
c,
,=0
0・Sy
mme
tr
ic
l
oad
-co
mpe
nsa
ti
on-t
hou
gh
me
cha
nic
all
y mo
re
c
omp
lic
ate
d-g
ive
s b
ett
er
l
oad
c
har
act
eri
st
ics
.
He
re
t
he
t
wo
o
pen
ing
s a,
an
d a2
bo
th
de
pend
o
n尺 a
s
we
ll
a
s
on
x.
The
op
ti
mal
de
sig
n es
r u
lt
s
in
41
刀己。广r
'az'
e
一UL
LOC
pt'
PSS
ua'
e
plot
pre
ssur
e
一,
W
X1
_Y
}
y
'
Ll -
i
一 ,
t
c
L-
一 S
t
t
cul
pp
ar
oYt
em b
rc
ut
P
一 nt
"nt
brane
瓢黔
t
7
J 1
0
一
’. L\ I t
一 -一J
弓
二U
0-
ivt
g
M OM "
P
如J
、
c
)
Fi
g.2.
7
Sy
mme
tri
c
lo
ad-
comp
ens
ati
on
c
、阿「 /, x.凡 、 x 凡、
(
2.9)
— !aol1十丁丁十 一
Y = 一广 ,} 二一) 一ao
l1一 丁丁一 ̄二一J
A
VPL U
PS U
PS
Fi
g.
2.
7
a)
i
l
lus
tr
at
es
t
he
d
esi
gn
c
onc
ept
.
By
s
upe
rpo
sit
io
n t
he
i
nf
lu
enc
e o
f b
oth尺 a
ndx on
al
a
nd
a2
i
s
efe
cte
d.
No
w c
omp
ens
ati
on
i
s
ef
fec
ti
ve
f
or
l
ar
ger
l
oad
a
lso,
as
ca
n b
e s
een斤'
OM
t
he
Fi
g.
2.
7
b)
,
de
ri
ved
f
rom
Eqn.
2.
9)
wi
th夕=o
x 1
一0.
5
(2P
L
I
P
,)
’一了
1一(
2P
L
/P
s)
’ 2只 、
, 1
(_ 厂+丁丁(
,
2只 、;
二 二)一+.
…
U 2
PL/
P 8
代 16 代
A
v
ery
c
ompa
ct
e
xamp
le
o
f a
s
ucc
ess
ful
a
ppl
i
cat
ion
i
n p
rac
tic
e i
s
sho
wn
i
n c
) wh
ere
a
c
irc
ula
rly
s
uppo
te
r d
le
f xi
bl
e
membr
ane
bet
wee
n t
wo
noz
zle
s a
cts
ik
l e
con
tr
ol
la
f pper
.
The
membr
ane
a
nd
t
he
2
noz
zle
s
in
cor
por
at
e
th
e s
eco
nd
st
age
o
f
an
open
-cent
er
3
-way
va
lve.
The
membr
ane
a
nd
t
he
po
rt
op
eni
ngs
a
re
i
nf
lue
nce
d ir
fstl
y b
y t
he
p
ilo
t s
tag
e p
res
sur
es
o
f t
he
s
erv
ova
lve
,
an
d s
eco
ndl
y
by
t
he
c
yi
ln
der
p
res
sur
e P=O
.SP+尺 a
ct
in
g o
n o
ne
s
id
e o
f
th
e me
mbr
ane
a
rou
nd
t
he
n
ozz
le.
Th
e l
oad
pr
ess
ure
ma
y c
han
ge
t
he
i
nl
et
a
nd
o
utl
et
p
res
sur
e d
rop
s,
bu
t a
t
th
e s
ame
t
ime
t
he
i
nl
et
a
nd
o
utl
et
o
pen
ings
a
re
i
nlu
f e
nce
d i
n t
he
o
ppo
sit
e d
ire
cti
on.
He
nce
, wi
th
a
pr
ope
rly
d
esi
gne
d
me
mbr
an
e,
t
he
i
nl
et必a
s
we
ll
a
s
th
e o
ut
le
t Q2
ar
e
al
mos
t
i
nd
epe
nde
nt
o
f几,P
or
E
F.I
n
t
he
f
in
al
d
esi
gn,
th
e c
ir
cul
ar
s
uppo
rt
o
f t
he
me
mbr
ane
h
as
b
een
e
st
abl
is
hed
b
y s
tee
l
ba
ll
s,
l
eaka
ge
be
ing
pr
even
ted
b
y o-
ri
ngs.
I
n a
ppic
l a
ti
ons
i
nvolv
ing
al
i
nea
r e
lec
tro
-hyd
rau
lic
s
tep
ping
mot
or
a
nd
f
or
a
co
pyi
ng
s
hap
er,
42
c
are
ful
d
esi
gn
o
f t
he
l
oad-
comp
ens
ate
d 3
-wa
y v
alv
e r
esu
lt
ed
i
n
a
me
mbr
ane
wi
th
a
d
iame
ter
价26
mm
and
a
thi
cknes
s 0.
9 mm.
2.
4 SYMMETRIC
M OTOR
CONTROLLED
BY
A SYMMETRIC
4-
WAY
VALVE
As t
her
e i
s a
n al
mo s
t
perf
ect
a
nal
ogy
wit
h a
n sy
a mmetri
c
moto
r c
ontr
oll
ed b
y a
3-
way
va
lve,
d
eriv
ati
ons
devel
op s
imil
arl
y.
Usi
ng
t
he
too
ls
of
Sec
tio
n 1
.8,
t
he
be
havio
r of
t
he
symmet
ri
c
motor
o
f Fi
g.2.
8
is
def
ine
d勿
(
P,一P
2
)A二1
:
F二FQ+F
,+W}+M},
。一。{
2(
P
,
P
,一。F
}
2P
,
o
_
p
i
一A
E
4
',
、
一
_
a3
。
C
d
F
2
P2
一C
d
2
(
P
S
P+
-
P2
)A
E
LZ
“・
As
sumi
ng
a
s
ymmet
r
ic
va
lve(
a,=a3
,
a2=a4)and ap
ist
on pos
it
io
n wi
th几 =Lz=L,
we
f
nd只+凡二只.
i Comb
ini
ng
t
hi
s
wi
t
h
P一P 2二Y
,F /A,we f
ind
凡-
艺-
F -
1 -2
一伪‘
J
只 =生+IF二1 +立
P
z=生_
!Ps
一
纽
一
一
尸目
PS
2
2A只 2
2Ps Ps
2
He
re
a
gai
n,
,F
Y
/A
i
s
t
he
l
oa
d
pr
es
su
re几.
Subs
tit
uti
on
yi
elds
C
a阿「 厂
y =-.、
I
甲可
—
I马 ,
I
〔
甲可飞 L二
i一 下于 一a,jt十二升 !
一二一
二料 . (
2-10)
A V
P
L V
r, 一V S」 2E -
Ex
cep
t f
or
a
f
ac
to
r
2
i
n
th
e l
oa
d p
res
su
re几=
=E
F/
A,E
gn.
(
2.1
0)
i
s
id
ent
i
cal
t
o
Eg
n.(
2.
2)
.
Co
nsequ
ent
ly
,
3-way
-va
lve-c
ont
rol
l
ed as
ymmetr
ic mo
tor
s a
nd
4-way-
valve
-con
tro
ll
ed
symmet
ri
c
mot
ors
b
eha
ve (
in
a
s
tat
ion
ary
co
ndit
ion
as
we
ll
as
dyn
amica
ly)
e
x a
ctl
y t
he
same,
p
rovi
ded
t
he
}Y L2
Fi
g.2.
8
Sy
mme
tri
c
mo
tor
c
ont
rol
le
d b
y a
4-
way
v
alv
e
asymmetr
ic mot
or
has
a pi
st
on a
rea
and
a s
pool
v
alv
e d
iamete
r t
wice
as
l
arge
as
sy
mme t
ri
c motor
and pr
ovided the pi
stons ar
e h
alf
-way a
longthei
r cyli
nders.
As 3-
way-
val
ve-co
n t
rol
led
symmet
a r
ic motor
sa re more
simple
and e
asi
erto manufa
cturetha
n 4-way-
val
ve-
control
l
ed
s
y mmetr
ic mot
ors i
s s
urpri
si
ng t
hat
asymmet
ri
c mot
ors
are
so se
ldom us
ed.
They
des
erve
mo re
po
pul
ari
ty.
Co
nse
que
ntl
y,
i
l
nea
riz
ati
on
a
nd
d
eri
vat
ion
of
s
yst
em
par
ame
ter
s f
oll
ow
e
xac
tly
t
he
s
ame
in
l e
s
a
s i
n s
ect
ion
2.
2.
Fo
r a
s
ymme
tri
c
cr
it
ic
al-
cent
er
v
alv
e we
c
an
wr
it
e
x> 0 a
l=。
。=4
a-
I二二b
x a2=a4=0,
x.
x<0 .a___
.
al=a3=0; a2“a4二一!
-- I
x二一b
x.
xn
1a
x
S
ubs
ti
tut
i
on
i
nt
o
Egn
.(
3.1
0)
y
iel
ds
Li
nea
rizat
ion
gi
ves
夕
一
A
b。-
P
P
S
x
1
-}
X
x
I
P
L}
一
2
LE
人・
wh
ich
c
an
b
e b
rou
ght
in
to
t
he
s
tan
d
夕
・
_
A
b
a
C
c
・
d
r
-
P
d
x
-
f
o
2
1
x
P
I
r
几
{
一
m
_
2
L
E
_P
L
、崖
r‘“
・F 一
。「 Y川
一co
一
Y二A司x一. }
L
Ch」
wher
e
凡-
。二
,‘ 一
b_ I
-
P 2五A
阮.
,
入 ,= 一 C、_}
一二 c,
,= c = — _
(
2.1
1)
A “VP
U
-
L
As
c
oul
d
be
p
red
ict
ed,
th
e v
el
oci
t
y g
ain
K,
is
e
qua
l
to
i
t
sco
unt
er-
par
t
fo
r a
n sy
a mme
tri
c mo
tor
,
whi
l
e c
.
an
d c
o
ha
ve
b
een
d
oub
led
(
see
Bqn
.
(2
.7)
).
2.
5 ASYMMETRIC
MOTOR
CONTROLLED
BY
A SYMMETRIC
4-
WAY
CRITICAL-
CENTER
VALVE
I
t wil
l
be
prove
d he
re
that
a
n a
symmet
ri
c l
inea
r mot
or
i
s i
nco
mp a
ti
blewi
t
h a
symmetr
ic
4-
wa y
c
rit
ical
-ce
nte
r v
alve,
even
under
n
o-l
oad
c
ondit
ions.
Ac
cor
ding
t
o Se
cti
on1.
8
and
Fig.
2.
9 we
f
ind,
f
or
the
asymmetr
ic
motor
,
th
at
君A;
一PZ
A2=IF二Fe+F
,十砂 +My,
夕
=a
A
,
Cd
2
(
P
s
p一
-
P,
)a
A
2
,
Cd
2
p
P
,一
4
EF
,
科
。
,_ 2
P2。
、_ 2
(P,
-
P
2).
L2。
Y二T Ud,
l— 一下万户
Ud、
{一一一-- t0 '2・
A2
V P
A2
V
P
c
Th
e i
mpl
i
cat
io
ns
o
f
a
s
ymme
tr
ic
4
-wa
y c
ri
ti
cal
-c
ent
er
v
alv
e a
re:
Ei
t
her
a,
=a3>0
wi
th
a2=a
4=0
o
r a
,=a
3=0
wi
th
a2=a4>0
.
L
山
,
L1
州件 一 一  ̄ ,.,
址(
气)
一(
a3{
「a
"
了 可 }Q3瓦 PS
P
S !’ !
Fi
g.2.
9
As
ymme
tri
c
mo
tor
c
ont
rol
le
d b
y a
4-
way
v
alv
e
T
he
p
res
sur
es只 a
nd凡 wi
l
l b
e
c
al
cu
lat
ed
f
or
t
he
s
imp
le
s
it
uat
i
on
I
F=0
a
nd
a,
,a2,
a3,
a4,
只an
d凡 a
re
a
ll
c
ons
tan
ts.
At
a,=a3>0
a
nd夕>0
t
he
p
res
sur
es君 a
nd几 c
an
b
e f
oun
d ro
fm
a2=a4>0.a
At nd夕<0
t
he
p
re
P
,
A
s
s
,
一P
2
A
ur
e
2
an
d
s
c
a
A
'
.
P
n
S
-
P
一
be
f
o
a
A
u
3
2
n
P
2
d
.
ro
f m
气一
n
a
厂丁-
“, 「,,--一 ̄竺一
-
d
只A,二凡A2
.
\
/P二A ̄
}
P
s一P・
“ 2
The
r
esul
ts
ha
ve
bee
n s
ummar
ize
d i
n t
he
t
abl
e (
2.1)
.
To
s
how
t
he
a
sto
nis
hing
e
fec
t o
f t
he
s
pee
d夕o
f a
n u
nlo
ade
d s
yst
em
o
n i
ts
p
res
sur
es,
i
n Fi
g .
2.1
0
P
a
nd
P
Z h
ave
b
een
p
lo
ted
f
or
Aa
/AZ=2
a
s
a
f
unc
ti
on
o
f
th
e
pi
st
on
s
pee
d
y.
Th
e f
oll
owi
ng
c
onc
lus
ion
s c
an
t
her
efo
re
be
d
rawn:
(
1)
E
xce
pt
f
or
A,
/AZ=1
,th
e c
omb
ina
ti
ons
s
how
e
nor
mou
s p
re
ss
ur
e
ju
mps
a
ro
und
y=0,
c
aus
ing
"
impl
osi
ons
" o
r "
exp
los
ion
s"
o
f o
il
o
wing
t
o i
ts
c
ompr
ess
ibi
it
l y
.
Con
seq
uen
tly
s
moo
th
o
per
at
ion
a
rou
nd夕=0
i
s
impo
ssi
bl
e.
(
2)
At
A1
/AZ
>1.
71
t
he
pr
ess
ure
s c
anno
t s
tay
wi
thi
n t
he
s
afe
ty
r
ang
e 工P 、P、旦
6 '
6
P
e
ve
n
u
nd
er
n
o-l
oa
d c
ond
it
io
ns.
Co
nsi
der
i
ng
t
he
o
ppo
si
ng
e
fec
ts
o
f t
he
l
oad
_F
Y o
n君a
nd凡,
mot
or
s
wi
t
h A,
/AZ<1
.71
c
ann
ot
o
per
at
e
wi
t
hin
a
s
af
e
pr
es
su
re
r
an
ge
e
it
he
r.
45
(
3)
Fo
r a
ccu
rat
e a
nd
s
moo
th
c
ont
rol
t
he
i
nco
mpa
tib
ili
ty
of
a
s
ymme
tri
c v
alv
e a
nd
a
n a
symme
tri
c
mot
or
i
s c
lea
r.
To
a
voi
d p
res
sur
e j
umps
a
nd
e
nsu
re
s
moo
th
o
per
ati
on
a
n a
symme
tri
c v
alv
e
(
nor
mal
ly
no
t a
vai
la
ble
) mu
st
b
e u
sed
, wi
t
h p
ort
o
pen
ings
p
rop
ort
ion
al
t
o t
he
c
orr
esp
ondi
ng
pi
st
on
a
rea
's
仇"
-
n
a
a,二a
3
d
人
a4
一
一
(
2.1
2)
Al
AZ A2
Tabl
e 2.1
}
}
IF=
,
0
P=P
‘
=。 { A,,}_
I — = 1
11.
}
71 }2 生 =1 1.71 2
} }A,
I } 只=P,=0 AZ
夕>0 二= 式
a,二a3>0
a,=a,=a
P,
,
+(
会)
’ 0.
50 0.
167
10.
111 凡_
只 . ,
A,
A,、
、
1十 t一一 )-
0.
50 0.
285 0.
222
A,
夕<0
a,=a3=0 只 l
A,
)
.
, 0.
50 0.
487 }0.
444 二二
立 0.
50 0.
833 0.
889
a==a,>0. P,
,
・(
A
,
)
3 P
S,・
(
A}
z
),
_
一t1
v
旅 14
1.-一
只
Jump
比.
,
t1,=(
,b`1
t)t) -钊
r_
1
,
.1
1
1)
1
17
!
P
,
1;
备 儿
叼
1
名二1
1
,
i31
,
'
‘日
‘.
一一一一一一一一一l
;r
1. 喇 乡
印丫万
且
Y
1
,>0
电
,
’<0
Fi
g.
2.
10
J
ump
s i
n
th
e c
yli
nde
r
pr
ess
ure
s
ar
oun
d夕=0,
due
t
o t
he
sy
a mmet
y
r o
f t
he
mo
tor
(A,
:
?
-A
L Z)
.
Thi
s combi
nat
ion
wi
ll
be
a
nal
yze
d i
n s
ect
ion
2.
6.
I
n t
he
for
ego
ing
an
aly
si
s i
t
has
b
een
as
sume
d t
hat
艺F=0. I
t c
an
be
sh
own
esi
aly
tha
t
at
IF=c o
nst
ant
,
on
ly
t
he
lo
cat
io
ns
of
t
he
p
res
sur
e
ju
mps
wi
l
l
ch
ang
e,
but
no
t
th
eir
ma
gnit
ude
s.
2.
6 ASYMMETRIC
M OTOR
CONTROLLED
BY
AN ASYM M ETRIC
4-W AY
VALVE
S
inc
e as
ymme
tri
c v
alv
e i
s i
nco
mpa
ti
ble
wi
th
a
n a
symme
tri
c mot
or,
a
n sy
a mme
tri
c 4-
way
46
么
.一人
口4
一
c
ri
t
ic
al
-c
ent
er
v
al
ve
i
n w
hic
h丘 =五 a
nd
一
wi
ll
be
a
nal
yze
d he
re.
Ac
cor
ding
t
o
A,
A2 A2
Fi
g.2
.9,
th
e mo
tor
be
hav
ior
i
s
de
fin
ed
b
y
P
,
A,
一P2
A:二IF二Fe+Fc+My+W}
,
・a
A
,
C
d
2
(
P,
P
-
P,
)
一a
A
2
,
Cd
俘一_
4
E“ (
2.1
3)
Th
e i
mpl
i
cat
ion
o
f a
n sy
a mme
tri
夕
一
c
4
a
A
3
2
C
-
d
d
wa
2
P
2一
a
A
y
c
r
4
2
C
d
i
{
2
(
t
i
P
,
P
-
c
P
2
)
a
l
十
L
E
-
2
“
c
ent
er
v
alv
e i
s:
, ., a,
a
“王 _
“3 、 八 .、 _ _ ___ a,
八 分 2
_ ad ‘__‘
分4 、 n _
:几_ , _ 。 _ 。
n
Ei
ther 一一二.= 一一二 ̄> u wi
th a,= a,
,= u
or 一 =— > u
wit
h a,= a,= u
1.1
" 2
+ ,
A'2
At
L,=L,=L,
the
p
res
sur
es月 a
nd几 c
an
b
e s
ol
ved
ro
fm E
qn.
(2
.13
)
只 A2
”一
艺FI
P¥ P
2
A
,一艺F
,Ps (
2.1
4)
只 A,+A2
P
s
A
,+A2
We
now
int
roduce
t
he
"
cor
rect
ed" t
o
ta
l
l
oa
d叉F
',a
nd
t
h
e l
o
ad
p
re
ss
ur
e
PL
of
an
a
sy
mme
t
ri
c
mot
or,
cont
rol
led
by
a
4-
way
va
lve
又;
一
,=y;一
A,
典P}
一
一
F,
+F,
+My
+Wy
--'
华PS
, 2 一 z
(
2.1
5)
凡=
J
F' (
2.1
6)
0.
5(A,+AZ)
Th
e
di
fer
en
ce
i
n
ar
ea
A,一AZ
"
cau
ses
",
i
n
a
wa
y,
a
c
ons
ta
nt
p
re
-l
oa
d
Fs
r=0
.5(
A,一AZ
)P
S,
s
o
th
e
in
t
rod
uc
ti
o
n o
f艺F
'an
d P
L
ma
ke
s
se
ns
e.
Co
mbi
n
in
g
Eg
ns
.(
2.
14
),
(
2.
1
5)
a
nd
(
2.
1
6)
几一
外el
ds
只一
1 一2
1 一2
I
F' 几 Pz
十
Ps
+
一
一
-
一
(
2.1
7)
只(
A,+A:
) 2
P,’P
S-2
Su
bst
it
uti
on
i
nto
Eqn
.
(2
.13
) g
ive
s
C
d阿[ 厂可 厂万 1 L二 (
2.1
8)
Y= 一丁一,
}— !a,,I
1一二丁
 ̄一a2,
11+.
二一 I
一二.
二几
A,
V
P
L V
r,
V
r」 :
Z
.
Th
is
e
qua
ti
on
e
xac
tl
y e
qua
ls
t
he
s
olu
tio
n (
2.1
0)
d
eri
ved
e
arl
i
er
f
or
a
s
ymme
tri
c mo
tor
wi
t
h a
s
ymmet ic
r val
ve,
s
o f
ur
the
r de
rivat
ions
wi
ll
be
s
imi
lar
.
Subs
tit
uting
a,
a,
l
a,_
。,}x
b, _ a
t x>0
1
‘月 . ‘ 吕....J. . 占 1  ̄ _ , 、
—
沼
= —J =i
l
— i
j
—盛 = —J x and “,
‘
二 “,
,
=V
八, 八, i x __ _ I八 , 八,
凡一
执一
气一
孟
‘ 1 ‘盆目丹 I 五 皿
人 47
人
人
一
一
一
一
一
a,=a3=0
an
d at
x< 0
外el
ds
夕
一
-
A
b,
,
C
d
P
p
s
x
1-
I
x
x
,
pP
i
,
{
一_
2
L
E
P
t
'
,
・A
Cd
P
S
x一
21
X
P
I
s
PL
一2
E
wher
e
夕
一「一F
C
h'
一c
0
,
艺F'
=艺F一0
.5
(A,
一A2
)P
S
K
_一鱼 。,.
Pr
A; 一V
P
,
b,一
b2
A, A2
(
2.19)
E(
A,十A2
)
c
‘二(
A}
j+A2
)'
l
's
co
}
x}
,‘
Comp a
ris
on wit
h t
he
r
esul
t
s of
Sect
ion
2.
4 s
hows
t
hat
the
pa
ramet
ers
of
a
s
ymmet
ric
mot
or
wi
th
a
symme t
ri
c
4-way
v
alve
ar
e v
ali
d f
or
an
as
ymmet
ri
c mot
or
wit
h a
n a
symme
tri
c
4-
way
val
ve
by
r
epl
acing
L
F
by艺F‘
二艺F一
0.5
(A
,一A
2)
Ps
,
A
b
y0.
5(A,
+A2
), (
2.
20)
叼A
b
y
b1
/式二b
2/A
2.
He
re
b=a
al
a
x i
s
th
e wi
dt
h
of
th
e p
oto
r
pe
nin
g a
rea“
・
2.
7 PUM P-
CONTROLLED
HYDRAULICS
Fig.2.1
1 shows const
ruct
ion deta
ils
of a pract
i
c a
l syste m. The pump d
isplacement conro
t l
shown t h
ere us
es manual
input (
cont
rol h
a ndl
e) t
o a sma ll val
ve-c
ontroll
ed servo s
ystem, and
me c
h anic
al f
eedback ro
f m the
pu mp swash-pl
ate
a n
gle. Ei t
her o
r both t
he input
and feedback
fu
n cio
t ns ca
n be
perf
ormed el
ecri
t cal
ly
in
the
e l
ecro
t hydraulic s
ystems
o f
th
is
type.
We
now st
udy i
n more d
etai
l
v a
lve-c
onro
t l
led hy
d r
aul
ic s yst
ems such as
th
at used t
o s
tr
o ke
the
pump in many pump-contr
oll
ed syst
ems an
d for
the d
ire
c t
c onrol
t of l
o ad
motio n
i
n my r
iad ot
her
appl
ications
.
Wh ereas
in
pump-contr
oll
ed s
ys t
ems the
fl
uid powe r
su
pp ly
must be
incl
uded in
the
syst
e m mo d
el,
anal
ysis
of va
lve-c
on t
rol
led s
ystems can
p roc e
ed wit
ho ut
consider
ati
on of power
s
upp
ly
d
eta
il
s
if
on
e a
ssu
mes
t
he
e
xis
ten
ce
o
f a
p
owe
r s
our
ce
o
f c
ons
tan
t s
upp
ly
p
res
sur
e
p,
,
i
rr
esp
ect
iv
e o
f low
f de
man
d.
Po
wer
s
uppl
i
es
t
hat
a
ppr
oxi
mat
e t
hi
s b
eha
vio
r a
re
a
vai
la
ble
i
n
s
evera
l di
ferent
f
orms that
tr
ade of
f comple
xit
y,
cost,
eff
ici
ency,
and s
tat
i
c/dy
namic
pr
essur
e
r
egul
ati
on ac
curac
y.
Fi
g.2.12a
) s
hows t
he si
mples
t
supply
,
the
reli
ef-
valv
e t
ype.
Th
e s
pri
ng-
load
ed
r
eli
ef va
lve i
s compl
etely s
hut u
nti
l p
res
sure
rea
c h
es the l
ow end
of th
e re
gul
ati
ng
range,
wh
ere
upo
n i
t
op
ens
s
uff
ic
ient
l
y t
o
by
pas
s
an
y pu
mp
low
f n
ot
r
equ
ire
d b
y t
he
s
erv
ova
lve
.
The
lu
f i
d
po
wer
(
vol
ume
low
f r
ate
t
ime
s r
eli
ef
v
alv
e p
res
sur
e d
rop)
o
f t
he
b
ypa
sse
d lo
f w
i
s
co
mpl
et
ely
l
ost
48
匀欲1
11
)‘
飞芍 、
1乙之
国二 瞬口
1
)、1
-定 P灿七 七1巴
Fi
g.2.
11
Co
nst
ruc
tio
nal
d
eta
il
o
f p
ump
-co
ntr
oll
ed
h
ydr
aul
ics
49
r,
-一 _;_
_ t_刁一 一 ̄一 Yet.cl
ue
一T
RI
g
ul
an
ng
e
g
n
a
r
卉
Loud
f
?oi
+
Bi
pas
s
fl
ow,
QB
I
`4
-一一一一 CI
Re
lie
f Lul
ve 一 一 ̄ ̄砂-叫卜 .
o
pelti
ng
与
Pump刀。、
,
a)
b)
c)
Fi
g.2.
12
Powe
r s
uppl
i
es
f
or
v
alv
e-c
ont
rol
hyd
rau
lic
s
a
nd
c
onv
ert
ed
t
o h
eat
,
wh
ich
ma
y r
equ
ire
i
nst
al
la
ti
on
o
f a
n o
il
c
ool
in
g s
yst
em.
Whe
n t
he
s
erv
o
z
ero
ef
fic
iency
.
sy
stem
re
qui
res
no
lo
f w,
all
t
he
p
ump-g
ene
rat
ed
powe
r i
s c
onve
rte
d t
o
he
at,
gi
ving
Suppl
y p
res
sur
e p,
var
ie
s by
abo
ut
13% o
ver
t
he
regul
ati
ngrange
f
or
st
eadycondi
tions
and
r
esp
ons
e t
o t
ran
sie
nt
lo
f w
r
equ
ire
men
t i
s q
uit
e f
ast
; t
hus
t
hes
tan
dar
d a
ssu
mpt
ion
of
c
ons
tant
p,
us
ed i
n s
ervo
-syst
em ana
lys
is
i
s u
sual
ly
re
asona
ble
. Pu
mp
s
ize
(
gal
/
min)
mus
t b
e c
hos
en
t
o
ma
tch
the
l
arg
est
ant
ici
pat
ed
ser
vo-
sys
tem
demand.
an
accumul
ator
i
n
Si
nce
s
erv
o-s
yst
em
lo
f w
dema
nds
a
re
b
y n
atu
re
i
nte
rmi
tt
ent
,
ad
dit
ion
o
f
50
F
ig.
2.
12
b)
a
ll
ows
u
se
of
a
s
mal
l
er
pu
mp
s
inc
e p
art
o
f
th
e r
equ
ir
ed
ra
tnsi
ent
lo
f w
i
s
no
w s
uppl
i
ed
by the acc
umu la
tor,
which
is
recharged duri
ng servo
idle
peri
ods.
Th e
sma lle
r pump wast
es
less
energy duri
ng peio
r ds when a
ll
the low
f is
bypasse
d, inc
reas
ing efici
ency a n
d p
erhaps al
l
owing
deleti
on of
the hea
t excha
ng e
r.
Pump pressu
re pulsat
ions
(re
lat
ed to
pu mp sp
ee d
and numb e
r of
cyl
ind ers
) a
re al
so r
educed by
the accumu l
ator.
Futh
r e r
i
ncrese
a i
n eficiency is
provided
by t
he
unloaded-val
ve type of s
up p
ly o f
Fi g
.2.12 c).
He r
e a
n accumulato
r i
s man d
atory si
nce t
he
unloader va
lve d
isconnect
s
the
pump ro
f m th
e load
wh en t
he
p r
ess
u r
e-sensing i
lne
report
s t
hat
set
pressure
has
been r
eached.
Although the
pump con
tinues
to
pump at
ful
l lo
f w,
li
ttl
e energ
y i
s us
ed
sin
c e t
he chec
k valve
isol
at
e s
t
he p
u mp ro
f m syst
em pre
ssur
e and t
he unloader val
ve provi
des
a
bypa ss
to t
he re
servoi
r at
near
ly zero gauge pres
sure.
Unloader
v a
lves have a regu
lati
ng ra
nge
si
mi lar
to
rel
ief
valves,
t
he pump cycing
l bet
we en lo
a d
and unl
oad to
ma intai
n syst
em pr
essur
e
ne
ar
t
he
s
et
po
int
.
Fi
g.
2.
12d
) s
hows
a
pr
ess
ure
-co
mpe
nsa
ted
pu
mp
t
ype
of
s
uppl
y.
He
re
a
v
ari
abl
e-d
isp
lac
eme
nt
p
ump
u
ses
a
pr
ess
ure
s
ens
ing
s
che
me
t
o a
djus
t
it
s di
spl
ace
men
t s
o a
s t
o
ke
ep
s
yst
em
p
res
sur
e
ne
ar
t
he
s
et
va
lue
a
nd
s
uppl
y
on
ly
t
he
p
ump
low
f n
eede
d by
t
he
l
oad
,
gi
vi
ng
hi
gh
ef
fi
cie
ncy.
2S
FROM
MATHEMATI
CAL
MODEL
TO
TRANSrtR
FUNCTI
ON
The
ma
the
mat
ic
al
mo
del
de
riv
ed
f
or
v
ari
ous
hy
dra
uli
c
se
rvo
sys
te
ms
r
ela
tes
t
he
s
pee
d夕t
o
th
e
v
al
ve
di
s
pl
ac
eme
nt
二an
d
t
o
t
he
to
t
al
l
o
ad艺F
丫
F 丫户
夕=Km
(
x一Ld-
一)一通已一.‘
Ch
CO
As
i
ndi
cat
es
i
n
Fi
g.
2.
13
t
he
i
neti
rlf
a
or
ces
My a
nd
t
he
v
isc
ous
f
ri
ct
io
n砂 c
an
b
e d
eiv
r e
d
mo
tor
d
is
pla
ceme
nt
y
.
The
e
xte
na
r l
f
or
ces
F
e a
nd
t
he
Co
ulo
mb
f
i
rct
i
on
Fc
c
an
b
e
di
stur
banc
e s
igna
ls
=Fe+F,+My+吟 .
API
=刃
F 凡 十F,
十
十
J
I护十”.
c
el
l C口
+ K,
,! 十
+
Fi
g.
2.
13
Bl
ock
di
agr
am
f
or
"
phys
ica
l"
mod
el
o
f h
ydr
aul
ic
s
erv
o
Th
e v
el
oci
t
y g
ain
Km,
th
e h
ydr
aul
i
c s
t
ifn
ess
c
h a
nd
t
he
o
il
s
pin
rg s
ti
fne
ss
c
o d
eri
ved
f
or
v
ari
ous
val
ve-
mot
or
combi
nat
ions
ar
e s
umma
rize
d i
n t
abl
e 2.
2.
As
i
ndi
cat
ed
i
n F
ig.
2.
13,
th
e c
ont
rol
lo
op
i
s
cl
ose
d by
51
E二 r一 y (
2.23)
d
an
x=K,
G,
(s)
E (
2.24)
.
th
wl
卿GS
(s
)=1
Tab
le
2
.2
S
yst
em
pa
rame
ter
s
Km,
Ch
a
nd
c
..
f
or
h
ydr
aul
i
c s
erv
o's
As
ymme
tri
c (
A,=2A2
=A) Sy
mme
tri
c
(Al
=A2
=A)
val
ve 3-
way 4-
way
Cent
er Open Cri
tical Open Cri
tical
K, 2
b_厉 b_厉 2
b 厉 b
(
p
— C、.{
一二一 一 C、、I
= — C、‘I 一 C.
,_】
一二
A 一VP: A 一VP: A 一VP: A “丫P
PS
A 2P,
A 2P,
A
Ch
P
S
A 一
U
,}
一x
I
i U
1
xI
x
Co E
A>E
A 2
EA>兰
EA
L
S L
S
K,
G,
(s)
=xl
e i
s t
he
t
r
ans
fe
r f
unc
ti
onof
t
he
mechani
sm (
for
i
nst
anc
e a
leve
r or
an
e
lec
tro
-hy
dra
ul
ic
s
erv
ova
lve
) t
ra
nsf
ormi
ng
t
he
e
rro
r s
ign
al
E
i
nto
t
he
v
alv
e d
ispl
ace
men
t
x.
Combi
nat
ion
of
t
he
above
equa
tions
yi
elds
IF
+K
.
I
F
+夕
=K }
KSGS
E
CO Ch
F_+ F_ F p
'
十P C
竺 ;
十(M
K ,
十-
w )y
、(1
十w
.
_
m)
K y
,‘
5
、
、.了
=K_K_
(G}
E一一
一乞一一二 )一 (
2
.
CO
Ch CO
Ch K.
r
ch CO
h
t
e
Obvi
ous
ly,n ew coef
fici
ent
s i
nthe
di
ff
ere
nti
al
equa
tion
ar
e des
irabl
e i
n or
der
t
o expr
ess
d
yna
mic
c
har
act
er
o
f
th
e s
yst
em
mo
st
d
ir
ect
l
y.
As
n
ew
c
oef
ic
f i
ent
s
ar
e
ch
ose
n
co
o,杏,
Kv,
Ch
a
nd
以,,
Eqn.
(2.
25)
cha
nge
s
int
o
歹一
_
_‘_ F+F_
、 户_
十左
嘴
+2
}上+夕
二K.,I
U,
P--L
-
I “
.
一
C。
卜一
I C。
. (
2.26)
嘶
\
1
1 /
Her
e
C*二KS
ch, (
2.21)
C
。二。
。(
1
1
十
+
-w
压)
Ch
二。
。
(
2.28)
Ks
Km (
2.29)
Kv= “K,
Km
二 wK,
I 卜 --
一 一泣
C
h
一厂 』
 ̄|
乙 一
(
2.30)
一
0)
n=
}M
52
。 K
么 = 一一 1
/
t--
M 七。十 ̄
w
- 产二=二= 二一
Kv
C
o十止竺- (
2.31
)
2
Ch’ 一 2
V
MC, 2(
o,C
h
M.
o
Ch,
Co
a
nd
K a
re
t
he
hy
dra
uli
c
st
ifn
ess
,
th
e o
il
s
ping
r s
ti
fne
ss
a
nd
t
he
v
elo
cit
y
ga
in
o
f t
he
c
ompl
et
e
s
yst
em,
r
esp
ect
i
vel
y.
1
t
sh
oul
d b
e
n
ote
d t
hat
i
n
p
rac
ti
ce
wK,
r
,/
ch<
<I,
s
o
Co“c
,
an
d K,
,“KS
Km・
Mos
t i
mpo
rta
nt
f
or
t
he
dy
nami
c be
hav
ior
a
re
t
he
u
nda
mpe
d na
tur
al
f
re
que
ncy以,
and
t
he
r
el
at
iv
e d
amp
ing杏,
rep
res
ent
i
ng
t
he
h
ydr
aul
i
c ma
ss-
spr
i
ng
s
yst
em
o
f
th
e a
ct
uat
ed
ma
ss
a
nd
t
he
oi
l
spri
ng.
Eqn (
2.26)
i
s depi
cte
d vi
sua
ll
y in
th
e blo
ck di
agr
am
of Fi
g.2.1
4,
which
can
al
so
be
de
riv
ed di
rec
tly
f
ro m
Fig.2
.13
by e
lement
ary
bloc
k dia
gra
m man
ipula
ti
o n.
Fi
gur
e 2.
14
shows
a
c
ont
ro
l
lo
op
wi
t
h a
n i
nt
eg
rat
or
i
n
se
ri
es
wi
t
h a
s
eco
nd
o
rde
r
sy
st
em.
Ex
tena
r l
l
oad
Fe
an
d
F,
ar
e
di
st
urbi
ng
t
he
s
yst
em.
Th
e t
ra
nsf
orma
ti
on
of
t
he
"
phy
sic
al"
mo
del
s
hown
i
n F
ig.
2.
13,
wi
th
i
ts
p
ara
met
ers
Km,
Ch,
C
o,M
a
nd
w i
nt
o
th
e "
con
tr
ol
"
mo
del
sh
own
i
n
Fi
g
.3
.14
wi
t
h i
t
s s
yst
em
p
ara
met
er
s K,
,
,Ch
,
Co
,Co
o
an
d杏wi
l
l
en
abl
e
us
t
o
st
ar
t
dy
nami
c
an
al
ys
is
a
nd
s
ynt
he
si
s
i
n
ch
apt
er
3
.
Some ina
f l
r
emar
ks
r
egar
ding
the
l
inea
ri
ty
of
hyd
raui
lc
sys
tems
must
be
mad
e.
Figs
2.13
and
2.
14
su
gg es
t f
ir
mly
al
i
nea
r sys
tem,
i.
e.
a s
yst
em ac
cord
ing t
o al
ine
ar d
ife
ren
tial
eq
uati
on.
Ho
wev
er
,
ev
ery
s
yst
em
t
re
at
ed
s
o f
ar
i
s
fa
r
fr
om
l
i
nea
r.
No
t
on
e o
f
th
e p
ara
met
er
s
K,
,Ch
an
d
co
i
n
Fi
g.
2
.13
i
s
a
r
eal
co
nst
ant
.
The
l
in
ear
iz
ati
on
p
roc
ess
i
s
ch
armi
ng,
of
c
our
se,
bu
t
it
sug
ges
ts
t
oo
mu
ch.
Neve
rth
ele
ss,
Fi
gs.
2.1
3 a
nd
2
.14
a
re
e
xtr
eme
ly
u
sef
ul
i
f
ha
ndl
ed
wi
th
c
are
.
FR+F,
1一
兰心
心
1
K VI+
S
G I
.
}
} 。2 __ S
氢+
2“可2
Fi
g.2
.14
Bl
ock
di
agr
am
f
or
"
con
trol
"
mo
del
o
f h
ydr
aul
ic
s
erv
o
Fo
r e
xampl
e,
o
ne
h
as
t
o
re
membe
r t
hat
t
he
o
il
s
pri
ng
s
ti
fne
ss
c
.
va
ri
es
c
ons
ide
rabl
y wi
th
t
he
p
ist
on
po
sit
ion
(
see
Fi
g.
1
.11
).
Th
us
a
nal
ysi
s b
ase
d o
n al
i
nea
r mo
del
h
olds
o
nly
f
or
s
mal
l
o
sci
ll
at
io
ns
ar
oun
d one
pos
it
io
n.
In
a
not
her
pos
it
ion
,
ano
the
r
c.
i
s
val
id.
In
pra
cti
ce
o
ne t
akes
t
he
p
osi
ti
on
whe
re
c,
i
s
a min
imum.
The
ve
loc
it
y g
ain
K o
f
a勿draui
l
c mot
or
depe
nds
on
the
l
oad
APL
(
see
Fi
g.
2.
3).
If
n
ece
ssa
y,
r a
dequ
ate
c
orr
ect
ion
s mus
t
be
ma
de.
Re
gar
din
g t
he
h
ydr
aul
ic
s
ti
ff
nes
s
ch,Ta
ble
2
.2
s
hows
a
h
igh
ly
n
on-
li
nea
r c
har
act
er
.
On
ly
i
n
open-
cent
er
val
ve sC
s
i ,
, a
ppr
oxi
mat
el
y,
a
co
nst
ant
i
s t
h
e u
nd
er
l
ap
r
e
gi
on
(
I
x)<U)
,bu
t
i
n
cri
tica
l-cent
er
c
val
,
,
pr
o
ves
i
s
po
rt
io
nlt
a o
l
x}
一‘SO
Ch=- a
t
X=0.In
cal
cula
ti
ons
re
gar
din
g
s
tab
il
it
y
and
d
amp
ing
,
a
l
ar
ge
c
.
up
t
o Ch=- wil
l
al
ways
p
rove
t
o
be
a
s
afe
ass
umpti
on.
I
n t
hes
e c
omp
lex
mat
te
rs
on
e c
oul
d e
asi
l
yfo
rge
t t
he
mai
n n
on-
li
nea
riz
ing
f
act
or.
Th
is
i
s
Co
ulo
mb
ri
fct
io
n F,,r
eve
rsi
ng
i
t
s s
ign
wi
th
t
hes
pee
d夕.Thu
s,
i
f
no
n-l
in
ear
c
ont
ro
l
an
aly
sis
h
as
t
o be
ca
rri
ed
out
, t
he
Coul
omb
f
rict
ion
s
houl
d beconsi
dered
fi
rst
.
54
CHAPTER
3 DYNAM IC
ANALYSIS
AND
SYNTHESI
S
3.
1 INTRODUCTION
F
or
t
he
d
yna
mic
a
nal
ysi
s
of
h
ydr
aul
i
c s
erv
o me
cha
nis
ms
t
he
d
iff
ere
nti
al
e
qua
ti
on
v
'
'_
一二.
,,+ V.
二 + LL
. =
,,「
= A A
_ F+F1
U _E 一 一
F十F
I一 -. (
3.1)
L
r ‘、 扩 }i
}
n t n
(l)
W _ I L ‘ I
V n
一 o
} L ,
. J u
de
riv
ed
in
s
ect
ion
2.8
will
b
e us
ed.
Fi
g 3.1
sh
ows
a
gai
n t
he
b
loc
k di
agr
am
wher
e r=clos
ed
loop
input
si
gnal
,
£=r一y=o
pen
loop
inp
ut
si
g na
l,
Y=o
u
tput
moti
on,
凡 =ex
te
rnal
fo
rce,a
nd
FQ=Co
u
lomb
f
ri
cti
on.
FB
+F,
三} 厄
Ck
I
1C,
9
石1
卜曰.
5
Gs匕
_
}
s
2
p十
,c
S
-
s0
+1
Fi
g.3.
1
Bl
ock
d
iagr
am
f
or
c
ont
rol
an
aly
sis
Th
e s
yst
em
p
ara
met
er
s a
re
t
he
v
elo
ci
ty
g
ai
n
K,
,
,th
e h
ydr
aui
l
c s
ti
ff
nes
s
Ch
an
d t
he
o
il
s
pr
in
g
s
ti
f
fne
ss
Co
of
t
he
c
omp
le
te
s
ys
te
m a
s
we
ll
a
s
th
e n
at
ur
al
f
re
que
ncy
(
00
a
nd
t
he
d
amp
in
g杏o
f
th
e
o
il
-s
pr
in
g-ma
ss
s
yst
em.
Th
ey
a
re
d
eri
ve
d f
ro
m t
he
"
phy
si
ca
l"
p
ara
met
er
s
KS
,K,
,Ch
,Co
,M
a
nd
w
(
see
Eg
ns.
2.
27-
31
a
nd
Fi
g.
2.
13
i
n Se
cti
on
2.
8)
.
Th
e t
ra
nsf
er
f
unc
ti
on
Gs
re
pre
sen
ts
t
he
d
yna
mic
p
ato
r
f
the
mechani
sm r
ela
tin
g t
he
er
ror
si
gnal
E=
=r一y t
o t
he
va l
ve d
ispla
cement
x.
Dy
namic
ana
lysi
s a
nd adj
ust
i
ng c
ontro
l a
cti
on wi
ll
be
bse
a d mainly
on fr
eque
ncy
res
pon
se.
A
d
iic
f ul
ty he
re is
the no
n-l
inea
r beh
avior
of Coul
omb fr
ict
ion, ma k
ing con
ven
tio
nal
li
nea
r
re
f qu
enc
y a
nal
ysi
s
impo
ssi
ble
. I
n s
pit
e
of
i
ts
l
arg
e i
nlu
f e
nce
o
n s
yst
em
be
havi
or,
Coul
omb
ri
fcti
on
ha
rdl
y a
fec
ts
t
he
a
dju
stme
nt
of
t
he
c
ont
ro
l ga
in
i
n p
rac
ti
ce.
Th
is
s
ta
teme
nt
wi
ll
be
come
mo
re
plaus
ibl
e whe n,
at
the
end of
thi
s cha
pte
r ,
i
t will
be pro
ved mathe
ma t
ical
ly i
n two d
iferen
t
sit
uat
ions
that
Coulomb
fri
cti
on
does
not
i
nlu
f en
ce the
sys
tem s
tabi
li
ty
i
n prac
tic
e.
Th
is
at
leas
t i
s s
ome jus
ti
fi
cat
ion
for
our
approach,
which
ig
nores
Coulomb f
ric
tio
n (
except
fo
r
one t
ri
vial
ca
se)
at
the
begi
nning
of
thi
s
chapte
r,
enabl
ing us
t
o pr
oceed
to
clas
sica
l
in
l ea
r anal
ysi
s
and s
y n
thesi
s.
This
does
not
mean t
hat
Coulomb ri
f c
ti
on has
no
inlu
f enc
e on s
ystem behavi
or i
n
gener
a l
, o
n the cont
rar
y.
In many a
ppli
cati
ons, Coulo
mb ri
f ct
ion
ma y
d e
ter
iorat
e the sy
stem
c
omp le
tel
y wi
th
res
pect
t
o s
peci
al
sp
eci
fi
cat
ion
such
s
a a
ccur
acy
(owi
ng t
o d
ead
zon
e),
ba
ndwi
dth
(
owing to dyna
mic dead zo
ne) or
si
lp-
sti
ck mot
ion (
demonst
rat
ing i
ts
elf
by a
ccel
er
ati
on
d
ist
urbanc
es c
aused
by
i
rre
gula
r ri
fct
ion
beh
avior
).
55
3.
2 PROPORTI
ONAL
CONTROL
The op
en
l
oop
t
ra
nsf
er
f
unc
ti
on
H(
s)=Y/£c
an
be
f
oun
d Eg
n(3.
1)
orFi
g.3
.lby
l
et
ti
ng
bandwi
dth
of
t
he
Fe=0 a
nd
F “Q.
Ass
umi
ng
f
urt
he
r,
GS
(s)=1
,wh
ic
h i
s
al
l
owe
d wh
en
t
he
s
erv
ova
lve
i
s
mu
ch
h
igh
er
t
han
t
he
na
tur
al
f
re
que
ncy姚 ,
we
f
in
d
Y Kv (
3.2)
H
(
s)=
S
! {
2
_
C s  ̄ 。 s
一a
+L
S一
刃一十1
C
U
(先
Eq
n.
(
3.2
) s
hows
a
n i
nte
gra
tor
i
n s
eri
es
wi
th
as
eco
nd
o
rde
r s
yst
em.
I
n t
he
o
pen
l
oop
0
0
f
req
uen
cy
re
spo
nse
t
he
ampl
i
tud
e r
at
ion
a a
nd
phse
a l
ag价c
an
be
f
ound
by
l
ett
i
ng
s=j
w a
nd
八
0H
0 a
夕一
Kv
田
.
一
一
 ̄
.
了
、
了
、
(
3.3)
一
-
一
八
、
.
」
E
、
一 ,  ̄ ̄  ̄,, 一 分一 一 - . ,一
了
c
o,
I
(1一 C
c
断
)
-
2)
2
、(
2
}
鱿
(
t
)
)
2
0=Ar
g H(
j()二 一一
o
0
兀
一 arct
an
2
}
'
0
(
C
O
0 (
3.4)
‘
一
C
0D
)
0
2
2
Th
e Bod
e di
agr
am
i
s
sh
own
i
n Fi
g.3
.2
a
),
t
he
Ny
qui
st
p
lot
i
n b)
.
Bo
th
di
agr
ams
s
hou
ld
be
s
tudi
ed
c
are
ful
l
y. 0
0
A
e 价=-7
phas r
oc
cur
s a
t。“叭 ,whe
r C二Kv/
e
G 2甄 .
To
b
e s
ta
bl
e
th
e
Ny
qui
st
pl
ot
s
houl
d note
nci
rcl
e
th
e po
int
一10
,.
Thu
s t
he
s
tabi
l
ity
c
ond
iti
on
i
s
Kv
(
3.5)
<2}
WO
a
nd
c
ons
equ
ent
lythe velo
city
gai
n
Kvi
s l
i
mit
ed
by
th
e pr
oduc
t ofnat
ura
l re
fquenc
y叭a nd
d
amp
ing .T
o t
ransl
ate t
his
i
ntophy
si
cal
p
ara
met
er
s,
WO
an
d杏o fEqns
(2.
30)
, (
2.31
) of
Se
cti
on2.
8 mus
t be
s
ubs
tit
uted.
The
r
esul
t
is
C _ w
K.
,
I
1一 - ,I< — . (
3.6)
Ch一 M
S
o t
he
ra
ti
o o
f o
il
sp
ri
ng
s
ti
ff
nes
s an
d hy
dra
uli
c
sti
ff
nes
s a
s wel
l
as
t
he
r
at
io
o
f v
isc
ous
fr
ic
ti
on
c
oefi
ci
ent
wa
nd ma
ss
M
ar
e
ext
remely
i
mpor
tan
t.
The
ir
c
onsequ
enc
es
wi
ll
be
di
sc
usse
d l
at
er
.
T
o
pr
o
vi
de
n
ot
on
l
y
st
a
bi
li
t
y,
bu
t
a
de
qu
at
e
d
am
pi
ng
a
s
we
l
l,
th
e
so
-
ca
ll
e
d
M
=
1.
3-
c
ri
t
er
i
on
c
oul
d be
u
sed
, i
mpl
yin
g t
hat
t
he
c
los
ed
l
oop
a
mpl
it
ude
r
ati
o ne
ver
e
xce
eds
1
..
5.
Ob
eyi
ng
t
is
h
c
ri
te
rion
t
he
cl
ose
d l
oop
st
ep
res
pons
e
normal
ly
sh
ows a
n ov
ers
hoot
o
f a
ppro
ximat
el
y 2
3% (f
or
a
s
eco
nd or
der
s
yst
em
thi
s
isex
act.
)
Thu
s i
n t
he
pol
ar
di
agr
am b)
t
he
Nyqui
st
pl
ot
i
s
not
al
lo
wed t
o
i
nte
rse
ct
t
he
M =1
.3-
cir
cle
. To
a
voi
d ompl
c i
cat
ed
c
alc
ula
ti
ons
i
t
is
go
od
e
ngi
nee
r's
p
rac
tic
e t
o
f
ocu
s a
tt
ent
i
on
o
n t
wo
i
mpo
rta
nt
po
int
s
of
t
he
M
=1
.3-
cci
rcl
e.
The
y l
ead
t
o t
he
we
ll
kn
own
56
口
口7
月
仍!l一
’、
『、 十
‘‘了 二
..口
户口洲曰
ii
i
/
户,‘尸
l
粼)
了 红
户 ,碑
1
一
白 硬叫 〕
/产
一
|
. ̄‘ .砚 L
- ̄一一 一  ̄
二‘ ̄.户 户门 ̄
巨三 已二二 ‘
{」
夕门
曰 为 + rl
}
尝阵 、、、 川 阶] / i
汽口口 \ /尸洲/ 门
万 下一n
习声/于 :}
级)
一沪
心 、
b} \
O
.
一
:
二_
.1
v ,份
、
尸一 它
.、
旗 \
U入 \
N
之
沪 IO
i
I
I
/
枯
1
+勺/
I
1
、勺之勺、
I l
l \
、
J
r
/
X夕尸
之、、灿七之势之』U护端 舒之,
多
t
z 卜,O , ,
l
I a -
泛r "
,
`r
M
Q o
0
M
/ I I
}
、
O p 粥 b 勺
N ^'
1卜 劝
、 ,
l
l }
lP 、,
哭 人0
爱
s
月门 O O 尧 夸 O N -
.
门 门
l
刀’
’
/ 义0
1
食 髯氯
尹、
气0 0
O 之 之
万 八
两
,灿
d
、气
0
协 气
火口一.
伪 0・
之知卜
城
队"
、
只兹
卜 月口.一  ̄
O
}I
3
O
b
+
冲]
补
,0 1
1
i
t
1
,
?
气J,
、"
.目
口创. 一 林. I
一 弘 、、
伙
0
甸‘
气1.1只\
、.
、
护00、.
、、
\
、 、
尹}口曳,岁 〕Q 叼』匀 几气一弋
‘..‘1
:
二多
巨口 丫:/一1
.护
尸尸尸产 厂芬
二口
门门
vi
/
/ a
口日U:/ ; 门门 /
// } 0
l 套 之
乡罕
.
一
1
一
一
{
l I
}
I }
l 二
I ‘
l
!
0
/
I
I
I
、 0
‘
二}3
1
4
1汾I L}14
。户 内-
0
代J ̄一 毛 O O 自百
'
CI
8
遗
b
Q j
T ・
O
仑
Fi
g 3.
2
Pr
opo
rti
ona
l
co
ntr
ol
57
1
)
ampl
i
tud
e ma
rgi
n
of
a
bou
t 1
00%,
me
ani
ng
a<
0.5
a
t价= -9r
0
0
2)
ph
ase
ma
rgi
n
of
a
bou
t
45
0,
me
ani
ng功<
_-
0.751
r a
t a二 1.
Resp
ecti
ng bot
h mar
gin
s i
mp l
ies
nor
ma l
ly
that
t
he
cl
ose
d loop
re
f quency
re
sponse
wil
l
show
an
ampl
itu
de rat
io
of
maxi
mally
1.3
and t
hat
the
st
ep
r
espons
e wil
l
have
an ov
ersho
ot
small
er
th
an
23%.
Foll
owing
the
r
ule
s o
utl
ined
above,
t
he s
yst
ems
wil
l
be s
ufic
ient
ly
damped.
F
or
l
ow
v
al
ues
of
杏,wh
ich
a
re
u
sua
l
in
h
ydr
aul
i
c s
er
vo'
s,
th
e a
mpl
i
tud
e ma
rgi
n
is
d
omi
nan
t.
At
0
)二WO
i
s价二一1 a=K,
r, /2
枷。,
Thu
s a
c0
.5
g
iv
es
K
一二 (‘
二 、
.
,1
2-
EO1
二二 (
3.7)
叭
’
L Ch
」M
Fo
r t
he
(
unu
sua
l)
l
ar
ge
v
alu
es
of
gthe
p
has
e ma
rgi
n
is
d
omi
nan
t.
At
W , FI
T-
1 .. 0
3
— =一白十V白一十1
田o
0
i
s中- --1r
and
4
a
0
=疏
2
}
w0
-
,
1
2
-_
K
}v
+
}
2
+1
}
s
o a
-
< 1 gi
ves
K
v、
Do
n
2
;
}
扼
-J
_
?}
;、
.f
2
+l, (
3.8)
The
re
sul
ts
of
Eqns.
(
3.7
) a
nd (
3.8)
h
ave
bee
n pl
oted
in
Fi
g.3
.2
c)
t
oge
ther
wi
th
exa
ct
r
esul
ts
of
th
e M=1.
3 -
cri
te
rio
n i
tse
lf,
o
bta
ined
by
co
mpu t
er.
I
t
sh
ows
an al
mos
t
perf
ect
a
gre
ement
b
etween
t
he
cr
iteri
a.
Ta
king
v
alu
es
o
f凡 /
au
, ro
f m
Fi
g.
3.
2
c)
i
n d)
a
ci
d
e)
,
t
he
c
los
ed
l
oop
a
mpl
i
tude
r
at
io
(
<1
,
.
3)
a
nd
uni
t
st
ep
r
espons
e (
maxi
mum
over
shot
<23%)
ar
e s
hown
Sa
tis
fac
tor
y r
espo
nse
s a
re
ob
tai
ned
wit
h 0.
3<咨<03.
At杏<0.
3 t
he
s
tep
r
espo
nse
i
s v
ey
r
n
erv
ous
a
nd
t
he
f
req
uenc
y r
esp
ons
e s
hows
a
n a
mpl
i
tud
e di
p l
owe
r t
han
0
.7.
At
杏X0.
7 t
he
s
tep
r
espons
e is
s
lowi
ng down a
nd t
he
bandwi
dth
(
i.
e.
t
he
f
requ
enc
y u
p t
o wh
ich
a
mpl
i
tude
r
ati
os
d
o
n
ot f
all
b
elow
0.7
) d
e c
rea
ses
ra
pidl
y.
Co
ncl
usi
on:
h
ydr
aul
i
c c
ont
ro
l s
yst
ems
wi
t
h a
n a
deq
uat
e
da
mpi
ng
c
oef
fi
ci
ent
0.
3<百<0.
7
an
d a
we
ll
a
djus
te
d v
elo
cit
y
ga
in
K,
, s
how
a
b
andwi
dt
h C
)h
e
qua
l
to
i
ts
n
atu
ral
re
fque
ncy
C
oo・
3.
3 DYNAM IC
LOAD
Fo
ll
owi
ng
t
he
c
alc
ula
ti
on
r
ule
o
f Fi
g.
3
.3
a
),
t
he
e
ffe
ct
o
f a
s
inu
soi
dal
ly
v
ayi
r n
g l
oad
FQ
on
t
he
o
utp
ut
y
a
t r=0
a
nd
F,=0
c
an
b
e f
oun
d d
ir
ec
tl
y.
Kv一
CO一
5一
「K
1
一
} {
sz
叭
+2
杏一上+1
+
吼
低
I-
— + 一一
y L
Lch
、
s
s
co
J
l叫o
.
(
0o 」 1
护一
Fe 1
未
K Ca 」。、 、, _ 「 _z
+2
}一二+1
」
S I S
1+
0
(
o’
‘(
0
o
L(
嘴
+2
杏二+1 叭
叭
}
(K.
, C,}(0
)}
I
I一 ・一 1
+1— 1
C.
v } i
)(
0I C
t
i I
1w
0
(
3.9)
一拼.
=I, “一 、
弓‘ “ 一‘
凡 1
1(K. __0
)
2丫 (
1
)
2(
_(A
)
2丫
}
}0
(
o一2
‘0
)
02
!十W
0
2\
‘一
U)
0
2
T
he
s
ta
bil
i
ty
c
ond
it
io
n Kv/
loo<2
杏,d
eri
ve
d e
arl
i
er,
c
an
b
e r
eco
gni
zed
e
asi
l
y s
in
ce
Co
yl
Fe二- a
tK,
/)o=2
0 } a
nd
(
o二No.
2 ++ H
s
Y
尸 _ 万 H H H
Hr Y=了H
-H艺十1
+H
H护
时
凡 +
F
,
1-心
二
Co
尸  ̄ S十 K v
I+
到‘
片月m
}+a
sm
1
r
b
)
l
I
Co
剑
-x - t
F,’2
鱼-
,C
}
h - 尸/ 李
笋
ni
几 .,
Ch -
一
一一洲/ .
夕 气
__
_
_
且J
了 1
_
1价不一 卜
、一
_军二
一F-
一
次
次
忿
3i
:
}-
于于
7- 一
,犷.
丁 一-
‘
、 \
不 - 「一
口
一 一 饭 F
、
x
t
-r官目尸吧 J喇,声护
次4 「一 }
、\
\_
‘
、
汉
}
I
{
/ K
i
V
”‘
=
一一
一B
i
}
J
'
WY
i
Q
i
Y_
-
.
r一L4
{
_
__
! \
!_
_立_
饮oa 次Ua叹(
7S吹r 心 众娜 疾‘ 度J 之 口 2
WQ一— ‘
Fi
g.3
.3
Dy
nami
c l
oad
F
or
t
he
a
nal
ys
is
of
th
e
lo
ad
b
eha
vi
or
we
a
ss
ume
a
ve
lo
ci
ty
g
ai
n Kv=0.
4wo
a
deq
uat
el
y
ad
ju
st
ed
t
o
an
a
cc
ep
ta
bl
e
da
mpi
ng =0
.4,
pov
rid
in
g
go
od
r
es
po
ns
es
o
f y
o
n;a
t凡=
0.
Wi
t
h
th
es
e
da
ta
h
owe
ve
r,
th
e
l
oa
d
be
hav
io
r
ma
y
st
i
ldi
fe
r
co
nsi
de
ra
bl
y,
as
is
sh
own
i
n
Fi
g.
3.
3
c)
wh
ee
r Co
刘凡i
s
pl
ot
t
ed
a
s
a
f
unc
t
ion
o
f
r
f
equ
enc
y
at
va
ri
ou
s
va
lu
es
o
f
Co/
q.Ac
cor
di
ng
t
o
Eq
n.
(
2.
31
)
i
n
Se
ct
io
n2.
8,
59
一
犷
K 一 一 w Kv 一 二 t
)
.2C
CO斗 — 。
一、_tc}
W
一点 、
/M C(
、十一,声二二二二 一一=0.
4 (
3.1
0)
2C`
,” 2y
I
MCa 2u)
o Ch
2Mo)o Ch 2M叭
b
ot
h
vi
sc
ous
f
ri
ct
i
on。
,an
d t
he
r
at
io
Co/
Ch
c
ont
r
ibu
ti
ng
t
o .S
in
ce
i
t
i
s
pr
ef
er
abl
e
fo
r
th
e
i
nf
lu
enc
e o
f
th
e l
oa
d凡o
n t
he
o
utp
ut
t
o
be
a
s
sma
ll
a
s
po
ssi
bl
e,
a
h
ydr
aul
i
c s
t
ifn
ess
Ch=0
0
i
s des
ira
ble
(
the
,
al
ter
nat
ive
i
s PI
-cont
rol
)
see
Sect
ion3.
8).
ht
t
hat
cas
e t
her
e i
s no
l
oad
e
rror
a
t l
ow
f
requ
enc
ies
.
1. Co 少 o一
Wi
th吼 二。 ^ =t
11M -_ 二0.
(
0-
-}
o 户 KV
Wh
en
t
he
re
i
s,
u
nfo
rt
una
te
ly
,
no
v
is
cou
s f
ri
ct
io
n,
Ch
sh
oul
d
be
d
ecr
eas
ed
i
n
or
dert
o pr
ovi
de
a
dequat
e da
mp i
ng.
But,
a
s Fi
g.3
.3
c
) s
hows
v
ery
c
lea
rly
, t
his
i
s do
ne
a
t t
he
e
xpe
nseof
t
he
l
oad
e
rro
r a
t l
ow
fre
quen
c i
es.
Co夕 co
Wi
t
h Ch
#
G
-,1
1tH
F
(
,
L
w
-1
0
,
y
,
cor
res
pondi
ng
t
o t
he
l
oad
e
rror
der
ive
d ear
li
er.
3.
4 DYNAMIC
DEAD
LONE
DUE
TO
COULOMB
FRICTI
ON
I
n j
ust
on
e tr
ivi
al
case
t
he i
nf
luen
ces
of
Coul
omb ri
fcti
on
on
t
he
dyna
mic
beha
vior
c
an be
d
elved e
xact
ly
. Thi
s i
s
the
ca
se o
f h
ydr
aul
ic
l
ock wh
ere
Coul
omb
f
ri
ct
ion
pr
eve
nts
any
out
put
mo
ti
on.
Th
e o
utpu
t〕,b
eam
zer
o,
Fi
g
, .
3.
3 b)
d
ire
ctl
y p
rov
ide
s t
he
s
olu
ti
on.
I
f
y=0,t
hen
i
n=0
a
s
we
ll
,
t
hus
a
t凡 =0
we
f
i
nd
"
v _
二=0
a
nd
"
1
一{
一C
F
h S
C0
;=:二互 +一
F`. a
t y=0.
Cy
K Co
This i
s a
l
ine
ar di
ff
ere
nti
al
e
quat
ion.
Thus
a
t a
s
inus
oidal
i
nput
s
igna
l r
or£ t
he
Co
ulo
mb
fr
ict
ion
wil
l
be s
inu
soid
al
t
oo,
s
tay
ing
with
in
i
ts
ma
ximur
n bor
der
s
一
}F(
.
}m
ax、F
c<+
,
F,
}。
。.
On
e mi
ght
s
ay
t
hat
Co
ulo
mb
f
ri
ct
ion
j
ust
t
ake
s t
he
v
alu
e t
hat
i
s n
ece
ssa
ry
t
o s
top
a
ny
mo
tio
n.
Us
in-
t
he
n
ota
ti
on孟 二I
F-{ we
f
in
d i
n
th
e f
re
aue
nev
do
mai
n
一 厅 一 二万一万一.
,万
F,
I
(低 } (0
o )
y-
-0 a
t”=£<才
r ̄1
11
r
,
y
-}+1
一K一} (
3.11
)
“‘
) Vl"h
J
\-`V
l
At
l
ow
f
requenci
es
一
凡一认
y=0
whe
n
l
lmr=l
ime< 口-A 臼-40
60
r
epr
esent
ing
t
he
wel
l-
known
s
tat
ic
dead
z
one.
At
hi
ghe
r
fr
equ
enc
ies
(0)>KV
Co/C,
,)
、
,=0
when r=E<
Th
us
t
he
d
ead
z
one
a
ppa
ren
tly
i
ncr
eas
es
f
requ
enc
y a
nd
c
oul
d t
her
efo
re
be
c
all
ed
t
he办na
mic
de
ad
z
one.
It
s
exi
stenc
e has
be
en
c
onf
irmed
e
xper
iment
all
y s
ever
al
t
imes
The
ot
her
way
ar
ound,
one
coul
d s
ay
t
her
e i
s a
re
f quency
co一
,妇
/ 1|||.、、
,』
、
、
吼
1 .1 1
-
),
0to
P=K, (
3.1
L)
户
wh
ere
a
t 0
)>(
05
tp
n
o mo
ti
on
wi
l
l
oc
cur
.
He
nce
,
Co
ulo
mb
f
i
rct
i
on
c
oul
d
be
a
s
er
io
us
t
hr
eat
to
t
he
bandwi
dth
of
a
hydr
aul
ic
s
ervos
yst
em.
3.
5 DAM PING
METHODS
二甄
I
n S
ect
i
on
3
.2
i
t
wa
s p
rov
ed
t
hat
f
or
go
od
p
ropo
rti
ona
l
co
ntr
ol
a
da
mpi
ng
c
oef
fi
cie
nt
K, 几-
二二-
气 二 — 乙 ,1
1
v.C1
)+
W
.
C,
r+ w
2ch’
2V
MCo Ch 2M叽
r
ang
ing
f
rom
0
.3
t
o
0.
7
is
d
esi
r
abl
e.
As
t
he
t
wo
c
omp
one
nts
o
f杏a
re
p
rop
ort
i
ona
l
to
t
he
r
at
io
o
f
o
il
s
pr
in
g s
t
if
fn
ess
t
o h
ydr
aul
i
c s
t
if
fne
ss
Co
/
Ch
a
nd
t
he
v
is
cou
s f
i
rct
i
on
c
oef
i
fci
ent
w,
r
esp
ect
ive
ly,
o
ne
o
r b
oth
o
f t
hes
e f
act
ors
mus
t
be
p
res
ent
.
In
Se
cti
on
3
.3
t
he
de
sir
abi
l
ity
o
f a
l
ow
co
/Ch
t
o
pr
odu
ce
a
s
mal
l
lo
ad
s
ens
it
iv
it
y a
t
lo
w f
re
que
nci
es
b
eca
me
c
le
ar
(
i
n c
i
rti
cal
-
cen
te
r
v
alv
es
CQ/Ch=0
a
rou
nd
x=0,s
ee
t
abl
e 2
.2
,)
.
He
nce
a
n e
xce
ll
ent
d
amp
ing
me
tho
d d
epe
nds
c
omp
let
ely
o
n t
he
v
isc
ous
f
ri
ct
ion
c
oef
fi
ci
ent
w,ind
ica
ted
i
n t
he
b
loc
k d
iag
ram
of
Fi
g.3
.4.
Un
for
tun
ate
ly,
i
n p
rac
ti
ce
i
t
is
h
ard
ly
p
oss
ibl
e t
o d
esi
gn
o
r t
o p
red
ict
be
for
eha
nd
t
he
a
mou
nt
of
v
isc
ous
f
ri
cti
on.
Ca
ref
ul
a
nal
ysi
s o
f s
hea
r i
n o
il
f
il
ms
o
f b
ear
ings
,
fo
r i
nst
anc
e,
s
hows
v
isc
ous
f
ri
cio
t n
t
o b
e n
egl
ect
ed
c
ompl
et
ely
.
Ye
t
vi
scou
s f
ri
ct
ion
i
s
pr
ese
nt
n
orma
lly
,
ju
st
a
s a
c
ompo
nen
t
of
t
he
t
ota
l
fic
r t
ion
,
si
nce
Co
ulomb
f
ic
r t
ion
a
nd
v
isc
ous
f
ri
ct
ion
f
requ
ent
ly
g
o ha
nd
i
n
ha
nd.
To i
ll
ust
rat
e t
his
, we
ref
er
to
many e
xperi
ment
s i
n t
he
hydra
uli
c r
ese
arch,
whe
re
in
olde
r
s
ervomotor
des
ir
i
ns
Coul
omb f
ri
ct
ion
a
nd vi
scous
f
ri
ct
ion
wer
e pr
ese
nt
a
.
,
sw e
ll
as
an
ac
cep
tabl
e
d
amp
ing
c
oef
f
ici
entg.T
his
v
al
ue
o
f百wa
s a
ppa
ren
tl
y c
aus
ed
b
y }
is
c
ous
f
r
ict
i
on,
a
s
l
oad
-co
mpe
nsa
ti
on
(
ac
cor
di
ng
t
o S
ect
i
on
2
.3)
wi
t
h C =冈 wa
s p
rov
ide
d.
1
n l
at
er
d
esi
gns
i
n
whi
ch
Co
ulo
mb
f
ri
ct
ion
wa
s e
limi
nat
ed
lmos
a t
compl
et
ely
b
y i
nst
all
i
ng
h
ydr
ost
ati
c
be
ari
ngs
.
ve
ry
p
oor
da
mpi
ng,
oe
fi
ci
ent
s
(
as
l
ow
a
s杏二0.
05)
wer
e
me
asu
red
.
6l
丫
(
fh
na
uuc
l
pr
ess
rn
c户ef
ba以
丫
r
necl
rtaucal bl
pas
s nr
iderl
ap
汁巨d
el
ecfr
irul
厂e
\
百一
4P
,
厂 i
is
co
us
f
rt
ef
io
n
r
o
1八,
+ ”,
r
十
凡
Gs h
0
i
s
ll l
l
‘ 一 一 _ _
 ̄
E ac
cel
era
tio
n户e
dbac
A _ 』
一 ..山.目.月.....内目.户‘,,.,.. 一
a)
pi
lot
v
ale
e
f
l
a
p
pe
r
n
a
z ,-‘
zl
e
, we盛
Ps
d,
三
吐 .
1
_J
pr
ess
uref
eed
boe
k
! 明 .  ̄ ̄ 份. . 一 一
R QL
bs
pass
一 一  ̄ 一一 一一一 一一一一 一一 资 明卜.,一 .
 ̄‘- .
SS
'1
}Sr
lt
.li
iOf09
' _」
J
止二二-
I
PL
弓 只
也 当
R r一 气
产一 、 R扣
办7
nnmi
c
咋荟叫
,。-,勺 。 弓 二 ̄p,车r ,
,卜,导 川卜... - 一 -
,I
Q
几 卜 ,- .
、目 .一
.产 口I Ts十 l P
TS十 1 、.- 侧产
Qn2
pressurefeedbaek
b)
+,]
几7 . 右
K,
, J,
,一山.
一
了
十
一
+
‘三
}
+
叭一
十
,2(
5+e,)
气
一 夕
办 ecel
erahc
口 IJ
曰
fe
erl
hnc6 7
、二,
、
c
9
Fi
g.
3.
4
Da
mpi
ng
me
tho
ds
I
f
in
suf
fi
cie
nt
v
isc
ous
f
ri
ct
ion
i
s
pr
ese
nt,
ap
pli
cat
ion
of
Eq
n.
(
3.1
0)
wi
ll
s
ugge
st
i
ncr
esi
a n
g t
he
r
at
io
C
O l
qin
o
rd
er
t
o o
bt
ai
n s
ufi
ci
en
t
da
mpi
ng.
S
in
ce
Kv/
)
t i
s
co
nne
ct
ed
t
o }
t
o
e
nsu
re
a
goo
d c
ont
rol
b
eha
vio
r,
a
nd
Co
(
whi
ch
i
s
pr
opo
rt
ion
al
t
o t
he
p
ist
on
a
rea
A)
h
s
a t
o
g
ove
rn
bo
th
l
oad
c
apa
cit
y
an
d b
and
wid
th,
Ch
mu
st
b
e r
edu
ced
t
o
ge
t
ade
qua
te
d
ampi
ng.
Howe
ver
,
62
t
he
c
ons
equ
enc
e
is
b
ad
l
oa
d
st
i
ffn
ess
.
Fi
g3.
4
b)
sh
ows
s
eve
ral
me
tho
ds
r
edu
ce
Ch.
A
v
ery
d
ire
ct
me
tho
d i
s
th
e i
nt
rodu
cti
on
o
f op
en-
cen
ter
v
alv
es
wi
th
a
l
ar
ge
u
nde
rl
ap
U,s
inc
e
Ch
i
s
pr
opo
ti
r o
nal
t
o PS
A
/
U
(
se
e T
abl
e2.
2)
.
Th
e u
nac
cep
tab
le
d
isa
dva
nta
ge
o
f s
uch
a
l
ar
ge
u
nde
rla
p Ui
s,
h
oweve
r, t
he
e
normous
oi
l c
ons
umpti
on.
Even i
n t
he neut
ral
po
sit
ion
of
t
he
u
nde
rla
p val
ve,
t
he
o
il
cons
ump
tio
n c
an
be
much
hig
her
tha
n t
hat
c
orr
esp
o nd
ing
to
t
he max
imu
m
s
pee
d夕of
t
he
mot
or,
req
uir
in
g
A夕。、.
A
fa
r
be
tt
er
me
tho
d t
o
re
duce Ch
i
s
th
ere
for
e
th
e s
o-
cal
l
ed
b
ypa
ss
me
tho
d a
ppl
i
ed
i
n
sy
mme
tr
ic
mot
ors
wi
th
4-
way
v
alv
es,
s
ee
Fi
g.
3.
4 b)
. A
l
ami
nar
lo
f w
r
esi
st
anc
e R
pr
opo
rti
ona
l t
o t
he
l
oa
d-p
res
sur
e
PL=P
,一P
a,wi
l
l
pr
odu
ce
a
"
le
aka
ge-
lo
f w"
么 =几/
R,
g
ivi
ng
t
he
mo
tor
(
i
n t
he
o
pen
l
oo
p s
it
ua
ti
on)
a
n a
ddi
t
ion
al
s
pee
d o
f匆 =QL
/A=AP
,了RA.Ex
ami
ni
ng
t
he
b
lo
ck
d
iag
ram
o
f F
ig.
3.
4
a)
,
we
c
an
s
ee
t
hat
t
he
by
pas
s
pr
odu
ces
a
"
shu
nt"
b
loc
k wi
th
H,.p
ara
ll
el
t
o
t
he
b
loc
k 1
/
c
h‘
匆1
K. 1 Aj
Hm一
AAPL AK.△凡
T
he
d
amp
ing
c
oef
fi
ci
ent
杏,d
ue
t
o
by
pas
s,
c
an
b
e c
al
cul
at
ed
b
y r
epl
aci
ng
1
/
c
h i
n
Eq
n.
(
3.
10)
b
y 1
/c.+H,
,
,.Th
e b
ypa
ss
me
th
od
d
oes
n
ot
h
ave
t
he
d
is
adv
ant
age
o
f e
nor
mou
s e
xt
ra
o
il
co
nsumpti
on, and
in
hy d
raul
ic
serv
o's
with
mechanical
f
eedback
i
t i
s t
he on
ly
da mpi
ng method
t
hat
c
an be
inst
all
ed i
n a
re
lat
ive
ly s
hor
t
peri
od o
f t
ime .
A
seri
ous
di
sadvant
age
is
t
h e
se
nsit
ivi
ty
of
t
he
la
min a
r low
f re
sis
tance
R
to
temper
atur
e
changes a
nd i
ts
di
ffi
cul
t
adju
stment
i
n pra
cti
ce.
Qu
ite
a dif
fer
ent
but
vey
r i
mp ort
ant
da
mp i
ng method
is
call
ed
pre
ssure
fee
dback,
pro
vidi
ng an
e
xtr
a s
erv
o-v
alv
e di
spl
ace
ment
Ax
pr
opo
rt
io
nal
to
t
he
l
oad
p
res
sur
e尺.
A me
cha
nic
al
v
ers
ion
o
f
p
res
sur
e f
eed
bac
k i
s i
ndi
cat
ed
s
che
mat
i
cal
l
y i
n Fi
g.
3.
4 b
).
As
wi
th
t
he
b
ypa
ss,
i
ts
e
ffe
ct
c
an
b
e
c
har
act
er
iz
ed
i
n
a
t
ra
nsf
er
f
unc
ti
on
H
m
p
ara
ll
el
t
o
1
/
ch
:
A 一A
H.二 Ax _A,AP
L
/C,_
-
F
2
一
C
AAP[
2AA尺
wh
ere
A i
s t
he
a
xia
l s
pool
a
reapr
ess by 几 a
ed nd
s
pri
ngs
of
s
ti
ff
nes
s
cv.Me
cha
nic
al
p
res
sur
e fe
edbac
k ca
ll
s f
or
a ve
ry
speci
al
valve
,which
has
to
be
des
igned,
manufac
ture
d,
tes
ted,
e
tc.
This
me t
hod
is
t
her
efor
e not
v
ey
r a
ttr
act
ive,
s
ince
t
her
e i
s
no
chance f
or
mass-pr
oduct
ion.
A
v
ery
us
eful
equi
val
ent
i
s
ele
ctri
cal
pr
essur
e f
eedbac
k,
whic
h pr
ovi
des
feedbac
k of
th
e el
ect
ric
all
y
me
asu
red
lo
ad p
res
sur
e凡 ‘P ,
一PZ t
o t
he
i
npu
t o
f a
n e
le
ctr
o-
hyd
rau
li
c s
erv
ova
lve
. T
he
me
tho
d i
s
i
ndi
ca
te
d i
n
Fig.
3.
4
a)
by
He
,
Pr
ess
ure
f
eed
bac
k s
ti
ll
h
as
t
he
d
isa
dva
nta
ge
o
f i
ncr
eas
ed
l
oad
s
ens
it
ivi
ty.
Thi
s d
isa
dva
nta
ge,
ho
wev
er,
c
an
b
e a
voi
ded
by
i
nt
rod
uci
ng
d
yna
mic
p
res
sur
e f
eedb
ack
wi
th
a
t
ra
nsf
er
f
unc
ti
on
H
63
i
n
Fi
g.
3.
4
a)
pa
ral
l
el
t
o]/
Ch.
H 二K 卫_
姗 十 1
At
lo
w f
re
que
nci
es
H,=0
a
nd
t
he
l
oa
d-s
ti
fne
ss
i
s
th
ere
for
e
no
t
sp
oil
ed.
At
hig
her
re
fque
nci
es
,
wh
ere
t
he
e
ff
ec
t
of
CO/Ch
is
l
es
s
(s
ee
S
ect
i
on
3
.3
),
H,
,=Kv
wi
l
l
pr
ovi
de
a
deq
uat
e
da
mpi
ng
i
f
1
/2《 u)
o o
r
0)
0,
r》 1.
Dyna
mic
(some
times
cal
led
der
iva
tiv
e)
pres
sur
e f
eedbac
k cal
l be
est
abl
ished
eit
her
mechani
cal
ly or
ele
ctr
ica
ll
y.
Fig
.3.
4b) s
hows sche
ma t
ic
all
y two hydr
o-me
chani
cal
f
il
ter
s
co
nn e
c t
ed
t
o t
he
cy
lin
der
compa
tme
r nts
.
The
fi
lt
ers
i
ncor
porat
e a
(
lami
nar
)
low
f r
esi
st
ance
Ra
nd
a
v
ey
r s
mal
l
(g
as
o
r l
f
uid
) a
ccu
mul
at
or
wi
t
h a
c
apa
ci
ty
Ca.
The
t
i
me
c
ons
ta
nts
i
ndi
ca
te
d i
n
th
is
f
ig
ure
a
re
r=R.
C..Ca
ref
ul
a
nal
ys
is
of
i
t
s e
ff
ect
wi
ll
r
esu
lt
i
n
a
f
unc
ti
on
Hm
i
n
Fi
g.
3
.4
a
)
1 r
s
Hm
2
R,矛Km’
,
rs
+1
To
d
esi
gn
i
n pr
act
i
ce
a
me
cha
nic
al
dy
nami
c
pr
ess
ure
f
eed
bac
k u
sin
g "
res
tr
ict
ed"
a
ccumu
lat
ors
,
i
s
no
t
to
o e
asy
.
To
d
et
er
min
e R,
,
and
c
h
(i
nt
ro
duc
ed
b
y p
res
sur
e
fe
edb
ack
) we
a
ssu
me
n
o i
ni
ti
al
d
amp
ing
a
t a
ll
,
so
wi
t
h w=0
‘一Kv
.
CO
two Ch
De
sir
ine
K.
.
‘.声
/
u)二C(
二0.
4)
we
r
equi
re
C,=0.
F l t沙
5C =0.
.,
5c.
As
、
s
umi
ng
a
s
ymme
,
tr
ic
mo
tor
, 二 ,舀 ,J 几夕 一 -
F
ig.
1.
11
c)
gi
ve
s
c.
。二4A到s,
so
C
,一0
.5
c。
一2A
E S
To
d
ete
rmi
ne
R we
ir
fst
l
ook
a
t
hi
gh
f
requ
enc
ies
(
(or>>1)
I
H
2
尺
R
n2
A2
K} Ch
Co
mbi
nat
ion
g
ive
s
工
j
,1, - 月月』
‘‘ 一L ,
C人 Ch
2A到:
二
二
一
}
R,=
J,
5
K, 2A2K, 2
A2凡
,
2
A2
To
d
ete
rmi
ne
t
he
a
ccumul
at
or
v
olume砚 we
ha
ve
t
o
fi
nd
}
几 一E
卜 r }月刹
a  ̄一
、价 一二 丫
E
!一
J一 5
一
一 5
崎t=叭R,
,
C。二叭
』 ,.
-, A
份n
K,
,
As
卜甘
n
"
n
0
0
洲
C
e
(
Th
e r
at
io
a
ccu
mul
at
or
v
olu
me
V t
o
cyl
i
nde
r
vo
lume
匀
V
K.
.
币一二— ’
0o丁
(
AS 田
64
Ha
vi
ng
K/0Q“0.
( 4
an
d r
equ
ir
in
g
O)
OT>1
,th
e
;ac
cumu
lat
or
v
olu
me
wi
l
l
ro
ugh
ly
b
e t
he
s
ame
a
s
th
at
o
f
th
e
cy
li
nd
er
.
Th
er
ef
or
e
"d
ia
phr
agm”-
-
acc
umu
la
to
r
or
g
as-
ac
cumu
la
to
rs(
Cq=Vu
/1
1
P,)
a
re
mor
e a
ppr
opr
iat
e,
r
esul
t
ing
i
n c
ons
ide
rab
ly
s
mal
le
r v
olume
s.
Th
oug
h e
leg
ant
p
rac
ti
cal
s
olu
tio
ns
a
re
k
nown,
we
ge
ner
all
y pr
efe
r t
he
e
lec
tri
cal
e
qui
val
ent
f
or
mec
han
ica
l d
yna
mic
p
res
sur
e
fe
edba
ck,
whi
ch
g
ive
s s
car
cel
y
an
y h
ard
war
e
pr
obl
ems
.
Pr
oba
bly
t
he
mo
st
a
tt
rac
tiv
e d
amp
ing
me
tho
d f
or
e
lec
tro
-hy
dra
uli
c s
erv
osy
ste
ms
i
s
ob
tai
ned
b
y
a
cce
ler
ati
on
f
eed
bac
k a
s s
hown
i
n t
he
bl
ock
d
iag
ram
of
Fi
g.3
.4c
).
I
t i
s ba
sed
o
n e
lec
tri
cal
ly
me
asu
red
a
cce
ler
ati
ons
o
f t
he
p
ist
on
(
in
r
ota
y
r mo
tor
s de
ri
ved
ro
f m
t
acho
gen
era
tor
s
ign
als
).
Re
arr
angi
ng
t
he
bl
ock
di
agr
am
t
o
Kv
「 2
〕
Is
_。 s _I
H(
s)=
几
wu
2+
2‘W
O+‘
{ (
3.1
3)
K K-
s2
1
+-下万一二
IS
一 -万
 ̄。 S _I
S
I
- +LC
,一 +1I
LC
0o C
OO 」
r
es
ul
ts
i
n
an
i
nc
re
ase
A百of
th
e
da
mpi
ng
c
oef
f
ici
en
t杏
△杏=0.
5KaKvt
vo
(
3.
14
)
Thi
s v
ery
el
egan
t
da
mpi
ng
met
hod
hs
a b
een
appl
i
ed n
umer
ous
t
imes wi
th
co
mple
te
s
ucc
ess
.
Yet
s
ome manuf
act
ure
rs
of
e
xpe
nsi
ve a
nd comp
lex s
yst
ems
see
m r
eluct
ant
t
o us
e a
cce
ler
ati
on
f
eedb
ack
a
nd
s
ome
ti
me
pr
efe
r
obs
ole
te
d
amp
ing
me
tho
ds.
Th
e v
ali
di
ty
o
f Egn
s.(
3.1
3)
a
nd
(
3.1
4)
i
s imi
l ted
o
f c
our
se
b
y t
he
v
ali
dit
y o
f t
he
mat
hema
ti
cal
mo
del
a
s s
hown
i
n Fi
g.3
.4a
) a
nd
t
he
a
ssu
mpt
io
n G(
s)=1.I
n mo
re
c
ompl
ic
ate
d s
yst
ems
,
in
cor
por
at
in
g c
ompl
i
ant
s
tr
uct
ure
s,
e
tc.
(
see
s
ect
ion
3.
10)
t
he
me
tho
d c
ann
ot
be
us
ed
wi
tho
ut
s
eri
ous
c
ompl
i
cat
ions
.
Ho
wev
er,
t
his
i
s
tr
ue
f
or
t
he
ot
her
da
mpi
ng
met
hods
s
a we
ll.
3.
6 ELECTRO-
HYDRAULIC
SERVOVALVE
I
n Fi
g.3
.5
a
) t
he
wi
del
y
us
ed
t
wo-
sta
ge
e
lec
tro-
hyd
raul
i
c s
erv
ova
lve
i
s
de
pic
ted
s
che
mat
i
cal
l
y.
Th
e v
alv
es
a
re
c
urr
ent
ly
u
sed
t
ype
of
s
erv
o v
alv
e i
n i
ndu
str
y,
wi
th
t
he
mo
st
p
opu
lar
v
ers
ion
b
ein
g
ma
nuf
act
ure
.
The
a
rma
tur
e i
s
he
ld
i
n a
ba
lan
ced
p
osi
ti
on
b
y a
p
erma
nent
magn
eti
c f
ie
ld
c
aus
ed
by
t
he
u
ppe
r a
nd
l
owe
r po
le
p
lat
es
s
urr
oun
din
g i
t.
Th
is
f
iel
d c
an
b
e u
nba
lan
ced
by
a
n op
pos
ing
ie
f l
d
gene
rat
ed
by
t
he
a
rma
tur
e coi
ls.
Thi
s unbal
ance
d f
iel
d ca
use
s t
he
a
rma
tur
e t
o r
otat
e a
bout
t
he
bas
e
o
f t
he
be
yl
r l
ium
c
opp
er
le
f x
ure
t
ube
wh
ich
i
s
an
cho
red
t
o t
he
v
alv
e b
ody
. As
s
hown
i
n F
ig.
3.
5,
a
t
ube
-wh
ich
a
cts
a
s t
he
f
lap
per
a
nd
a
lso
c
ont
ai
ns
t
he
c
ant
il
eve
r s
pri
ng-
is
f
it
te
d i
nsi
de
t
he
le
f x
ure
t
ube
.
As
sumi
ng
t
hat
t
he
e
lec
tri
cal
s
ign
al
t
o t
he
f
orc
e mot
or
wi
ll
c
aus
e a
c
loc
kwi
se
t
orqu
e o
n t
o t
he
a
rma
tur
e,
t
his
t
orq
ue
wil
l r
ota
te
t
he
la
f p
per
t
ube
i
n a
cl
ockwi
se
di
rec
tio
n.
I
t s
hou
ld
65
r  ̄ ̄
加,
qne
矩T
t
e
mot
一
or
J
,
0
+
Bs
+
c b)
ctn an a e
b
f
lapper-
nozz
le(
a)
(
pil
ot
st
age)
fe
edbac
b s
p,r
ng
一 p
lot
pressr
a
e
叨o
olta,
)
,
e,tr,crm n
+)
In
ter
act
ion
s
poo
l一刀a
ppe
r
f
l,
j1
'
p
R
a,
h I
f
+
'
S一 1一
1
1
C
,’o
尸s Ct
t
ot叮t
ie
mot
or
Ka
,
ht
P
-
-" 儿 z+ 刀‘"
-
Cr
R +b R +b s
pool
as
s
ei
r
o;,
otor
、十B(
R+b
)奋+、
F
/b
, Cb
d
onun
au,
t介e
dbad
pad,
c)
(
hr
R,b
) C
h 乙、+o(
R +1
r)
e)
Fi
g.3.
SA
nal
ysi
s
of
t
wo-
st
age
e
lec
tro
-hy
dra
uic
l s
erv
ova
lve
b
e p
oin
ted
o
ut
h
ere
,
th
at
t
hes
e mo
veme
nts
a
re
v
ery
s
mal
l.
Thi
s
ro
tat
ion
o
f t
he
la
fppe
r c
han
ges
t
he
c
lea
ran
ce
b
etwe
en
t
he
la
fppe
r a
nd
t
he
n
ozz
le
f
ace
s,
i
ncr
eas
in;
t
he
l
eaka
ge
f
rom
t
he
r
igh
t-h
and
noz
zle
and
decr
easi
ng
it
fr
om the
lef
t-
hand
one.
Thes
e noz
zles
ar
e s
u p
plie
d wi
th
lu
f id
via
f
ixe
d
or
if
ice
s i
n a
f
il
te
r l
ocat
ed
at
the
val
ve'
s p
res
sure
con
nect
ion,
a
nd t
he
lu
f i
d c
hanne
l t
o
th
e n
ozzl
es
is
al
so
connect
ed
to t
he e
nds of
ten
ma i
n va
lve
spool
ch}
ai
ge t
o t
he nozzl
e l
eak
age,
cha
nges
the
66
ga
ll
ey
r pr
ess
ure
behi
nd
t
he n
ozzle
s.
So,
as
t
h e
l
eakage
fr
om t
he
ri
ght-
hand
noz
zle
is
i
ncrese
a d,
th
e
pr
ess
ure
in
t
he
rig
ht-
han
d gal
ler
y is
r
educed.
The
res
ult
ing
fo
rce
imbal
anc
e i
s t
owards
the
ri
ght,
ca
usi
ng t
he
spo
ol
to
move
in t
hat
dir
ect
ion
. Thi
s s
pool
movement
co
nnect
s
the
ri
ght
- h
and pi
st
on
p
rop
ort
i
ona
ll
y t
o
th
e pr
ess
ure
po
rt,
a
nd
c
onn
ect
s
le
ft
-ha
nd
p
ist
on
t
o r
et
un.
r Th
e s
poo
l
mov
eme
nt
al
so
ta
kes
wit
h i
t
th
e fe
edback
cant
il
ever
spr
ing,
whi
ch r
eact
s ba
ck on
to
the
armature
i
n opposiio
t n
to
t
he
gene
rat
ed
ma gne
tic
fi
el
d.
This
r
educes
the
res
ult
ant
cl
ockwise
t
orque
on t
he la
f p
per,
c a
us i
ng
it
t
o move
to t
he r
ight
and r
eadj
ust
the nozzl
e l
eakag
es,
wh i
ch i
n t
urn changes
the gall
ery
pre
ssu
res
.
Wh en
th
e net
for
ces
of
low
f r
eact
ion,
ar
mature
t
orque,
ca
nti
leve
r spr
ing,
and gall
ey
r
pr
essu
res
bal
ance,
the s
pool
will
st
op moving
in t
he sp
oo l
posi
ti
on re
qu i
red by t
he i
nput
el
ect
ri
cal
si
gnal
.
The block
dia
gram (Fig.
3.
6) i
ll
ust
rat
es t
he cl
osed l
oop of
the va
lve, which
ca
uses
a
ny
ext
raneous
di
st
urba
nce
t
o be
cor
rect
ed.
.住rl
naw re
1
10tHdona丈
二!
t
orque
几a pper S
poo
ifor
re
de
fecti
on ba丈a,
:亡‘ Ou
t
p之‘
亡
i
l
(
p
ur)
竹t口f夕叶t
ellt
圆
Fi
g.3.
6
Bl
ock
d
iag
ram
f
or
t
he
a
cti
on
o
f
S
a
Cal
PI
la
f p
t
公t
pe
t
i
;
i
t
o
r
t
y
ei
r
vr
'
qu
e
pe
s
evo
rva
lve
F i
rs
t
of
a
ll
an
el
ect
ri
cal
s
igna
l
is
conve
rt
ed
in
to
a
mec
hani
cal
s
igna
l,
a
s t
he
c
ontr
ol
cur
ren
t i
r
esul
t
in
a
t
orqu
e
T whi
ch
cause
s a
r
o t
aio
t n
0
o
f t
he
a
rmat
ur
e
toge
ther
wi
th
t
he
la
f ppe
r-a
rm.
The
(
simpli
ie
f d)
bl
ock
diagra
m of
th
e t
orque
motor
i
s
gi
ven
in
Fi
g.
3.5
b)
.
Th
e
fir
st
st
age
(ote
f n ca
ll
ed pi
lot
s
tage)
of
t
he
val
ve c
onsi
sts
of
a s
ymme t
ric
la
f pper-
n oz
zle
s
ystem,
fed
by t
he cons
tan
t s
upply
pres
sure
vi
a t
wo
rest
ri
ct
ions.
I
n t
he s
tat
ionar
y si
tuat
ion an
e
xc u
rsi
on
9R of
the la
f ppe
r r
esul
ts i
n two di
fer
ent
pil
ot
pres
sure
s.
Assuming a symme t
ric
c
ons
tr
uct
io
n,
a
ll
o
ri
ic
f e
a
rea
's
b
ein
g e
qua
l a
t0=0,a
n "
und
erl
ap"
u
wi
th一u<朗 <u,t
he
p
ilo
t
pr
ess
ure
s c
an
b
e d
eri
ved
d
ire
ctl
y.
OR
PP
A 目
=
Pp
l
-
P
p2
二 1 _ 1 一一止生_ 二
9R
(
3.1
5)
P
,
Ps
1
+(
1一
丝),1
+(
1+
丝),1
、工(
8R
)4 “ “ 4 “
u
A
s
lo
ng
a
s
I
9R
I<0
.8u
i
t
s
ch
ar
ac
te
i
rs
ti
c
i
s
a
lmo
st
pe
rf
ec
tl
y
l
in
ea
r.
I
n t
he
dynamic
sit
uati
on t
he fou
r o
if
r ice
s of
the
pi
lot
s
tag
e co
ntr
ol
th
e oi
l
flo
ws to
and
fro
m the
"c
ylin
der"
compart
me nts
on bot
h si
des of t
he mai
n spool
of
the
servo
valve,
by whic
h t
he
se
rvomot
or i
tsel
f i
s c
on t
rol
led.
Accordi
n g
to t
he
mod el
of
hydr
aul
ic s
ervomoto
rs de
rived
in
Chapt
er
2,
t
he main
spool
i
tself
sh
ould
be co
nside
red
as
a v
ey
r smal
l
serv
omo t
or.
Fi
g.3.
5 b)
gi
ves
t
he
b
loc
k d
iag
ram,
co
mpis
rin
g t
he
we
ll
-k
nown
p
ara
met
er
s
Km,
Ch,C
a,M
and
w.
67
Th
e d
omi
nant
f
eedb
ack
pa
th
i
s
pr
ovi
ded
by
t
he
f
eedb
ack
s
pri
ng.
An
e
xcu
rsi
on
x
o
f t
he
mai
n
v
alv
e t
oge
the
r
wi
t
h a
f
l
app
er
a
rm
r
ot
at
io
n 8
r
esu
lt
s
i
n
a
f
or
ce
[
x+e(
R+b
)]c
b
at
th
e
en
d o
f
th
e
f
eed
bac
k s
pri
ng.
Th
e r
esul
t
ing
t
orqu
e a
cts
a
s a
d
ire
ct
f
eedb
ack
i
ndi
cat
ed
i
n t
he
l
owe
r f
eed
bac
k
p
ath
of
Fi
g.
3.
5 b
).
The
s
ti
ff
nes
s c,o
f t
he
f
eedb
ack
s
pri
ng
ha
s a
v
it
al
f
unc
ti
on,
wh
ich
wi
ll
b
e
d
isc
uss
ed
a
gai
n
la
ter
on
.
By
r
eac
ti
on
t
he
f
eed
bac
k f
or
ce
[
x+8(
R+b
)]c
b
of
t
he
s
pr
in
g s
hou
ld
b
e r
ega
rde
d a
s
a
l
oad。
。
t
he
s
poo
l.
An
ot
he
r
"s
pri
ng"
f
or
ce,
c
f
x,i
s
ca
use
d b
y a
xia
l
lo
fw f
or
ces
i
n
th
e p
ort
op
eni
ngs
o
f
th
e
s
poo
l.
The
u
ppe
r
le
ft
f
eed
bac
k pa
th
i
n F
ig.
3.
5
c)
re
pre
sen
ts
t
he
i
nte
rac
tio
n b
etwe
en
ma
in
v
alv
e a
nd
l
f
app
er
,
a
s
th
e
lo
ad
p
res
su
re明 i
mpl
i
es
a
f
ee
dba
ck
f
or
ce
o
n
th
e
la
fppe
r.
It
s
i
nf
lu
enc
e
sh
oul
d
be
smal
l.
The wh
ole
bl
ock d
iagr
am, s
impli
fi
ed
as
i
t i
s,
s
ti
ll
i
s t
oo c
ompl
ica
ted
t
o be
deal
t
with
her
e i
n
d
eta
il
.
Bri
ely
f we s
imply
st
ate
her
e t
hat
t
he
l
ower
f
eedbac
k pa
th
i
s
domina
rt.
Con
seque
ntl
y a
t l
ow
f
requenc
ies
t
he
t
ransf
er
f
unct
ion
becomes
,
1、
1 x=
M 一
K,
3.
( 161
5
-70
i (
R -
-b)
c,,
I
t i
ll
ust
rat
es
t
he
wel
l-
-
known
f
act
t
hat
t
he
cl
osed
l
oop
t
rans
fer
f
unct
ion
of
a
cont
rol
s
yst
em with
H
(
s)>>l
e
qua
ls
(
app
roxi
mat
el
y)
t
he
r
eci
pr
oca
l
of
t
he
f
eedb
ack
p
ath
t
ra
nsf
er
f
unc
ti
on.
Thi
s knowledge e n
ables us to explai
n how s
e r
vova l
ve charact
eri
sti
c c
ou l
d d
eter
ior
ate
compl e
tel
y by
a cl
earance
or
backlash
betv
N
een
the ball
at
t
he end of
the f
eedback s
prin
g and
the
slo
t i
n the
spool
.
Us i
ng Egn.(3
.16),
Fig
3.5 c)
and d)
sho w how the
presenc
e of
su c
h a
clear
ance
causes a val
ve charact
eri
st
ic with av ert
ica
l "jump" .
This is ver
y per
ilo
us in clos
ed loo
p
ser
vos y
stems,
as
a comparable
jump in t
he vel
ocity
ga i
n chara
cteri
st
ic of
the
servo wi
ll
appea
r,
making t
he sy
ste
m co mplet
ely
unst
able
at
low spe
e ds.
I
n
the t
able
be
low some import
ant da
ta of
t
he Mo og val
ve have
b e
en summa r
ized.
On th
e bas
is
of
ma nufa
ctur
er'
s s
pe ci
ic
f at
ions o
n e
could
repr
esent
the s
ervovalv
e tr
ansfer
f
u n
cti
on in
prac
tic
e b
y
a f
ir
st o
r s
econd
order s
yst
em, dependi
ng on
the
fr
e quency r
ange i
nv o
lved.
一
十!
Tabl
e11
Pr
ope
rti
es
of
Moog
s
ervov
alves
一j }
S
er
ie
s一
Q
i
n
dn
i
hr
un
30
1.1一7.6 230 Hz
!
31
6.
5一26 230
Hz }
{
32
26一53 :
50 Hz }
34
} 35
48一72
72 一1 13
Zf
Ol
)HZ
i
z
{
}
} 76 上一,
6.
5-
1
13 一_
2.
3-
39 50-9c}Hz ;
3.
7 DERI
VATI
VE
CONTROL,
MODES-
(
VEHI
CLE
STABI
LITY
AUGMENTA'
T'
ION)
One
o
f t
he
mos
t c
ommo
n a
ppl
ica
ti
ors
o
f de
riv
ara
、。c
ont
rol
, u
sua
ll
y i
mpl
eme
nte
d o
n t
he
68
co
ntr
o l
l
ed var
iab
le,
i
nvolv
es th
e i
mp r
ovement
of
vehi
c le
ha
ndli
ng dynamics
by
fee
dback
means.
The
conce
pt was
pr
obabl
y or
igina
lly
devi
sed
for
ai
rcr
aft
but
is
no
w use
d on
all
t
ypes
o
f vehi
cl
es.
Ai
rcr
a f
t
at
ti
tude
(an
gular
or
ient
ati
on)
cont
rol
i
nvolve
s thr
ee
axes
(pit
ch,
ro
ll,
a
nd ya
w)
of mot
ion,
howe
v er
pr
elimi
nay
r desi
gn usua
ll
y cons
ider
s e
ach axis
se
parat
ely.
A si
mplif
ied
Newton'
s l
aw
a
nal
ys
is
f
or
t
he
p
it
ch
a
xis
p
rod
uce
s
th
e a
ir
cr
af
t
t
ra
nsf
er
f
unc
ti
on
r
el
at
in
g p
it
ch
a
ngl
e
9c
t
o
el
ev
ato
r
a
ngl
e占*
.
却
KQ
(
z,
,
s+1
)
一
一
_2
(
3.1
7)
含
2示 ,
、
s(一 犷 咔・一 +I
)
、,、乙
山朋
田n
a
T
he
n
ume
ri
ca
l v
alu
es
o
f p
ara
met
er
s KQ
,2Q
,C
c1
,
,,
a
nd
}v
a
ry
f
or
m o
ne
a
ir
pl
ane
o
r mi
ss
il
e t
o
a
not
her
,
an
d f
or
a
g
ive
n ve
hic
le,
wi
t
h a
lt
it
ude
a
nd
Mac
h nu
mbe
r.
Ae
rody
nami
c de
sig
n f
or
d
rag
r
edu
ct
io
n o
ft
en
r
es
ul
ts
i
n
ve
ry
p
oor
(l
ow)
v
alu
es
o
f
ai
r
cra
ft
d
amp
ing杏 Us
e
of
d
eri
va
ti
ve
c
ont
r
ol
c
onc
ept
s a
ll
ows
u
s t
o r
adi
cal
ly
i
mpr
ove
t
his
po
or
s
tab
ili
ty
wi
tho
ut
c
han
gin
g t
he
a
irp
lane
i
n a
ny
wa
y,
t
hus
p
res
erv
ing
t
he
de
sir
abl
e l
ow
d
rag.
Fe
edb
ack
c
ont
rol
i
s t
hus
a
p
owe
rfu
l t
ool
i
n v
ehi
cl
e
d
esi
gn
s
inc
e i
t
al
lows
t
he
s
imu
lta
neo
us
f
ulf
il
lme
nt
of
c
onli
fct
ing
p
erf
orma
nce
r
equ
ire
men
ts.
A
b
asi
c
pi
t
ch-
comma
nd
c
ont
ro
l s
yst
em
wo
uld
f
or
ce
t
he
a
ir
cr
af
t
pi
t
ch
a
ngl
e
ec
t
o f
ol
lo
w t
he
p
il
ot
's
c
ont
r
ol
s
ti
ck
r
ot
at
io
n8,
,,u
si
ng
a
p
osi
t
ion
g
yro
sco
pe
t
o me
asu
re
B
c a
s i
n F
ig.
3.
7.
T
he
e
lec
tro
hyd
rau
hc
s
erv
o t
hat
po
wer
ful
l
y p
osi
ti
ons
t
he
e
leva
tor
i
n r
esp
ons
e t
o v
olt
age
c
omma
nd
e,i
s
i
ts
el
f a
c
omp
let
e f
eed
bac
k s
yst
em;
h
owe
ver
, wh
en
p
rope
rl
y d
esi
gne
d,
i
ts
c
los
ed-
lo
op
r
espons
e can
be
a
dequat
ely
model
ed
as
f
irs
t or
der
. We
s
ee
f
rom
the
r
oot
l
ocus
t
hat
t
his
bas
ic
s
yst
em
i
s u
nac
cept
abl
e s
inc
e t
he
c
los
ed-
loo
p d
amp
ing
i
s a
ctu
all
y wor
se
t
han
t
hat
of
t
he
a
irp
lan
e
i
t
sel
f
,
wh
ich
wa
s
al
r
ead
y b
ad.
By
a
ddi
ng
a
r
at
e
gy
ros
cop
e t
o
pr
ovi
de
me
asu
reme
nt
o
f Sc
we
c
an
i
mpl
eme
nt
ir
fst-
der
iva
tiv
e c
ont
rol
(
see
Fi
g.3
.8)
.
Th
e c
los
ed-
loo
p s
yst
em
now
h
as
be
ter
d
amp
ing
t
han
t
he
a
irc
raf
t
al
one
,
gi
ving
h
and
lin
g q
uait
lie
s a
cce
pta
ble
t
o t
est
pi
lot
s.
S
inc
e t
he
air
pla
ne i
ts
elf
h
as
not
be
en
cha
nged,
the
re
ader
may
fin
d i
t
di
fic
f u
lt
to
be
i
l e
ve t
hat
t
he
i
mprove
d da
mp ng
i r
eal
ly
occu
rs.
This
"
magic
al"
as
pect
o
f de
riv
ati
ve
con
trol
ca
n be
dis
pel
led
by
r
eme
mbe
ri
ng
t
ha
t
th
e e
le
vat
or
i
s
a
"
gen
era
l-
pur
pos
e
to
rqu
e p
rod
uce
r";
ro
tat
i
on
o
f c
5
,,
ca
use
s
a
pro
por
t
ion
al
p
it
chi
ng
t
or
que
o
n t
he
ve
hic
le.
The
d
eri
va
ti
ve
co
ntr
ol
s
imp
ly
s
ends
a
0c
s
ign
al
t
o
th
e
e
lev
at
or
so
t
ha
t i
t
pr
odu
ces
a
t
orq
ue
oppos
it
e
and
pr
opo
rt
io
nal
to
.c,th
e s
ame
k
ind
of
t
or
que
p
rod
uce
d b
y v
isc
ous
d
amp
ing
a
s t
he
a
irc
raf
t
ro
tat
es
t
hro
ugh
t
he
a
ir.
Once
t
his
i
s
cl
ear
,
on
e c
an
a
lso
69
/
尹
El
erat
or i二
洲尸
aero
心warni
c
for
ce
Cent
er 乓
户洲 o
f ma
ss
已下 ‘ 叫 竺二
Pi
lot'
s
st
ick Air
craft
and
Srnnr
ning
El
ect
ro pi
tch
pot
ent
iomet
er anrpl
i
frer 为.r
ranri
cs
hrdrar
dic
se刀,>
0
KQ(
Tas+1)
尸osi
tion
那 y
ro
Fi
g.3
.7
Ai
rcr
aft
p
itc
h-a
ngl
e
co
ntr
ol
wi
t
hou
t s
tabi
l
ity
a
ugme
nta
i
ton
s
ee
wh
y t
hi
s
sc
heme
wi
ll
no
t wo
rk
o
n a
mi
ss
il
e
tha
t
le
ave
s t
he
a
tmo
sph
ere
,
si
nce
t
he
el
eva
tor
i
s
n
ow i
ncapabl
e of
produ
cing any
to r
que.
We c
an now, ho
we ver
, us
e gas
jet
s as o
ur t
orque
p
roduci
ng devi
ce,
an
d der
ivat
ive
contr
ol c
an a
gain
produc
e damped moti
on e
ven
though
vehic
le
v
isc
ous
da mpin
g i
s
zero
i
n empty
space.
For
a
vehi
cle
(i
l
ke t
he sp
ace
shutle)
t
hat
ope
rat
es
both
in
s
pace
and i
n t
he at
mosphere
, a c
o nt
rol s
cheme bl
endin
g both ae
rodynamic
contr
ol s
urf
aces
(
elevat
or,
r
udder
,
et
c.)
and
gas
j
ets
i
s
nece
ssar
y.
70
K
(
r}s+1
)
Fi
g.3.
8 Der
iva
tive
c
ont
rol
a
ugment
s a
irc
raf
ts
tabi
li
ty
3.
8 BLOCK
DIAGRAM
FOR
HARDWARE
COMPONENTS
Bl
ock
diag
ram f
or c
ont
rol
s
ys t
ems
obv
ious
ly r
epr
ese
nt
hardwar
e c
omponent
s.
However
,
the
r
ela
tio
ns b
etween
the
par
amet
ers in
t
he t
heor
eti
cal
bl
ock d
iagr
am and
th
e c
har
act
eri
st
ics
of
ha
rdware
c
omp o
nent
s a
re
some
times
ha
rd
to
fi
nd.
The
ref
ore
,
it
makes
s
ense
t
o make
t
wo t
ypes
of
bl
ock
di
agr
am,
one
f
or
t
he
t
heor
eti
cal
cont
rol
beha
vior
a
nd
one
f
or
t
he
har
dwar
e.
As
bot
h
di
agr
ams
s
hou
ld
c
ove
r t
he
s
ame
s
yst
em,
the
r
ela
ti
ons
c
an
e
asi
ly
be
f
oun
d.
To
i
ll
ust
rat
e t
his
,
Fi
gs.
3.
9
a)
a
nd
b
) pr
ese
nt
t
wo
bl
ock
d
iag
rams
a
cco
rdi
ng
t
o c
ont
rol
th
eor
y,
o
f a
s
i
mpl
e
li
nea
r
el
ec
tr
o-h
ydr
aul
i
c s
erv
osy
st
em
wi
t
h a
cce
le
rat
i
on
f
ee
dba
ck.
Th
e e
xt
ra
d
amp
ing△百
c
an
b
e f
oun
d b
y e
qua
liz
ing
bo
th
l
oop
ga
ins
(
see
a
lso
Se
cti
on3
.5,
Fi
g.
3.
4
c)
).
△杏=0.
5KQ
K(o
o
I
n Fi
g.3.
9
c)
t
he
"
har
dwa
re"
b
loc
k d
iag
ram
i
s
pr
ese
nte
d.
I
n a
g
ive
n s
yst
em
we
h
ave
t
he
f
oll
owi
ng
c
omponent
s:
.s
erv
ova
lve
ampl
fe
i r
t
o c
onv
ert
t
he
c
ont
rol
e
rro
r s
igna
l£(
in
Vo
lt
s)
t
o t
he
s
erv
ova
lve
i
npu
t
c
ont
rol
cu
rre
nt
x
, (
in
Amp
?
r
es)
.
.s
erv
oval
ve
c
onve
rt
ing
x
, i
nt
o lo
fw x
, (
in
i
n’I
s)
.
hi
c
omme
rci
al
c
ata
log
s a
vai
la
ble
da
ta
mi
ght
i
ndi
cat
e mi
ll
ia
mps
a
nd
g
all
ons
p
er而n
ute
. Th
ese
d
ata
mus
t b
e c
onv
ete
r d
i
nto
ISO-
uni
ts.
.po
sit
io
n t
ran
sduce
r(K3),
giv
ing
Vol
ts
p
er
mete
r.
.a
c
cel
erat
i
on t
ran
sduc
er(
K4 )
,giv
ing
Vol
ts
p
er
m/s’
.
・mo
tor
,
wi
t
h p
is
ton
a
rea
A (
i
n m
z
)an
d ma
ss-
sp
ri
ng
s
yst
em
p
ara
met
er
s (
00
a
nd杏.
T
he
n
at
ur
al
fr
eq
ue
nc
y (
D
o=沂石
不 m
igh
t
b
e
de
t
er
mi
ne
d
by
ca
l
cul
a
ti
ng
(
o
r
es
t
ima
t
in
g)
C
a
f
ro
m F
ig.
l.
11.
I
t
i
s
al
s
o p
oss
ibl
e
t
o
me
asu
re
C
t
)o
a
ndge
xpe
ri
men
tal
l
y.
Th
e g
ain
V1
a
nd矶i
n
Fi
a.
3.
9
c)
c
an
b
e c
hos
en
re
fely
t
o
sa
ti
sf
y
th
e d
esi
re
d a
dju
stme
nt
o
f K,
,
a
nd
K,o
f
Fg.
3.
9a
)
.
Eq
ual
i
zin
g t
he
p
rod
uct
s
of
t
ra
nsf
er
f
unc
ti
ons
i
n
th
e v
ari
ous
l
oop
s i
n瑰.
a)
a
nd
F
ig.
c)
y
iel
ds
V
,
K,
K,
.工
A
.
K,一K
1
V
1
K1
K;
’面‘
K4
V
s=KV
K,
Th
e f
ore
goi
ng
r
esu
lts
c
an
b
e s
umma
riz
ed
i
n t
he
ina
f l
ad
jus
tme
nts
2)C
AK,
, K}
K3
式 。= 一 , V = V2二
K,
叭 K1
K,
K3 K4
Thi
s e
xampl
e j
ust
i
ll
usra
t t
es
h
ow
one
c
oul
d t
ry
t
o f
ind
t
he
a
dju
stme
nts
. Ho
wev
er,
i
t i
s no
t
a
lwa
ys
p
oss
ibl
e
to
t
ran
sla
te
t
heo
ret
i
cal
di
agr
ams
d
ire
ctl
y
int
o h
ardwa
re
di
agr
ams
.
By
f
ocu
ssi
ng
on
s
pec
ial
a
spe
cts
, ho
wev
er,
i
n ma
ny
c
ase
s g
ood
a
ppr
oxi
mat
ions
c
an
be
ma
de
t
o e
nabl
e a
deq
uat
e
pos
sibi
li
ti
es
f
or
ca
lcul
ati
on
or
es
tima
tion.
_一一竺工一一-
V
l买+:
;s十,
I
U'
}
W
"
]
w 4
72
1
ts
n
ne
芝
曹
. 、‘
o
n
mp
.
犷
才
.卜 尸 1 ! J
o
e
c
‘勺〕,
心,晚
、二三 ,
』之七 ̄
J
星
妇二十x
r
引.
wa
‘
.,、
产1 夸
尸! 之
rd
叽
ha
r
| 」 岛
祝
+刁
o
刀‘
卜,
〕是 月
,,l丫
Jl t
f吸
、
m
f
筋
f共 十
ara
g
、
卜对小
?卜储一 夏 、
曹! 、
孙、 冬之 之
i
k
个+ 全
k ,、
d
二
S
,
,、
c
1
、!
o
1
. 护
挂 .
1 ..
上 日
 ̄
‘ 巨
! I
L
卜
. 一
‘ 口
.JJ., 工 .< 一
三.1怪...1 ‘V 一
.书几笼! }』 』 一
几
Bl
川
如翻 I
9
se wewe曰1.
se
l」 、
家1.
气 ,.
11,.口 七暑 冬、
二、
.
趁曳诱势 叫之之、
势1好
,1、
、t
、
g.
Fi 3
3.
9 PI-
CONTROL
For
some
a
ppl
ica
tio
ns
of
hydra
uli
c s
erv
osyst
ems
a
PI-
cont
rol
ler
mig
ht be
use
d in
ord
er t
o
i
ncr
eas
e t
he
op
en
loop
gai
n a
t l
ow re
f qu
encie
s a
nd
to
de
crea
se c
onst
ant
-lo
ad e
rror
s a
s wel
l
as
"
acc
ele
rat
i
on"
e
rr
ors
.
The
c
hoi
ce
o
f
th
e "
es
r e
t
ti
me"
Z,
of
t
he
PI
-c
ont
rol
le
r
sh
oul
d
be
c
ons
ide
red
c
are
ful
ly
,
as
t
he
t
ran
sfe
r
fu
nct
ion
2i
s+1
H(
s)= (
3.1
8)
公i
s
c
ont
ai
ns
a
do
ubl
e i
nte
gra
tor
,
ca
usi
ng
s
eri
ous
s
tab
ili
ty
p
rob
lems
. To
s
ee
t
his
we
i
gno
re
f
or
a
mo
men
t t
he
s
eco
nd
o
rde
r
sy
st
ems
(
so叭 =0)a
nd
c
hoo
se
t
he
v
al
ue
o
f Ko
n
t
he
b
asi
s
of
a
p
has
e ma
rgi
n o
f
45
0。
0.
6
几, ... . .1
-
从
“\声 之乏 义 t
=‘
尸口,.侧
气
0.
{ ) \
a2 z彩 冲 1口
)1
(
艺
尸 夕
尸.,一一一 一
一 一
尸/
.ofr= 16
尸/
I 刀
丫
洲
1
. l
1 .一- ̄ 口口目叫 一 , ̄
,
碑洲
l
尸护
A ?;
— /{
,
8 Wo
T,,
..曰.臼....口.口...
一L I
/
0 04
1
1.04 0.1 0.
2 0.
4 饮6
t '
4
2
.
‘
4
tr 口
巨.舀 .
t
+
fi
口,
,J
才
.厅百
‘
一尸产一 么 分入
!
一 一 ...印油....自..‘
‘币二 二..自.卜,... 口口.曰曰.户
「-
‘
] 四长 日‘
一一
灯 0.4
\ }
八
入 。
」
t
1
一
、 }
一
1价_,
s.叹娜 F¥‘,_ \\一
i 花 招 认 v
!梦 、
00
, rj
=
2S 时 _ }、 \
叹2 火
\
\
坎丁
口-
b
)
Fi
g.3.
10
PI
-cont
ro
74
The
result
is
K,
T,:0
.5
万 o
r
Kv/
。。_0
>.
7
/
。。:
‘ (
3.
19
)
s
o t
he
g
ain
Kv
s
hou
ld
b
e c
hos
en
n
ot
t
oo
s
mal
l
i
n
or
der
t
o
pr
ovi
de
a
n a
deq
uat
e
da
mpi
ng・
Re
sul
ts
o
f c
omp
ute
r
ca
lcu
lat
io
ns,
u
sin
g t
he
M =1
.3-
cr
it
er
ion
a
nd
Egn.
(
3.1
9),
h
ave
b
een
pl
ot
te
d
i
n
Fi
g.
3.
10
a),
wh
er
e K.
/
0)
o
as
a
f
un
ct
io
n o
f }
wi
t
h NJ
a
s a
p
ara
met
er
s
hows
t
ha
t
up
per
b
ound
ari
es
a
s we
ll
s
a l
owe
r bo
und
ari
es
h
ave
be
en
r
esp
ect
ed.
As
ac
cur
ate
kn
owl
edg
e of
t
he
v
alu
es
o
f Q
)0
a
nd否c
ann
ot
b
e
ex
pec
te
d,
o
ne
s
hou
ld
b
e
ve
yc
r
ar
ef
ul
i
n
ad
ju
st
in
g K,
,
an
dT,
.To
a
voi
d
n
ega
ti
ve
i
nf
lu
enc
es
o
r
ri
sks
d
ue
t
o
th
e I
-a
cti
on峨Zr
>25
i
s
re
comme
nde
d.
F
ig.
3.
10
b
) ma
kes
mo
re
c
lea
r
th
at
o
ne
s
hou
ld
c
hoo
se
t
he
v
alv
e o
f K,
,
/wo
c
orr
es
pon
din
g t
o
t
he
uppe
r bounda
ry
i
n o
rder
t
o obt
ain
a
maxi
mum
bandwi
dth.
Si
nce
PI
-co
ntr
ol
i
ntr
odu
ces
r
eal
ri
sks
wi
th
r
ega
rd
t
o s
tabi
i
lty
,
in
t
he
l
arg
e ma
jor
it
y o
f a
ppl
ic
ati
on
P-
cont
rol
i
s
pr
efe
rabl
e.
3.
10
FIRST
ORDER
SYSTEM
IN
SERI
ES
To
g
ive
am
or
e c
ompl
et
e p
ict
ure
o
f t
he
d
yna
mic
be
hav
ior
o
f h
ydr
aul
ic
s
erv
osy
ste
ms,
t
he
i
nf
luenc
e of
a
f
irs
t
or
der
s
yst
em
i
n s
eri
es,
i
mpl
ying
a
ra
t ns
fer
f
unct
ion
1 K,
H(
s)= (
3.20)
。+1 「.
c
2 ,
_.; 1
S
I一一
一5-
+2}
= +1!
L叭一
oo
c J
i
s di
scuss
ed.
Such
a f
ir
st
or
der sy
stem
could
be i
nt
roduce
d in
a"nat
ural
" way (
for
i
nsta
nce,
b
y
b
ringi
n g
int
o a
ccou
nt so
mei nf
luence
of
the
se
rvovalve
ra
t nsf
erf
uncti
on) or
del
ibe
rat
ely
by a
s
impl
e
RC-
net
wor
k i
n s
eri
eswi
th
a
n el
ect
ro-
hydr
aul
ic
s
ervoval
ve.
The
i
nfluence
of
t
he
t
imeconstant Z
on
t
he
ma
ximum
a
ll
owa
ble
v
elo
cit
y g
ain
K,,
ha
s b
een
i
nve
sti
gat
ed
s
yst
ema
ti
cal
l
y.
Al
lowi
ng一u
st
a
s
be
for
e-a
clos
ed
l
oop
ampl
i
tude
rat
io
lee
曰1
、
=1
H
01
0・
H
0, n ever
e
xce
edi
ng
1
.3,
t
he
v
elo
cit
y ga
in
Kco
uld
b
e c
alc
ula
ted.
Th
e r
esu
lt
s
R
g
a
re
pre
sent
ed
in .
3.11
a),
s
howi
ng a rem arkable i
ncrease
of Kv
/叭a
tpoor
d
ampi
ng
c
oefi
ci
ent
s(心<0.
2).As
si
gni
ng va
lue
to
the
bandwi
dth 1
/
r
of
t
he
fi
rstor
der
sys
tem
c
ons
ide
rabl
y l
owe
r t
han
t
hen
aturalf
req
uenc
y叭 (
i.
e.
0c
)
o'
t>2)
,th
e v
elo
cit
y g
ain
c
oul
d b
e
i
ncr
eas
ed
eas
il
y b
y a
f
act
or
2.
Thi
s
impr
ove
ment
i
s
ma i
nl
y du
e t
o
the
ext
ra
p
hase
la
g o
f t
he
fi
rs
t
o
rde
r s
yst
em.
At叭Z>>1 t
he
i
nlu
f e
nce
of
t
he
sec
ond o
rder
s
yste
m co
uld
eve
n be
neg
lec
ted
,
s
in
ce
Kv
co
ul
d
be
a
s
la
rg
e
s
a 1
.4
/ro
,
r
K,
,
/1
,0
0二1
.4
/o)
oz.
App
roa
chi
ng
t
ha
t
ex
tr
eme
s
it
ua
ti
on
h
owe
ver
,
th
e b
andwi
dt
h叭of
t
he
c
ont
rol
l
ed
s
yst
em
i
s
de
cre
se
a d
s
har
ply
.
75
众6
二!
wo!0
.4
人
飞
B
\
尸一扁、二乙: ̄ ・一
/忆, 州乌
丁
几_-
v
舅
日二‘  ̄ ,晰、
]
.
/
咬一
之 、,、
不谈蒸
一 ,,户
/ 、、、
、、
\
、、
0.
2 名 / 卜、
乙犷
、、
/ 一  ̄
、 、、
、 、
荞 、、
、、
/一乙V 丫—  ̄.
 ̄  ̄一 ̄ ̄
1
伙
一  ̄ 肠曰 ̄
— ,
/冷 不 L 臼
r
d;
,
犷
a 一一 f
一
04
{
_
_
0.
0.
04 改1 .夕
0 (
L了 叹‘ 1 _ 2
‘ 一
妙
O
夕
}H}
1l十 介 .由
口
13
二口口‘口尸口升云启二架、 }
一八
一
1
「
一 、目 口厂
1
_
__
_ \
叹
‘
「
一 「
一-—
}
_ 1 \
丫_
叹
了
mor-2.
S
叹2 口{
}
}
0.
1 L
__
0.
0.了 0.
1
. 改a 坎4 改‘ r _ 2
印
一 确一 , ̄,.卜
h
) 气
Fi
g.3.
11
Fi
rst
or
der
s
yst
em
i
n s
eri
es
F
iog
.3.
11
b
) s
hows
t
hat
f
or
v
al
ues
o
f t
he
d
amp
ing
c
oef
fi
ci
ent杏a
s l
ow
a
s 0
.1,
a
s
yst
em
ba
ndwid
th
app
roa
chi
ng
the
na
tur
al
re
fquenc
y肠 i s
pos
sib
le.
Thus
compa
red
wit
h p
rop
ort
io
nal
co
ntr
ol
(
see
Fi
g.
3.
2),
the
i
nf
lue
nce
of
a
p
rope
r
fi
rs
t
or
der
s
yst
em is
a
po
sit
iv
e o
ne.
3.
11
COMPLI
ANT
STRUCTURES
A way t
o ch
ange dyn
amic prope
rti
es of
hydra
uic
l ser
vosyst
ems dr
ama ti
cal
ly i
s by th
e
int
roduc
ti
on of
compli
ant
st
ruct
ure
s.
Fi
g .
3
.12
a)
gi
ves
a (
si
mpliie
f d)
example,
where
the
pis
ton
is
act
uati
ng am ass
-sp
rin
g sys
tem.
Such as
i
tuat
ion
can
be e
xpecte
d i
n l
arg
er c
o nst
ruct
ion
s where
par
t of
t
he a
c t
uat
ed mass
i
s
connect
ed
mo r
e or
l
ess
le
f xi
bly
to
t
h e
pi
ston.
Anal
y si
s o
f Fi
g.
3.12
a)
shows
76
1 H1
-S
raz
il
ic
s
erv
o‘、
t位’
之VI声apostle
s
s-s
r
)尹rng 琴}
sr}r
rrs
王井
}'
i
t
'
V
C
,
rA
夕 一计
环穷 夕
_
」 巴少司
介:
川
1一
co
C孙 J
石s=
凡I ,:
. }3 k
,、十C
1 多’
,
二司f
1
J;
,祥
!
-
[i
t订J
la
h,
n
十;
),
’ 十
li
}h
',
n
.F
l J
石s=
+A(
t
s+C
,
百
了-
心
J
石、:
、 !:
。C,
+1
二何-
漆握一 I
V
,+
-C上+1
田6 分“
‘C
Ol
+
1
)
d
E
_
}
h
}
G,
(
t
)-C
J}r
(
Oo(
Vl
刁
l
,
曰卜一
.
11,
Fi
g.3
.12
Co
mpia
lnt
St
ruc
tur
es
77
F=C,
(y
。一Y,)
+K,
(y。一夕1
)=M,
.
Y,
(
K,
s+C,
)
Yo=(M
,
s2+K,s+COY
, }
Yo K,
s+C,
Y
,
MI
S
2
+K,
s+C,
Th
e b
loc
k d
iag
rams
b
),
c
),
d
) a
nd
e
) i
ndi
cat
e h
ow
t
he
t
ra
nsf
er
f
unc
tio
n c
oul
d be
f
oun
d.
Th
e
r
esu
lt
i
s
a
r
at
her
c
omp
lex
t
ra
nsf
er
f
unc
ti
on
wi
t
h t
hre
e n
atu
ral
f
re
que
ncy码,C
t)
2
an
d o
)3.
I
t
s
re
f qu
enc
y r
esp
ons
e i
s
usua
ll
y r
ath
er
wil
d,
es
pec
ial
l
y when t
wo o
f t
he
n
atu
ral
re
f qu
encie
s (
for
e
xampl
e码 an
d 0
)2)ar
e vey
r c
los
e, as
ha
ppens
fr
equen
tly
. Then
in as
mall
fr
equenc
y range
(
aro
und码 a
nd
0
c)
2)he
avy
c
han
ges
i
n t
he
o
pen
l
oop
a
mpit
lude
r
ati
o a
nd
ph
se
a l
ag
c
oul
d b
e
e
xpe
ct
ed,
e
spe
ci
al
l
y a
t
lo
w v
al
ue
s o
f杏;a
nd/
or杏:
.Th
e Ny
qui
st
p
lot
i
n g
) g
iv
es
a
t
ypi
ca
l
example
sho wi ng
a char
act
eri
sti
c c
url.
As
t
he sys t
ema t
ic a
nalys
is
o f
co
mpl i
ant
st
ructu
re wil
l
le
ad us
t
oo f
ar now,
we res
tr
ict
our
sel
ves
in
a more qualitat
ive
wa y
t
o one
impo r
tant
sp
a ect.
Becaus
e c
omplian
t s
truct
ures
coul
d det
eri
orat
e
the
dynami c propert
ies
of
a ser
vosyst
em co
mp l
etel
y th
ey ar
e s
ome t
imes i
ndica
ted s
a "par
si
a ti
c"
mass-s
pri
n g s
y st
ems .
To i
lmit
it
s i
nf
lue n
ce, t
he nat
ural
f
reque
ncy of
the par
si
a t
i
c syst
em it
sel
f
should
be de s
igned hi
g h
s
a possi
ble,
prefera
bly 1
0 ti
me s
hi
gher
(o
r e
v e
n mo r
e)
than
the
desi
gned
bandwidt
h of
the complet
e
syste
m.
A
compa rable c
ompl ex
si
tuat
ion
is
depic
ted
in
Fig.3.
120,ari
sin
g for
i
n s
tance
when an
act
uat
or
sta
nds on a lo
f or,
ha v
ing a
nat
ural
re
f quency i
ts
elf.
Bot
h t
ypes
of
comp l
iant
st
ruct
ures
re
f quent
ly
occur
in hy drauli
call
y act
uated
ma chi
ne tool
s.
One s
imple
sol
uti
on of t
he contr
ol
proble
m i
s
s
hown
i
n t
he
next
sect
ion.
3.
12
LF-
CONTROL
OF
COMPLEX
HYDRAULIC
SERVO'
S
I
n Fig.
3.
13 a
pr
o p
ort
ional
l
y co
ntrol
le
d hydra
uli
c s
ervo
wit
h a
par
si
a t
ic ma
ss-s
pri
ng
sys
tem
is
re
prese
nted
by i
ts
s
impli
fied
bl
ock
diag
ram corr
esp
onding
to
Fig.
3.1
2 e
).
Fig.
3.13
b)
s
hows t
he
way
in
wh ic
h many
hydr
a u
ic
l al
ly
act
uat
ed NC machi
ne t
ool
s a
re co
ntr
oll
ed.
In
t
hose
mac
hines
t
he
sc
rew spi
ndle a
nd machi
ne t
ool
sl
ide
repre
sent
th
e ma
ss -
spr
ing sys
tem and
ar
ota
ry
hydr
aul
ic
mot
or
t
he
act
uat
or.
I
n Fi
g.
3.
13
b
) a
t
acho
f
eedb
ack
l
oop
ha
s b
een
i
ntr
odu
ced
s
a we
ll
s
a a
n i
nte
gra
ti
ng
c
ont
rol
l
er
(
H,=K,
/s)
.Ne
gl
ec
ti
ng
t
he
e
xt
ena
r l
l
oa
d,
t
he
b
lo
ck
d
ia
gr
am
c
an
b
e s
imp
li
fi
ed
i
f
th
e
na
tu
ral
re
fque
nci
es
o
f GS,Gi
a
nd
G2
a
re
v
ey
r l
ar
ge
c
omp
are
d wi
t
h t
he
d
esi
r
ed
b
and
wid
th
o
f t
he
c
omp
let
e s
yst
em.
Th
at
b
andwi
dth
us
ual
ly
i
s v
ery
l
ow
i
n NC
mac
hine
s.
I
n t
his
l
ow
ba
ndwi
dth
c
ont
ro
l s
yst
em,
t
ra
nsf
or
med
t
o d
iag
ram
c)
, we
may
a
ssu
me
GS
U00,Gi
(j
o))
a
nd
G2
(J
0)=1
a
t
0)
=V;=K,
KK.
So
t
hi
s
LF
-co
ntr
ol
s
yst
em
i
s
si
mpl
i
fi
ed
t
o
in
te
gra
to
r
in
s
er
ie
s
wi
th
a
f
ir
st
o
rde
r s
yst
em,
f
ed
b
ack
p
ropo
rti
ona
ll
y.
A
maj
or
pu
rpo
se
o
f t
his
me
thod
o
f c
ont
rol
i
s
t
ha
t v
el
oc
it
y t
i
me
c
ons
ta
nt;,=K,
/
K,
R (
i
nve
rs
e o
f t
he
v
el
oc
it
y g
ai
n.
)
is
d
et
er
min
ed
c
omp
let
el
y b
y c
ons
tant
a
nd
v
ery
r
eli
abl
e l
ine
r e
leme
nts
.
It
c
an
b
e a
dju
ste
d e
asi
ly
a
s we
ll
a
s v
ery
pr
eci
sel
y t
o a
79
Pr
opor
ti
onal
cont
rol ect
ert
tnl
l
out
!
c;
n
plc
x s
econd
or
derWs
teni
s
1
S
r
L
u
I
M
b
G
e
)
(8
)
+j
G
r
(叭。!
G
).・
W3.
04)
或 ,
,
Cn
产
/ -C
1 1十
+_ j
n
h
T
R K
S
'
力.
“矛刀C
YIII
6X I
pos
it
ion
fe
edbac
k
l
unI
Gvc
o)I
=1
恕G
I(
iw)1
l
= l
il
l
(,
f
(j
w)
I
=
1 口p
- o  ̄
‘
)
,
Mr
th
t
achol
oop
a
nd
i
nte
gra
tin
g c
ont
roler 几l
ad
!-
G
r
+
, b,卜二吞司丁I-
l
C YI
_
hr一
G N
e
r ‘ I
t
acbol
oop v
eloci
t
y,
f
eedbac
k
K,
s
po
sit
ion
fe
edbac
k
b)
Ne
gle
chng
e
xtr
nal
l
oad
丛命
仪.
(
"
"
G, 1
'I
+_
瓦
b
,K
s
,
瓦
I
i,
+
G
G
t
、1 1
I
c,
Gw)
I一I
1一几
t
K
aI
匆
五
自
人
注s
sur
nrng
-
-
I
气6W)
I
一1
..
J
一
G (
cc川 =I
I
J"
I
-- . 卜
十 r,
.V
(Z,
S+ 1)
Kr 1
公、
.= 公r= _,
.
_
- 4 f, - .., ‘、里 r
万
T
R K.
h,.
凡
Fi
g.3.
13LF-Cont
rol
d
esi
red
f
ixe
d va
lue,
a
nd
t
hat
is
p
recis
ely
wha
t
shoul
d b
e d
one i
n
NC mac
hin
e wi
th
a
se
rvo
for
ea
ch
a
xis
of
moti
on.
It
s i
mpo
rta
nce ca
n be d
emonst
rat
ed
by as
sumin
g an
NC mac
hine
mil
li
ng wit
h
79
c
onstant
speed al
ong
a st
rai
ght
line,
i
n which t
wo ser
vo's part
i
cipat
e.
The v
elocity
err
o r
s of
th
e
s
ervo'
s do not
caus
e any
wo r
kpiece
er
ror
if
the
veloc
ity g
ains
are
exact
ly
equal.
I
n that
par
tic
ular
c
ase t
he comb i
nat
ion
of
the
velo
city
err
ors
re
s ul
t
s i
n a
cutt
er pat
h t
angi
ent
al
to
the work
piece.
For
t
his
reas
on the
vel
ocit
y g
ain of
a
ll
ser
vo's
should
be
examined and
adju
ste
d e
xpe r
ime n
tal
ly
to hav
e
e
qua l
va l
ues wit
hin
a
few
p e
rcent
ages.
Th
e i
mp li
cat
ion
of
thi
s method
of
LF-control
i
s t
hat
the
ma xi
mum obt
ain
able bandwidt
h is
not
r
eached a
t all
,
but
in
many NC ma ch
ines
a l
arge
bandwidth i
s c
onside
red
to
be f
ar le
ss
imp or
tant
t
han equal
velo
cit
y ga
ins.
Thus
Fig.
3.13
shows a
safe
and pra
cti
cal
sol
uti
on
for a
complex cont
rol
pr
obl
em.
3.
13
LOAD
ACTUATORS
Hy
d
raul
ic
se
rvosyst
ems fi
nd wides
pread appli
cati
ons
in
ma te
rial
s
and
fati
gue
tes
ti
ng e
qu i
pment
wh e
re t
he f
orces(=load bei
n g i
ntr
oduce d) a
re ess
enti
al.
There i
s a
remarka
ble l
ack of
cl
ear
i
n fo
rmati
on
conceni
r ng t
he dynamic
behav ior
of
load
act
uators
. One
of
the
reso
a ns
for
this
i
s
the
enormous
var
iet
y of
t
he mechanical
s
tru
ctur e
s i
n which
loa
d actuat
ors
a
re
incor
porat
ed.
Ob v
ious
ly,
t
h e
str
uct
ure o
f a hy
drauli
c pres
s,
of f
atigue tes
ti
ng equi
pme nt f
or
aero
plane
wings,
and of
a
simple
te
nsi
le
stre
ngth p
robe tes
ter
are
co mp l
etel
y di
ffe
rent.
No r
mall
y,
each s
pecial
e
qu i
pment
r
e qui
res
tho
rough anal
ysi
s,
reveali
ng i
n ma ny ca
ses di
str
ibute
d para
me t
er s
yste
ms a
s well
s
a
l
ump
ed
p
ara
met
er
(
mass
-s
pri
ng)
s
yst
ems
.
F
ig.
3.
14
a
) ep
r r
ese
nts
a
s
impl
if
ied
mod
el
of
a
l
oad
a
ctu
ato
r i
n a
c
omp
lia
nt
me
cha
nic
al
s
tr
uct
ure
.
Fi
g.3
.14
b
) gi
ves
t
he
blo
ck
di
agr
am i
n whi
ch t
he
se
rvo
syst
em i
ts
e l
f
as
wel
l s
a t
he
mech
ani
cal
st
ru
ctu
re
can
be r
eco
gni
zed
e
asi
ly.
I
nt
rod
ucing
two
"con
cent
ra
ted" ma
sse
s Mo an
d M,,t
wo
s
pri
ngs(
C,
a
nd
C2
)an
d t
wo
d
amp
ers
(k,
an
d
k2
)we
f
in
d:
M,
Y1=C
(
Yo一Y
1)+k
,
(Y
。一y
1)一CY,
一k2
Y2
画1
s
2+(
、,
、k2
)
:+C
,+C
2
.
1
!,
一(k
,
s+C
,)
,。
Y1 k
,s+C,
Yo 一M,
,,+(
k,+k
2)s
+么+C2.
The
l
etwa
f rd
f
orc
es
Fo
a
ndF,
o
n t
he
pi
st
on
a
nd
t
he
ma
ss
M,
,re
spe
cti
vel
y,
a
re:
F
O=k
,
(i
>
。一Y
i)+C,
(,。
一Y1
)=(
k,
s+CO(
Y。一Y
1)=(
MIs
2+k
2s+C2
)Y
I
汽=k2
夕,+C2
Yt=(
k2s+CZ
)
Yi
・
Th
e t
ran
sfe
r f
unc
ti
on
v,
/
v.
a
s we
ll
a
s t
he
了 1 才 了 U
f
orc
es
F,
a
nd
F,
a
re
i
ndi
cat
ed
d
ii
ect
ly
i
n di
agr
am
b)
.
V I 班  ̄
I
t
sh
oul
d
be
n
ote
d t
hat一i
n r
eac
ti
on一t
he
f
orc
es
F.
a
nd
F a
ct
i
n
ri
ght
war
d d
ir
ect
io
n o
n t
he
st
ructure.
Th
ere
a
re
a
t
le
st
a f
our
p
oss
ibi
li
ti
es
f
or
f
eedb
ack
:
yo
a
nd
y,
b
y me
ans
o
f a
d
isp
lac
eme
nt
t
ran
sdu
cer
,
80
Ct 几
巧
1
1,
一芍
s
en,
osy
st
ent
汤el
f ac
tuat
ed
nj
echani
cal
st
ruct
ure
b)
十
上
1一
C,
-tn
G
_
C
i
h A
1s2
+ir
s 气‘十Ct
5
.+
J
a
+
竺仁
c)
肠一
肠
b
sGs
C"
'
d1
Fi
g.3
.14
Si
mpl
i
fie
d mo
del
o
f a
l
oad
a
ctu
ato
r
81
Fp
a
nd
F,
b
y me
ans
o
f
a
f
orc
e t
ra
nsd
uce
r
(l
oad-
cel
l
).
The ch
oice
t
o b
e made
de
pends
l
argel
y o
n t
he
natu
re
of
the
ap
p l
ica
tio
n a
nd t
he
v
alu
es
of
t
he
var
ious
par
ameter
s.
Si
nce
gen
era
liza
ti
ons
ar
e ha
rdl
y pos
sib
le,
we t
ake
as
exa
mpl
es
two
e
xtr
eme
ly
si
mp if
l i
ed
cas
es.
(
1)
I
n
th
e c
se
a t
ha
t C2=0 (
and
c
ons
equ
ent
l
y y
,
=0
),
d
iag
ram
b
)
ca
n b
e s
imp
li
i
fed
t
o
c)
a
nd
d
).
i
f GS
=1
,
th
e r
es
ul
ti
ng
d
iag
ram
e
) i
s
si
mil
ar
t
o t
ha
t
of
a
u
sua
l s
erv
o wi
t
h p
osi
t
ion
f
eed
bac
k.
Fr
om
t
he
c
ont
rol
p
oin
t o
f v
iew
i
t
ha
rdl
y ma
ter
s wh
eth
er
t
her
e i
s a
po
sit
i
on
or
a
for
ce
feedback.
(
2)
I
n t
he
c
se
a t
hat
C2=0
a
nd
k2二0,
we
f
in
d t
he
c
omp
li
ant
s
t
ruc
tur
e
be
ing
d
eal
t
wi
t
h i
n
Sect
io
n 3.
11.
No w posi
tion
an
d f
orce
fe
edbac
k di
ff
er
ess
ent
ial
ly.
Thes
e t
wo s
impl
e c
se
a s sh
ou l
d make
ver
y cl
ear
th
at
t
he n
atur
e of
t
he
me ch
ani
cal
s
tr
uct
ure
h
s
a a
n
ov
erwhel
ming
in
lu
f enc
e on
the
o ve
ral
l
dyna
mic be
havi
or.
3.
14
COULOMB
FRI
CTION
AND
STABI
LITY
As
s
ta
bii
l
ty
a
nd
da
mpi
ng
a
re
c
ruc
ial
p
rob
lems
i
n c
ont
ro
lin
lg h
ydr
aul
ic
s
erv
osys
te
ms,
i
t
is
wo
rthwh
ile
t
o i
nve
sti
gat
e wh
eth
er
Co
ulo
mb
f
ri
cti
on
c
ont
ri
but
es
t
o s
tabi
i
lty
a
nd
da
mpi
ng
o
r n
ot.
S
imp
ly
by
i
nt
ui
ti
on
(
or
b
y mi
sl
ead
ing
r
efe
ren
ces
)
on
e mi
ght
e
xpe
ct
o
ne
b
ene
fi
ci
al
c
ons
equ
enc
e o
f
Coul
omb
ri
f ct
ion,
s
a i
t
be
hav
es
i
n mo
re
o
r l
ess
t
he
s
ame
wa
y a
s v
isc
ous
f
ri
cti
on,
whi
ch
do
es
pr
ovi
de
s
tab
iit
ly and
da
mpi
ng.
F,
+ F
1-
三
与
co
£ 十一
h
,.户
5丢+,
s
I
之 十= .
之夕‘
,
Fi
g.
3.
15
Hyd
raul
i
c s
erv
osy
ste
m wi
th
Cou
lomb
f
ri
ct
ion
To
get
a
one-s
ided
but
ne
veth
r el
ess
cl
ari
fyi
ng i
nsi
ght
in
th
ese
matt
ers
,
we make
the
f
oll
owing
hypot
het
ica
l e
xperi
ment
wit
h a
hydr
aul
ic s
erv
osyst
em acc
ordi
ng t
o F
ig.
3.1
5,
in
wh i
ch
Coulo
mb
fr
ict
ion
i
s t
he onl
y no
n-l
inea
ri
ty.
We a
ssume t
he sy
ste
m to
be a
djus
ted
so
tha
t i
t
is
st
abl
e al
so
a
t凡 二0,
con
se
que
nt
ly
K,
,
/0)
0<2
奋
We
gi
ve
t
his
s
yst
em
ac
ons
tan
t s
pee
d i
npu
t s
igna
l r=v
ro
f m
t=一 t
o t=+的 and at
t
he
s
ame
t
i
me
i
ncr
eas
e
th
e v
el
oci
t
y g
ai
n
"i
ni
fn
it
el
y
sl
owl
y"
t
o
K/%>2
0 }.
A
sys
tem
wit
hotC
u o
ulo
mb
ri
fct
io
n(Fc
=
0)
wo
uld
s
howi
nst
abi
i
lty
, e
xac
tly
wh
en
p
ss
a i
ng
K/0
)0=2
} ma
nif
est
ed
by g
radu
all
y a
ri
si
ng
os
ci
ll
at
ion
s wi
th
ev
er
i
ncea
rsin
g
a
mpl
it
ude
s,
s
ugg
est
ed
(
in
a
n e
xag
ger
ate
dwa
y)
i
n瑰 .
3.1
6 b
).
Fi
g.3
.16
S
tab
iit
ly a
nd
Cou
lomb
ri
fct
ion
Le
t u
s a
ssu
me
f
or
a mo
men
t t
hat
a
sy
ste
m wi
t
h Co
ulo
mb
ri
f c
tio
n ema
r ins
s
tab
le
a
t
Kv
/co
o
>2
百an
d t
he
n p
rov
e t
ha
t
th
is
ss
aump
ti
on
i
s
wr
ong
a
nd
t
he
eb
r y
p
rov
e i
nd
ir
ec
tl
y
th
at
Cou
lomb
ri
fct
ion
ha
s n
o s
tabi
li
zi
ng
e
fec
t.
T
he
i
ni
ti
al
ss
a u
mpt
i
on
i
mpl
i
es
t
hat
th
e s
ame
e
xpe
ri
men
t
as
be
foe,
r wi
t
h a
s
yst
em
wh
ere
凡 #0,
s
hou
ld
s
how
ar
espo
nse
i
l
ke
Fi
g.
3.1
6 b)
.
St
abi
l
it
y d
ue
t
o Co
ulo
mb
ri
fct
ion
i
mpl
ie
s a
t v
ey
r l
est
a
t
ha
t
th
e s
er
vo
a
t K,
,
/0%
>
2心c
ont
i
nue
i
t
s r
es
pon
se
wi
t
h夕>0
p
er
man
ent
l
y.
Ho
wev
er
,
su
ch
a
r
es
pon
se
i
s
imp
oss
ib
le
b
eca
use
i
t
s i
mpl
i
cat
i
on
i
s
th
at
Fc=+凡 p
erma
nen
tl
y;
s
ee
F
ig.
3.
16
b
).
Bu
t a
c
ons
ta
nt
f
ri
ct
io
n f
or
ce(
凡 =+凡=
con
sta
nt)
i
mpi
l
es
a
p
erf
ec
t
i
ln
ear
s
yst
em
b
eha
vi
or
,
whi
ch
i
s
inh
een
r t
ly
un
sta
ble
,
se
e F
ig.
3.1
6 a
).
S
o ou
r i
nit
ial
ss
a ump
tio
n wa
s wr
ong.
Th
e mos
t ik
l e
ly
r
esp
ons
e o
f as
yst
em
wit
h Co
ulo
mb
f
ri
ct
ion
t
o 户=v=c
ons
tan
t,
wh
ere
Kv
X0
)0
i
s
so
mewh
at
l
ar
ge
r
th
an
2杏,i
s
il
l
ust
r
ate
d i
n
c)
,
wh
ic
h s
hows
a
h
unt
i
ng
(
st
i
ck-
sl
i
p)
mo
ti
on
wi
th
p
eri
odi
cal
s
to
ps(夕=0)
.Ana
log
s
imu
lat
io
ns
c
onf
ir
med
t
hi
s
pos
si
bil
i
ty.
Al
th
ough
o
ne
c
oul
d s
ay
t
hat
a
r
esp
ons
e a
cco
rdi
ng
t
o c
) i
s
st
abl
e,
be
cau
se
o
sci
ll
at
ions
ha
ve
l
i
mit
ed
a
mpl
it
ude
,
83
s
uch
a
r
esp
ons
e i
n
pr
act
ice
i
s
si
mpl
y n
ot
a
cce
pta
ble
.
Th
us
we
c
onc
lud
e t
hat
Cou
lomb
f
ri
cti
on
wi
ll
not
i
mpr
ove
s
tabi
li
ty
i
n pr
act
ice.
3.
15
ANALOG
SIM ULATION
OF
COULOM B
FRICTION
Th
e i
nf
lue
nce
o
f Co
ulomb
f
ri
ct
ion
o
n d
yna
mic
b
eha
vio
r o
f h
ydr
aul
ic
s
erv
osy
ste
ms
c
an
b
e
a
nal
yze
d b
y a
nal
og
s
imu
lat
ion.
To
pe
rmi
t
th
is,
Fi
g.3.
1S
a
) a
nd
b
) h
ave
be
en
c
omb
ine
d a
nd
wo
rke
d
o
ut
c
omp
let
ely
t
o t
he
"
ana
log
bl
ock
d
iag
ram"
o
f Fi
g.3.
17
a
).
To
a
voi
d i
mpr
ope
r pa
rame
ter
a
dju
st
men
ts
,
t
he
t
wo
c
omp
one
nts
o
f
th
e d
amp
ing
c
oeic
f i
en
t杏 h
ave
b
een
s
epa
rat
ed.
Ac
cor
di
ng
t
o
Eg
n.(
3.1
0)
K .Cn
w
C =否,
十杏2
.
F
U .
二二 — — 十
2(
Ch 2M叭
vo
Fi
g.3.
17
b)
s
eems
t
o be
s
uffi
cientf
or
anal
ogs
imul
ati
on,
as
Coul
omb
f
ric
tion
a
ccor
ding
t
o Fi
g.
3.
16
b)
i
s
in
clu
ded.
Bu
t
it
i
s
not
,
be
caus
e
at夕=0
t
he
Co
ulo
mb
f
ri
ct
io
n i
s
no
t
de
ter
min
ed.
F:
十界
F
,
:} F
C
o盛+、c
n
于黔 一>0_j
F
,=+I
F
夕<口分 兀‘一}
,
I
F'
'
,
a
1
I
1n
uc
,
+
W
u+
- 矛呵
(
t
)
p 似。 /。。 c
o"
!
_
一;(
5
1
+}
_
) }
力)
84
F,
.
亡1
1
+二研
o
h
r卜+
- +一 )
’
,
界一与
了.
、,
一 = con
留留 一stunt
r
oll 臼。
:
咨,
r
v
s
o ?
}r
v
o}
w
o 4
t
)o
+
-
E/
i
ta
C
O
s
O
"
E
2
f
1
;广夺
)
(
))
Fi
g.
3
.17
Ana
log
s
imu
lat
ion
o
f Co
ulo
mb
f
ri
ct
ion
I
n p
hys
ic
al
r
eal
it
y,
ho
wev
er,
Co
ulo
mb
f
ri
ct
io
n a
t夕=0
i
s d
ete
rmi
ned
by
t
he
f
act
t
hat
i
t
wi
ll
s
imp
ly
t
ake
wha
tev
er
va
lue
i
s ne
ces
say
r t
o k
eep夕=(
)
.As
l
ong
a
s r
equi
re
d ri
fct
io
n i
s
一凡 <凡 <+
Fr,t
he
s
pee
d y
wi
l
l r
ema
in
z
ero
. As
s
oon
a
s凡 =+Fc
(
at
d
Fc/
dt>0)
o
r
Fc=一
元 (
at
dFc/
dt<0)i
s
re
ach
ed,
th
e s
ys
te
m s
t
art
s
mo
vi
ng
a
gai
n.
Th
is
p
roc
edu
re
h
as
b
een
imp
le
men
ted
in
Fig
.3.
17
c
),
whe
re
FC=几I
a
s
l
ong
as 夕#
0
and
F
c=凡:
whi
l
e
y
=
0.
To
s
imu
lat
e ac
ons
ta
nt
a
nd
pe
rma
nen
t s
pee
d(户=v=c
ons
tan
t,
s
ee
Se
cti
on
3
.14
) s
ome
r
ear
ran
geme
nts
mu
st
be
ma
de,
s
ee
d
iag
ram
d)
.
In
t
his
s
et-
up
t
he
s
tat
eme
nts
ma
de
i
n S
ect
i
on
3.
14
a
re
c
ompl
et
ely
c
onf
ir
med.
An
alo
g e
xpe
rime
nts
wi
th
j=v=c
ons
tan
t s
howe
d t
wo
s
tab
ili
ty
b
oun
dar
ie
s.
T
he
b
oun
dar
y
be
twe
en
s
ta
bl
e
an
d s
emi
-
uns
ta
bl
e
is
r
ea
che
d a
t
K,
/0)
o=2
}.Th
e
bo
unda
y
r b
etwe
en
s
emi
-uns
tabl
e a
nd
c
omp
let
el
y u
nst
abl
e i
s d
ete
rmi
ned
by
t
hre
e d
ime
nsi
onl
ess
q
uan
ti
t
ie
s,
K/(
O
0,百a
nd
C
oo
瓦/
V
Co.
The i
nlu
f en
ce
of Cou
lomb f
ric
ti
on on s
inu
soi
dal
ly
act
uat
ed sy
ste
ms wa
s i
nvest
iga
ted a
lso
.
Ex
per
iment
s c
onf
ir
me d
o
ur s
usp
ici
on t
hat
Coul
omb
fr
ict
i
on ha
s n
o efect
ive
s
tab
ili
zi
ng i
nlu
f ence
.
玩t
he
Se
cti
on
3.
16
o
f t
hi
s
ch
apt
er
a
n e
lega
nt
me
tho
d i
s
in
tro
duc
ed
t
o p
rov
e t
hi
s
th
eor
et
ic
all
y.
At
t
he
s
ame
t
i
me,
more
i
nsi
ght
met
hod
i
s
ob
tai
ned
i
nto
s
ome
v
ey
r pe
cuia
l r
p
hen
ome
na.
3.
16
DESCRI
BING
FUNCTION
OF
COULOMB
FRICTION
I
n t
he
me
tho
d o
f t
he
Des
cri
bi
ng
Fu
nct
ion
(
DF-
met
hod
) a
ll
s
igna
ls
a
re
s
upp
ose
d t
o
be
s
inu
soi
dal
.
85
Th
is
ma
y b
e r
eais
lti
c i
n in
l e
ar
s
yst
ems
, bu
t i
n n
on-
li
nea
r s
yst
ems
i
t
is
a
n a
ppr
oxi
mat
ion.
Th
e
n
on-
in
l e
ar
e
leme
nt
c
onv
ert
in
g s
pee
d夕 i
nt
o Cou
lomb
ri
fct
io
n eq
r u
ies
r s
pec
ial
a
tt
ent
io
n.
As
sumi
ng
t
he
i
npu
t
si
gna
l y
o
f t
he
n
on-
i
lne
ar
e
le
men
t
to
b
e s
in
uso
ida
l,
t
he
o
utp
ut
Fc
wi
l
l
be
b
loc
k-t
ype
(
see
Fi
g.
3.
18)
,
wh
ich
c
an
b
e e
xpa
nd
i
n
a
Fo
uie
rr
se
ri
es.
六。,.
al
^ J
F=万F
,.
Fi
g.
3.
18
De
scib
r i
ng
f
unc
tio
n o
f Co
ulomb
ri
fcti
on
I
nput
: y=y
s
in
wt
1 1内」
1 15
4二
+
+
Ou
tpu
t: Fc=一 户c (s
tn
o
x s
in
3wt si
n 5wt+…)
.
兀
86
I
n t
he
DF-
met
hod
o
ne
t
ake
s t
he
f
ir
st
ha
rmo
nic
o
nly
,
ig
nor
ing
t
he
h
ighe
r h
armo
nic
s,
wh
ich
s
hou
ld
h
ave
c
ons
ide
rab
ly
s
mal
ler
a
mpl
it
ude
s.
S
o t
he
DF
-me
thod
g
ive
s a
l
i
nea
riz
ed
t
ran
sfe
r f
unc
ti
on
Fc一
Fc_生Fc
夕
4
-
一
二 ;
兀叻
、。。。
not
a
ti
on宾=4元/
二=c
ons
t
ant
,
we
ge
t
(
3.21)
The
DF
-ap
pro
xima
tio
n i
s a
lso
j
ust
if
ied
by
t
he
s
o-c
all
ed
ha
rmo
nic
ba
lan
ce
(
see
F
ig.
3.
18
b
)).
At
o
ne
g
ive
n s
inus
oid
al
s
ign
al
y,t
he
"
exa
ct"
Cou
lomb
ri
fcti
on
a
nd
i
ts
f
ir
st
ha
rmo
nic
(
hav
ing
9
0 0
pha
se
l
ead
t
o
y)wi
t
hdr
aw
t
he
s
ame
a
moun
t
of
e
ner
gy
ro
f m
t
he
s
erv
osy
ste
m.
I
ntr
oduci
ng
the
DF-
fri
cti
on
int
o our
dife
rent
iale
qua
tio
n o
f t
he
ope
n l
oop,
we
get
・
y一
F +F,
F
e+F}
凡
y
+
y
=2C = £
O
)
o
2
Kv 叭Kv Ch
KyCo
}
u
)
o
S
z
3
K
v一
;i
l
)
S
O
K2
v
・
+
-
K
S
vy
一C
Fe
o
[
Co
h
+
KS
v
一f
c
o
l
-!
C
C
h
+
K
S
v
Al
l
si
gna
ls
be
ing
s
inu
soi
dal
,
th
e o
per
at
or
s
c
an
be
r
epl
ace
d
=
一
b
F
c
o
e
y
[
C
O
h+
副一
jo
).
C
F
o
`
y
[
C
Th
O
h
+
副S
us
a
w
y
.
te
f r
s
ome
c
alc
ula
ti
ng
wi
t
h FQ=0,
we
g
et
_
,w2 。 P
-.「。‘
_(1
)
2、C P
-飞
一二一2;
- ,一二厂 .
一 ̄七+A
- I
1-一一
二}+二二.
二六 .
} (
3.22)
y
wo
K,
Kv
Co
y
LKv
l
)o)‘*C
O OY
J
Her
e a
not
too
compl
icat
ed
equat
ion h
as b
een de
riv
ed f
or t
he
inv
e r
se
o
pen l
oop
tr
ans
fer
f
unc
tio
n.
I
t
is
vey
r pr
ofit
abl
e t
o make
use
here
of
an i
nver
se
Ny qu
ist
pl
ot,
becau
se i
t
gi
ves
t
he mos
t
si
mple
el
r a
ti
on
be t
ween
open
loop
an
d cl
o s
ed l
oop
pr
es e
nta
tio
n (s
ee F
ig.
3.
19)
.
In
t
he c
ase
of
-1 一£
oH
H
0
‘一.1’
1+H
子, H
-
一
=1十 一 =2+
1+
H
H
Fi
g.3
.19
I
nve
rse
Ny
qui
st
pl
ot
87
Eg
n.(
3.2
2)
t
he
a
dva
nta
ge
i
s
ev
en
l
arg
er
a
s t
he
i
nlu
f e
nce
o
f Cou
lomb
f
ri
cti
on
on
i
nve
rse
t
ran
sfe
r
f
unc
ti
on
i
s
l
in
ear
(
se
e
Fi
g.
3.
20)
.
I
t
sh
ows
f
i
rst
of
al
l
t
he
l
oc
us
f
or凡=
0(t
hi
ck
d
sh
a e
d i
l
ne,
st
ar
ti
ng
a
t 0)
.
At
a
g
ive
n re
f q
uen
cy
t
he
v
ect
or
OA=£/y
and
t
he
v
ect
or
CA=r/y
r
epr
ese
nt
t
he
i
nve
rse
t
ran
sfe
r
fu
nct
i
ons
o
f t
he
op
en
l
oop
a
nd
c
los
ed
l
oop
r
esp
ect
iv
ely
.
E
qua
t
io
n
(
3.
22
)
d
ir
ec
tl
y
in
di
ca
t
es
t
he
i
nl
f
uen
ce
o
f
F
I
coy
on
t
he
(
co
mpl
ex
)
v
ect
or
OA=£/y.
Owi
ng
t
o
Co
ulo
mb
f
ri
ct
io
n t
he
v
ect
or
OA
is
c
han
ged
i
nto
v
ect
or
OB,
whe
re
B
i
s
t
ra
vel
i
ng
a
lon
g a
s
tr
ai
ght
li
ne
(
th
e DF
-
l
ine
)
as
FC
i
ncr
eas
es.
Th
e
in
ver
se
Ny
qui
st
pl
ot
now
ma kes
poss
ibl
e a
dir
ec t
swit
ch
from t
he
open l
oop
t
o t
he cl
osed
loop,
si
mp l
y
by
repl
aci
n g
t
he
or
igi
n 0 b
y t
he p
o i
nt
C(一1,
0).
So
the
same
locus
a
nd t
he
same DF-i
lne
can be
use
d f
or
open
loop
an
d cl
osed l
oop.
Consequent
ly
th
e cl
osed
loo
p ve
c t
or
CA i
s
changed i
nto
CB by
Coulo
mb fr
icti
on.
Thus
Fi
g.3.2
0 enabl
es
u s
t
o an
alyze
t
he i
nf
luence
of
Coul
omb ri
f ct
ion gra
phic
all
y.
As
t
he in
f al
analy
sis o
f Egn.
(3.
22) wil
l
be made a
nalyt
ica
lly
, we res
tri
ct
our
sel
ves t
o th
ree
obs
erva
tions
on
Fi
g.
3.
20.
(
a)
T
o f
in
d t
he
a
mpl
i
tud
e y
o
f
th
e s
ig
nal
y
f
rom
a
g
ive
n r
" a
nd
a
g
ive
n
FC,
the
c
ons
t
ruc
ti
on
o
f
AA
DE
i
s
us
eful
,
wh
ere
AD
=I
a
nd
DE/
/CB.
r
DE 刀A
DA'
1 .
__. J
CBJ_ y _x,
Co
r
、
凡
— 二— 二— 二— s
o!刀F}二
CB
AB
B'
A' A'
B'
Th
e l
eng
th
o
f DE
bei
ng
kno
wn,
t
hepo
int
E c
an
be
f
oun
d,
a
nd
d
rawi
ng
t
he
i
l
ne
CB
/
/
DE
we
f
in
d t
he
p
oin
t
B
a
ls As}
o: CB}=P
yy
l
,th
e
amp
li
t
ude
y
c
an
b
e
fo
und
s
a
wel
l.
(
b)
Co
ulo
mb
f
ic
r t
i
on
wi
l
l
st
op
a
ny
mo
ti
on(
夕=0
)
if
CB=0,
and
a
s
DE/CB
t
hi
s
ma
y
oc
cur
i
f
IDE
I=I
DAI
.An
al
ys
i
s
of
Fi
g
.
3.
2
0
yi
e
ld
s
}
DE
{一}
C
B
j=
一曰一C
o
r一一
,
D
A’
“‘
F
c
1
〔
(
-
K
0
)
,
!・
:_
C
C
o
ha
S
OI
DE卜I
DA卜,
and
y=0
i
f
(
3.23)
I
n f
act
,
th
is
i
s
th
e dy
nami
c d
ead
z
one
,
de
riv
ed
e
arl
ier
i
n S
ect
ion
3.
4.
Th
e
on
ly
d
ife
re
nce
ro
fm E
gn.
(
3.1
1)
i
s
a
f
act
or
7
r1
4,d
ue
t
o
th
e DF-
app
rox
ima
ti
on.
(
c)
A
l
in
ear
s
yst
em
i
s
co
nsi
der
ed
t
o
be
u
nst
abl
e(夕/r=0
0)i
f
th
e po
int
C(
一1,
0)
i
s
on
t
he
l
ocus
.
I
f
F,=0
t
his
wi
l
l o
ccu
r h
er
e a
t
K,
/C
oo= 2
}
whe
n。=wo
.
I
f
F,
#
0
a
nd
K,
/
wo=2
心,a
gai
n y/
r=oa
t。=w0
i
s
a
(
st
ea
dy-
st
at
e)
s
ol
ut
i
on,
co
nir
fmin
g
th
at
Co
ulomb
f
ri
cti
on
i
s
no
t i
nlue
f nci
ng
s
tab
ili
ty
.
Fi
g.3
.20
De
scr
ibi
ng
f
unc
ti
on
i
nve
rse
Ny
qui
st
pl
ot
of
h
ydr
aul
ic
s
erv
osy
ste
m wi
th
Co
ulo
mb
ri
fct
ion
T
o s
tu
dy
s
ome
v
eyp
r
ecu
li
ar
p
hen
ome
na,
we
c
ont
i
nue
f
urt
he
r
an
al
ys
is
o
f
Eg
n.(
3.
22
)
an
aly
ti
cal
l
y.
Su
bst
it
uti
on
o
f£=r一y
yi
el
ds
。, 。 F
二=1一2‘,-二尸 ,
.. 「。} _
02、。 F
( ,.1
一下厂.朴二二
士+刀:丁-}t一一 二 !
+二子..
二.二:}.
Y wo
K,
,
r,(
-
o
y
L
K
}
,
l
c
Oo
_)‘
.、UO
YJ
Ta
kin
g t
he
mod
ulu
s g
ive
s
「
;,2
_「:1
1
Y。, 。P
C
1
!丁 1 ‘! 1一‘5
2
.「。!
,。,
、.o
c
C
o
P
宾1
C
t
z
.一 二丁 ̄’二 尸 育 I
r I
- I
I一 -一二 1
, .二罗..t万-万 !
L
YE
L
O
)
oA
Kv
Lo
yJ
L
K,
l
wo一)Ch
L
.0
Y
J
Mu
lt
i
pl
ic
at
i
on
w
it
h仔/
r了a
nd
w
ork
in
g
ou
t
i
fn
al
l
y
gi
ve
s
r,,:r/ 。\
2 ,l ,l
2〕 f
-二 ,2「
/ \
2 /_ \
2
1
!Y I
II. 、, 0 v-
-
1
tj一 !。 t t)一 I I !It, I
I1 to I
I L. I I
!份 I ・!I1一 1
.(
1 一— I十 一一二.I1一 一一- I !十 !一一止‘一I ・11— 1
+ 1一
一- 1 1
L
r」L
戈 一t
o
o
K,
) K,
`
t
t
oo)」L
COP
J L
(Kv
)LC
h)」 (
3.24)
,
。Y
P
C。[
Co‘
,(
t 4 -7 .—
1
)
2、(
,,,0
.-
)
2习 ,
I— !1一 一.
尸汀 1
一 i1一 GL
' — 日 一 1= V
‘C
O
F
Kv
LCh
LN0少L 一t
o
o
K,
.
刀
Ev
ide
ntl
y
th
e s
eco
nd
d
egr
ee
e
qua
ti
on
i
n y/
r a
nd
Fc
/
co
r r
epr
es
ent
s
co
nic
s
ect
i
ons
,
no
rma
ll
y
e
ven
e
ll
ips
es
h
avi
ng
ac
ent
er
c
oin
cid
ing
wit
h t
he
o
rig
in.
I
n Fi
g.3
.21
a
) s
eve
ral
a
mpl
it
ude
d
esc
ri
bi
ng
f
unc
ti
on
s a
re
d
rawn
u
p a
t d
ife
re
nt
re
fque
nci
es
a
t K /0
)0
=
0.2
5,杏=
0.2
5
a
ndCh=0
0・
S
ubs
ti
tu
ti
ng夕/r
=0
i
n
Egn
.(
3.
24)
we
a
gai
n
me
et
t
he
d
yna
mic
d
ead
z
one
.
* I
/_
_\2 ,\
2
夕=0
if c
A
o
r
-=
一c
、
}
!K!
0
3+!
C.
I .、 vj 、 一n /
Fr
om
Egn.
(
3.23
) t
he
s
-c
o a
lle
d s
top
re
f q
uen
cy
c
an
a
lso
b
e de
riv
ed.
/ ,1
1
2 /。 \
2
,
=。i
f
。一s
t
o
p一
、II
I
70
-
r1
一
'
t
Co
.
】 勺、 一‘ l
、 ’
./
(
3.25)
This
st
op re
f q
uency h
as be
en
deri
ved e
arie
lr a
lso (
in
Sect
ion
3.4
) a
nd i
s i
ll
ust
rat
ed
in t
he
ampit
lude
char
act
eri
st
ic
b)
.
一
Th
e
i
nte
rpr
etat
i
on of
t
he a
mpl
i
tude d
escr
ibi
ng
fun
cti
ons
i
s e
xpla
ined
i
n mo
e
r deta
il
,
acc
ord
ing
丽
F
igs
.
3.2
1 c
)
an
d d
).
Th
e
re
la
ti
on
b
etwe
en夕/
r"二、
d凡 /
car
wi
ll
no
t
st
op。 夕/r
=0
a
nd
凡,
Co
r=
如/、
,1
2十
(C
O/
丫朴.
As
C
ou
l
om
b
f
i
rc
t
i
o。一
凡<凡、
+兀m
i
gh
t
t
a
ke
s
mal
l
er
v
alu
es
t
han凡 =4凡 l
It
,if
it
is
s
uf
fi
ci
ent
to
s
to
p mo
ti
on,
th
e h
ori
zon
tal
ax
is
wi
t
h
互c0;
y一"
0
n
a
一
d
>
一
r
s
houl
d b
e s
een
s
a as
ubs
tan
tia
l p
art
of
t
he
p
ict
ure
.
A
v
ey
r pe
cul
i
ar
s
it
uat
i
on
i
s
now
p
res
ent
ed
i
n
d
),
wh
ere
(
acc
ord
ing
t
o
ou
r
in
te
rpr
et
at
io
n j
ust
ma
de)
t
hr
ee
d
if
fe
ren
t
va
lu
es
o
f夕/
r mi
ght
oc
cur
a
t
on
e s
i
ngl
e
va
lue
o
f凡 l
co
r
".An
alo
g s
imu
lat
i
on
r
eve
al
ed
h
owe
ver
t
hat
on
e p
art
of
t
he
e
ll
ip
se
d
oes
n
ot
p
rov
ide
a
s
tea
dy-
st
at
e
so
lut
io
n.
Fi
g.
3.
21d
)
sh
ows
h
ow 夕/r
c
an
c
han
ge
d
isc
ont
in
uou
sl
y
wi
t
h v
ayi
rng凡 /
CO户.
The
se
d
is
con
ti
nui
t
ies
h
ave
b
een
o
bse
rve
d d
uri
ng
a
nal
og
s
imu
lat
i
ons
.
A
ki
nd
of
p
ara
dox
is
i
nclu
ded
in
Fi
g.3
.21 d
) s
inc
e u
nde
r c
ert
ain
c
ond
iti
ons
t
he
s
yst
em
c
an
mov
e a
t
a
ri
fcti
on
l
eve
l
th
at
might
s
top
a
ny
mo t
ion.
Co
nsi
de
ri
ng
s
ta
bi
li
t
y
fo
r
th
e
lst
a t
i
me,
su
bst
i
tu
ti
on
o
f K/
0o=2
( } a
nd
i
n
Eq
n.
(
3.
23
)
gi
ve
s
90
{
刹
C
F
O
c
YZ
=了
砚
w
.
}
Z
1
.
c
n
}
2
.
(
Kv
{十(
Ch
{
I
ts
r
epr
ese
nta
ti
on
i
nFi
g.3.
21
e)
s
houl
d be
s
een
as
an
el
li
pse
wit
h one i
nfi
nit
ely
l
ongaxis
凡l
co
r
=0
b
ein
g a
n axis
o
f s
ymme
ty.
r Co
nfo
rm
t
he
f
ore
goi
ng
i
nte
rpr
eta
ti
on
i
s
号
“
N
d
>
.=r
}
r 0
aJ
0
63\
!
4J 改止
a以 8 甘
肚.
芝
t
f
可 东
、 ,,r
二共、
口户州
V尸
砂洲产 口, .
洲尸口
户口.
洲
七-一二r叫
坛
口口口口! 1
.口..,二二鱼吧竺竺 C -
J 〔二二二 二,,,气.日
矛、
-- .百.反 巴妞 r- ̄ ̄...口门脚‘‘ 」
之
欢月暇 之撼 卜L 仪
口
1 门限瓜 洲 -
0
.1
{}i\\.0 !
\’ .
0
6
15
/士
一!0,.4才{(
r‘
)
F
A
.
了一
",1_I
71.1
{)
“
1
6 ’} }
‘I
l
{
}
一
低l
1
l
山一
0.
0子 Ol 0.
: 0.
4
0.6
0.
8 1
此
L)
91
产r・
一<r
^
{阁
五卿
Fi
g.
3.
21
Ampl
i
tud
e de
scibi
r ng
f
unc
ti
ons
a
nd
re
fque
ncy
c
har
act
eis
ric
t
互c0?
里二0
0
.
y一
a
t典
0
一
-
<
r
Co
r
Un
les
s夕/
r
=
0 i
s s
een
a
s a
s
at
is
fa
ct
oy
r s
ta
bil
i
zat
i
on,
Co
ulo
mb
ri
fct
io
n d
oes
n
ot
s
ta
bii
l
ze
h
ydr
au
i
lc
s
er
vos
yst
ems
, s
in
ce
s
ta
bi
i
lt
y c
ond
it
io
n K,
,
/
Ct
)
o
<2C i
s v
ai
ld
f
or
s
ys
te
ms
wi
t
h
凡 二0
a
s we
ll
a
s凡 ¥0.Th
is
c
onf
i
rms
p
rev
iou
s s
t
ate
men
ts
a
s we
ll
a
s v
ari
ous
p
rac
ti
ca
l
expe
rime
nts
.
CHAPTER
4 PERFORMANCE
DIAGRAM
AND
DESIGN
OF
SERVOM OTOR
4.
1 I
NTRODUCTION
Th
e d
esi
gn
o
f l
i
nea
r s
ervo
mot
ors
i
nvo
lve
s t
he
d
ete
rmi
nat
io
n of
t
he
p
ist
on
a
rea
A,
the
s
tr
oke
S
,
t
he
v
al
ve
c
apa
ci
ty
Q.
a
nd
t
he
s
upp
ly
p
res
sur
e
PS.Th
e me
cha
nis
m t
o b
e a
ctu
at
ed
n
orma
ly
r
equ
ire
s
ce
rt
ai
n s
pec
iic
fati
ons
s
uch
a
s
th
e s
tr
oke
S
a
nd
t
he
a
ctu
ate
d ma
ss
M.Ot
her
s
pec
if
ic
ati
ons
,
s
uch
a
s
th
e ma
ximu
m s
pee
d
y.,
the
ma
ximu
m a
cce
ler
at
io
n y.
X
an
d t
he
r
equ
ir
ed
b
and
wid
th
o
r
s
pee
d o
f es
r p
ons
e,
a
re
s
ome
time
s l
ess
d
ete
rmi
ned
b
efo
reh
and.
I
n ma
ny
c
ase
s t
hey
a
re
s
ubj
ect
t
o
c
omp
romi
ses
a
nd
t
hei
r c
ons
equ
enc
e.
F
requ
ent
ly
, i
t
is
t
he
pr
ic
e o
f t
he
i
nst
al
la
ti
on
t
hat
ma
kes
speci
ic
f at
ions
mo r
e r
ealis
ti
c.
I
n t
his
chapte
r,
the rel
ati
onsh
ips bet
ween spe
ciic
f at
ions a
nd des
ign data
are
shown for
a
si
nusoid
ally
-act
uate
d i
lnear
motor
load
ed by i
neti
ra fo
rces a
nd
a co
nst
a nt
pre
-lo
ad.
An en
ormous
hel
p i
n the desi
gn process
is t
h e
perfo
rmance diag
ram, which
s hows some f
undamenta
l
re
laio
t nshi
ps
in
a re
f quency
dia
gram.
As
t
he desi
gn pr
o c
ess i
s s
imple
r fo
r a s
ymme t
ri
c li
near mot
or
control
le
d by a
cr
it
ica
l-c
ent
er
4-
way
s
erv
ova
lve
, t
his
mot
or
wi
ll
b
e t
rea
ted
f
ir
st.
I
t wi
ll
be
pr
ove
d t
hat
,
wi
t
h af
ew
si
mpl
e
c
orr
ecio
t ns
,
th
e r
esu
lts
a
re
v
ali
d a
lso
f
or
a
n a
symme
tri
c l
ine
ar
mo
tor
c
ont
rol
l
ed
by
a
n symme
a tri
c
4
-wa
y v
alv
e.
I
n p
rac
ti
ce,
t
hi
s
symme
a tr
ic
mo
tor
ha
s t
he
a
dva
nta
ge
t
hat
t
he
s
pac
e n
eed
ed
f
or
t
he
mo
tor
i
s
co
nsi
der
abl
y s
mal
le
r,
b
eca
use
t
her
e i
s
onl
y
one
pi
st
on
r
od.
But
t
hen
a
n symme
a tri
c
4-
way
v
alv
e i
s de
sir
abl
e t
o o
bta
in
s
moo
th
s
yst
em
beh
avi
or.
Thi
s i
s ad
isa
dva
nta
ge
(
sys
tema
ti
cal
ly
o
ver
loo
ked
i
n it
lera
tur
e),
be
cau
se
s
uch
an
asy
mme
tri
c v
alv
e i
s n
orma
ll
y n
ot
a
vai
l
abl
e.
Co
mme
rci
al
ly
,
it
ca
n b
e ob
tai
ned
on
ly
o
n s
pec
ial
r
equ
est
an
d a
t a
s
pec
ial
pr
ice
.
An
al
ter
nat
ive
i
s t
he
3-
way-
val
ve-
con
trol
l
ed
as
ymme
tri
c mo
tor
. He
re
t
he
l
owe
r na
tur
al
re
fque
ncy肠 (
see
Fi
g.
1
.11
)
cou
ld
b
e a
de
cis
ive
d
rawb
ack
.
Re
lev
ant
f
ormu
ls
a f
or
t
he
de
si
gn
o
f a
3
-way-
valv
e-co
ntrol
led
asymmetr
ic mot
or can
e a
sil
y be
fou
nd s
a the
des
ign
proc
ess
fo
llo
ws
e
xact
l
y t
he s
ame in
l es
as
f
or t
he
symmet
ric
i
lnear
motor.
The
s
ame is
tru
e for
4-way-
valv
e-c
ontr
o l
l
ed rot
a r
y mo
tor
s.
Here
, ho
wever
, t
he mome
nt of
i
ner
ti
a
J
i
s
su
bje
ct
t
o
de
si
gn
a
s we
ll
,
as
t
he
t
ran
smi
ss
ion
i
s
pa
rt
o
f i
t.
I
n ma
ny
c
se
a s
t
he
o
pti
ona
l
d
esi
gn
o
f t
he
t
rans
mis
sio
n t
o o
bta
in
a
mi
nimum
i
neti
ra wi
ll
i
ntr
oduc
e ma
jor
c
omp
lic
ati
ons
(
see
Rg.
1.1
1).
4.
2 RELATI
ON
BETWEEN
MAX.
STROKE
S=2
夕、AND
BANDWI
DTH
q
Fo
r we
ll-
des
igne
d s
erv
osy
ste
ms
(
wit
hou
t
"p
ara
siic
t "
ma
ss-
spr
ing
s
yst
em
o
r po
or
s
erv
ova
lve
s)
94
t
he
o
bta
ina
ble
b
and
wid
th
0
).
r
oug
hly
e
qua
ls
t
he
n
atu
ral
re
fque
ncy
(
00
c
aus
ed
b
y t
he
o
il
-s
pri
ng
wi
th
s
ti
fne
ss
co
a
nd
t
he
mo
vin
g ma
ss
M.Fo
r i
nst
anc
e,
Fi
g.
3.
2 s
hows
a
n o
pen
l
oop
t
ra
nsf
er
「_
2 _ 〕
f
unc
ti
on
H(
s)=K.
.
/,}s +2C
F ,
I 乙 时 武、
+1}wh
er
e
a
p
ro
月
p
er
c
hoi
ce
o
f K e
nab
le
s
0)
,
,二(
on
卜 朴 r , .
I n)。 【U八 !
L ’一U
u J
p
rov
id
ed
t
he
d
amp
ing杏 h
as,
o
r h
as
b
een
g
ive
n (
se
e F
ig.
3.
4)
,
va
lu
es
b
etwe
en
0
.3
a
nd
0
.7.
A
cco
r
di
ng
to
Fi
g
.
1.
1
1e)
,
wh
er
e
0)
a二}
_
C0/
_
M* c
o
_
>4
,
:/s
=2A
E /
y=,
we
f
i
nd
_2AE
2AE 2AE
(
4.1)
y
,
M‘一
Co-MC2 Mc
Oo j
)b
2
He
re
we
h
ave
a
f
und
ame
nta
l
re
la
ti
on
b
etwe
en
t
he
s
tr
oke
S=2夕
二 o
f t
he
s
er
vomo
tor
a
nd
i
t
s
b
and
wid
th叭 .
The
r
el
eva
nt
pa
rame
ter
s
ar
e
M,A
a
nd
E.
4
3 RELATI
ON BETWEE
N VALVE
CAPACI
TY
Q.
AND
MAXI
MUM
SPEED
y
.
A
given
ser
voval
ve
has
-det
ermi
ned
by
i
ts
maxi
mum por
t ope
nings
-a
gi
ven
maxi
mum
lo
fw Q=
X.
As
Qn,
X=A
y.
we
f
i
nd
Q.
=
A Y
. _A
> 叻 a
t
si
nus
oi
dal
mo
veme
nt.
So
八
y
s= .
二醚 0)
泥 *、
(
4.2)
4,
4 RELATION BETW EEN CAVI
TATION BOUNDARY AND
MAXMUM
ACCELERATI
ON
y
max
I
n s
ect
io
n 2.
4,
t
he
c
yli
nde
r p
res
sur
e只 a
nd凡 on
b
oth
s
ide
s o
f t
he
pi
st
on
we
e
r d
eri
ved.
As
sumi
ng
a
c
ons
ta
nt
p
re-
lo
ad
Fs
t a
nd
ne
gle
cti
ng
ri
fct
io
n h
ere
we
f
in
d
。
,一1
。士从,
wh
e
re
P
,=迎=
.
M
2 ‘ 2 “
y+
Fs
r “ A
A
。 2 。
To
a
voi
d cavi
tat
ion
and/
or
s
atur
ati
on 乙 <一 长 must
be
e
nsur
ed,
s
o
“
3
M,
y二 十Fst二二:—
Mc
02夕+只,<_
2
PL= 一 厂一
A
A
3
and
2_ _
一1
' A 一 F'
,
。_ }I0
3X二二二 3
V c —
了 ,.1 .夕 __z
(
4.3)
【U
MO
)
八
r
Pog]
n 一一甲一 ̄ RI
’
J}
“。
、・
漏\ 八
、
、 ‘、
^.r
Y
=co
;ISf
anf
= rel
l
st
all
f 全
力伪口x
、
,
人儿 -1
R2
M飞阅月划、|〕 山 叮
、、
秘从丫 土 田
人为
山夕 a2
Fi
g.4.
1
Pr
eli
mina
y
r pe
rfo
rma
nce
d
iag
ram
4.
5 PRELI
M INARY
PERFORM ANCE
DIAGRAM
Th
e o
bta
ina
ble
a
mpl
i
tude夕 o
f
a
g
ive
n mo
tor
-va
lve
c
ombi
nat
i
on
i
s
now
d
ete
rmi
ned
i
n
th
e l
ow,
mi
ddl
e
an
d h
igh
f
req
uen
cy
r
ang
es.
, 2AE
LF v<— 田<C
ot Y=Y. , (
4.4)
Ml
i
)b
2
MF ,,Y二 _Q = (
4.5)
Ys一 一 码 <田<叭 YI
>Y>
Y2
CI A(o
)
全 PA一F
-
H
F
y、粤 - 田 一
3
MC
02
2 <山<妈 Y2
O
C >Y
> Y3・ (
4.6)
I
n
th
e p
rel
imi
nayp
r
erf
orman
ce
di
agr
am of
Fi
g.
4.1
(
a)
(
wit
h
loga
ri
thmi
c s
cal
es
fo
r夕 and
o))
,th
e
i
l
nes
fo
r t
he
mid
dle
re
f que
nci
es
ha
ve a
sl
ope-
1,
t
he
HF-l
ine
s a
sl
ope
-2.
Thes
e l
in
es
a
t
th
e s
ame
t
i
me
i
ndi
ca
te
v
alv
es
o
f夕二a
nd
,
y
..
T
he
p
osi
t
ion
s o
f
th
e p
oin
ts
R,
(夕,
,A)
O ,RZ
(夕:,
Ct
})a
nd
R3(
Y3,
魏)a
re
c
har
act
er
is
ti
c
fo
r
t
he
p
erf
orma
nce
of
t
he
s
erv
osy
ste
m.
Co
nce
ni
r n
g t
he
p
oin
t R,
(夕;
,q)
,Eq
ns.
(5
.4)
an
d (
5.
5)
g
ive
Yl
=Yma
x
2AE
码=
Yma
x 二MO
)
b
z・
Qma
x Qmax
MO
)h
2 夕、 2EAZ AY.
wher
e
96
Ay
O
l
二;
_一Q
cn
a
x=、Y
2. 八
Co
nce
rni
ng
t
he
p
oin
t R2
(夕:
,峡)
,Eq
ns.
(4
.5)
an
d (
4.
6)
g
ive
.
r2_ 飞
云
刊 育只A一F I
夕2 MQ
2
二 as _ .
、
!',
fax
,2一下 一  ̄ 二
I
“
.S 一 “ I
J
Y2 -一一 = r.
. ,
勺
,,
A`I
,I乙 _ ,
一F
,
A一r"
.
. I
YU
M MQ.*
_ I
L3 “
’
」
wher
e
、,
2=
;_=Q
ma
x二q
y1
八
and
2 二
呈PA一F
..
3 2,
峡 Y2=Yma
x= =(
'
03
Y3・
M
P
oi
nt
R.
,
(矛,
.
J
0
-1
6)
.
ca
n b
e f
ou
nd
b
y s
ubs
t
it
ut
i
on
o
f C
t
),
=(U
,
,=、
厉面 丽万 i
、 护 口 J ,
nt
o
Eq
n - .沪 口 , - .‘J.闷j、 一
(
4.6)
_YD
Y3
=_
O
)b
2
,一
{3
PS
A
一二
」 2AE
4.
6 DESIGN
OF
SYMMETRIC
M OTORS
Al
l
th
e a
bov
e r
esul
ts
h
ave
b
een
s
umma
riz
ed
i
n t
he
d
esi
gn
c
har
t
of
F
ig.
4.
2.
I
t
en
abl
es
us
t
o
d
esi
gn
t
he
mo
tor
an
d v
al
ve
,
i.
e.
to
d
er
iv
e
A,
Ps,Q.
a
nd
N
p
ea
k ro
fm t
he
s
pec
ii
f
ed
E,
M,
O)
b,
Y
max
,ym
ax,y
n=,
and
F
sr,
Eq
n.
(
4.4
)
pi
st
on
ar
ea
,
一MO
)
b2
了1 二‘ 一夕
二 (
4.7)
2E
Eq
n.
(
4.
6)
g
ive
s
th
e s
uppl
y
pr
ess
ure
Ps= 3
E,
一M
Ft
] z 二
叭 Yma
x
(
4.8)
Y..
Y.
Eq
n (
4.5
) gi
ves
t
he
v
alv
e c
apa
ciy
t Q
. =A
y二二Ma
)
b2E
(
4.9)
[
3__
F,
1:
Ma
x
pe
ak-
powe
r
no
w b
eco
mes
Nn=Q.
P
s二 一 M 41
( 0)
Y
,十M-
」
Yn
=
So
t
he
des
ign
 ̄
da
ta
A.只 。口__and
N-
can
be
exor
一 J 一  ̄ “.田八
ess
ed
exvl
ici
tl
y i
n t
he
s
peci
icat
f ions.It
v
i
s
“ “
cl
ear
t
hat
,
if
t
he
s
pec
iica
f tions
l
ead
“
t
o una
cce
ota
ble
va
lves
of
A.只 .口__
_or
N_,a
compr
is
e ’ J  ̄ “翻;人 v ‘
must
be
made.
R,
,
R, 2AE
Y= —
Mc
2AE
o, Y
I
一Y
-一
可 Y二:
叼二
山 ,= 气丁 ̄- 二 一,丁.
一
口<u)
,,Y,=Y二
R,
R, Y
_x
A
Y一_
、 .。
夕。 Q、
,
Y
2・
-V一
Y
-
)
1
M
},
a
一一,r
0
1
Y
I一
)
0.
y2
“夕一二
}
A
ma
z
’
一
’A2
(
2
-
3
P
s
‘一
凡,
Y= —
Aw
),<。 <0
0 ),
.
.
A(一 只 A一F-)
Y,
一3
Yt>Y>Y2
0
j,=一 = A
.
f
-/l
Y}
1
}jv ,
0
)2夕I二(
1
)2
3
}3
R2
R, }
.
2P
,
一F
A
六
飞
二Y
-=一下讶一一
呈 PA一
F-
”
y 一‘
y
入
口
2_
_ 、
一
3
一
- 自 峥
M由 只
Y.
(}
PA一户,
,)
六 _ .r j
田2<田<。3 Y3 = 一下.丁 = 一一一一一丁下二丁一- 一
6卫)一
2A P
-
Yz>Y>Y3
De
sign
of
s
ymme
tri
c
、
一、一
、一}
M
2Y
A
.
E
Servormot
or
wit
h ,_M可 ・
n 一
a
m
yma
xo
x
—
/
n Mt
o
bz・
Y,X・
Y.A
M V - —
4
-wa
y s
erv
ova
lve
2E ’一 2E
__ 。 丑.、
JL(
y+二号) 。 _ 。
P
x二一 甲些.N,二Q.P=舟M(
端 十书)
.瓜
峨 ’夕ma
x 一 J
L
Vl
As
ymme
tri
c
servom ot
or wi
th Mynx+F
,
.,
-
A
a ,
- A
Z_
P
2
asymme仇 c 4-
way A _ 1刀
servoval
ve
Fi
g.4.
2 Des
ign
cha
rt
4.
7 FI
NAL
PERFORM ANCE
DIAGRAM
Ea
sie
r t
o h
andl
e t
han
t
he
p
rel
imi
nar
y p
erf
orma
nce
d
iag
ram
i
s t
he
f
ina
l p
erf
orma
nce
d
iag
ram
98
(
Fig.
4.
3)
,
whe
re
t
he
s
pee
d夕a
nd
t
he
re
f q
uen
cy
a
re
t
he
o
rdi
nat
es.
Now
t
he
LF-
li
nes
o
f c
ons
ta
nt
a
mpl
it
ude夕=夕/
o
) h
ave
a
s
lop
e +1
, wh
il
e t
he
HF-
li
nes
o
f c
ons
ta
nt
a
cce
ler
at
io
n夕二夕C
A
h
ave
a
s
lop
e -1
.
So
t
he
l
i
nes
of
c
ons
ta
nt夕 a
nd
t
hos
e o
f c
ons
tan
t夕 a
re
p
erp
end
icul
ar
.
Ju
st
k
nowi
ng
t
he
s
o-c
all
ed
2
5%-
ser
ies
:
1
一1.
25-
1.60
-
2.00-
2.50
-
3.1
5-4.
00
-
5.00-
6.3
0 -
8.00-
10-
.
(
wh i
ch
div
ide
s t
he
r
ang
e I
t
o 1
0 i
nto
i
ncrement
s
of
equal
-pe
rce
nt)
,
th
e p
erf
orma
nce
d
iag
ram
c
an
b
e put
t
oge
the
r e
si
a l
y
by
any
one
ro
f m l
ine
ar-s
cal
ed
pa
p e
r i
n a
ny
si
ze.
4.
8 DESI
GN
OF
ASYMMETRI
C SERVOMOTORS(
A,
#
AZ
)
The
res
ults
ob t
ain
ed so fa
r are
rel
ev a
nt for s
ymme t
ric l
ine
ar moto
rs with
symme t
ri
c 4-way
va
lves
.
In prac
ti
c e
, a
symme tr
ic motor
s mu st
someti
mes be u
sed,
for
i
nsta
nc e
to
avoi
d one
of t
he
t
wo pi
sto
n rods.
Such
a motor
mu st
be
comb ine
d with
an a
symme t
ri
c 4-
way v
alve
, a
s a
sy
mme tri
c
va
lve
is
i
nc o
mpa ti
ble
(s
ee Sect
ions
2.5
and 2.6)
.
I
t
can
be found di
rect
ly ro
f m the c
on c
lusions
in Se
ctio
n 2.6
th
a t
t
he a
b ove
resu
lts
al
so c
over
a
symme t
ri
c motors,
pro
v i
ded t
h a
t s
ome typical
corr
ect
ions
are
made.
In
Eg n.
(4.
7) t
he
pist
on ar
ea
A
s
hou
ld
b
e
re
pla
ced
b
y 0
.5(
A,+AZ
)an
d i
n
Eg
n.(
4.
8)
f
or
t
he
s
upp
ly
p
res
sur
e,凡,s
hou
ld
b
e
r
ep
la
ced
b
y
F
r
一0.
5(
A,一AZ
)P・
So
A=0
.5(
A,+AZ
)
M(
A
)b
z夕二 (
4.11
)
2E
只 =3
E・
My
m+1
n Fs
,一0
.5
(A,
一A2
)Pl (
4.1
2)
,, z,
M叭 Yn
Wx
Q
一 “ ̄
a
nd
N-
Y
shoul
d be
c
orr
ect
ed
cons
eauent
lv. “
.
右
扭留
入
R 、
i
'
=
m
产
丫
R1
.
}
m
'
a
x
}
l.
7
1
3
r=J’刃,“℃
人
/ \
、产
"'
I
少/
气
/ \
、声甲
\
万 ^ ^
多 =
PUnstall
t J
'-・
1
}3
/ 一 ,1
其 a,
山1 田 . 勿
子
(%J
Fi
g.4.
3
Fi
nal
pe
rfo
rma
nce
d
iag
ram
99
Eq
uat
io
n (
4.1
2)
ma
kes
c
lea
r t
hat
t
he
l
owe
st
p
oss
ibl
e s
uppl
y p
res
sueP
r S
wi
l
l b
e o
bta
ine
d
a
t
F5
,二0.
5(A,一A2
)P
S.Co
mbi
nat
i
on
o
f
Eg
ns.
(
4.1
1)
a
nd
(
4.
12)
gi
ve
s
ar
eas
A,
a
nd
A2
・
A
,
2…F
p
S
5
,
W
-2
b
2
E・
9
ma
x
1
3M
2
F
}
,
ma
x
・ (
4.1
3)
3Ey,
a
x (
4.1
4)
PS
mi
n=
Wb夕
叭2。、
Ma
ximu
m v
alv
e c
apa
cit
y
an
d p
eak
p rnow
owe bec
ome
(
4.15)
Qi
二 =A,
yma
x,
Ni
p=P,
人夕, ・
3
2[
M
}
-.
+F
,
-
T (
4.1
6)
In
co
mpari
son
wi
th
t
he s
ymmetri
c
motor
,
the
e
ner
gy
consumpt
io
n of
a
p
re-l
oad
ed
asy
mmetr
ic
mo
tor
c
oul
d b
e c
ons
ide
rabl
y l
owe
r,es
pec
ill
a y
when
t
he
"
inte
grat
ed"
po
wer
cons
umpti
on
(
in
k
Wh)
i
s cons
ider
ed.
In
man
y c
ase
s
an
"o
pt
imal
"
ar
ea
r
at
io
A,/A2,mak
in
g
FS,=0.
5(A,
一A2
)PS,i
s
not
po
ssi
bl
e
f
or
pr
ac
ti
cal
r
ea
son
s.
Ta
ki
ng
t
he
ra
ti
o入/ A:二;L a
s
a
sy
st
empar
amet
er
Eg
ns.
(
4.11
) a
nd
(4.
12)
y
ie
ld,
at
f
er
c
ons
id
era
ble
wo
rki
ng
o
ut:
耐,+F s
,.
2()
.+卫 a
t 几兰
3
耐二十2
凡 (
4.1
7)
Ps二3
E・ 3
耐 二一2
Fst
M(
Ob
2夕, 5几一1
"
M’j
呻
耐。:一Fs
,. +2F5
t
P
S=3
E・
2(}
.
+
1)
a
t几全 (
4.18
)
Mc
)b
2夕,
5一又 一3My
,一2 Fs
t‘
PS二
3EYf
k
%r 、
=3Mjy
. +
2F,
, (
4.1
9)
2 ,
3My二 一2Fs
t
叭
Yma
x
4.
9 DESIGN EXAMPLE
(
I
IN
Fi
g.
4.
4)
mot
or
t
o be
de
signe
d:
Sp
eci
fi
cat
ion
s o
f a
s
ymme
tic
r
E=1
09
N/
m 2 F=
0,
M “1
500k
g,C
t
)b“l
OHZ二6
3ra
d l
s
夕二 =lm 夕
,=2m ls ;
二二1
0m/
,=’
“・
Th
e d
esi
gn
d
ata
n
ow
b
eco
me
U
n
, 2
iv
A=
成,
二 ̄ b
,盯口x
1
50
0.6
32.
1.1
。一m2
=30
cm2
2E 2。
109
1
00
3・
109.
P=
3芍I
x
a
n 1
0
=7
5.
10
5聋=
75
b
ar
,
C
O
6
2夕二 632.
1 ___‘
I
I刀
_ Mo
V
o
n2y
.
,.
二二
Y
}X
二 一
1
50
063
2
.
1
.
2
_c
1
n-
3
m3
_
"c
n
二
d
二—
'
n3 = V l1V 一 = Juv —
一’ 2E
2.
IOy
s
min
Nr
=Q.
P,=6
.10
-3・
75・
105=4
500
0Wa
tt=4
5kW.
Furt
her
we
f
ind
2 22
2
ra
d=0
.
32
。,Y
2一李=
叭、
— = 0.
4 m
=40
cm,
呻 欲
1 10
“
肠一叭
-
1
0 _r
ad 10
1
2,
-
一
二 — = 5 =0.
8 Hz
, Y3= - ̄二二= — m = 2.
5mm.
63`
400
Th
e r
esu
lt
s(I)a
re
pl
ote
d i
n
the
p
erf
orma
nce
di
agr
am
of
F
ig.
4.
4.
I
nt
r
.,o
duc
ing
_a
c
ons
ta
nt
p
re-
l
oad
F
,s
.
r=Mg
_ =1
__500
0
_N,
the
_d
_e
s
_i
g
功n
o
n3f
mo
to
_r
_a
nd
v
_a
_l
ve
_(
A,
Q.)
r
eA
ml
a
i/A
n
t
hZ=5
e
sa
m g
e i
v
,
bues
t
PA;
,
a=
n5
0c
d m一
NP 1
mu
s;
=
t
b75
e
db
oa
r
u,
b
l(
e
dh二
,
se=6
e
Eq
nU
sU
.
(4.
8),
,
a
n Ni
d
(
4.
1p
0
)=
.75kW
Th
e c
ons
ta
nt
p
re-
lo
ad
Fs
t=1
500
0N
ma
y l
ead
t
o
th
e d
esi
gn
o
f
an
a
symme
tr
ic
mo
tor
wi
t
h a
n
sy
a mme
tri
c 4
-wa
y v
alv
e.
Eq
ns
(
4.1
7),
(4.
18)
a
nd
(
4.1
9)
l
ead
t
o:
rm n
-
4
0 c
m 2 100bar 480
3
0 c
m 2 150bar 360
Thi
s
ex
ampl
e
sh
ows
ve
yc
r l
ea
rl
y
th
e
eno
rmous
i
nf
lu
enc
e
of
t
he
ch
oi
ce
of
A1
/AZ
on
t
he
su
ppl
y
pes
rsur
e
PS
,t
he
va
lv
e
ca
pac
it
y
Q. and
t
he
p
eak-
po
wer
Nn t
o f
ul
fi
ll
th
e
spe
ci
i
f c
at
io
ns.
4.
10
"I
NVERSE"
DESI
GN
EXAMPLE
(
II
I
N FI
GAA)
I
t i
s a
lso
po
ssi
ble
of
c
our
se
t
o c
alc
ula
te
t
he
p
erf
orma
nce
o
f a
n e
xis
ti
ng
mot
or,
f
oll
owi
ng
t
he
"
inv
ers
e"
d
esi
gn
p
roc
edu
re.
As
an
e
xampl
e,
we
e
nvi
sag
e a
s
ymme
tri
c
mo
tor
,
ha
vin
g
M=1
05k
g,Y}x=0.
045
m, A=9・
10-
4m2
.
5 一6
/
爪
0
J
月
 ̄
一
-
j
E=
10
9呱2
,PS
一21
01
05
NI
m2,Q
.
勺
S
-
/ .
1
W e
now
ind
f
田I=
Qr
= _2
0.
58
竺 一3
.2
8Hz
,
芍R
= s
)2=
0
2
P A2=1
29
.
6丝 一2
0.
6H
z,Y
2=3MQ赢 =0.00714m=7.14mm,
3MQ. s
2
P,A3
1
01
自匕 . OnO OO心 00甘 心石
心钊尸 00尸
/八\夕尸
\ .
成权/\/\/\
\
N
ae
N-
O
m
0
2
e
. ̄ 户,- 一 - 州,
; , ,
一O
1O
-t
n
-
<
一 太舜火
09
行0
丫」 叉淤 V1 又 1
0幻
又交
0, 喊 尹
火又《 只 E }
'
0
.
月 的刊
匆
阳
火丫 \ 4 ;EE
t
o
/ E又丫兴
0八
坦
b洲丫丫
\
火只 《 卜0 }
l
峨刊p
行9
/
、 0
只又 只只火只火
0目
e
代.
0哈 峨 石
/
r一
汉又 又只只又
、
嘛
h 助八
,
/
火 t 城 火义又火
心代仍 叹已
0代
-
I 理
丫义丫 d
i
t
a
0l义又N 峨 到钾
t
o
}
Pr
a
只 ,
丫试又 义只又丫
a 幻创户
. 乓9 叭
0
4 丫又 义义 义 黑双只
.。 n:
呼 心气
\/
/ \又丫义只义又义
的.
灼 咬入
州只只丫义又 ̄
寸
挥
幻尸n. 哟叫. 八
叫
\/
/
o
'
` \/
/\\/\/\/\/\声认/
,
幻刊.
月口0.
尹
?尸
们钊 L
N
q
L 一
_
口〕
n
幻
_
.1
_
呀
的
尸
r,
目
代
一
O
N
‘
公
尸
.
幻
N
_
含 名 蕊
,.
Fi
g.4.
4 Des
ign
e
xampl
e
102
只夕m
a
x 八八
八八,
,。_
_
_ 八。,
。__
__
_
_
、
=}
M2
A
Y
.
E =6
17丝 =9
8Hz y3=—
3E
= v .
V IJV J 1-
)rn = v .j
i-)r
rcrri
2P,
A
。、二Q
; -‘
A
二0
.9
26
m
s
y二x=
3M =
12
0畏=12g.
Th
e r
esu
lts
(
I)
hav
e be
en
pl
ot
ted
al
so
i
n t
he
pe
rfo
rma
nce
d
iagrm, g.4.
Fi 4.
Taci
tl
y i
t was
as
sumed
her
e t
hat
t
he mas
s M ismoving
i
n hor
izon
tal
pl
ane
. Wh
en
i
t i
s
mo
vin
g v
ert
i
cal
l
y,
a
c
ons
t
ant
p
re-
l
oad
f, sintroduced.
,=Mg,i The
res
ult
s r
ema
in
t
he
same
e
xampl
e t
hat
2P,
A 12
0竺 =
Y=X+g=
= 一 一 _2
1
2g so
Y= =ll
g.
3M S
Co
nse
que
nt
ly
, C
c
)2
=1
8.
9Hz
,夕:=
7.7
9mm,
a
nd夕
3
=0
.2
89mm.
Th
e d
sh
a e
d i
l
ne
i
n F
ig.
4.
4
s
hows
t
he
i
nf
luence
of
g.
4.
11
CASE
STUDIES
AND
SPECI
AL
TOPICS
4.
11.
1 AN
ANTI
ARCRAFT
GUN
DIRECTOR,
A
HYDROMECHANI
CAL
TYPE-
2 SERVO
W ITH
M ECHANICAL
COMPENSATION
Fi
g.4.
5 s
hows
a
n a
nti
air
cra
tg
f
un
p
osi
ti
oni
ng
s
erv
o s
yst
em
i
n wh
ich
t
he
p
osi
i
ton
c
omma
nd
i
s
p
rov
ide
d b
y a
h
uma
n t
rac
ker
.
(An
oth
er
v
ers
ion
o
f s
uch
s
yst
ems
u
ses
a
r
ada
r t
rac
ker
t
o r
epl
ace
t
he
h
uma
n,
h
owe
ver
t
he
g
un
po
siio
t n
ing
s
erv
o r
ema
ins
e
sse
nti
al
ly
t
he
s
ame
.)We
s
houl
d f
ir
st
n
ote
t
hat
t
he
c
ompl
et
e gu
n d
ire
cto
r mus
t
pr
ovi
de
t
wo
a
xes
(
azi
mut
h a
nd
e
val
uaion
t ) o
f c
onrol
t to
po
int
t
he
gu
n pr
ope
rly
. I
t
is
u
sua
ll
y a
cce
pta
ble
f
or
i
niia
t l
d
esi
gn
p
urpos
es
t
o c
ons
ide
r e
ach
a
xis
s
epa
rat
el
y,
s
o f
rom
her
e on
our
di
scus
sion
i
s of
t
he
a
zimut
h axi
s onl
y.
(
Final
de
sign
s
houl
d c
heck
f
or
any
po
ssi
ble
pr
obl
ems
t
hat
mi
ght
a
ris
e ro
f m
s
imu
lta
neo
us
mo
io
t n
o
f b
oth
a
xes
, s
uch
a
s t
he
a
zimu
th
mo
men
t o
f i
ner
i
ta
be
comi
ng
v
ari
abl
e
wi
t
h e
leva
ti
on
a
ngl
e.
)
I
n a
ma
n-ma
chi
ne
s
yst
em
l
ike
t
hat
of
Fi
g.4
.5
t
he
d
esi
gne
r s
tri
ves
t
o r
eli
eve
t
he
hu
man
op
era
tor
o
f a
ny
t
ask
s t
hat
a
re
be
ter
p
erf
or
med
by
ma
chi
ney
r s
o t
hat
c
onc
ent
rat
ion
c
an
b
e f
ocus
ed
o
n t
hos
e
f
unc
ti
ons
be
st
s
uit
ed
t
o h
uma
n p
erf
orma
nce
.
Compu
tat
i
on
o
f t
ar
get
le
ad
a
ngl
e
cou
ld
be
a
ssi
gne
d t
o
t
he
gu
nne
r's
mi
nd
(
duc
k h
unt
er
d
o i
t
al
l
th
e t
i
me)
bu
t t
his
s
yst
em
pr
ovi
des
a
ma
chi
ne
a
id.
Lea
d
a
ngl
e
de
pen
ds
o
n t
ar
get
r
ang
e,
a
ngul
ar
v
elo
cit
y
COI
OS
of
t
he
in
le o
f s
ight
,
an
d g
un
b
all
is
ti
cs.
If
t
he
g
unn
er'
s
ins
tr
uct
ions
a
re
s
imp
ly
t
o k
eep
t
he
c
ros
s ha
irs
o
n t
he
t
arg
et
(
not
t
o l
ead
i
t)
,
th
is
t
ask
c
an
b
e a
ccu
rat
el
y
pe
rf
or
med
a
nd
we
ma
y a
ssu
me
t
ha
t
th
e a
ngu
la
r
ve
loc
it
y
of
t
he
9,s
haf
t
is
)L
O o
s,
wh
ich
we
me
su
a r
e a
s s
hown
wi
t
h a
r
ate
g
yro
sco
pe.
We a
lso
a
ssu
me
t
hat
a
n o
pti
cal
f
ind
er
(
det
ai
ls
no
t g
ive
n)
p
rov
ide
s r
ang
e i
nfo
rma
tio
n.
Fi
xed
b
all
is
ti
c da
ta
i
s c
omb
ine
d wi
th
t
he
v
ary
ing
r
ange
a
nd
C
ALO
S
da
ta
i
n
a
c
omp
uti
ng
me
cha
nis
m a
uxi
li
ay
r t
o
th
e r
at
e
gy
ro,
th
e o
utpu
t
of
t
he
c
ompu
tat
ion
s
1
03
b
ein
g a
n a
ngul
ar
r
ota
tio
n t
hat
d
ele
f c
ts
t
he
u
ppe
r p
ris
m t
hro
ugh
a
n a
ngl
e e
qua
l t
o t
he
c
omp
ute
d
l
ead
a
ngl
e.
Th
e o
pti
cal
s
yst
em
t
hus
a
ll
ows
t
he
gu
nne
r t
o l
ook
r
ight
at
t
he
t
arg
et
bu
t t
o
po
int
(
sha
t
f
a
ngl
e
Ov)a
t
it
s
pr
edi
ct
ed
f
ut
ur
e
po
si
ti
on
wh
ere
t
ar
get
an
d a
rt
il
l
ery
s
hel
l
sh
oul
d
co
ll
id
e.
T
he
d
ata
s
yst
em
j
ust
de
scr
ibe
d i
s
no
t
wi
t
hin
t
he
s
erv
o l
oop
bu
t
is
o
f
in
ter
est
s
inc
e i
t
sh
ows
whe
re
t
he
s
erv
o
书集・
c
omma
nd
o
rig
ina
tes
.
 ̄
Tar
get
Ra阳刃:
, and Lead
aneie, 洲产 -
.....
. .......... .....
缪仁一7
//‘/
_
/ _
//
/
了
角 dr
ardi
c
G un
么,
一ac C
B
Synchr
o .
Synchro Sy
nchr
o
transnt
i扮er di
fferent
ial tr
ansmi
tter
Fi
g.
4.
5
40
mm
a
nti
ai
rc
raf
t
gu
n d
ire
cto
r
Th
e s
erv
o s
yst
em
i
ts
el
f
pr
ovi
des
f
urt
her
a
id
t
o
the
o
per
at
or
i
n
th
at
t
he
p
owe
r t
o
dr
iv
e t
he
gu
n
t
ur
re
t
is
s
upp
li
ed
b
y a
h
ydr
aui
l
c s
yst
em
r
at
her
t
han
g
unn
er
mu
scl
e.
Co
mma
nd
a
ngl
e 0
V i
s
me
as
ur
ed
b
y a
s
ync
hro
,
as
i
s
gu
n a
ngl
e
0,
.A
s
ync
hr
o d
ife
re
nt
ia
l
mo
to
r r
ec
ei
ve
s 0
,
an
d
0.
i
nf
ormat
ion
and
pr
oduce
s a
s
hatr
f
ot
at
ion
equal
t
o t
hei
r
di
feen
r ce.
This
l
ow-
power
er
ror
s
igna
l
a
ctu
ate
s a
va
lve
-co
ntr
oll
ed
se
rvo
t
hat
st
rok
es t
he
var
iab
le
d
isp
lac
eme nt
pu
mp
supp
lyi
ng
a r
ota
ry
h
ydr
aul
i
c mo
tor
ge
are
d t
o
th
e g
un
t
urr
et
.
I
nt
er
nal
de
tai
l
s a
re
s
hown
i
n
Fi
g.
4.
6
an
d 4.
7.
Th
e S
ync
hro
1
04
d
ife
ren
ti
al
mo
tor
's
s
haf
t
ro
tat
i
on
i
s 9
v一9,
+9
B,wh
ere
9
,,一9
, i
s
pr
odu
ced
e
le
ct
ri
ca
ll
y a
nd
9
B i
s p
rod
uce
d me
cha
nic
al
ly
b
y as
pri
ng/
damp
er
p
osi
t
ive
f
ee
dba
ck
i
l
nka
ge.
Th
e p
osi
t
ive
f
eed
bac
k r
eco
ver
s a
n i
nte
gra
ti
on
l
ost
by
c
los
ing
a
l
oop
a
rou
nd
t
he
va
lve
/cy
lin
der
a
nd
mak
es
t
he
o
ver
all
sy
ste
m t
ype
2
wh
en
c
omb
ine
d wi
th
t
he
i
nte
gra
tio
n i
n
th
e p
ump
/mo
tor
.乃pe
2
i
s
de
sir
ed
.
Me c
hanical
de
l,
q
net
u-or
k
Fi
g.
4.
6
Pu
mp
s
tro
ke
c
ont
rol
wi
th
de
lay
ed
po
sit
ive
f
eed
bac
k a
nd
d
ith
er
s
in
ce
t
ypi
ca
l
ta
rg
et
mo
ti
ons
p
rod
uce
r
amp
i
l
ke
c
omma
nds
9
y a
nd
z
er
o t
r
ack
ing
e
rr
or
f
or
s
uch
c
omma
nds
me
ans
t
he
g
un
p
oin
ts
d
ire
ctl
y a
t
th
e p
redi
ct
ed
t
ar
get
po
sit
io
n,
no
t
be
hin
d i
t.
To
ke
ep
105
f
ri
cti
on
b
rok
en
l
oos
e i
s a
ll
pa
cts
o
f t
he
d
riv
e s
yst
em;
t
he
v
alv
e s
lee
ve
i
s g
ive
n ac
ont
inu
ous
l
ow-
ampl
i
tud
e,
hi
gh
f
req
uen
cy
(
33Hz
) "
dit
her
" mot
i
on
xd
b
y a
c
am
a
nd
pu
sh
r
od
a
tt
ache
d t
o
th
e
s
lee
ve
f
eed
bac
k l
i
nka
ge.
The
f
req
uen
cy
o
f s
uch
di
t
her
ing
i
s
no
t
cr
it
ica
l
s
a l
ong
a
s i
t
is
h
igh
r
ela
ti
ve
t
o s
yst
em
ban
dwi
dth;
33
Hz
wa
s u
sed
h
ere
s
inc
e i
t c
orr
esp
ond
ed
t
o t
he
r
ota
ti
ona
l s
pee
d o
f t
he
e
lec
tri
c
pu
mp
d
riv
e mo
tor
,
wh
ose
s
haf
t
co
uld
t
hus
be
u
sed
f
or
t
he
d
ith
er
c
ams
haf
t
An
aly
sis
o
f t
he
s
yst
em
b
loe
k d
iagr
am
(
Fig.
4.
7)
r
eve
als
s
ome
i
nte
res
ti
ng
f
eat
ure
s.
F
irs
t
no
te
t
hat
t
he
s
eco
nd-
ord
er
dy
nami
cs
u
sua
ll
y i
ncl
ude
d i
n v
alv
e%yl
i
nde
r a
nd
p
ump
/mo
tor
h
ydr
auic
ls ha
ve
been
ne
glect
ed
f
or
s
impl
ici
ty
and
t
hat
we
i
gnor
e t
he
di
ther
.
We
ca
n wr
it
e
{
1(
1
9'
-1
9,
,十瓦
K
P
ms
(
K
2
b
j
,
s+
1
)。
・
」
一K
s
KP
m
c
。
・}
K
sV
E
一K
s
P
m口
・
!
K
二
(
,
T
e
Z
1
K
b
j
p
v
+
m
1
r
一
/
"
v
{
Kv
e
K
1
P
mr
・制2
+
z
+
K
P
1
m
r(
K
r一
K
b
fr
)
s+
1
8e(
4
.
2
1)
I
f
we
se
t
K,一K,r二0, t
he
s
te
ady-
st
at
e
er
ror
f
or
a
r
amp
6V
be
comes
z
er
o,
as
d
esi
re
d.
Howev
er
i
f
2j
b wer
e
zer
o,
we
woul
d
lo
se
t
he s t
er
m c
ompl
et
ely
i
n
Egn.
4.
21,
gi
vin
g a
bso
lu
te
i
nst
abi
li
t
y;
t
hus
t
he
de
la
y i
n
the
po
si
ti
ve
fe
edb
ack
is
e
ss
ent
ia
l.
To s
ee
th
at
se
tt
i
ng Ki=Kf ,
r makes
t
he
s
yst
em
t
ype
2
we
n
eed
t
o
re
duc
e t
he
mi
nor
l
oop
s.
r
Wve
crl
hnder 乃r
nr
p ne
otor
一  ̄.、 ̄...叭 产 .曰..护....气
工丫 臼
0
or + BJ Re一陇 +)
& x
8
j, 二。 I
K1
.
C 1,
. 区
P
m匕二
卜 +.
一 !
一厂
儿 N,
t
h,
I)el
ared Ad
pow n"
e
f
'
eedhnc
ir A
_i
f
协s+1
Fi
g 4.
7
Gu
n d
ire
cto
r bl
ock
di
agr
am
1
(
0,
一
II
L\
。
。
十Kf
'二
;l
一、,
二
;{
KV
}
一二
, ‘ 二 J
J
,
L
t'
r 1
/
1
I }
_
I
(
4.22)
xp
K,
(2
'
bs+
f 1
)
(
s)二
6,一8, s
仃.s+1
)
where
106
△
z户 }
,
K
r
T,二 K,
= (
4.23)
1
十:.
bK
v
c凡 1
+T
bK
J
,
cK,
We
c
an
n
ow
r
edr
aw
t
he
b
loc
k di
agr
am
a
s i
n F
ig.
4.8
, wh
ich
ma
kes
c
lea
r t
hat
we
h
ave
t
ype
2
.
Al
so
n
ote
t
ha
t s
in
ce
Z
l<T
.,
t
her
e i
s
al
so
l
ea
d c
omp
ens
at
io
n,
a
ugme
nt
in
g s
ta
bi
li
t
y.
I
n f
ac
t,
o
ur
p
res
ent
model
i
s a
bsol
ut
ely
st
abl
e f
or
any
pos
it
ive
lo
op
gai
n.
I
ncl
usi
on
of
t
he
n
egle
cte
d dyn
amics
i
n t
he
v a
lve
cy
/ lin
der
and/
or pu
mp /
mo t
or
hydr
auic
l s
wi
ll
of
cour
se
make
a
bsol
ute i
nst
abi
li
ty
pos
sibl
e.
01
- +
凡(
协S+1
) 朴 K
pi
, g
r
s(
r,+1
) s
9e
Fi
g.4.
8
Bl
ock
d
iag
ram
wi
th
mi
nor
l
oop
r
edu
ced
4.
11.
2 THE
HAYSTACK
HILL
RADI
O TELESCOPE
POI
NTI
NG
SYSTEM;
CONDITIONAL STABI
LITY/
SATURATION PROBLEM S SOLVED
WI
TH
A DUAL-
M ODE
SERVO
I
n 1
960
t
he
U.
S.
Ai
r Fo
rce
, t
hro
ugh
i
ts
p
roj
ect
ma
nag
er
MIT
Li
nco
ln
La
bor
ato
ry,
s
oi
lc
ite
d
p
ropo
sal
s f
or
d
esi
gn
a
nd
c
ons
tr
uct
ion
o
f a
l
arg
e a
nd
p
rec
ise
r
adi
o t
el
esc
ope
t
o b
e i
nst
al
le
d o
n
Ha
yst
ack
Hi
ll
o
uts
ide
Bo
sto
n,
Ma
ssa
chu
set
ts
.
Pr
opo
sal
s
ro
f m
va
rio
us
b
idde
rs
we
re
e
val
uat
ed
a
nd
t
he
c
ont
rac
t
wa
s a
war
ded
t
o No
rth
Ame
ric
an
Av
iat
ion
(
Col
umb
us,
Ohi
o Di
vis
ion)
.
On
e pa
rt
o
f t
he
p
ropo
sal
wa
s de
vot
ed
t
o t
he
e
lec
tro
hyd
rau
lic
c
ont
rol
s
yst
em
f
or
s
tee
ring
a
nd
po
int
ing
t
he
a
nte
nna
d
ish,
and
we
b
ase
o
ur
di
scus
si
on
o
n t
he
de
sign
p
res
ent
ed
i
n t
his
p
rop
osa
l.
De
sig
n a
nd
c
ons
tru
cti
on
p
roc
eed
ed
o
ver
s
eve
ral
ye
ars
,
wi
th
t
he
a
nte
nna
be
ing
e
rec
ted
a
nd
p
ut
i
nto
s
erv
ice
i
n 1
964
,
wi
t
h
RF
axis-
-
石l
evat
ion
Fi
gA.
9
El
eva
ti
on-
ove
r-a
zimu
th
a
nte
nna
c
onf
igu
rat
ion
107
o
per
ati
on
c
ont
i
nui
ng
t
oda
y(1
985
).
Ou
r di
scu
ssi
on
wi
ll
a
lso
i
ncl
ude
p
robl
ems
t
hat
a
ros
e du
rin
g
f
ina
l d
eve
lopme
nt
a
nd
r
equ
ire
d mo
diic
f a
ti
ons
o
f t
he
o
rig
ina
l d
esi
gn.
At
t
he
t
ime
of
i
t
s e
rec
ti
on,
t
his
a
nte
nna
wa
s t
he
mos
t
pr
eci
se
i
nst
rume
nt
o
f i
ts
s
ize
i
n t
he
wo
rld
.
A so
-calle
d el
evat
ion
-over-
azi
muth
confi
gur
atio
n was c
hosen
to
posi
ti
on t
he
12 0-f
t-d
iamete
r
di
sh(
see Fig.
4.
9).
Fi
gure
s 4.1
0 a
nd
4.11
show mechan
ical
de
tai
ls
of
t
he dr
ive
ar
rangeme nt
,
two
hydr
auli
c motor
s
(fed
i
n pa
ralel
by
a
s
ingl
e s
erv
o val
ve)
be
ing
use
d t
o d
riv
e ea
ch
axi
s .
Digi
tal
Fi
g.
4.
10
Me
cha
nic
al
a
rra
nge
men
t
of
d
riv
e s
yst
ems
s
haf
t a
ngl
e e
nco
der
s wi
th
ar
e
sol
uti
on
o
f 0.
001
0 a
re
u
sed
t
o s
ens
e t
he
po
sit
ion
o
f e
ach
a
xis
.
Be
ari
ngs
f
or
t
he
e
lev
ati
on
a
xis
a
re
g
rad
e z
ero
Ti
mke
n t
ape
red
-ro
ll
er
be
ari
ngs
, wh
ere
s
a t
he
e
lev
ati
on
a
xis
t
hru
st
a
nd
r
adi
al
l
oad
s a
re
c
arr
ied
b
y af
our
-po
int
a
ngul
ar
c
ont
act
b
all
b
ear
ing
(
141
-i
n.
d
iame
ter
)
wi
t
h t
wo
r
ows
o
f b
all
s.
(
Aft
er
t
he
c
ont
ra
ct
wa
s wo
n a
nd
d
esi
gn
p
roc
eede
d,
i
t
wa
s de
cid
ed
t
hat
a
hy
dro
sta
ti
c o
il
b
ear
ing
woul
d i
nst
ead
b
e u
sed
f
or
t
he
e
lev
ati
on
a
xis
.)
Th
e
108
r
equ
ire
d po
int
ing
a
ccu
rac
y i
s
10.
005
0 a
t
th
e RF
(
rat
io
f
requ
enc
y)
a
xis
f
or
a
ny
c
omb
ina
ti
on
o
f 0
t
o
1
0/
s,
t
r
ack
in
g v
el
oc
it
y a
nd
0
t
o 0
.01
20/
s
2
t
r
ack
in
g a
cc
el
er
at
i
on,
mu
st
b
e l
es
s t
ha
n
士0
.00
50
s
i
nce
d
is
h
ge
ome
tye
r
r
ror
s
(c
aus
ed
b
y
gr
avi
t
y a
nd
t
emp
era
tu
re
i
ndu
ced
d
is
to
rt
io
ns,
et
c.
)
ca
n cause
a而s di
rec
ti
on of
the
elec
tro
ma gn
eti
c l
ine
of
si
ght (
RF a
xis
)even
when t
he
shaf
ts
a
re
pr
e c
ise
ly
posi
ti
one
d.
Se
l
ecti
on
of
motor
s a
nd ge
ar
b o
xes r
equi
res
e
sti
ma t
ion
of
the
t
orq
ues
th
at
mu s
t
be
provi
ded
f
or
ri
fc t
ion,
win
dage,
a
nd acc
ele
rat
ion.
Ma ss
moments
of
ine
rti
a f
or
th
e e
levat
ion
and
azi
muth
ax
is
we
re
e
st
i
mat
e
d a
t
3
x
1
06
a
nd
4
x
1
06
l
b
f一户一s
2,r
es
pe
ct
i
ve
ly
,
wh
er
ea
s
de
ad
w
ei
gh
ts
w
er
e
1
20,00
01bf
a
nd
240,
000
lbf
.
Sin
ce t
he a
zimuth
axi
s
th
us
has
t
he
most
s
eve
re
fr
ict
ion
and
i
ner
ti
al
l
oad
s,
the
pr
opo
sal
c
arr
ied
out
a
de
sign
for
i
t
onl
y a
nd
sug
ges
ted
t
hat
if
t
he
sa
me c
omp o
nent
s
were
us
ed
f
or
t
he
e
lev
ati
on
a
xis
i
ts
p
erf
orma
nce
s
hou
ld
a
lso
be
a
deq
uat
e.
Th
is
de
si
gn
ph
il
oso
phy
a
lso
mini
mizes
the s
ize
of
the
spar
e par
ts
in
vent
ory
that
mu s
t be
kept
on ha
nd.
Note
a
lso
t
hat
,
at
t
he
pr
oposal
st
a g
e, e
ngine
eri
n g
desi
gn efot
r i
s f
und e
d by th
e pro
po si
ng
company
and
cann
ot
be
r
ecover
ed
unless
the
de
sign
competi
ti
on i
s
won,
thus
th
e compani
es must
r
isk
s
ome
of
th
eir
c
api
t
al
i
n t
he
att
empt t
o get
busin
e s
s.
I
nqui
rie
s at
t
he be
ari
ng manufa
ctur
er
gave
△
Tf=f
ric
tion
t
orque (
4.24)
=(
0.00
2 t
o 0.
003
) (
thr
ust
lo
ad)
(
bea
rin
g r
adi
us)
F
r
om
pu
n
gl
几 广T
For
ce
o
re
s
e
mot
r
v
o
i
or
r
Serro
val
ve
SG娇 Tachomet
er
cucoder
龟
角"
draul
ic
场v
lr
auhc
1
uotor
.
peed _
re汤l
eer
Bul
;ear 一
从e
e
r11
{
g
-
t
11
p幼 e
r 山”
切or
ator Di
git
al Com mand
r
e;八t
er
To
readou
t
Fi
g.
4.1
1 De
tai
ls
o
f s
erv
osy
ste
m h
ard
war
e
109
A
cons
erva
tive
des
ign
val
ue
was
t
ake
n a
s
T
f=(
0
.0
2)
(
24
0,
00
0)
(
80
in
.
)ft )
一32
,
00
0.
f
t一
1b
, (
4.25)
11
2)
An
ten
na
s
pec
iic
f a
ti
ons
r
equ
ire
a ma
xi ma
mu c
ce
le
ra
ti
on
o
f
10
/s2,t
hu
s t
he
t
or
qu
e T
f
or
accel
erat
ion
i
s
-
L ,
a=J ,
a=(
4J‘
XI
U一一,
,一f一
Of t一s,
一.
(
)
}
二二1
0
一
二万/
s
,2
- 、_
一万}
=’_
/
U,_
0_
0__
0
ft一l
b
f (
4.26)
又〕/
.3一
/
ra
d)
S
inc
e t
he
a
nte
nna
wi
l
l b
e i
nsi
de
a
p
rot
ect
ive
f
ibe
rgl
ss
a r
ado
me,
no
wi
nd
g
ust
to
rque
s a
re
p
oss
ibl
e;
ho
wev
er
r
ota
ti
on
o
f t
he
d
ish
i
n s
ti
ll
a
ir
d
ose
c
aus
e a
d
rag
t
orq
ue.
One
i
ntu
iti
vel
y be
li
eve
s t
his
i
s
s
mal
l,
bu
t s
inc
e t
he
di
sh
i
s v
ey
r l
arg
e,
yo
u c
an
no
t b
e s
ure
wi
tho
ut
a
c
alc
ula
ti
on.
The
wo
rst
c
se
a
wou
ld
be
a
zimut
h r
ota
ti
on
a
t ma
ximu
m v
elo
cit
y wi
th
t
he
a
nte
nna
p
oin
ting
h
ori
zont
al
,
s
a i
n Fi
g.
4
.12
.
Sp
eci
f
ica
ti
ons
ca
ll
f
or
a
ma
ximu
m v
elo
ci
ty
o
f3.
3%s
.
Th
e
dr
ag
t
or
que
i
s
as
sume
d t
o
be
c
aus
ed
b
y s
tag
nat
ion
p
res
sur
e o
f t
he
a
ir
a
gai
nst
t
he
d
ish
s
urf
ace,
whi
ch
i
s t
rea
ted
a
s a
la
f t
p
lat
e (
wit
h a
c
orr
ect
ion
f
act
or
o
f 1
.28
f
or
t
he
a
ctu
al
s
hape
).
,e __
:
_ 」
______
__。 、,,。 。
。*厂py 2飞、。_
_、 _
、
2
,产
、..尹
L、= wc
nuage
or
t que ‘,
1
1,1-
40
1-
I一一
,
.一.
I
V
--f
Ark (
4
:
‘
”欠 ‘ )
,曰
0八
、
、.尹
T
,y=2
.68
x
1
0-G
R
5
208
0
ft
一1b
f
f
or
R=60
ft (
4
Th
e a
zimu
th
d
riv
e wi
ll
t
hus
h
ave
t
o b
e c
apa
ble
o
f p
rod
uci
ng
a
ma
ximu
m t
orqu
e of
T
..=乓+T
.+几=1
04,
08
0
ft
一1b
f (
4.29)
V 二 尸“
h二2,
fR2一T
2
Fi
g.4.
12
An
ten
na
wi
nd
a
ge
c
alc
ula
tio
n
110
We
c
an
n
ow
s
el
ect
a
mo
tor
a
nd
g
ear
box
t
o me
et
t
hes
e n
eeds
.
Ge
arbo
x c
ompl
鲜it
y}
c
os-
t
,
an
d
backlash
usu
all
y
incre
se
a with
gear
ra
tio,
t
hus
we
prefer
a
high
-tor
que mo t
or.
In
1960
t
he s
ele
ctio
n
of h
igh-t
orque
hydr
a u
lic
motors
s
uita
ble f
or
pr
eci
sio
n ser
v o
cont
rol
wa s
not
as
gr
eat
a
s i
t
is
t
oday,
thus
ther
e were
no
t many c
andidat
es
to c
onsi
der
and
the
choic
e was
narrowed
to
an
ei
ght-
cyl
inder
radi
al pi
ston
desi
gn wit
h ap
rov
en record
of
se
rvo per
formance i
n ma chi
ne
tool
dr
ives
and a
d
is
pl
ac
eme
nt
d,
,
of
2
2
in
3/
r
ev.
Th
is
mo
tor
is
c
ont
r
oll
ab
le
o
ver
a
r
ang
e
of
1
/6
0,
00
0
to
5
00
r
pm.
Whe n r ot
at
i ng
at 100
rpm i
t can ber e
ver
sed
i
n 0.0
015s
with
onl
y 0.
000
6 o
f a
rev
olu
tio
n
m ovem ent ter t
af he s
ignal
t
o reverse . Sy
s
tem
suppl
y pr
ess
ure
was ch
ose
n s
a 10
00 ps
i,
a
conservat
ive
low v
alue
conduci
ve
t
o l
ong
l
if and one
e f
or
whi
ch
many
st
andar
d hyd
raul
ic
c
ompone
nts
a
re de
sig
ned.
Usi
ng
t
he
ma x
imu
m powertr
ans
fer
re
sul
t,
we
want
t
he
mot
or
p
ressu
re
d
rop
t
o be
66
7 psi
at
t
he
ma
ximu
m t
orque
c
ondi
ti
on.This
gi
ves
fo
r bo
th
mot
ors
on
t
he
azi
muth
dr
ive
t
ot
al
mo
to
r
t
or
au
。二2 m 2
些d x
667
x
22
:二 一 :二
354
f
t一l
,
b,
,
(
4.30)
1
2
x
21t 乙兀
A
t
ent
ati
ve
g
ear
r
ati
o
is
t
hen
N=型塑9
=2
94to
1 (
4.31)
354
它二0勺灿御认』、从钾艳
产
扮 勺。
沁 1
Const
ant
v
eto
cit
y
f
、
deg
ree
s
/
secon力
亡ons
tant
acre佗r
ation
at二1。:
,2
,山tl
0
t2 t.sec
公
v
夏I
SO
是
a
a
,
Fi
g.4.
13
Du
ty
c
ycl
e
fo
r 1
800
p
osi
ti
on
c
hang
e
111
Co
nsu
lta
ti
on
wi
th
g
ear
b
ox
man
ufa
ctu
rer
s i
ndi
cat
ed
t
hat
f
our
me
she
s o
f p
rec
isi
on
h
eic
lal
g
ear
s
wi
th
98
% e
fic
ien
cy
p
er
me
sh
wi
ll
be
n
eed
ed.
^co
rre
cte
d g
ear
r
at
io
i
s
th
en
294
N = =320 t
o 1 (
4.32)
0.
984
Si
nce
a
ctu
al
ge
ar
r
ati
os
c
onv
eni
ent
t
o p
rodu
ce
d
epe
ndo
n t
ool
ing
a
nd
o
the
r ma
nuf
act
uri
ng
cons
ider
ati
ons
,
the
f
inal
choi
ce
ha
d t
o a
ccommoda
te
t
his
and
was
325
t
o 1
.
Wi
th
mot
ors
a
nd
g
ear
b
oxe
s c
hos
en,
we
n
ow
t
un
r t
o s
erv
ova
lve
s
ele
cti
on,
wh
ich
ge
ner
all
y i
s
d
ete
rmi
ned
b
y ma
ximum
lo
f w
r
ate
r
equ
ire
men
ts
a
t a
ctu
ato
r ma
ximu
m v
elo
cit
y.
The
ma
ximum
a
ngu
lar
v
elo
cit
y o
f 3
.30
/s
us
ed
i
n t
he
wi
nda
ge
t
orqu
e c
alc
ula
tio
n wa
s n
ot
d
ire
ctl
y gi
ven
i
n t
he
s
pec
iic
f a
ti
ons
b
ut
wa
s de
riv
ed
ro
f m
t
he
f
oll
owi
ngr
equ
ire
ment
.
The
a
zimu
th
d
riv
e mus
t
be
c
apa
ble
of n
maki g
a
18
00
p
osi
ti
on
c
hang
e i
n o
ne
mi
nuewi
t th
ma
ximum
a
cce
ler
ati
on
h
eld
t
o 1l
0/s
2.
Thi
s
mot
ionc
an
b
e a
cco
mpl
is
hed
i
n v
ari
ous
wa
ys,
s
o o
ne
mus
t c
hoos
e s
ome
r
eas
ona
ble
d
uty
c
ycl
e,
such
s
a t
hat
in
Fg.
4.1
3,
f
or
c
alc
ula
ti
on
pu
rpo
ses
.
Fr
om
t
hat
ig
f ur
e we
ma
y wr
it
e
10ti
2 , 10
t矛
18
00=共二二
二+t(60一2t
,)+气种三 (
4.33)
s`
2 ‘’ “ s`
2
t
i=3
.3s,
max
imu
m v
elo
cit
y=3.
301
s
Ma xi
mum
mot
or
v
eloci
ty
i
s
then
179 r
pm and
the
serv
ova
lve
lo
f w r
ate
(
for
t
wo mot
ors
)
is
3
4.1
ga
l/
min.
Whe
n t
he
loa
d is
runn
ing a
t cons
tan
t v
elo
cit
y t
he
mo t
or s
uppl
ie
s onl
y ri
fct
ion
and
wi
ndage
t
orques
(
no
ac
cel
erat
ion)
and,
f
or
bot
h mot
ors
34080
mot
or
tor
que
_ ’116
t一l
f b,
护
j
(
4.34)
mo
tor
pr
ess
ure
d
rop=1
98
ps
i.
Wit
h a
19
8 ps
i
dr
op a
cross
ea
ch
motor
,
th
e dr
op
acro
ss
bot
h
port
s
of
t
he
s
erv
oval
ve
is
8
02 p
si.
One
ca
n now
cons
ult
s
erv
ov a
lve
cat
al
ogs
and
choos
e a
val
ve
t
hat
wil
l
pr
ovi
de
at
l
eas
t
34.
1 g
pm wi
th
a
n 8
02
ps
i pr
ess
ure
d
rop
whe
n wi
de
o
pen.
Th
oug
h n
ot
n
ece
ssa
y
r f
or
v
alv
e s
ele
cti
on
we
c
an
e
st
imat
e t
he
nee
ded
va
lv
e po
rt
l
fow a
rea人 fro
m s
ta
nda
rd l
f
uid
mech
ani
cs
ori
i
fce
-l
fo
w
e
quat
i
ons,
usi
ng
a
n
oil
de
nsi
ty
o
f
0.
000
084(
lb
f-
s2
)
/i
n.
a
an
d d
is
cha
rge
c
oeic
f i
ent
of
0
.63
.
二
,
一,・
,
一 。
.
6
3A
,糯蕊.
、,
一。
.
0
97
i
n.
2 (
4.35)
At
this
poi
n t
_o
ur
"bare
bone
s"
syst
em c
ompri
ses
a
ty
pe-1
el
ect
ro
hydr
aul
ic
se
rvo
as
i
n
Fig.
4.1
4.
Ou
r n
e x
t
s t
ep i
s
to
est
imate
th
e l
oop
gai
n ne
ede
d i
n t
his
sy
ste
m t
o mee
t
a s
te
ady-
sta
te
er
ror
g
oal
乡ve
n in
t
h e
speci
fi
cat
ions.
Th
is
spe
ciic
fati
on
re
quir
es
a t
ra
cki
ng
err
or
at
t
he
RF ax
is
of
no
more
t
han
10.
0050
when
t
he
a
nte
nna
has
a
mot
ion
gi
ven
by
二、2
}2
6
9
-=O
x+竺生=(
10
/
S)
t+(
0.
012
/
S2)
生 (
4.36)
2
、 ‘ ’ 2
Si
nce
t
hi
s
mo
ti
on
i
s
qu
it
e
sl
ow,
we
si
mpl
if
y
th
e s
yst
em
mod
el
t
o
th
e p
ure
i
nt
egr
ato
r o
f
Fi
g.
4.
14
b
),
wh
ich
t
he
n
gi
ves
er
ro
r
9E
a
s
112
1
d9,
.
1,
d}=一 一 二一 t
o)
+at, (
4.37)
K dt
K 、
Th
e
er
ro
r
co
mpo
nen
t at
(
ca
use
d b
y t
he
c
ons
ta
nt-
ac
cel
er
at
io
n c
omp
one
nt
o
f
9,)
wou
ld
1
1)
h)
Fi
gA.
14
Ba
sic
s
yst
em
bl
ock
di
agr
am
wi
th
s
impl
i
ie
f dversion
i
nc
re
se
a wi
t
hou
t
bo
und
,
ho
wev
er
t
he
0
.01
20
/s
2
ma
xi
mum
a
cce
le
ra
ti
on
c
ont
i
nue
s o
nl
y
un
ti
l
i
t
c
ont
ri
but
es
a
v
elo
cit
y
eq
ual
to
t
he
g
ive
n C
.
0o
f
10
/s.
Thus
t
he
ma
ximu
m e
rro
r i
s
1 201s
(
4.38)
d
E
ma
x=K(
。+田)
=
K
Th
is
p
oin
tin
g e
rro
r i
s t
he
"
vec
tor
s
um"
f
or
bo
th
a
zimut
h a
nd
e
lev
ati
on
a
nd
i
s wo
rst
wh
en
t
he
e
lev
ati
on
a
xis
po
int
s
ho
riz
ont
al
ly
,
gi
vin
g
t
ot
a
l
e
rr
or
=扣2
E
.
el
+2
eE
.
az
=v9
E.
a
z (
4.39)
whe
re
we
have
as
sume
d t
he
a
zimut
h and
el
eva
tion
er
ror
s t
o be
equal
.
We
ha
ve
t
hen
K_(
20
/
0.
s)
,
005
F2
_56
7
1 (
4.40)
I
f
we
u
se
K
=5
67,
all
t
he
a
ll
owa
ble
e
rro
r i
s
us
ed
u
p by
t
he
c
ont
rol
s
yst
em,
le
avi
ng
no
ne
f
or
t
he
d
ish
s
tru
ctu
ral
e
rro
rs
me
nti
one
d e
arl
ie
r.
Co
nsul
t
ati
on
be
twe
en
t
he
s
tr
uct
ura
l a
nd
c
ont
rol
-sy
ste
m
e
ngi
nee
rs
i
n t
he
el
r a
ti
ve
s
eve
rit
y o
f t
hes
e t
wo
e
rro
r s
our
ces
r
esul
t
ed
i
n a
d
eci
si
on
t
o d
ivi
de
t
he
e
rro
r bu
dge
t a
bou
t e
qua
ll
y a
nd
ad
es
ign
g
oal
of
K=1
200
wa
s s
et.
Th
e s
tru
ctur
al
e
ngi
nee
rs
a
lso
pr
ovi
ded
a
n e
sti
mat
e o
f t
he
a
nte
nna
s
tr
uct
ure
's
f
ir
st
na
tur
al
f
req
uenc
y o
f v
ibr
ati
on
s
a 5
11z
. To
113
r
edu
ce
t
he
p
oss
ibi
li
ty
o
f c
ont
rol
-s
yst
em
mot
ion
s e
xci
ti
ng
t
his
f
req
uen
cy,
a
cl
ose
d-l
oop
p
eak
o
f 1
.3
a
t 0.
5 Hz
, wi
t
h r
api
d a
tt
enu
at
ion
b
eyo
nd
0.
5Hz
, wa
s s
et
a
s a
not
her
d
esi
gn
g
oal
,
co
nti
nge
nt
u
pon
t
his
n
ot
v
iol
at
in
g a
ny
o
the
r s
pec
ifi
cat
ion
s.
At
t
his
poi
nt
t
he
ba
sic
har
dwar
e i
tems
ha
d a
ll
bee
n s
elec
ted
and
i
t wa
s pos
sibl
e t
o c
heck
d
yna
mic
s f
or
p
oss
ibl
e
si
mpl
i
fic
ati
on.
Th
e s
haf
t
enc
ode
r,
DI
A c
onv
ert
er
,
an
d pr
eampl
i
fie
r p
rod
uce
d
a
1
000
Hz
a
mpl
i
tude
-mo
dul
at
ed
a
c s
ign
al
t
hat
h
ad
t
o b
e d
emo
dul
at
ed
a
nd
l
ow-
pas
s f
il
te
red
t
o
p
rodu
ce
a
s
uit
abl
e d
o c
ont
rol
s
ign
al.
Beca
use
o
f t
he
hi
gh
a
c re
f qu
enc
y,
t
he
l
ow-
pas
s f
il
te
r c
oul
d
us
e as
mal
l
T(
0.01
s)a
nd
t
hus
t
hes
e d
yna
mic
s c
oul
d b
e n
egl
ect
ed
r
el
at
iv
e t
o c
ont
rol
s
yst
em
ba
ndwi
dth
o
f a
f
ew
h
ert
z.
S
erv
ova
vle
d
yna
mic
s f
rom
v
avl
e
cu
rre
nt
t
o v
avl
e
ope
ning
we
re
o
bta
ine
d
ro
f m
t
he
v
avl
e
ma
nuf
ac
tu
re
ss
a
ec
ond
o
rde
r
wi
t
h 0n=5
( 0Hz,
aga
in
n
egl
i
gib
le.
Fa
r
th
e h
ydr
aul
i
c
s
yst
em,
th
e n
atu
ral
re
f q
uen
cy叭 wa
sc
al
cul
at
ed
f
rom
、一
、I
/
,
2
_
}
M!
7
B
7
d一1
4
ra
d
V
/
・
JVI
(
4.41)
wher
e
△
MB= bulkmo
dul
us=1
00,
000
ps
i
△
J=i
nert
ia r
efer
red t
o mot
or
s
haft ; .J
rad
o
J ____ __ ,r w
尸月。日
口r
ee
s一二 一
乙
儿
4X1
06
X1
2
J二0.
004+ =500i
n一1
b,
S2
(
325
)2
J
vc
=
vol
ume o
f
oi
l
under
compr
ess
ion
vc
=5
5in
.3
(
two
mot
ors
)+45
in.
3
(
est
imat
ed
t
ubi
ng)
F
rom
p
ast
ex
per
i
enc
e,
h
ydr
aul
i
c s
yst
em
d
amp
ing
wa
s
es
ti
mat
ed
a
t杏二0.
3.
Th
e
ba
sic
sys
tem t
hus
appe
ars
as
i
n
Fi
g.
4.1
5
a
),
for
whi
ch
Ro
uth
cri
t
eri
on
g
iv
e
K<2
枷。=8
.4;
whe
re
as
we
r
eq
ui
re
K=1
200
t
o
me
et
t
he
s
te
ady
-s
ta
te
e
rr
or
s
pe
ci
fi
ca
ti
on.
Th
is
pr
ese
nts
a
ch
all
engi
ng
c
omp
ens
ati
on
p
rob
lem.
Us
ing
f
req
uen
cy-
res
pons
e d
esi
gn
met
hod
s
(
Bod
e/Ni
cho
ls
pl
ots
),
v
ari
ous
c
omb
ina
tio
ns
o
f l
ag
a
nd
l
ead
c
omp
ens
ato
rs,
pl
us
f
eed
bac
k o
f
二_ ! h"
I
01 + V
E. 扭 I 以r
t6,
、I二几十 — +1
1
1-
t
Basi
c
a)
Servoral
re
currcn r
1‘
t刀c
ompe
nba万Ol
r
/ a 43
91
.+ OE .
80((5
s+1) 0.
1
1厂了7
oc
}
了口4 0
be
(SOS+1)
I于- + - + 11 (3
= 5!
Iv6
14
Gearbox
0.
18(
0.05
8+I) 0.
6夕,。
a 79s
(
0.01‘+1) }
(U
.
N
I
.s+1)
“
Lead
勺pr
oxi
mat
e Tachom et
er
compensanan second
dea-
rranre 召enerar
or
C
'ompe
rrs
ated
句
Fi
gA.
15
Ba
sic
s
yst
em
b
loc
k d
iagr
am
wi
th
s
imp
lif
ied
v
ers
ion
115
de
riv
ati
ves
o
f t
he
c
ont
rol
l
ed
v
ari
abl
e we
re
t
ri
ed,
fi
nal
ly
l
ead
ing
t
o t
he
s
yst
em
o
f 1
 ̄
ag.
4.1
5 b)
.
Th
e
Ny
qui
st
pl
ot
(
not
e t
he
d
ist
ort
ed
s
cal
e)
s
hows
t
he
c
ompe
nsa
ted
s
yst
em
t
o b
e c
ond
iti
ona
lly
s
tabl
e
bu
t wi
th
amp
le
(
10
t
o 1
) ga
in
ma
rgi
n f
or
bo
th
i
ncr
eas
ing
o
r d
ecr
eas
ing
g
ain;
wh
ere
as
t
he
c
los
ed-
loo
p f
or
a
nd
me
et
t
he
de
sig
n g
oal
s
gi
ven
e
arl
i
er.
Comp
ens
ati
on
i
n f
eed
bac
k pa
th
us
es
v
ey
r
s
tr
ong
(
appr
oxi
mat
e f
our
th)
d
eri
vat
iv
e s
ign
als
b
ut
t
he
t
ach
ome
ter
g
ene
rat
or
v
end
or
p
romi
se
d a
v
ey
r c
lea
n (
0.04
% r
ippl
e)
t
acho
met
er
s
ign
al.
Anal
og
c
omp
ute
r s
imul
at
ion
(
dig
ita
l
si
mul
at
ion
wa
s
u
nco
mmon
i
n 1
960)
o
f t
his
l
ine
ar
s
yst
em
mod
el
wa
s pe
rfo
rme
d a
t t
his
po
int
t
o c
hec
k t
ran
sie
nt
p
erf
orma
nce
s
pec
ifi
cat
ion
s,
s
inc
e t
he
p
enc
il-
and
-pa
per
d
esi
gn
wa
s b
ase
d on
ly
o
n s
tea
dy-
sta
te
a
nd
s
inu
soi
dal
be
hav
ior
.
Al
l
sp
eci
fi
cat
ions
we
re
me
t,
s
o t
he
l
ine
ar
de
sign
wa
s a
cce
pte
d.
(
1909
ps
i)
110
岛、
女勺
・之灿
』
长J之景吟
石
0
n如‘
之
0l o
」劲 r
or
s
peed
b户
Fi
g.
4.1
6 S
erv
ova
lve
n
ull
pr
ess
ure
s
ens
it
ivi
ty
a
ndmot
or
f
rict
ion
behav
ior
A
in
f a
l
st
age
o
f
ana
lys
is
i
ncl
ude
d i
n
th
e p
rop
osa
l
wa
sa s
tud
y of
t
wo
n
onl
in
ear
it
ie
s t
hat
we
re
f
el
t
t
o be
p
ote
nti
all
y s
igni
fi
cant
,
se
rvo
val
ve
n
ull
p
res
sur
e s
ens
it
ivi
ty
a
nd
g
ear
box
ba
ckl
ash
. Wi
t
h t
he
mot
or
an
d l
oad
s
tat
io
nar
y,
dy
r f
ri
ct
io
n pr
eve
nts
moti
on
unt
il
t
he
ser
vov
alve
ope
ns
suf
fi
cie
ntl
y t
o
de
vel
op
th
e n
eed
ed "
bea
r ko
ut"
pr
ess
ure
.
In
an
ide
al
val
ve,
a
ny p
res
sur
e ro
f m z
ero
t
o t
he
suppl
y
pes
r su
re can e o
btai
ned, under
st
all
ed-
load cond
itio
ns,
with
an i
nfi
nite
simal va
lve openi
ng.
Becau
se of
i
nte
rnal
l
eakage
in a
r
eal
val
ve,
however,
near
t
he
nul
l posi
ti
o n
t
here
is
proport
ion
ali
ty
bet
ween va
lve
openin
g and the
pre
ssur
e devel
oped ac
ross
a
sta
lled l
oad.
This
is
cal
le d
t
he nul
l
pre
ssur
e se
nsit
ivi
ty
and re
su l
ts
i
n a
stea
dy-s
tat
e posit
ion
err
or
since
a fi
nit
e err
or
signal
to t
he
se
rvoval
ve i
s r
equi
red t
o devel
op e
nough pr
essur
e t
o brea
k t
he f
r i
ct
ion l
oose.
Err
ors bel
ow thi
s
11
6
cri
ti
cal
va l
ue wil
l not
be correct
ed, t
hus the
type-1
st
e ady-s
tat
e e
rror
beh avio
r ass
ume d
in t
he
li
near
mo del
will
not
be ac
hie v
ed in
real
ity.
Experi
ence with
servoval
ves of t
h e
typ
e use
d in
thi
s
syst
em hs
a shown that
i
t t
ake s
5
t o
8% of
the t
otal
va
lve str
oke to a
chieve full
supp
ly pr
essur
e
acros
s the loa
d. Full s
tro
ke i
n th
e p r
e s
ent val
ve re
qu ir
ed 40 mA of c
ur r
ent,
so,
taki
ng a
conser
vative 1
0% v
alue f
o r
the null
range, our
null
p re
ssure sens
it
ivi
ty i
s 250 psi/
mA (s
ee
Fig.
4.16
a).
To
es
ti
ma te
the
stat
ic
pointi
ng err
or
c a
used by t
his
comb i
nati
on o
f l
oad sta
tic
fri
cti
on a
nd val
ve
null
behavior
,
we need an umber f
or syste
m tot
al
sta
tic f
rict
ion.
We earie
l r computed
beari
ng
fri
cti
on
(and ig
n o
re d
motor f
ric
tio
n) and si
nc e
th
e motor
has been sel
ect
ed n
ow, we can
get
ro
f m
the
ma n
uf ac
tur
er and i
ncl
ud e
it
in o
ur pr
es e
n t
er
ror
est
ima t
e.
Fig.4.16
b) s
ho ws me as
ured
fri
cti
on
tor
que f
or our
mo t
o r
,
the
breakout
fri
ct
ion of 160
in.
-l
bf
be i
ng t
he numb e
r pertin
ent
to
our p
resen
t
cal
cul
ati
on.
32,
000
4=1
l
oad
f
rict
ion
(
ref
err
ed
t
o mot
or
s
haf
t)
= 16
f
t一鸣
3
25(
0.
98)
t
o
t
al
f
ri
c
t
io
n
at
mo
t
or
s
ha
t
f=
警+‘,
・7
3”
一l
bf
m
o
to
r、r
eq
ui
r
e
d
t
o
b
r
ea
k
lo
o
se
=瞥7,
一,5
0
ps
i
I
n
Fi
g.
4.
15
b
)
we
n
ow
n
eed
t
o
co
mpu
te
h
ow
l
ar
ge
8E
mu
st
b
e t
o
ca
use
a
v
al
ve
c
urr
ent
i
of
1
mA.
(
9E)
(32
5)(
280
6)=1
mA
OE=1
.1X1
0一r
ad=0.
000
630
Since
the
r
esol
uti
on
of t
he
shaft
enc
oder i
s
0.001"
a
nd th
e al
lowable
point
ing
er
ror
i
s
10.0050
,
thi
s
err
or
wasj
udged
acce
ptabl
e.
Fo
r t
he
gear
-back
lsh
a st
udy, t
he
four-
me s
h gear
bo x
wit
h ba c
klsh
a i
n each
mesh
of g
ears
was
approxi
mated
s
a a s
ingle
me s
h of
gears
(32
5 to 1
) wit
h as
i
n gle
back
lash.
Due
to
th
e la
rge dy
r
fri
cti
on va
lue
and
the s
low moti
ons i
nvolve
d th
e ri
f ct
ion-
control
led
backlsh
a model
was j
udged
a
ppl
i
cab
le
f
or
a
d
esc
ri
bi
ng-
fun
ct
io
n a
nal
ys
is
.
Wi
t
hq a
nd
GZ
d
efi
ned
s
a i
n F
ig.
4.
17a
),
t
he
d
esc
ri
bi
ng
f
unc
ti
on
g
rap
h o
f Fi
g.
4.
17b)
p
red
ict
s
a
imi
l t
cy
cle
o
f a
mpl
it
ude
0.
440
a
nd
3
.2
r
ad/
s
fe
r qu
ency
at
t
he
mot
or
sh
aft.
Al
t
hough
th
is
i
s
only
0.
00050
a
t t
he
ant
enn
a s
haf
t,
imi
l t
c
ycl
i
ng
o
f
any
kin
d wa
s ju
dge
d unac
cept
abl
e a
nd a
"fx"
i ha
d to
be dev
ise
d.
For
tun
atel
y,
t
he u
se
o
f
h
igh-
re
f q
uenc
y d
it
her
t
o q
uen
ch
s
uch
i
l
mit
c
ycl
es
i
s
a
we
ll
-kno
wn
t
ech
niq
ue.
A
n
onin
lear
a
nal
og
s
imulat
ion was used
both
to
check t
he d
es c
ribi
ng-f
uncti
on p
redi
ctions
a n
d t
o desi
gn
a nd
eva
luat
e
t
he di
the
r e fect
.
Limit-
cycle
amplitu
de and re
f quency
clos
ely
ch e
c k
ed the
analy
ti
ca l
pr
edi
ctio
n
a
nd a
60 Hz d
ithe
r at
t
he
mo tor
shaf
t
was found
to
quench
an es
tabl
ished imi
l t
cyc
le
in 2
s.
I
n th
e
a
ctual
sy
ste m,
d i
the
r wa
s int
roduced el
ectr
ical
ly
at t
he s
ervova
lve, us
ing co
nvenien
tly
avai
labl
e
6
0-Hz l
ine powe r
s
a t
he d
ithe
r s
ou r
ce.
I
ha
ve now present
ed
in a
bbrevia
ted
form t
h e
es
senti
al
f
eatur
es
o f
th
e desi
gn p
roposal
su
bmitte
d,
i
n c
omp et
ition
with
other
s,
to
th
e proj
ect
sponsor
.
It
was
sel
ected
s
a the most
pro
mising,
a
cont
ract
117
wa
s a
war
ded
,
an
d d
eta
il
ed
d
esi
gn
a
nd
c
ons
tru
cio
t n
we
nt
f
orwa
rd
ov
er
a
p
eri
od
of
a
bou
t
fo
ur
y
ear
s.
(
.t 叹,
(
},
一 一 ̄-一人 ̄--一-一---- 尸.一 八.一一、产一 一 、
便
0 1i
,
4
,
+ A
目
_}
万让
二例
=S0G(
(
50
5s
a
-
S+1)
1)
竺坐 +11
1挤
=
3
1
11
7e
1
One-ro-one ge
a
i
n sh
ivi
th
011
s;阴05
s +1) r
ll
i of
the
ba办l
n+h
(
2(l
ls+1
)(08
3S十1
): (
b二(
(
1.7
5“)
(
"r愁h6
))
G,
1
b
)
Fi
g.4
.17
Ge
ar
ba
ckl
ash
d
esc
ri
bin
g-f
unc
tio
n s
tud
y
It
is
qu
it
e
commo n
i
n pro
jec
ts
of
t
his magni
tude
t
hat
unan
tic
ipat
ed prob
lems wil
l
ari
se,
r
equir
ing
desi
gn
modi
ic
f a
ti
ons
.
The t
achomet
er ge
nerat
or,
on
which
all
t
he f
eedback
compensat
ion
depen
ds,
tur
ned
out
t
o be
noi
si
e r
t
han e
xpe
cted,
req
uiri
ng
dra
sti
c r
educt
ion in
th
e "amount"
of
deri
vat
ive
c
ont
ro
l
pos
si
ble
.
Thi
s,
t
oge
the
r wit
h a
r
edu
cti
on
in
hydr
aul
ic
na
tura
l
re
f quenc
y rom
f 1
4 t
o 6
r
ad/
s
(
due
t
o a
l
ar
ger
vol
ume
of
oi
l
u n
der
compr
ess
ion
tha
n o
rig
ina
lly
es
ti
ma t
ed)
gr
eat
ly
re
duced
t
he
a
llo
wab
le
l
oop
g
ain
.
To
ma
int
ain
t
he
r
equi
re
d a
ccur
acy
wi
th
r
edu
ced
l
oop
g
ain
i
t
wa
s n
ece
ssa
ry
t
o
go
t
o
ty
pe
2,
ra
the
r t
han
t
he
o
rig
ina
l
ty
pe
1
.
A
t
ype
-2
s
yst
em
t
hat
me
t
al
l
t
he
s
pec
iic
fati
ons
un
der
t
he
ne
w c
ond
iti
ons
was
d
esi
gne
d.
Th
is
t
ype
-2
s
yst
em,
ik
l e
t
he
o
rig
ina
l t
ype
-1
s
yst
em,
wa
s
c
ondi
t
iona
ll
y st
abl
e,
and
i
t was
di
scove
red
that
unde
r s
ome o
per
ati
ng
condi
t
ions,
s
uff
ici
ent
s
at
ura
ti
on
occu
rred
t
o
re
duc
e l
oop
ga
in b
eyon
d th
e g
ain
mar
gin,
ca
usi
ng
abs
olu
te
i
nsta
bil
i
ty.
(This
p
robl
em
woul
d
ha
ve
b
een
p
res
ent
in
t
he
o
rig
ina
l
ty
pe-
1 s
yst
em
a
lso.
)
Th
is
s
ta
bil
it
y pr
obl
em
(
caus
ed
by
t
he
s
atu
rat
io
n n
onl
i
nea
ri
ty
) wa
s ixe
f d b
y u
se
o
f a
not
her
(
int
ent
iona
l)
n
onl
ine
ar
f
eat
ure
,
a
du
al-
mod
e c
ont
rol
l
er
t
hat
swi
tc
hes
t
he
s
yst
em
f
rom
t
ype
2
t
o t
ype
118
七 1,勺勺h自
00峨叭匀
ms
灿的、0卜1.与‘勺,民
‘0乞 之
e
1+ 必100
bl
t
斗 o
』队 七、气
卜 心
你 勺尧勺夸
o
卜
助工
IJD勺娜,仁.
六旧
r
y
p
鑫 勺女tl
勺.
t’,专P
t
i
之注‘1!
毯‘ 叭荟、
l
升
找 工
i
1知久番
l ab
匆 11氏
、-u卜
1勺勺
示 八-匕补
t
+
s
之勺』七宾
‘之 1)七 』
七切 冬之之勺
‘:泥
na
曳1勺勺00
・
芝 、、
‘、
七李〉
为卜
o
+
iti
几| ||洲|卜|
d
{
『||
二丁 」
ole
v o
cn
M
, 常卜
{
o
、『+ 晚恤勺0 + ! 、呜
s
1十 趁 叭
、
宕 本+
砖
.
‘
闪 卜 你
卜 1
0 0 勺()落
o t
穿
1+ 沪叭
厂
119
十
才
峨叭叭
W
处电叭美
rv
含+代t泛 5
e
g
s
+
1
叱
chn
i
七 Ilt)七1
10勺
wit
3二州£言
8 l
Dua- de
mo s
+
g
Fi4.
. 1
1 whenever s
yst
em error
exceeds a cri
ti
ca l
val
ue where sat
urati
on begi
ns t
o reduce l
o op gain
danger
ou s
ly.
Figu r
e.
4. 18 shows thi
s final confi
gurat
ion. Wh en l
ar
ge dis
turbanc
es and/or
comma nds ca
use mo me nt
aril
y l
a r
g e
e r
ror an
d s
ubsequent sat
urati
on,
the
swi t
ching l
ogic that
monito
r s
t
he s
ize
of
the err
or s
witches
the
o p-
amp contr
oll
er to
the
lower,
"
str
aight
throu
gh "
signal
pat
h t
ha t
de
fin
es
a type -
1 c
losed-
loop sys
tem with
a Nyquis
t pl
ot t
hat
does
not
show i
nst
abiit
ly for
ed
r uced gai
n.
This s
table s
yste
m b r
ings t
he er
ror
back t
owar d
zero,
and
when i
t i
s s
ma ll
enough to
no
longer
cause
satur
a ti
on-
induced gai
n reduct
ion,
t
he cont
ro l
l
er swit
ches
t
o t
he upp e
r s
ignal
path,
which
a d
ds an
el
ec t
ronic
int
egrat
o r
plus
some l
e a
d compensati
on, def
ini
ng
a t
ype-2 sys
tem of
high
accur
acy and
bri
n ging the
err
or st
a b
ly t
owa rd ze
ro.
Final
d e
sign of
such
syste
ms wit
h mul t
iple
nonin
l eari
ti
es
(sa
tu r
ation
plus
a s
wi tc
hing cont
roll
er)
is
be
st accompli
shed
by compu t
er
simulati
on ,
once l
i
ne ar
sy
stem an al
yti
cal
studi
e s
have "ro
ug hed
out"
the two in
divi
dual
(type 1
and ty
pe 2)
desi
gns.
Th
e probl
em of co n
dit
ional
stabil
ity wit
h s
atur
ati
on is not imi
l te
d to t
h e pre
sent sp
e c
ific
e
xamp
le
a
nd
t
he
s
wit
chi
ng
c
ont
rol
er
a
ppr
oac
h us
ed
i
n t
he
Hays
tac
k a
nte
nna
i
s a
c
and
ida
te
s
olu
tio
n f
or
s
imi
la
r p
rob
lems
.
Ot
her
po
ssi
ble
a
ppr
oac
hes
ha
ve
b
een
i
nve
sti
gat
ed,
Fi
g.
4.1
9 s
howi
ng
t
he
b
asi
c
sy
ste
m u
sed
a
s a
n e
xamp
le
by
Tr
uxa
l.
(
;
1(s
) G ,
(s
)
州
丰卜
+ E C
万似 0-s+1) 0.
066而 +1
(
s+1) s(
0.
:;s
'
+ 1
)(O
.OOS
s+1
)(0
.00:
3
8
+1
)
B
Li
near
gam “10
弓(
1
A0
)
ha
s
ma
xan
um
p
has
e
l
ag
o
l'
7S0
Fi
g.
4.
19
Co
ndi
ti
ona
ll
y
st
abl
e
typ
e-
l
s
yst
em
wi
th
s
at
ura
ti
on
B
节 口叫
J乙
Fi
g.4.
20
Re
duc
ti
on
i
n p
hase
l
ag
r
emo
ves
c
ond
iti
ona
l
st
abi
i
lty
Th
e yp
t e
-1
s
yst
em
s
hown
t
her
e h
as
t
he
c
ond
iti
ona
ll
y s
tab
le
Ny
qui
st
p
lot
o
f F
ig.
4.
20
a
) wh
en
l
oop
g
ain
i
s
pr
ope
rl
y s
et.
I
f
we
c
oul
d c
aus
e t
hi
s
cu
rve
t
o c
han
ge
s
a i
n Fi
g.
4.
20
b
) wh
en
s
atu
rat
io
n
120
?和
o
ccurs
, t
he
lower
ef
fect
ive
gai
n would
no
t c
ause
i
nst
abil
it
y.
The
red
uct
ion
in
phse
a l
ag
neede
d t
o
s
hif
t
the
cur
ve a
s s
hown is
ob
tai
n e
d in
a
s
witch
ing
con
tro
lle
r b
y r
emova
l of
a
n i
nte
grat
or,
givi
ng
a
9
00
reduct
ion
at
evey
r re
f quenc
y.
I now
wish
to
sh
ow Truxal
's
al
te
na
r tiv
e met
hod f
or
ach
ievi
ng
t
he
desi
red
la
g r
educt
ion.
To
i
mpl
eme
nt
t
his
me
tho
d we
mu
st
be
a
ble
t
o
re
con
fi
gur
e t
he
s
yst
em
wi
th
a
mi
nor
l
oop
a
s i
n
Fi
g.4.21
. I
n des
ign
ing t
his
minor
lo
op we make u
se of
th
e we
ll
-known
beha
vior
of
any
cl
osed-l
oop
sy
ste
m wit
h d
esi
red
va
lue
V c
ont
rol
le
d v
ari
abl
e C,
forwar
d pa
th
G,
a
nd f
eed
bac
k p
ath
H.
V l
ee 火
式.
,
E
I _ I
MI
巾
.、 -
-+
i
-
Ki卜
六
}
二甲.. l ee ...ee ee .J
H,
F
4
9
.
21Mi
nor
-l
oop
f
eedb
ack
t
o r
educ
e pha
se-
lag
d
uin
r g
s
atu
rat
ion
I
f
amp
it
l u
de
r
at
io
of
GH
(
i叻 i
s
sma
ll
r
ela
ti
ve
t
o 1
.0
C -V
G G
二G
1+G厅
1+0
I
f
ampl
it
ude
r
at
io
o
f GH(
io
c)
) sl
i a
rge
r
ela
ti
ve
t
o 1
.0
C -V
G
G 1
二二 一 孚书 一
1+GH 1+0
H
I
n F
ig.
4.
21,
t
he
mi
nor
l
oop
c
ont
ai
ns
t
he
s
atu
rat
ion
, wh
ichwe
t
rea
t a
ppr
oxi
mat
ely
s
a a
v
ari
abl
e
0
二
}
一
g
ai
n
0<Kd
f引.
0,u
si
ng
t
he
d
esc
i
rbi
ng
f
unc
ti
on.
F
or
n
on-
sa
tu
rat
i
ng
o
per
at
io
n
Kd
f
叨
V‘
n
wh
ere
s
a i
nc
re
as
in
g s
at
ur
at
io
n c
aus
es
Kd
f
-+
0
.
F
or
n
ons
at
ur
at
in
g o
per
at
io
n we
wa
nt
舀
b
eha
vio
r t
o b
e i
den
tic
al
t
o t
hat
of
Fi
g.4.
19
t
hat
is
M, ‘ K(
0.02i
o)
+
1
)
论
几、
|
=乡(
Iw)“
七
(
io)
+1)
s
inc
e t
his
s
ys
-
t
em
wa
s s
ati
sfa
ctor
y when
uns
atur
a
一
ted.
When
s
atur
ati
on
ca
uses
K-
-
->
0.we
ee
t 叨 丫
M I _
=户(
l))
o 。K
,
K
2
K
df二
K,
K2
Kdf (
4.43)
七
1
+K2
K,
.
H,
whi
ch
we
s
ee
ha
s z
ero
p
has
e l
ag,
75
0 l
ess
t
han
t
he
u
nsa
tur
ate
d s
yst
em;
thu
s t
he
de
sir
ed
c
urv
e s
hit
f
f
ro
m F
ig.
4.
20
a
)
to
4
.2
0 b
) ma
y b
e o
bta
in
ed.
S
in
ce
Kd
.=0.
0 i
s
re
al
iz
ed
o
nly
f
or
i
nf
in
it
el
y
l
arg
e s
ignal
s,
we
wi
ll
no
t a
ctual
ly
ach
ieve
a
750
i
mp r
ovement;
howe
ver
sufic
ient
imp r
ove
ment
for
s
tabi
li
zat
i
on ma y
b
e poss
ibl
e.
Since
the
des
cri
bing f
uncti
on i
s a
n appr
oximai
ton,
ina
f l
des
ign
s
hould
be
check
ed
and
re
in
f e
d wit
h s
imul
ati
on
studi
es,
however
a
pr
eli
minay
r de
s i
gn
can
pr
oceed
1
21
as
fol
lows.
To
p
res
erv
e t
he
o
rig
ina
l
be
hav
ior
f
or
u
nsa
tur
ate
d o
per
ati
on
we
r
equ
ire
凡KZ
Kd
f K(
0.02i
cp+1
)
when Kd
f二1 (
4.44)
1
+
Kz
K(
f
H,
(i
a)
) i
o)+ 1
So
lvi
ng
f
or
H,
(
i(
o)
g
ive
s
H,
(i
or
))二一 }
,
K
,
KK
^
2一
。
.
02
).
(
0
+}
K KZ
一
(
0.02i
(o+1
)
,公一- (
4.45)
a
nd
s
inc
e K,
Ka mu
st
e
qua
l K
to
ma
tch
t
he
o
ri
gi
nal
ga
in
0.
98
(
i
(v)
K,
H,
(i
o))= (
4.46)
0.
021
(.
o
+
1
Thi
s gi
ves
Mz KK,(
0.02i
o.
)+1)
(
it
w)=
川
、 产
(
4.47)
K
(
0.
98Kd
f+0.
02)
iw+1
For
uns
atur
ated
ope
r
‘
at
ion
(K-=1
) t
his
be
comes
、 1
9
MI
K(0.
02i
o)+1
)
一犷, tc
})
(i ma
x
l
ag=7
50 (
4.48)
石 i
(t
)+1
whereas
maxi
mum sat
urat
ion (
Kdf
-a
0
)gives
M.
,__
一户 (
i(O)=KK 、 "J
max
lag=00 (
4.49)
石
Us
ing
t
his
p
rei
l
mina
ry des
ign a
s a
s
tar
ti
ng
poi
nt,
s
imu
lat
i
on
s
tudi
es
wou
ld
expl
ore
t
he
a
ppo
rti
onme
nt
o
f g
ain
K n to
i Kt
a
nd
K2
a
nd
s
yst
em
b
eha
vio
r f
or
v
ari
ous
t
ype
s a
nd
s
ize
s o
f
c
omma
nd
i
npu
ts,
t
o s
ee
whe
the
r a
cce
pta
ble
pe
rfo
rma
nce
i
s a
tta
ine
d.
J
ust
a
s f
or
t
he
s
wit
chi
ng
c
ont
rol
l
er
a
ppr
oac
h,
t
hi
s
sc
heme
i
s
po
ten
ti
al
ly
us
efu
l f
or
c
ond
it
ion
al
s
ta
bil
i
ty/
sat
ura
ti
on
p
robl
ems
i
n
ge
ner
al,
no
t
jus
t
th
is
s
pec
ifi
c
exa
mpl
e.
122
CHAPTER
5 ROBUST
CONTROL
SYSTEM S
In
previ
ous c
ha pt
ers we desc
r i
bed
the benefi
ts
of us
ing ac ompensator t
o ac
hieve the
desi
red
per
formance o
f a feedbac
k s
ystem.
Now we tur
n our
aten t
ion to
the issue
of
robu st
ness.
The
desi
gn of
high
ly accura
te contr
ol s
ystems in
the pr
esence
o f
signi
ficant u
ncert
ain
ty requi
res
the
desi
gner
to se
ek ar
obust
syste
m. I
n thi
s chapter
we util
ize f
ive metho ds
for
a r
o bust
desi
gn ,
inc
ludi
ng ro
ot
locus,
fre
quen c
y res
ponse,
and ITAE methods f
or a
robust
PID sys
tem.
Ma ny
re
ce n
t
cont
rol desi
gn me t
hodologies focus on robust
ness-
stabil
it
y robustness and p er
formance
robus
tnes
s-in
the pres
ence of
un c
ert
ainty
. The rec
ent
design methodologies
addr
ess the
fact
t
hat
our p
hysica
l s
ys t
ems a
n d t
he exte
rnal e
nv i
ronme nt
in
wh ich t
h e
y ope ra
te ca
nnot be modeled
pre
cisel
y,
may change i
n an u
npredic
table
ma nner,
a
nd ma y
be subj
ect t
o sign
iic
f ant
dist
u r
bances.
Howe ve
r,
we should
also
reaiz
l e t
hat
clas
sical
desi
gn te
chniques may also
produce
robust
control
syst
ems.
Contr
ol engine
ers who
a r
e a
wa re
of
these
iss
ues
c a
n design
robus t
PID cont
rolle
rs,
ro
b us
t
lea
d-lag
contr
olle
r s
,
and so f
orth.玩t his
chap t
er
we als
o d i
scuss
the use of
in
ter
na l
mo del
and
pseud
o-quanti
tati
ve fee
d b
a ck
me t
hods to ac
hi e
ve ro
bust contr
ol.
Th e
c hapt
er
concludes wi
th a
PID c
ontrol
le
r des
ign fo
r t
he Sequent
ial
De si
gn Example:
Disk Drive
Re ad
Syste
m.
5.
1 I
NTRODUCTI
ON
A co
ntr
ol
s
yst
em desi
gne
d us
ing
the
methods
and
concept
s o
f t
he p
reced
ing
cha
pter
s a
ssu
mes
k
nowle
dge
of
t
he
mo d
el
of
t
he p
lant
an
d c
on t
rol
ler
an
d cons
tant
pa
ramete
rs.
Th
e pl
ant
model
wi
ll
a
lwa
ys
be
an
i
naccu
rat
e r
epr
esen
tat
ion
of
t
he act
ual
phys
ical
syst
em bec
ause
of:
. Par
amet
er
changes
. Unmode
led
dynamics
. Un
mode
led
t
ime
d
ela
ys
. Cha
nges
i
n
eq
uii
l
bri
um
p
oin
t
(op
era
ti
ng
p
oint
)
. Se
nsor
noi
se
. Un
pre
dic
ted
di
st
urb
anc
e i
nput
s
D(
s)
Di8
turbc
arce
Fi
g.5.
1
Cl
ose
d-l
oop
s
yst
em
s
tru
ctu
ral
d
iag
ram
The
go
al
of
ro
bust
sy
ste
ms
des
ign
is
to
r
eta
in
ss
a u
ran
ce
o
f s
yst
em pe
rfor
mance
i
n s
pit
e o
f mo
del
i
nac
cura
cie
s and
chang
es.
A sy
ste
m i
s r
obu
st
when
t
he sy
ste
m ha
s ac
cept
abl
e cha
nges
in
p
erf
orma
nce
d
ue
t
o mo
del
c
hang
es
o
r i
nac
cur
aci
es.
123
A robus
t c
ontr
ol
sys
tem e
xhi
bit
s t
he
de
sir
ed
p
erf
orma
nce
de
spi
t
e t
he
p
res
enc
e o
f s
ign
ifi
cant
p
lan
t (
proc
ess)
u
nceta
r i
nty
.
The
sy
stem s
tr
uctur
e t
hat
incor
porates
pot
enti
al
syst
em unce
ta
r int
ies
i
s shown i
n Fi
g.5.1
.
This
model
i
nclu
des
the
se
nsor n
oise
N(s)
,
the
unpredi
cte
d di
stu
rbanc
e i
nput
D(s),
a nd
a
pl
a n
t G(s
)
with
pot
ent
ia
ll
y unmodele
d dynamics
or
pa r
amete
r chan
ge s
.
The un
mod el
ed
dy n
a mic
s a
n d
para
me t
er
cha
nges
may b
e si
gni
fic
an t
or
v
ery
lar
g e,
a
nd for
t
hese
sys
tems,
t
he c
hall
enge i
s
to
crea
te a
des
ign
t
hat
r
etai
ns
t
he
des
ired
per
for
mance.
5.
2 ROBUST
CONTROL
SYSTEMS
AND
SYSTEM
SENSITIVI
TY
Desi
gnin
g hig
hly a
ccur
ate
sy
stems
in
the
presen
ce o
f s
igni
ic
f ant
plant
uncer
tain
ty i
s a
cl
ss
a i
cal
f
eedback
desi
gn pro
blem.
The
theor
eti
cal
ba
se s
fo
r t
he s
olut
ion
of t
his
proble
m date
bac
k to
the
works
of
H.S.
Black
and
H.W Bo de
in
the
earl
y 1
930s,
when t
his
pro
blem wa s
ref
e r
red
to
as
the
se
nsit
ivi
ti
es d
esign
probl
em. A si
gnif
ica
nt amount
of l
it
era
tur
e has been pubis
l hed s
inc
e the
n
r
egardi
ng t
he de
sign
of
sys
tems s
ubje
ct
to l
arge
pla
nt
uncer
tai
nty.
The
de s
igner
seeks
to
obta
in a
s
yst
em t
hat
per
forms
adequat
el
y over
a
l
arge
ra
nge
of
un
cert
ain
pa
ramet
ers
.
A syst
em
i
s s
aid
t
o be
r
obus
t
when i
t
is
dur
able
,
h a
rdy,
a nd
r
esi
i
lent.
A
con
tro
l sys
tem i
s r
obust
wh en
(1
) i
t
hs
a low
se
nsit
ivi
t
ies,
(
2)
it
i
s s
tabl
e o
ver
t
he
rang
e of
pa
rame
ter
v
ari
at
ion
s,
a
nd
(
3)
t
he
pe
rf
orma
nce
c
ont
in
ues
t
o
me
et
t
he
s
pec
if
ic
ati
ons
i
n
th
e p
res
enc
e
of
a
set
of
c
hanges
in
the s
yst
em par
ameter
s.
Robus
tne
ss
is
th
e s
ensi
ti
vi
ty
to
efect
s t
hat
a
re
not
co
nsi
dered
i
n t
he a
nalysis
and
des
ign
phr
se
a -f
or
exa
mp l
e,
di
st
urba
nces,
meas
ueme
r nt
nois
e,
and
unmode
led
dynamics
. The sys
tem s
houl
d be ab
le t
o wit
hst
and
the
se ne
glect
ed
efect
s when
pe
rfo
rming
t
he t
ask
s f
or whic
h i
t
was
desi
gned.
匀s
t
em
s
ensi
ti
v勿i
s def
ined
a
s
气
T 一a
T 一a
d 一』
Sa
T二 (
5.1
)
d
上aa/a
wh
ere
a i
s
th
e p
ara
met
er
,
an
d Ti
s
th
e t
ra
nsf
er
f
unc
ti
on
of
t
he
s
yst
em.
Root
se
nsi
ti
v勿
is
def
ined
a
s
Sa
'= (
5.2)
When
t
he
ze
ros
of
T(s)
a
re
i
ndependent
of
t
he
par
amet
er
a,we
s
howe
d t
hat
S
a
T
一艺
s
a
r
,
・(s+r)
1
=1
(
5.3)
f
or
a
n n
th-
ord
er
s
yst
em.
For
e
xampl
e,
i
f
we
h
ave
a
c
los
ed-
loo
p s
yst
em,
s
a s
hown
i
n Fi
g.
5.
2,
wh
er
e
th
e
va
ri
ab
le
p
ar
ame
te
r
is
a,
the
n
T(
s)=1
/[
s+(
a+1
)]
,an
d
一a
Sa
T二s+a + 1 (
5.4)
Fu
the
r rmo
re
t
he
r
oot
is
t'
,=+(
a+1
),a
nd
124
一}
'
a
r
t二一a (
5.5)
Fi
g.
5
.2
A
ir
fst-
orde
r s
yst
em
Fi
g.
5.
3
Se
con
d-o
rde
r s
yst
em
Therefor
e
Sa
T二一S
a"
,S十a +1 (
5.6)
us e
Let
xami
ne
t
he
s
ens
it
ivi
ty
of
t
he
s
econd-
orde
r s
yst
em
shown
i
n Fi
g.5.
3.
The
t
rans
fer
f
unc
ti
on
o
f t
he
c
los
ed-l
oop
s
yst
em
i
s
K
T(
s)= (
5.7)
52 s+K’
s+
The
s
yst
em
s
ens
it
ivi
ty
f
or
K
i
s
C
.
'
T
S(
s+1)
 ̄ K (
5.8)
1 S) S2+S+K
+G万(
40
1
;= 1/1
20
20
1ogI
SI
dB
0
/\了
/ 又
20
1ogj
刀
/ 弋
2
才卜
工
/ \
子
了卜
工
.口1
)
( 0.1 1
1
.5
1 10 100
功
F
ig
.
5.
4Se
ns
it
i
vi
ty
a
nd
2
01
og
IT
(i
o)
)l
fo
r
t
he
s
ec
on
d-
or
de
r
s
ys
te
m
in
Fi
g.
5.
3
Th
e a
sympt
ot
ic
a
ppr
oxi
mat
io
ns
a
re
s
hown
f
or
K=1/4
125
A
B
ode
p
l
ot
o
f
t
he
a
s
ymp
t
ot
es
of
2
01o
gj
T(
jo
e
))
i
an
d 2
01o
gj
S(
ja
))
f
is
s
ho
wn
i
n
Fi
g.
5
.4
f
o
r
K=1
/4
(
cr
it
ic
al
d
amp
ing
).
No
te
t
hat
t
he
s
ens
it
iv
it
y i
s
sma
ll
f
or
l
owe
r f
r
equ
enc
ie
s,
wh
il
e
th
e
t
ra
nsf
er
f
unct
i
on
pr
imar
il
y
pas
ses
l
ow
fre
que
nci
es.
Also
not
e
for
t
his
c
a s
e,
T(
s)
=1一S(s).
Of
cou
rse
t
he
se
nsi
ti
vi
ty
S
onl
y
rep
res
ent
s
ro
bust
nes
s f
or
smal
l
changes
i
n g
ain.
If
K
cha
nge
s
ro
f m
1
/4
wi
th
in
t
he
r
ang
e K=1
/16
t
o
K=1
,th
e r
es
ul
ti
ng
r
ang
e o
f s
t
ep
r
esp
ons
e
is
s
hown
i
n
Fi
g.
5.
5.
Th
is
s
yst
em,
wi
th
a
n e
xpe
cte
d wi
de
r
ange
o
f
K,
may
n
ot
b
e c
ons
ide
red
a
deq
uat
el
y
ro
bus
t.
A
r
obu
st
s
yst
em
wou
ld
be
e
xpe
cte
d t
o户e
ld
e
sse
nti
al
ly
t
he
s
ame
(
wit
hin
a
n a
gre
ed-
upo
n v
ari
ati
on)
r
espons
e t
o a
s
ele
cted
i
nput
.
1.
4
l
厂 =了
/
产、、 .. 目..
甲.. 一 一 ..
1.
0
一
一 一 ,
且 尸竺二二二二蕊刁曰
.
#
*I+
-
了
)
,
(
1l
1 了2
Fi
g.
5
.5
S
tep
r
esp
ons
e f
or
s
ele
cte
d g
ain
K
EXAM PLE
5.1
SENSITIVITY
OF
A CONTRO1
,ZRD
SYSTEM
C
ons
i
der
th
e
sy
st
em
s
how
n i
n
Fi
g-
5-
6,
wh
er
e
G(
s)=1
/S
2
,an
d
a
P
D c
ont
r
oll
e
r
G,
(s
)=b
,
+b2
(s
).Th
en
t
he
s
ens
it
iv
it
y
wi
t
h r
es
pec
t
to
c
han
ges
i
n
G(
s)
i
s
s2
,
.+
T= _ (
5.9)
v G ,
1+ GG,
(s
)
s
2
+b
2s+b
l’
1
R(
s) b
l+
h
=
s s- I
树
Fi
g.
5
.6
A
s
yst
em
wi
th
a
PD
c
ont
rol
le
r
126
b
?5+b
l
and T(5) (
5.1
0)
一万砚蔺而’
Cons
ide
rthe nor
nla
l cond
ition 否=la
nd。。
=橱 .T
he
nbZ
=2。。
toachi
eve杏=1.
。e
re
ew
r
o
f 。m
ayp
,o
tZOl
og
1S
IandZ
Olo
g}
T}
onaB
ded
e ‘a
,,ss
a h
own‘
nF
ig
・5
・7
・No
e
t
ha
ttt
hef
req
uen
cy叭 i
sa
nin
ic
d a
toro
nthebo
und呷 be
twe
ent
hef
equ
r enc
yreg
ioni
nwhi
cht
he
s
ens
it
ivi
tyi
sth
eimP0
rta
ntd
esi
gnc
it
r e
io
r na
ndt
her
egi
oni
nwh
icht
hes
tab
ili
y m哗i
t nis
1
哪or
ta
nt,
Thu
sifwes
Pec
iy叭 P
f r0Pe
rl
yt
ota
kei
nt
oco
nsi
der
io
t
a nt
hee
xon
t t
ofmo
dein
lgero
r
ndt
a hef
equ
r enc
yofe
xema
t ld
isrb
u
t nc
a e
,wec
ne
a Pe
x c
thes
t yst
emt
ohvea
a nac
ceP
tbl
a ea
mou
nt
o
fro
bus
ue
t s
s.No
et
t h
t G。
a (5)i
sap
rop
or
io
tnal
一de
iv
r iv
t
a e(
pD)
con
rol
t l
e.
r
2O
1}{}
!1
1 丁
201
}!
..月...曰.
口.
兮1
刀
、!1}皿}匕一
1} {}
}. . .. 目
4
夕 《
... ......
B
I
t
硼1姗( 洲从 }
2
月仕
1
{
一
1一
1
N
划1
}
夕 } 1
平}
I
!
20
1‘
烤}5
们“
子
1
,
下
月
姗}一
{
一一洲
}
} {
{
一
」 }
一
} ,{
{
刀
办
]
L
-
.01
)
(
01 1 10 100
口初 ,
,
Fi
g.5
.7S
ens
it
ivi
tya
ndT(
5)f
rt
o hes
ecnd
o 一
。记e
rsy
semi
t nFig
s.6
E叉滩加仔凭刀反ZSYS不瓦U ”份了万 A 兄r
G HT
-月连ND一
尸尤孟四万ZERO
Co
nsi
dert
hes
yst
ems
howni
nFigs.
8,wh
eet
r hePl
nth
a s az
a o i
r
e nther
igh
t一ha
ndPl
ne.
a The
cl
osed一
looPt
ra
nsr f
e
f unct
ioni
s
K(
5一1
)
T(
5) (
5.11
)
5
2+(
2+尤)
5+(
1一尤).
Th
esys
emi
t sst
bl
a ef
r ag
o in一2<K<1.
a Thes
te
ay一
d s
tee
t
a o i
r s
1一ZK
e打 = (
5.1
2)
1一K
十 v 一1
.
( )
R仁少 K 1丫s
)
t
介+1
)了
Fi
g.5.
SAs
eco
nd一
or
ders
ysem
t
127
一,
V一尸分一间口
// ‘ 一 — 16 -一  ̄・
{
/- }
2
t
i
'
/t) 方子
_/
7-
/
j
。
饮 才
0 浮 /0
Is
Zn
't
" ,
却
r(set)
Fi
g.
5.
9
St
ep
r
espo
nse
o
f t
he
s
yst
em
i
n Fi
g5.
8 wi
th
K-1
/2
a
nd
e
.,=0,
whe
n K=
=1
/2.
The
r
es
pon
se
t
o
a
n
ega
ti
ve
u
ni
t
s
te
p
in
put
R(
s)=一1
/s
i
s
sh
own
i
n
Fi
g.
5.
9.
No
te
t
he
i
ni
ti
al
u
nde
rsho
ot
a
t t二1
se
cond
.
Th
is
s
yst
em
i
s
se
nsi
ti
ve
t
o
ch
ang
es
i
n
K,
as
r
eco
rde
d i
n Ta
ble
5.
1.
Th
e pe
rfo
rma
nce
o
f t
his
s
yst
em
mig
ht
b
e c
ons
ide
red
b
are
ly
a
cce
pta
ble
f
or
a
c
hang
e o
f
ga
in
of
o
nly士1
0%.
Thu
s t
hi
s
sy
ste
m wo
uld
n
ot
b
e c
ons
ide
red
r
obus
t.
Th
e s
te
ady-
st
at
e
e
ror
r o
f t
his
s
yst
em
c
han
ges
g
rea
tl
y
as
K
c
han
ges
.
Tabl
e 5.
1 Resul
ts
f
or
Exampl
e 5.
2
0.50。10%27
0.550.2211%30
K
0.
25 0.45 0.75
l
ess
l 0.67 0.18 1.
0
Under
shoot 5% 9% 15%
15
24
45
Set
tl
ing
t
ime(seconds
)
5.
3 ANALYSIS
OF
ROBUSTNESS
Co
nsi
der
t
he
c
los
ed-
loo
p s
yst
em
shown
i
n Fi
g.5.
1.
Sy
ste
m g
oal
s i
ncl
ude
mai
nta
ini
ng
s
mal
l
t
ra
cki
ng
e
rro
r[e
(t)=r(
t)一y(
t)]f
o ni
r
a np
ut
r
(t)
a
nd
k
eepi
ng
t
he
o
utpu
t
y(
小 ma
ll
f
or
a
d
is
t
ur
ba
nc
e
d
(
t
).T
he
s
en
so
r
n
oi
se
n
(t
)
mu
st
be
sm
al
l
w
it
h
r
es
pe
ct
to
r
(t
)
s
o
t
ha
t
i
rl
>>l
n
l・
The
s
ens
iti
vit
y血nc
tion
i
s
S
(s)
=[1
+G.
(s
)G(
s)
]一
,
a
nd
t
he
c
los
ed-
loo
p t
rans
fer
func
ti
on
i
s
G.
(s
)G(s
)
T
(
s)=
1
+G,
(s
)G(s)
wh
en
G,
(s
)=1
.We
t
hen
ha
ve
S(
s)+T(
s)=1. (
5.1
3)
Th
us
i
t i
s i
mpor
tan
t t
o ma
ke
S(
s)
s
mal
l.
For
phy
sic
alyr
l e
ali
zab
le
sys
tems
, t
he
l
oop
128
g
ain
L(
s)=G, G(s)
mu
st
b
e s
mal
l
fo
r h
igh
f
req
uen
cie
s.
Th
is
me
ans
t
hat
S(
jo)
) a
ppr
oac
hes
1
a
t
hi
gh
fr
eque
ncie
s.
An
ad
dit
iv
e pe
rtur
bat
i
on
c
har
act
eri
zes
t
he
s
et
o
f p
oss
ibl
e pl
ant
s a
s f
oll
ows
(
her
e we
a
ssu
me
GG
(s
)=1
):
Ga
(s
)=G(
s)+A(
s),
whe
re
G(s
)
is
t
he
n
omi
nal
p
lant
,
an
d A(
s)
i
s
th
e p
ert
ur
bat
io
n t
hat
i
s
bo
und
ed
i
n ma
gni
t
ude
.
It
i
s
a
ssu
med
t
hat
Ga
(s)
an
d
G(
s)
h
ave
t
he
s
ame
n
umb
er
o
f p
ole
s i
n
th
e r
ig
ht-
han
d
s
-
pl
ane
(
i
f a
ny)
.
The
n t
he
s
yst
em
s
tab
ili
t
y wi
ll
no
t
ch
ang
e i
f
I
A(
jo)
)
I
<I
1
+G(
to)
I
f
or
a
ll
o)
. (
5.
14
)
Th
is
ss
a u
res
st
abi
i
lty
b
ut
not
dyna
mic
per
for
mance
.
A
mul
ti
pli
cat
ive
pert
urb
ation
gi
ves
t
he
pl
ant
Gm
(s
)=G(
s)
[
1+M(
s)
].
Th
e p
ert
ur
bat
i
on
i
s
bo
und
ed
i
n
ma
gni
t
ude
,
an
d a
gai
n
it
is
ss
aume
d t
hat
Gm
(s
)
an
d
G(
s)
h
ave
t
he
s
ame
n
umbe
r o
f p
ole
s i
n t
he
r
igh
t-ha
nd
s
-
pla
ne.
The
n t
he
s
yst
em
s
tab
ili
ty
wi
ll
no
t
ch
ang
e i
f
}
_ 1 }
I
M U0
.
0 <1
1
+户
二刃二一二
If
or
a
ll
.
C
A
.
(
5.1
5)
} G(
jw)
Egn.
(5.
15)
i
s ca
ll
ed t
he
ob
r ust
s
ab
t i
liy
t c
rit
eri
on. Th
is
is a
t
est
fo
r r
obust
ness
wit
h r
esp
ect
to
a
mult
ipl
i
cat
ive
per
tur
bat
ion.
Th i
s
for
m o
f p
ert
urbat
ion
is
ote
f n
use
d bec
ause
i
t s
ati
sfi
es
t
he
i
ntui
ti
ve
pr
opeti
res
of
(1)
bei
ng s
ma l
l
at
low re
f quen
cies,
whee
r th
e n
ominal
pla
nt
mo del
i
s u
sua
ll
y wel
l
k
nown,
an
d (
2)
b
eing
l
ar
ge
a
t
hi
gh
re
fque
nci
es,
wh
ere
t
he
no
mina
l
mod
el
i
s
al
ways
i
nex
act
.
EXAMPLE5.
3 SYSTEM
WITH
MULTI
PLICATIVE
PERTURBATION
Cons
ider
t
he
s
yst
em
of
一
Fi
R.5.
1 wi
th
G-
 ̄
(
s)=1
.
G_=K.
a
nd
Y , 护 r ‘
1
70,
000(
s+0.
1)
G(
s)
s
(s+3
)(S
2 +1
0S+1
0,0
00)’
Th
e s
yst
em
i
s u
nst
abl
e wi
th
K=1
,bu
t a
r
educ
ti
on
i
n g
ain
t
o
K=0.
5
wi
ll
s
tab
iiz
le i
t.
No
w
c
ons
ide
r t
he
e
fec
t o
f a
n u
nmo
del
ed
po
le
a
t
50
r
ad/
s.
I
n t
his
c
ase
,
th
e mul
t
ipl
ica
ti
ve
p
etu
r r
bat
ion
i
s
de
ter
mine
d rom
f
50
I
1+M(
s)1
=
s+50
o
rM(
s)=-
s
/(
s+5
0).
、 ma
gni
t
ude
bo
und
i
s
t
he
n
I
M(jC
o)}a
nd
I
1+1
/
KG(
jco
)
I
a
re
p
lot
t
ed
i
n
Fi
g
5.
10a
),
wh
er
,
M
‘
’・
,
e
i
=
1
j
r
-
t
i
o
j
+
(
0
5
s
s
0
e
en
t
hat
th
e c
ri
te
ri
on
o
f
Eq
n.
(5
.15
)
is
not
sat
is
ie
f d.
Thus
t
he
syst
em ma
y no
t
be
s
ta
ble.
I
f
we
us
e a
l
ag
co
mpensa
tor
G,
(s),wher
e
129
0.
15(
s+25)
G}
(s
)=
s+2.
5
lo-
, 护.一 〔生
弋
卜叫门,.分, ,,月卜‘..叫 〔
, 分 一
三兰卜一于州十宁 卜,护一 ̄ ̄闷,-护-卜,资今卜日卜. ̄.闷 一
1 ,州, 一 一,二『 一一一一一闷 「 ̄ 分 份, 代.  ̄ — , 1 」尸 , ,r甲 ,
旧 〔 声仁别并;
悦
二乙, 月 圈
胜签签;
‘ _ 生一 ,孟一 1  ̄仁」 」 U. _ _ 」盆_ ,t L i ‘】 , 几 T r j力.J 诀 rrn
r 丁飞 代 甲 .二 【
下 二片 、
. ,二
口陌 二 丁 叮
)r 门 匕 老 白1. L 寺班百〔八 L .刁
巨 巨荆摧 〔一 月曰
. r二.1
绷
L Lf1 __ ..- Z _ 」 一一二
眨1 I
卜
月卜’ I U(
. 一
!
日M -
二一 rI团 〔
a
__ r
」二门r
,巨 卫
I
- _ 口-
:
卫工笙 尸 厂 1 ̄ 〔
{. 口 n l〔L I口
-
,一-
一-
I刀r口 厂’
汗叮I t
_「
_ ,
F
;t}
(ar 0
}) { 口」
_上比
l住口
鑫} 羁 鉴 熟奎教:
三E
生 爵 万井萎4
l
4津鑫
、
嗯遏 之警翔
匕三_
」「二〔〕二lf
ll]
旧 【二 口 11沙书李1
」 〔二〔工二
下压万2 洲
口 } I 【日一
「1卜 j
【
丁卫「
二Il
丁 生
区正日 泣 日 !卜
1今1
{
‘ 经 〔工山 }
{ 目砰 {汁一日
「
-曰洲 {
一 [口.
! 坦业1 几
.‘乙口既l 1}
口 江}s盯口
刀
刀
怕Jl
巨 鱼险 更
酬 .口.曰月 ,r ̄ ̄ 吧〔三己r 件二气芳 1 11 r L.JJ
1
.一 -
七所 .
习 卜,
r飞 二I0
.r月尸户怪 习
口L一 肥里,
,,,,万性咬r 娜. ̄ 目,尸 .二
亡二r,r,,州口
Tl
I创 (TI写ilI
日
}・
I.
I
_. . I
【.- 乞
石 Ll习
I[ 幻〔 万_‘曰 耳)jl
. 口 洲r二 〔仁 以别口
l l LP
门二吓吓〔 「日 11
1!川 曰
巨 ,
L 曰 口二
1产 } 朋 口
r_ 11 11l
i l目f
J
1口 口’
!
- }
一
幽}’疚}
]!2 F团r.
汇一
{ 耳}
}! }口
山}
10 1
0'1
" 1
V`
1
/11 II
I: it
)
w
(Tf
lt
t y
ed
)
,
口
万盆
巨王绝
脸 p1
4{书容:
粼鉴班P
-团件排 二二片冬引 共 鑫二二二区,
目困 丰葬二了目I
-捧 }. 二 , 下 爪
「买于尸叶什非一 ̄ 尸任叮I ,
I 巨 月渐 淤
一 赚子浏巨一参于
肚
仁_
I
土仁 滩1+},
C
.
'}
(
r.仁巨
}
} 息汇 遨i
口
1
巨 :气 于 .爪
〕
勺胜仁目翻井留井 共 门翔咚目井
巨至 至理睡呈 至照 巨至 生肛 能井 !
皿居溉 书卜军一 t翻
之,
巨三
日日{
书妇} 日日呸 科l
份一 j 卜一曰卜
于附份.甲日任既 日卜  ̄日日十出
,
「二二曰 「{王1
1}日 曰 「团 I
Llt
( 口 曰 母五
11龟
/ 口 「匹 I{
礴 口 田
!箩I
忘、
厂一
口日皿L ̄ 口口
江皿 绝熟典 毅 丛 羹艇 夔I
。
介
、
卫
护
考 性
厅扫陈}
口里里层 一
每1 甲三三 荟圈阵 呸 禅 止谧
F二多二三三乏 月圈阵翅
尽 !
称台合
理 目翻 留 自兰鉴 里
兰鉴 性拼里件 口
卜,, 任 村月抢 一 一 干 闷召卜分份 闷 门 亏于妇日卜- ̄ 州 日区 卫叶开.-刊巨 ,卜日曰
任子日日F
?计十一日日序洲 卜一勺不卜
升传十‘日吕洲 托 服
十州日田 日
【
二 二口 口 {;
1l
(D 厂 口庄 l
州 匕产 厂 j忿n{
!
!活 [
]「口T}
1
11 r]1I !}
1 〕 F笼({
1
! 才匕 口[1
1一 }}
{} 口日
比洲} 厂}
}f}
}
}{i
-,
[
叮{}
1
r
一职
曰 !
}N1
I
rt一
: 1
o
-1
l
o",
1
/j
'
I
ll
= 尸
1
0
'
!
r
) (
r(al
'
Se
'[
少
1
1)
Fi
g.5.
10
The
r
obu
st
s
tab
ili
ty
c
rit
eri
on
f
or
Ex
amp
le
5.
3
t
he
l
oop
t
ra
nsf
er
f
unc
ti
on
i
s
1+Gr
(s
)G(
s),a
nd
we
r
es
hap
e t
he
f
unc
ti
on
G,
G(j
o))
i
n t
he
re
f q
uen
cy
r
ang
e 2<。<25
The
n we
h
ave
t
he
a
lt
ere
d ma
gni
tud
e
a
s p
lot
ted
i
n Fi
g.5
.10
b).
Her
e t
he
r
obu
stne
ss
i
n e
qua
lit
y
is
s
ati
sfi
ed,
an
d t
he
s
yst
em
r
ema
ins
s
tabl
e.
The
cont
rol
obj
ect
ive
is
,
to
sel
ect
a pr
eil
f t
er
G-
(
s)
and
a compe
nsat
or
几
G_
(
s)
in
a ," 、 产 ‘ L
t
wo-degree-
of-
ree
f dom f
eedbackc
o nf
igurati
on (
Fig.5.
1) s
o t
hatthe t
ransi
ent
,
st
eady-
st
ate
, a
nd
f
req
u e
n c
y -
domain sp
eci
fi
cati
onsare achieved and the cost of
feedbac
k measur
ed by t
he
130
ba
ndwi
dth
of
t
he
c
omp
ens
at
or
G,
(
jo)
) i
s s
uic
f i
ent
ly
s
mal
l.
Th
is
b
andwi
dth
c
ons
tr
ai
nt
i
s
n
eed
ed
mai
nly
be
caus
e o
f no
ise
t
hat
i
s
in
evi
ta
ble
i
n mea
sur
ing
t
he
s
yst
em
out
put
.
A
l
arg
e n
ois
e
a
mpl
i
fi
ca
ti
on
c
an
s
at
ur
at
e
ei
t
her
th
e l
at
te
r
st
age
s
of
G.
(s
)
or
th
e
ea
rl
y
pl
ant
st
age
s.
5.
4 SYSTEMS
W ITH
UNCERTAI
N PARAMETERS
Many
sy
ste
ms ha
ve s
eve
ral
par
amet
ers
t
hat
ar
e c
onsta
nts
b
ut
u
nce
rta
in
wi
thi
n ar
ang
e.
Fo
r
e
xamp
le,
cons
ide
r
a
sys
tem
wit
h a
c
har
act
eri
st
ic
e
quat
ion
S
"
+a
n_
,S
n-
1
+a
,_2
Sn
-2
+.
.
.+a
p
=0
(5
.1
6
))
wi
th
known
coe
fici
f ent
s wi
thi
n bounds
a:
‘aj‘月,
,i=0
'.
..
.
n,
wh
ere
a
n=1
.
To
as
cert
ai
n t
he
sta
b i
i
lty
of
t
he s
yste
m, one
might
h
ave t
o i
nves
ti
gat
e a
ll
poss
ibl
e c
o mb i
nat
ions
of
pa
ramet
ers.
Fotu
r natel
y
it
i
s
possi
ble
to
i
nves
tig
ate
a
li
mited
number
of
wors
t-
case
polynomial
s.
The
anal
ysi
s
of
onl
y four
pol
ynomial
s i
s
sufic
ient
,
and
they
are
re
adi
ly
defne
i d f
or
a th
ird-o
rder
sy
ste
m wit
h a
ch
ara
cteri
sti
c
equa
tio
n
S
3+a
2s2+a,
s+a
。二0. (
5.
17
)
The
n t
he
f
our
p
oly
nomi
al
s
ar
e
q
,
(s
)=S
3+a2
s2+P'
s+00
,
q
2
(S
)=S
3+p2
s2
+a,
S+ao
,
q
3
(S
)
=S
3+吞z
S2+/
3,
S+ao
,
q
4
(S
)=S
3
+a2
s2
+a,
s+00
.
On
e of
t
he
f
our
po
lyn
omi
als
r
epr
ese
nts
t
he
wo
rst
c
ase
a
nd
ma
y i
ndi
cat
e
un
sta
ble
p
erf
orma
nce
o
r a
t
l
eas
t
the
wor
st
per
for
manc
e f
or
t
he
s
yst
em
i
n t
hat
ca
se.
EXAMPLE
5.
4 THIRD-
ORDER
SYSTEM
WITH
UNCERTAI
N
Co
nsi
der
a
t
hir
d-o
rde
r s
yst
em
wi
th
u
nce
rta
in
c
oef
fi
ci
ent
s
su
ch
t
hat
8‘a.‘6
0 =
* ao=8
,刀。=6
0;
1
2‘a,‘1
00
=
* a,=1
2,月;=1
00;
7‘a
2‘2
5 =
* a2=7
,月:=2
5.
Th
e f
our
po
lyn
omi
als
a
re
q,
(s
)=S3+7S
2+l
oos
+60,
q2
(S)
=s
3+255
S+
212s+8
,
q
3
(s
)=s
3+2
552
+lo
os+8
,
q4
(S
)=s
3
+7
S2+1
2s+6
0.
131
We t
hen
pr
ocee
d t
o c
hec
k th
ese
fo
ur
pol
yno
mial
s u
sin
g the
Rout
h-Hurwit
z c
rit
eri
on
a
nd
h
enc
e
de
ter
mine
t
hat
t
he
s
yst
em i
s
st
abl
e
for
a
ll
t
he
r
ang
e of
un
ceta
r i
n pa
ramet
ers.
EXAMPLE
5.5
STABILITY
OF
UNCERTAI
N SYSTE材
Co
nsi
der
a
u
nit
y f
eed
bac
k s
yst
em
wi
th
a
pl
ant
tr
ans
fer
f
unc
tio
n (
nomi
nal
c
ond
iti
ons
)
4.
5
G(
s)
S
(S+1
)(S+2
)
Th
e n
omi
nal
cha
rac
ter
ist
i
c e
qua
ti
on
i
s
th
en
q(
s)=S
3
+3
s2+2
s+4
.5=0.
Us
ing
t
he
Ro
uth
-Hu
rwi
tz
c
rit
eri
on,
we
ind
f t
hat
th
is
s
yst
em
i
s
nomi
nal
l
y s
tab
le.
Howe
ver
,
if
t
he
s
yst
em
ha
s unce
rtai
n coef
fi
cient
s s
uch
t
hat
4‘a
o‘5
=
=>
ao=4,
月。=5
;
1‘a,‘3
=
* a,=1
,刀:=3
;
a
nd
2‘a
2‘4
=
* a2=2
,刀2=4
,
t
hen
we
mus
t
ex
ami
ne
t
he
f
our
p
oly
nomi
als
q,
(s
)=S
3+2S2+3
s+5
,
q2
(s
)=S3+4
S2
+1
S+4
,
q
3
(S
)=s
3
+4
S2
+3
s+4
,
q4
(S
)=S
3+2S
2+i
s+5
.
Us
in
g t
he
Ro
uth
-Hu
rwi
t
z c
ri
te
ri
on,
q,
(s
) a
nd
q
3
(S
) a
re
s
ta
bl
e a
nd
q2
(S
) i
s ma
rgi
na
ll
y
s
ta
ble
.
Fo
r q4
(
s)
we
h
ave
IJ
5
1
2
5
2
5
1
5
一3/2
0
5
5
The
ref
ore
t
he
s
yst
em
i
s
un
sta
ble
f
or
t
he
wo
rst
c
ase
, wh
ere
(
x2
=
mini
mum
a,
=mi
ni
mum,
a
nd
风=ma
ximu
m.
Th
is
o
ccu
rs
wh
en
t
he
pl
ant
ha
s c
han
ged
t
o
G(
s)
= ̄一三一一
s
(s+1
)(
s+1
)’
No
te
t
hat
t
he
t
hir
d po
le
h
as
mo
ved
t
owa
rd
t
he
j
o)
-
axi
s t
o i
ts
imi
l t
at
s=一1
an
d t
hat
t
he
g
ain
ha
s
i
ncr
ese
a d
t
o i
ts
l
imi
t
at
K=5.Of
te
n we
a
re
a
ble
t
o e
xami
ne
t
he
t
ra
nsf
er
f
unc
ti
on
G(
s)
a
nd
pr
edi
ct
t
he
wor
st-
cas
e condi
ti
ons
.
5.
5 THE
DESIGN
OF
ROBUST
CONTROL
SYSTEMS
Th
e d
esi
gn
o
f r
obu
st
c
ont
rol
s
yst
ems
i
s b
ase
d o
n t
wo
t
asks
:
de
ter
min
ing
t
he
s
tru
ctu
re
o
f t
he
c
ont
rol
l
er
a
nd
a
dju
sti
ng
t
he
c
ont
ro
ll
er'
s
pa
rame
ter
s
to
g
ive
a
n "
opt
i
mal
"
sys
te
m p
erf
orma
nce
.
Th
is
132
desi
gn
process
is
normally
done
wit
h "ss
a ume d
co mplet
e knowled
ge" of
the
pl
ant.
Furt
hermore
the
pl
an t
i
s n
o r
mally
descr
ibed
by a
l
inea
r t
ime -
invari
ant
conti
nuous
mo del
.
The s
truc
tur
e of
the
cont
rol
ler
is
chose
n s
uch th
at
the
s
yste
m's
response ca
n meet
cert
ai
n per
formance
cr
it
eri
a.
On
e poss
ible
obj
ecti
ve i
n t
h e
d
esig
n of
a
c ont
rol s
yste
m is
t
hat
the
contr
oll
ed
syste
m's out
put
shoul
d exact
ly an
d ins
tanta
neous
ly r
epro
duce i
ts i
nput.
That i
s,
the
syst
em's t
ran
sfer
funct
ion
s
houl
d be
uni
ty:
T
(
s)
=器・
‘・ (
5.1
8)
I
n ot
her
wo
rds
,
th
e s
yst
em
s
houl
d b
e pr
ese
nta
ble
o
n a
Bo
de
g
ain
v
ers
us
re
f q
uen
cy
d
iag
ram
wi
th
a
0-
dB
ga
in
o
f i
nin
f i
te
b
and
wid
th
a
nd
z
ero
ph
ras
e s
hif
t.
I
n p
rac
ti
ce,
t
hi
s
is
no
t p
oss
ibl
e,
s
inc
e
e
vey
r sys
tem wil
l
cont
ai
n i
nd uc
tiv
e -
and
capaci
ti
ve-t
ype component
s t
hat
st
ore e
ner
gy
in s
ome
f
orm.
It
is
thes
e el
ement
s and t
heir
int
ercon
necti
ons wit
h ener
gy-di
ssi
pati
ve co
mponent
s t
hat
p
roduc
e the s
yst
em's
dynamic r
esponse c
harac
ter
ist
ics.
Such s
yste
ms r
eprod
uce s
ome i
nputs
a
lmost
exact
ly,
whi
le
ot
her
input
s a
re not
r
epro
duced
at
all
, s
igni
fyi
ng
that
t
he s
yste
m's
ba
ndwidt
h
i
s l
ess
t
han
i
nf
ini
te.
On
ce
we
r
eco
gni
ze
t
hat
t
he
s
yst
em'
s d
yna
mic
s c
ann
ot
b
e i
gno
red
, we
n
eed
an
ew
des
ign
o
bje
cti
ve.
On
e po
ssi
bl
e
de
sig
n o
bje
cti
ve
i
s
to
ma
int
ai
n
th
e ma
gni
t
ude
r
espo
nse
c
urv
e a
s la
ft
an
d a
s
cl
ose
t
o u
nit
y f
or
s
a la
rge
a
ban
dwidth
as
pos
sib
le
for
a
gi
ven
pla
nt
a
nd
cont
rol
ler
co
mb i
nat
ion.
Anot
her
i
mport
ant
g
oal
of
a
c
ont
rol
sy
ste
m des
ign
is
t
hat
th
e efe
ct
on
t
he
o
utpu
t o
f t
he
sys
tem
du e
t
o
di
st
ur
ban
ce
i
s
mi
ni
miz
ed.
Thu
s we
wi
sh
t
o
mi
ni
miz
e Y(
s)
/
D(
s)
o
ver
a
r
ang
e o
f
re
f q
uen
cy.
Co
nsi
de
r t
he
c
ont
ro
l s
ys
te
m s
hown
i
n F
ig.
5.
11
, wh
ere
G(s
)=G,
(s
)G2
(s
) i
s t
he
p
lan
t,
a
nd
D(
s)
i
s
the
d
ist
ur
ban
ce.
We
th
en
h
ave
Y(
s) GCG,G2
(
s)
T(
s)= (
5.1
9)
R(
s) 1
+GcGIG2
(
s)
and
Y(
s) G2
(
s) (
5.20)
D(
s) 1
+Gc
G,G2
(s
)
D(
s)
R付 + +
I I刊
G;
.
fs) G,
(
r) --- ̄争
Fi
g.
5.
11
A
sy
ste
m di
sturbance
Not
e t
hat
bot
h t
he
r
efe
rence
a
nd
di
stur
bance
t
rans
fer
f
unct
ions
ha hes
ve
t ame
denomi
nat
or;
i
n
o
the
r wo
rds
,
th
ey
ha
ve
t
he
s
ame
c
har
act
eri
st
ic
e
qua
ti
on-
tha
t
is
,
1
+Gc
(
s)G,
(s
)G2
(s
)=1
+L(
s)=0. (
5.21
)
Fu
rth
ermo
re
we
r
eca
ll
t
hat
th
e s
ens
it
ivi
ty
o
f T(
s)
wi
th
r
espe
ct
t
o G(
s)
i
s
133
砚
1
(
5.22)
1
+Gc
GiG
(
s)’
a
nd
the
cha
ract
eri
st
ic
e
quat
ionis
hei
t nlu
f en
cin
g f
act
or
on
t
he
s
ens
it
ivi
ty.
Eqn.
(5
.22)shows t
hat
f
or
l
ow se
nsit
iv
it
y S
we
re
qu i
rea
hig
h va
lue
of
loop
gai
n
L(to
),but
i
t i
s
known t
hat
high
gain
dc
coul a
use
i
nst
abi
li
ty
o
r
po
or
r
espo
nsi
vene
ss
of
T(
s).
Thu
s t
he
d
esi
gne
r
se
eks
t
he
f
ol
lo
win
g:
T
(
s)
wi
t
h wi
de
b
and
widt
h
an
d f
ai
th
ful
re
pro
duc
ti
on
o
f
R(
s).
2.
Lar
ge
l
oop
g
ain
L(
s)
i
n
or
der
t
o
mi
ni
miz
e t
he
s
ens
it
iv
it
y
S.
3.
Lar
ge
l
oop
g
ai
n
L(
s)
a
tt
ai
ned
p
ri
mar
il
y
by
Gc
(s
)G,
(s
),s
in
ce
Y(
s)
/
D(
s)”1
/GC
G,
(s)
.
Hi
gh
g
a
ins
fa
r g
ood
pe
r i
l
rma
nce
成
/(
Ca1
11
11
1and
)
g
n
i
w
l
o
f
漾哪f
2
0 l
og
I
G
C:
(
|.
Stabl
e crossover
(
gai
, n and
pha
se
ma
rgi
ns)
Mi
nimum
per
for
mall
ce
11-叭
b
ount
h卜
舞 Robus
tness
A
,-/
b
oun
ds
i ,
Law gril
ls
t
o r
educe
s
ensit
iri
t-
to
sens
or
noi
se a
nd model
uncert
ainty
Fi
g.
5
.12
Bo
de
d
iag
ram
f
or
2
01o
gIG}
GO0
))l
Se
tin
g t
he
d
esi
gn
of
r
obus
t s
yst
ems
i
n re
f qu
enc
y-d
oma
in
t
erms
, we
a
re
r
equ
ire
d t
o f
ind
a
p
rop
er
c
omp
ens
at
or
,
G,
(s
),s
uch
t
hat
th
e c
lo
sed
-l
oop
s
ens
it
iv
it
y
is
le
ss
t
han
s
ome
t
ol
er
anc
e
va
lu
e,
an
d se
nsi
tivi
ty
mini
mizat
ion i
nvolv
es f
indi
ng ap
roper
compensat
or suc
h tha
t t
he cl
osed-lo
o p
se
ns i
t
ivi
ty eq
ual
s o
r i
s ar
bit
rar
ily
cl
ose
to t
he mi
nimal a
tt
aina
ble
sensit
iv
it
y. S
imila
rly
, t
he gain
margin
probl
em
is
t
o in
f d
a pro
per
compensat
or
to
achi
eve
some p
rescr
ibed
gain
margi
n,
and gain
margin
ma xi
miz
ati
on i
nvolv
e s
fi
ndi
ng ap
roper
co
mp ensat
or
to
achie
v e
the
ma x
imal a
tt
ainable
gai
n ma r
gin.
Fo
r t
he re
f quency-
domai
n speci
ic
f at
ions
, we re
quie
r th
e foll
owing f
or t
he Bo de
d
ia
gra
m o
f
G,
G(j
o),
( sho
wn
i
n
Fi
g.
5
.12
.
1
.
Fo
r r
el
at
iv
e s
ta
bil
i
ty,
GC
G(j
co)
mu
st
h
ave
,
fo
r a
n a
deq
uat
e
ra
nge
o
f
Lo,n
ot
mo
re
t
han
-
20dB/
dec
ade
s
lop
e a
t
or
n
ear
t
he
c
ros
sov
er
re
fque
ncy叭・
2.
St
ead
y-s
tat
e
ac
cur
acy
a
chi
eve
d by
t
he
l
ow
re
fque
ncy
g
ain.
3
.
Ac
cur
acy
o
ver
a
b
and
wid
th
C
)B
b
y n
ot
a
ll
owi
ng
I
GcG(
CO
)!t
o
fa
ll
b
elo
w ap
res
cr
ib
ed
l
evel
.
4
.
Di
st
ur
ban
ce
r
ej
ect
i
on
b
y h
igh
g
ain
f
or
GC
(
jco
)
ov
er
t
he
s
yst
em
b
and
wid
th.
Us
in
g t
he
r
oot
s
ens
it
i
vit
y
co
nce
pt,
we
ca
n s
ta
te
t
hat
we
r
equ
ir
e
th
at
S二b
e mi
ni
miz
ed
wh
il
e
134
a
tt
ai
ni
naT(
s)w
it
h
domin
ant
ro
ot
s
th
at
w
il
l
pr
ov
id
e
th
e a
ppr
op
ri
at
e
r
es
pon
se
a
nd
mi
nimi
ze
th
e
e
fec
t
of
D(
s).
Ag
ai
n we
se
e
th
at
t
he
g
oal
is
t
o
hav
e
th
e g
ai
n
of
t
he
l
oo
p p
ri
mar
i
ly a
t
tai
ne
d
b
y GC
(
s)・
As
a
n e
xamp
le,
le
t
G,
(s)=K,
G,
(
s)=1
an
d GZ
(s
)=1
/
s
(s+1
)
fo
r
th
e s
yst
em
i
n
F
ig
.5
.1
1.
Thi
s
s
ys
te
m h
as
t
wo
r
oo
ts
,
a
nd
we
s
el
ec
t。g
ai
n
K
so
t
ha
t
Y(
s)
ID(
s)
is
mi
ni
mi
ze
d,
S分i
s
mi
ni
miz
ed,
an
d T(
s)
h
sd
a
es
ir
abl
e
do
min
ant
ro
ots
.
The
s
ens
i
ti
vi
ty
i
s
S
r_dr K_到、 .
dK ; dK}
K. (
5
.2
3)
and
t
he
char
act
eri
sti
c equat
ion
i
s
s
(s+1
)+K=
=0. (
5.24)
The
ref
ore
dK
/d
s=一(
2s+1
),s
inc
eK=-s
(s+1
).We
t
hen
o
bta
in
・
-
s
(s
+
1
)
一1
S
r K
一
一
(
5.25)
2s+1 S
I
s=r
Wh
en否<1
,th
e
ro
ots
ar
e
co
mpl
ex,
an
d
r=-0
.5+J
w.Th
en
一
'
U (
X卜I0
—.2
5+。,丫
Ii
s
(
5.26)
l
4w` )
,‘
月”
80
}
:
60 节 一 护1
:
』
}
S
e
ls
i
t
1
7
t
)'
勺乞会卜,苏 倪
尸 口
州尸 I
开i
甲
什
刀
了
;
甘 一一一
之,〕
火之
hers
为not
,、
山吸
0.
20
厂 一
0
/ { 2
1
{
3 子 S
0
K
Fi
g.
5.
13
Se
nsi
ti
vit
y
an
d p
erc
ent
o
ver
sho
ot
f
or
a
s
eco
nd-
orde
r s
yst
em
The magn
it
ude
of
t
he
sensi
ti
vi
ty
is
plo
tt
ed
in
Fig.
5.
13 f
or K=0.2 t
o K=5,Th
e p
erc
ent
ove
rsho
ot
to
a
st
ep
i
s
al
so s
hown.
I
t
is
be
st
t
o
red
uce
the
s
ensi
ti
vi
ty
bu
t t
o
imi
l t
K
to
1
.5
or
l
ess
.
We t
hen
at
ta
in t
he
majo
rit
y of
t
he a
tta
ina
ble
reduct
i
on in
s
en s
it
ivi
ty
whil
e ma
int
ain
ing
g
ood
pe
rfo
rman
ce
for
t
he
s
tep
res
pons
e.
I
n g
enera
l we
c
an us
e t
he
de
sign
proce
dur
e
as
fo
llo
ws:
1
.
Sk
etc
h t
he
r
oot
l
ocus
o
f t
he
c
ompe
nsa
ted
s
yst
em
wi
th
Gc
(
s)
c
hos
en
t
o a
tt
ai
n t
he
d
esi
re
d
l
oca
tion
f
or
t
he
domi
nant
r
oot
s.
135
2
.Ma
xi而z
eth
egino
a fG。(
5)t
or
edu
cet
hee
ec
f t
oft
hed
is
tu
ba
r nc
e.
3
.De
te
nl
r i
neS二a
nda
ai
t n‘
he而n
imu
mva
lueo
fhes
t ens
ii
ti汀c
v ns
o i
sen
t t
wit
ht
het
ns
a
r i
en
t
es
r P
ons
erequ
ied
r ,
asd
esc
ib
r e
dins
tePI
.
五男乃.
加仔生召5
.石5日切汀刀 于了TYAND C以村乃曰W沮刀 口N
此tu
sco
nsi
de
r由es
ys
emw
t hen
G(5
)=1
152
,H(
5)=1
,ndG。
a (5)i
st
ob
ese
l
ecdb
e
t y
fqu
e
r en
cyresP
ns
o emethds
o .T
heef
rorethecmP
o ens
atori
stobesel
ct
e edtoa
ch i
ee a
v n
pP
a r
oPia
re ga
t ina
ndPha
sem田名i
nwhil
e而ni而z
ings
ens
iiv
tit
ya
ndt
heeec
f toft
hedi
st
urb
nc
a e.
ThusweChoos
e
K(
5/2+1
)
G。(
5)= (
5.7)
2
slP+1
M触c
hos
eK=
=l0t
or
edu
cet
hee
ec
f to
fthed
ist
ba
r
u nc
e.oa
T t
t助naPh
sema
a gi
r nof
45‘
,
wes
el
ec
edz=2
t .oa
ndP二1
2.0.W
ethe
natt
int
a hec
oPe
m n
sedd
t
a iag
rams
howni
nRg
・5.
9
ndr
a e谬a
edi
t nFig
・5.
14.
ec
R a
llt
hatt
hec
los
edl
卿 b
nd
a wi
dhi
t s
叱 =1
.6叭.T
husw
ewil
l
i
ncea
rset
heba
ndwi
dhbyus
t in
gthec
oPe
m ns
to
arandi
Pr
m ov
ethef
id
e1i
yo
t fr
ePr
odu
cio
tn oft
e
h
i
nPu
tsi
gna
ls.
卜址又 }
灭丈
,
l
l
丈巡鳄一
}
- -
匕一
。 c,
、气
} 、 1 ,,. t l . 1
1
1 1
} 囚二
口 巨 匕各一
、、 、
。l.四,
I
’ l
.
{
,
t
又
一粇,
} 口
卜 、、
、、 ”J
gn
,
t
、如
’t
I ’
,
卜_
一,、
1 健
}
、
6n
,肖乃才
q 口砂d IJ夕
一砂。亡价J召e-
l 口口[口
} { }
} 副扣尸
卜一 声 l
}
1 _!
l 口口
卜
以
一 物叮 P
’
,
P,1
t矛若州
l
.
,
”七u护 1’—
,
,
日尸洲一
1目
!
2协r‘
}一日
l
沁.
.
尹J翰I
,
’ 一
’
.
1“
‘夏
一1沙口。
ng.
5.1
4Bo
ded
iag
ramf
r Ex
o 别rnP
les.
6
hes
T ens
iiv
tiyma
t ybea
sce
ai
t
r n
eda
t此a
s
}
,
:
、,
・}
石
磷而一
。 (
5.28)
1
36
P
has
e
of不
、办 \ /
0二
i 一一 一一卜卜 、 ̄ ‘
“\ D
.1RdB 一
“ / .
/一 —
一
了 州 — 一 、、
一尸尹
沐六口
目碑产一 朴 汀
一 ̄  ̄、、
\ L{
、、
尸户一
斌少尸 ̄ ̄一一一
犷 I
0
.
-
-I
下 { 了入 GG,
(j动
/
务 冲 -
B
寸)人
-77
入r
7Q4
w
1
.
厂
{
i 1 f l-
0
d
B饭
之、
龚 3 一 ・ t
I 入
d
B
q
1
%
.
2
1
Md
B2
 ̄
\
勺
邀 层万 、
,
一入 \
彝去 撰 公琪
,
举 9
 ̄七助身 氏
,
7
6
[中
-4
口口 百币 -
一‘
鬓1 参鬓 彝 城
趴 O
护J
2
!
1
龟
囊喜 {
\
鑫呈 T
,d
-
]0
" d
B ({
挂巨’
了- J
r ̄,1 一创闷卜一 J月
.
丫/
气洛‘
.、
井介
乡运 f
"
k 一
一0
N
曰
二亏一
\_
6--
-
孰
_
N
1
护日
飞
、 、
(
I }
「' 「 ] 一斗一一 尸V {
:土 i上上
、
1
一 、 !}
G 、 }
产书 ,(价,尸,,二.二.1S
Jla
gn而 de
七 万厂l
L
n
}”(
I i\丁
厂不 --
‘石一 电
州侧.口‘曰一
}{
}
_I {
_L{
一\ { {
’ }
-} 、
活 1
‘. .,魂口.州. ‘,曰.一一
月J‘, , .. .门 . 一
Loo
p ga
i”一
J 口I矛矛
{
— G 一
1十 G
【 .一 一 下洲,
「 曰 1 }
Lo
op砂 “
,.
,,
耳(G)
.,”de
gre
es
Fi
g .
5.1
5 Ni
cho
ls
c
har
t
fo
r Ex
amp
le
5
.7
T
o
es
ti
mat
e
I
Sn}
.we
r
ec
al
l
t
ha
t
t
he
Ni
ch
ols
ch
art
en
abl
e
! U !
us
:。o
bt
ai
n
137
、
_, _、} G‘、G(
。)
I
I
lj
0 l
=1---—
- -
- -
(
5.29)
}
1+Gc
G(
a))
T
hus
we
c
an
p
lot
a
f
ew
p
oin
ts
o
f GC
G(j
(o)
o
n t
he
Ni
cho
ls
c
har
t
an
d t
hen
r
ead
T(
(o
)
fr
om
t
he
chart
. Then
wh
ere
(
o,
i
s c
hos
en
a
rbi
t
rar
il
y a
s (
oc
/
2.
5.
I
n
}
'
S
G
r,
g
e
一
n
1
G
T
C
G
(
e
r
(
o
,
(
o
)
l
a
)
l
we
c
hoo
se
af
r
eq
uen
cy
b
elow
G
)c
t
o
(
5.30)
d
et
er
mi
ne
t
he
v
al
ue
o
f
I
S(
o)
,)
}。
Of
c
au
se
we
de
si
r
e
a
l
ow
v
al
ue
o
f
se
ns
it
i
vi
ty
.
Th
e
Ni
c
hol
s
c
ha
rt
f
or
t
he
c
omp
ens
at
ed
s
yst
em
i
s s
hown
i
n F
ig.
5.
15
. F
or
(
0,=。。
/2.
5=2
we
h
ave
2
0 l
og
T=2.
5dB
a
nd
2
0 l
og
GAG=9
dB.
The
ref
or
e
IT I 里 =
0.
47.
I
"
W o,
)= 1
- 1=
J
GCGJ 2.8
EXAMPLE
5.7
SENSITIVITY
WITH
A LEAD
COMPENSATOR
Let
u
s a
gai
n con
side
r t
he sys
tem i
n Exa
mpl
e 5.
6,
u
sin
g t
he
r
oot
l
ocu
s d
esi
gn
o
bta
ine
d
Exampl
e
5.
3.
The
co
mpe ns
ato
r was
cho
sen
as
8.1
(s+1)
GC
(
s)= (
5.31
)
(s+3.6)
f
or
t
he
s
yst
em
o
f Fi
g.
5.1
6.
Th
e d
omi
nan
t
ro
ots
a
re
t
hus
s=一1士j2.
Si
nce
,
t
he
g
ain
i
s
8.
1,
t
he
e
ffe
ct
of
th
e d
ist
urba
nce
is
reduced,
an
d t
he
t
ime
r
esp
onse
me
ets
t
he
s
pec
ifi
cat
ions
.
The
sens
it
ivi
ty
a
t a r
oot r may be o
b t
ained by ss
a umi
ng
th
at t
he
sys
tem,
wi
th
dominant
ro
ots
, may be
a
ppropr
iat
ed
by
the
se
cond-orde
r
s y
ste
m
K
K
T(
s)
s`+2
营0)
,S+K ,
2+2s+K
s
in
ce
4
C0,二1
.Th
e
ch
ar
ac
te
ri
st
i
c
eq
uat
i
on
i
s
t
hus
s2+2s+K=0.
力(
s)
r
ig
.
5.
16
Fe
ed
ba
ck
c
on
tr
ol
sy
st
em
wi
t
h。d
es
ir
ed
i
np
ut
R(
s)
an
d
an
u
nd
es
ir
ed
i
np
ut
D(
s)
T
hen
dKI
ds=一
(2s
+2)
,si
n
ce
K=一(
5
s+
22s
).T
her
ef
or
e
138
一1 一(
s
2
+2
s) s+2
Sr
二二 一 .一 一
(
5.32)
(
2s+2)
s 2s+2
wh
ere
r=一1+j
2.The
n s
ubs
ti
tu
ti
ng
s=r,
we
o
bta
in
(
S
K{・
0.
56
I
f
we
r
ais
e
th
e ga
in
t
o
K=1
0,we
e
xpe
ct
r 一1.
1士j2.
2.Th
en
t
he
s
ens
it
iv
it
y
is
1
S
K}・
0.
53
・
Th
us
a
s K
inc
rea
se,
th
e s
ens
it
iv
it
y
de
cre
ase
s,
b
ut
t
he
t
ra
nsi
ent
p
erf
orma
nce
d
ete
rio
rat
es.
5.
6 PI
D CONTROLLERS
The
PID
contr
oll
er
has
t
he
t
ransfer
f
unct
ion
G
。(
:)
一K,
十KZ
十K3
S. s
Th
e p
opu
lar
it
y of
PI
D c
ont
rol
l
ers
c
an
b
e a
tt
ri
but
ed
p
art
ly
t
o t
hei
r
ro
bus
t p
erf
orma
nce
i
n aw
i
de
r
ang
e o
f o
per
at
in
g c
ond
it
io
ns
a
nd
p
art
ly
t
o t
hei
r
fun
cti
ona
l
si
mpl
i
cit
y,
wh
ich
a
ll
ows
e
ngi
nee
rs
t
o
o
per
ate
t
hem
in
a s
impl
e,
s
tr
aig
htf
orwa
rd
man
ner
. To
i
mpl
eme
nt
s
uch
a c
ont
rol
ler
, t
hre
e
p
ara
met
ers
mus
t b
e d
ete
rmi
ned
f
or
t
he
g
ive
n p
roc
ess
: p
rop
ort
iona
l g
ain
, i
nte
gra
l ga
in,
a
nd
der
ivat
ive
ga
in.
Consi
der
t
he
PID
cont
rol
ler
_ _
_ K, 一 Ks2+K,
s+K,
。。
(s)=K1
+一 +K3 s=
J
l ‘
s
s
(
5.33)
_K3
(S
2+a
s
+
b
)_K3
(s
+z,
)(
s+z
2)
s
s
wh
ere
a
=
K,
/
K3
,
an
d b
= K2
/
K3.Th
ere
for
e
a
P
I)c
E
onr
t
oll
er
i
nro
tduc
es
a
ra
tnsf
er
f
unc
ti
on
wi
t
h
o
ne
p
ole
a
t
the
o
rig
in
a
nd
t
wo
z
ero
s t
hat
ca
n b
e l
oca
ted
a
nywh
ere
i
n
the
l
et-
fhan
d s
-
pl
ane
.
Rec
all
that
a
roo
t l
ocu
s b
eing
s a
t
th
e po
les
a
nd
e
nds
a
t
th
e z
eros
.
If
we
ha
ve
a
s
yst
em
a
s s
hown
i
n F
ig.5.16
wi
th
G,
(s
)=
(s+2)
(s+5)
a
nd
we
u
se
a
PI
D c
ont
rol
l
er
wi
th
c
ompl
ex
z
ero
s,
we
c
an
p
lot
t
he
r
oot
l
ocus
a
s s
hown
i
n F
ig.
5.1
'7.
As
t
he
g
ain
,
K3
,of
t
he
c
onro
tll
er
i
s
in
cre
ase
d,
t
he
c
omp
lex
r
oot
s
ap
pro
ach
t
he
z
eos
r .Th
e
c
los
ed-
loo
p ra
t ns
fer
f
unc
tio
n i
s
G(
s)G,
(s
)G,
(s
) 一
K3
(s+z
,
)(
s+z
,)。 , 、_K3
_ G'
(s
)
T(
s)= L7 _饭s )三 (
5.34)
1
+G(
s)G,
(s
) 了, . , 、/, 二
、。Y
f2八。"
T",]八J'
T'r
l)
、了  ̄ . 二 、 v 、
(
S+r
2)
1
构
Fi
g.
5.
17
Roo
t
lo
cus
wi
th
Z,=-6+JZ
b
eca
use
t
he
z
er
os
a
nd
t
he
c
omp
le
x r
oot
s
ar
e
ap
pro
xima
te
ly
e
qua
l(r
,”z
,)・S
et
ti
ng
G,
(
s)=1
we
have
K3
K3
T(
s)= (
5.35)
2 S+k-
S+r 3
whe
n凡 >>1,Th
e o
nly
i
l
mit
in
g f
act
or
i
s
th
e a
ll
owa
ble
ma
gni
tu
de
of
U(s
) (
Fig.
5.
16)
wh
en
K3
is
la
rge,
if
K3
is
1
00,
th
e s
yst
em
h
as
a
f
as
t
re
spo
nse
a
nd
z
er
o
st
ead
y-s
ta
te
e
rr
or.
Fu
rt
he
rmo
re,
t
he
eff
ect
of t
he
d
ist
urbanc
e i
s
reduc
ed
si
gniic
f an
tly.
I
n g
ener
al
we
note t
hat
PII
) co
ntr
oll
ers
ar
e part
i
cul
arl
y u
sef
ul
f
or
r
edu
cin
g s
tea
dy-
sta
te
e
rro
r
a
nd
i
mpr
ov
in
g
th
e
t
ra
ns
ie
nt
r
es
po
ns
e
wh
en
G(
s)
ha
s
on
e
or
two
po
le
s
(o
r
ma
y
be
ap
pr
ox
ima
t
ed
b
y
a
s
eco
nd-
ord
er
pl
ant
).
5.
7 THE
DESIGN
OF
ROBUST
PID-
CONTROLLED
SYSTEMS
Th
e
sele
cti
on of
t
he t
hree
coef
fi
cie
nts
of
PID co
ntr
olle
rs
is
b
asi
call
y as
ear
ch pr
oble
m i
n a
t
hree-
dimensi
onal
sp
ace. Poi
nts
i
n t
he s
ear
ch s
pace
cor
respo
nd
to
difer
ent
sel
ect
ion
s of
a
PI
D
c
on t
rol
ler
's
thre
e par
ameter
s.By c
hoosi
ng di
ff
ere
nt poi
nts
of
th
e parame
ter s
pace,
we
can
p
rod
uce
, f
or
e
xampl
e,
d
iff
er
ent
s
tep
r
esp
ons
es
f
or
a
st
ep
i
nput
.A
PI
D c
ont
rol
l
er
c
an
b
e
d
eter
mine
d b
y mov
ing
i
n t
his
s
ear
ch s
pace
on
a
tr
ial
-an
d-err
or
bas
is.
Th
e mai
n pr
obl
em i
n t
he s
ele
cti
on o
f th
e t
hre
e coef
fi
cien
ts
is
t
hat
t
hes
e c
oefi
cie
nts
d
o n
ot
r
ead
il
y t
ra
nsl
at
e i
nt
o t
he
d
esi
re
d p
erf
or
man
ce
a
nd
r
obu
stne
ss
c
har
act
er
is
ti
cs
t
hat
t
he
c
ont
rol
s
yst
em des
igne
r h
as
in
min
d.Se v
era
l
rul
es
a
n d
met
hods
have
be
en
pro
posed
to
s
olv
e t
his
pro
ble
m.
I
n t
his
s
ect
ion
we co
nsi
der
s
eve
ral
de
sig
n me
thods
u
sing
ro
ot
loc
us
a
nd pe
rfo
rmance
i
ndic
e s,
Th
e
f
ir
st
d
esi
gn
meth
od us
es
the
I
TAE pe
rfo
rmanc
e i
ndex.
(
To
reduc
e t
he
cont
ri
but
io
n of
th
e
l
ar
ge
i
ni
ti
le
a
rr
or
t
o
the
va
lue
of
t
he
per
for
mance
i
nte
grl,
a a
s wel
l
as
t
o e
mphasi
ze
er
ror
s
occur
ri
ng
140
a
ter
i
n t
he
res
pons he f
e,
t oll ng i
owi nde
x h
as
b
een
p
rop
ose
d:
I
TAE ・
fti
e
(t
)
{
dt,
・T
he
opt
imum
coef
fi
cie
nts
f
or s
tep
i
npto
u r
f
or
a
ra
mp
i
npu
t.
He
nc wesel
e ect
t
he
t
hre
e PI
D
coef
fici
ents
to
mini
mize t
he
I
TAE
per
for
man
ce
i
nde
x,
wh
ich
p
rodu
cesan excell
ent
t
ransi
ent
r
espo
nse
t
o a
s
tep
o
r a
r
amp.Th
e d
esi
gn
p
rod
uce
c
ons
is
ts
o
f t
hre
e s
tep
s:
1.
Se
lec
t t
he 0
),
,
o
f
the
cl
osed
-l
oop
sy
ste
m by
spe
cif
yin
g t
he
set
tl
in
g t
ime・
2.
Det
ermine
t
he
t
hree
c
oefic
f i
ent
s
us
ing
th
e a
pprop
ri
ate
opt
imum e
quat
ion
a
nd
t
he叭 o
f s
tep
I
t
o
ob
tai
n G,
(s
)
3
.
De
t
er
mi
ne
a
p
re
f
l
te
r
Gp
(s
)
s
o
t
ha
t
t
he
cl
o
se
d-
l
oo
p
sy
st
em
t
r
an
sf
er
fu
nc
ti
o
n,
T(
s)
,
d
oe
s
no
t
hav
e a
ny
ze
ros.
EXAMPLE反8
ROBUST
CONTROL
OF
TEMPERATURE
Co
nsi
der
a
t
empe
rat
ure
c
ont
rol
l
er
wi
th
a
c
ont
rol
sy
ste
m a
s s
hown
i
n
Fi
g.5.
16
a
nd
a
p
lan
t
1
G(
s)= (
5.36)
(
S+1
)2
I
f G,
(s
)
=1
,
t
he
s
t
ead
y-s
ta
te
e
rr
or
i
s
50
%,
a
nd
t
he
s
et
ti
l
ng
t
i
me
(
2%
c
ri
te
ri
on)
is
3
.2
s
eco
nds
f
or
a
s
tep
i
npu
t.
We
d
esi
re
t
o o
bta
in
a
n o
pti
mum
I
TAE
p
erf
orma
nce
f
or
a
s
tep
i
npu
t
an
d a
s
ett
li
ng
t
ime
o
f l
ess
t
han
0.
5
se
con
d.
Us
ing
a
PI
D c
ont
rol
l
er
we
ha
ve
G
。(
S
)・K
3s
2
+
K,
s+
K2. (
5.37)
Ther
efor
e t
he
cl
osed-
loop
t
rans
fer
f
unct
ion
wi
thoutp
rei
f
te
ri
ng
[
GP
(
s)
=1
]
i
s
Y(
S)_ G,G(s
)_ K3
s2+K,
s+K2 (
5.38)
T
,
(s
)=
s)
R( 1+G,G() S
s 3+(
2+K3
)S2
+( 1
+K,
)s+K2
Th
e o
pti
mum
c
oefi
ci
ent
s
of
t
he
c
har
act
eri
st
ic
e
qua
ti
on
f
or
I
TAE
a
re
o
bta
ine
d a
s
(
S3
+1
.7
5(
tx2
+2.
15
co
n
S+。刃
). (
5.39)
We
ne
ed
t
o s
ele
ct叭 i
n o
rde
r t
o me
et
t
he
s
ett
in
l g
t
ime
r
equ
ire
men
t.
Si
nceT
S=4/
枷。a
nd
su
杏i n
kno
wn
b
ut
n
ear
0
.8,
we
s
et
(
0=1
0 .
Th
en
e
qua
ti
ng
t
he
d
eno
min
ato
rof
Egn.
(5.
38)
t
o
(
Eqn.5.
39)
,
we
o
bta
in
t
he
t
hr
ee
c
oefi
ci
en
ts
a
s
K,=21
4,K3=1
5.5,
and
K2=1
000
.
Then
Egn.
(5.
38)
becomes
巧.
5x2+214s+1000 巧.
5(s+6.
9+j4.
1
)(s+6.
9一j4.
1)
T,
(S) (
5.40)
S3+1
7 .
5S2+215s+100
0 S3+17.
5x2+215s+1
000
Th
e r
esp
ons
e o
f t
his
s
yst
em
t
o a
s
tep
i
npu
t h
a novers
s
a hoot
of
32%,
as
r
ecor nTabl
ded
i e 5.
2.
We
se
lec
t a
pr
efi
lt
erG_ (s)
s
o t
hat
we achieve
the
desir
ed ITAE
r
esp ons
e wi
th
.
尸 、 一
T(
s)=
G,
GG,
(s
) 1
000
(
5.41
)
1
+G,
G()‘
s 、
3+1
7--
is
2+21
5s+1
000
Ther
efor
e we
r
equi
re
141
64.
5
GP
(s
)=s2+13. (
5.42)
8s+64.
5
i
n o
rde
r
to
e
limi
nat
e t
he
z
eros
i
n Eqn
.
(5
.40
) a
nd
b
rin
g t
he
o
ver
all
nu
mer
ato
r t
o 1
000
.Th
e r
espo
nse
o
f t
he
s
yst
em,
T(
s),t
o
a
s
te
p i
npu
t
is
i
ndi
cat
ed
i
n
Ta
ble
5.
2.
The
s
yst
em
h
sa
a s
a
m l
l
ov
ers
hoot
,
a
s
ett
in
l g
t
ime
of
l
ess
t
han
1/
2 s
eco
nd,
a
nd
zer
o s
tea
dy-
st
ate
er
ror
. Fu
rthe
rmo
re
f
or
a
d
ist
urb
anc
e
D(
s)=1/s,t
he
ma
ximu
m v
alu
e o
f
y(
t)
du
e t
o t
he
d
ist
ur
ban
ce
i
s 0.
4%
of
t
he
ma
gni
tud
e o
f t
he
di
st
urba
nce
.
Th
is
i
s
a
v
ery
f
avo
rabl
e
de
sig
n.
Cont
rol
ler G,
(s
)
=1
PI
D a
nd
G,
(s
)二1
PI
D wi
t
h G,
(s
)
Pr
eil
fte
r
Percent
overshoot 0 31.
7% 1.
9%
3.2 0.
20 0.
45
Se
tt
li
ng
t
ime
(
sec
ond
s)
St
eady-
stat
e er
ror 50.1% 0.
0% 0.
0%
52%n 0.
4% 0.
4%
{
y(
t)
ld
(t
)}
二im
a
m
EXAMPLE反9
ROBUST
SYSTEM
DESIGN
Le
t
us
r
eco
nsi
der
t
he
s
yst
em
i
n Ex
amp
le
5.
8
wh
en
t
he
pl
ant
va
ri
es
s
ign
ifi
can
tly
,
so
t
hat
G(
s)
=一K (
5.43)
(
Zs+1
)2
wh
ere
0.
5 5
x
_1
< ,1<
_K<
_2.I
t
is
d
esi
red
t
o a
chi
eve
r
obus
t
be
hav
ior
u
sin
g a
n I
TAE
o
pti
mum
s
yst
em wi
th
ap
reil
fte
r wh
ile
at
tai
nin
g an
ove
rsho
ot of
l
ess
th
an 4
% a
nd
as
et
tin
l g
t
ime (
2%
c
rit
eri
on)
of
l
ess
than
2s
econ
ds,
while
G(
s)
can
att
ain
any
val
ue
in
t
he
r
ang
e in
dic
ate
d.
We
s
ele
ct叭 =8
i
n o
rde
r t
o a
tai
n t
he
s
et
tl
in
g t
i
me
a
nd
de
ter
mine
t
he
I
TAE
c
oefi
ci
ent
s
fo
r K=1
a
nd
z=1
.Co
mpl
et
i
ng
t
he
c
al
cu
la
ti
on,
we
o
bt
ai
n
t
he
s
ys
te
m wi
t
hou
t
a
p
ref
i
lt
er
[GP
(
S)=1
]
a
s
1
2(
s
2+1
1.38
s+4
2.6
7) (
5.44)
T,
(s
)=
S
3+1
4S+
5
2137.
6s+51
2
and
1
2(5
2+1
1.3
8s+4
2.6
7)
Gr
(s
)= S
(
5.45)
We
s
ele
ct
a
p
ref
il
te
r
42.
67
Gp
(S
)=S2+11.
385+42.
67
(
5.46)
t
o o
bta
in
t
he
o
pti
mum
I
TAE
t
rans
fer
f
unc
tio
n
512
T(
s)=S3+14S2+137. (
5.47)
65+51
2
We
t
hen
o
bta
in
t
he
s
tep
r
esp
ons
e f
or
f
our
c
ond
iti
ons
:
r=1
' ,K=1;
i=0.
5,K =1;
z=1,
K=2;
142
a
nd
T=0.
5,K=2
Th
e r
esu
lts
a
re
s
umma
riz
ed
i
n
Ta
ble
5
.3.
Thi
s
is
a
v
ey
r r
obus
t
sy
ste
m.
Tab
le
5.
3
Re
sul
ts
f
or
Ex
amp
le
5.
9
wi
th叭 =8
Pl
ant
Condi
tions 公 = T= 0.
5
T二 1 '
r二 0.
5
K=1
K =1
K =2
K=2
Per
cent
over
shoot 2%
0% 0%
1
%
」
Set
tl
i
ng
t
i
me(
sec
ond
s)… 1
.
1
.25 .
0.
81
.l1
… 二
,」
0.
8二1
二「.
。0,
.
9, 卜
.
.
Th
e v
alu
e of
叭th
at
c
an
b
e c
hos
en
wi
ll
be
l
imi
te
d by
c
ons
ide
ri
ng
t
he
ma
ximu
m a
ll
owa
ble
u
Q),
whe
re
u(t
) i
s
th
e o
utp
ut
o
f t
he
c
ont
rol
ler
,
as
s
hown
i
n Fi
g5.
16.
If
t
he
ma
ximum
v
alu
e o
f e(
t)
i
s
1
, t
hen
u(
t)
wou
ld
no
rma
ll
y be
l
imi
ted
t
o 1
00
o
r l
ess
. As
a
n e
xamp
le,
c
ons
ide
r t
he
s
yst
em
i
n
F
ig
5.1
6
wi
t
h a
P
ID
c
ont
r
ol
le
rG(
s)=1
/s
(s+1
),a
nd
t
he
n
ece
ss
ayp
r
r
ei
fl
t
er
Gp
(s
)
t
o
ac
hi
ev
e
I
TAE
p
erf
orma
nce.
If
we
s
el
ect
叭 =1
0,
2
0,
a
nd
40,
th
e ma
ximu
m va
lue
o
f u
(t)
i
s
as
r
ecor
ded
i
n
T
abl
e
5.
4.
I
f
we
wi
sh
t
o
li
mit
u(
t)
t
o
a
ma
ximu
m e
qua
l
to
1
00,
we
n
eed
t
o
li
mit
叭to
1
6.
Th
us
we
a
re
imi
l te
d i
n t
he
s
ett
li
ng
t
i
me
we
c
an
a
chi
eve
.
u(t
)
maxi
mum for
R(s
)=1/
s 35 135 550
Setli
ng
t
ime
(
seconds) 0.
9 0.5 0.3
Letu
s c
ons
ide
r t
he
d
esi
gn
o
f a
PI
D c
omp
ens
ator
us
ing
f
req
uenc
y r
esp
ons
e t
ech
niq
ues
f
or
a
s
yst
em wi
th
a
t
ime
d
ela
y s
o t
hat
天讼-
Ts
G(
s) (
5.48)
(
Ts+1
)
Th
is
t
ype
of
p
lant
re
pre
sen
ts
ma
ny
i
ndu
str
ial
p
roc
ess
es
t
hat
i
nco
rpo
rat
e
a
t
i
me
d
ela
y.
We
us
e a
PI
D
compensa
tor
t
o i
ntr
oduce
t
wo
equal
zer
os
so
t
hat
G
,(
・
)一K
2
(
ar
s
+l
)'
. s
(
5.49)
The
d
esi
gn
me
tho
d i
s
as
f
oll
ows
:
1
. Pl
ot
t
he
u
nco
mpe
nsa
ted
Bo
de
d
iag
ram
f
or
K2G(
s)
/
s
wi
t
h ag
a
in
KZ
t
hat
s
at
is
ie
f s
t
he
s
tea
dy-
stat
e e
rror
r
equi
reme
nt.
2.
Pl
ace
t
he
t
wo
e
qua
l
ze
ros
a
t
or
n
ear
t
he
c
ros
sov
er
f
requ
enc
y,
(
0,.
3.
Tes
t t
he
r
esul
ts
a
nd
a
djus
t
Kz
or
t
he
zer
o l
ocat
ions,
if
nec
ess
ar
y.
EXAMPLE
5.10
PID
CONTROL
OF
A SYSTEM
WITH
A DELAY
Co
nsi
der
t
he
s
yst
em
o
f Fi
g5.
16
wh
en
143
天e-
0i
-
s
G(
s)= (
5.50)
(
OAS+1
)
wh
ere
K=20
i
s
se
lec
ted
t
o
ac
hie
ve
a
s
mal
l
st
eady
-st
at
e
er
ror
f
or
a
s
te
p i
npu
t,
a
nd
wh
ere
G,
(s
)=1・
We
wa
nt
a
n
ov
ers
hoo
t
to
a
s
te
p
in
put
of
l
es
s
th
an
5
%.
Pl
ot
ti
ng
t
he
Bo
de
di
agr
am
f
orG(
jo)
),we
f
in
d t
hat
t
he
u
nco
mpe
nsa
ted
s
yst
em
h
as
a
n
ega
iv
t e
p
hse
a ma
rgi
n a
nd
th
at
t
h e
s
yst
em
is
u
nst
abl
e.
We
wi
ll
us
e a
PI
D cont
rol
l
er
of
t
he
for
m of
Egn.
(
5.4
9)
t
o a
tt
ain
a
de
sir
abl
e p
has
e ma
rgi
n of
7
00
.
The
n
th
e l
oop
t
ra
nsf
er
f
unc
i
ton
i
s
2
0e一,
‘(
:,
s+1
)z
GG,
(
s) (
5.51)
s
(O.
ls+1
)
wh
ere
K2
K=20.
We
p
lot
t
he
Bo
de
d
ia
gra
m wi
t
hou
t t
he
t
wo
z
ero
s,
a
s s
hown
i
n
Fi
g5.
18.
The
p
hse
a ma
rg
in
i
s一3
20,
and
t
he
s
ys
te
m i
s
un
st
ab
le
p
ri
or
t
o
th
e
in
tr
odu
ci
to
n
of
th
e
ze
ro
s.
S
inc
e we
ha
ve
i
ntr
odu
ced
ap
ole
a
t t
he
o
rig
in
b
eca
use
o
f t
he
i
nte
na
r l
t
erm
in
t
he
PI7
c
omp
ens
at
or
,
we
ma
y r
edu
ce
t
he
g
ai
n K2
K b
eca
use
e
,,
i
s
no
w z
ero
.
We
p
lac
e
th
e t
wo
z
ero
s a
t
o
r n
ear
t
he
c
ros
sov
er,叭 =1
1.We
c
hoo
se
t
o s
et
T
l=0.
06
s
o t
ha
t t
he
t
wo
z
ero
s a
re
s
et
a
t
C
o=1
6.7,
Als
o,
we
r
edu
ce
t
he
g
ain
t
o
K2
K=4.
5.Th
en
we
o
bta
in
t
he
re
fque
ncy
r
es
pon
se
s
hown
i
n Fi
g.5.
19,
wher
e
4.
5(
0.
06s+1
)
z
e
0”
-
GG,
(s
) (
5.52)
s(
O.l
s+1
)
、、  ̄ }
、、 ,
口口口川 口口口
日口口}}
} 口口
巨『口琳 L 口巨
dB
口口口日} {
、、
口 L 口口
口厂厂}1
}日} 洲s〔[
口口口}{
{口{ 口口
、‘
口『「}}
}日} 口仁
....口. 门门...曰
日日尸}
... 舀二 【
一
}
}口
,
口
I!
一份一 「 一 . 1 , 匕1
曰日卜吸琳别口 et
;_
}
!}! 泛‘
’
」{}}
日日日{
}} 洲阅口
六 )
) 门口口}
}} 口N、
-,
川
口口}
}} 口口
}
_
」口口!
}} I 」口[
一{
10 必‘ 100
fl)
Fi
g.
5
.18
Bo
de
d
iag
ram
f
or
G(
s)/
s f
or
Ex
amp
le
5.
10
144
.
,二
J
口
,口‘
、‘、
、、、
、知‘
d8 , 、、、
、、、%
、、、
_)
1 、、
,、‘
} 洲
、‘
-120
肠卜、 场,、
1
园  ̄.弓
匕
} 日
、卜
-140
口,
. 10
CO
F
ig.
5.
19
Bo
de
d
ia
gra
m f
or
GcG(
s)
f
or
E
xamp
le
5
.10
T
he
n
ew
c
ro
ss
ove
r
r
fe
que
ncy
i
s
C
O,=4
.5,
and
t
he
p
has
e
ma
rg
in
i
s
7
00.
The
s
t
ep
r
es
pon
se
o
f
t
his
sys
tem
has
no
ove
rshoo
t a
nd
has
a
set
t
in
l g
t
ime
(
2% cr
it
erio
n)
of
0.8
0 sec
ond.
Th
is
r
espons
e
s
ati
sie
f s
the
r
equi
rement
s.
Howeve
r,
i
f we
wan
ted
t
o a
dju
st
the s
yst
em fu
rther
,
we
cou
ld r
ais
e
K2
K t
o 1
0 a
nd
a
chi
eve
a
s
ome
wha
t
fa
st
er
r
espo
nse
wi
th
a
n o
ver
sho
ot
o
f
le
ss
t
han
5%.
A s
a ina
f l
consi
dera
ti
on
of
th
e des
ign
of
r
obust
c
ontr
ol
s
yst
ems
usi
ng a
PI
D cont
rol
le
r,
we
t
un
r
t
o a
n s
-plan
t r
oo t
l
ocus
met
hod.
Th
is
desi
gn
appr
oac
h may
be
s
impl
y s
tat
ed
as
fo
ll
ows:
1
.
Pl
ace
t
he
p
ole
s
and
z
ero
s
of
G(
s)
/
s
o
n t
he
s
-pl
ane
.
2
.
Se
le
ct
a
l
oca
ti
on
f
or
t
he
z
ero
s o
f G,
(s
) t
hat
wi
l
l r
es
ul
t
in
a
n a
cce
pta
ble
r
oot
l
ocu
s a
nd
s
uit
abl
e domi
nant
r
oot
s.
3
.
Te
st
t
he
t
ransi
ent
r
espon
se
of
the
co
mp en
sate
d s
yst
em
and
it
era
te
s
tep
2,
if
ne
cessa
ry.
Th
e r
oot
l
o c
us de
sign
meth
od is
i
llu
str
ate
d by
the
f
oll
owing
de
sign
ex
amp l
e f
or
an
a
irc
raf
t
aut
opi
lot
.
5.
8 DESIGN
EXAMPLE:
AI
RCRAFT
AUTOPI
LOT
A typ
ical
a
irc
raf
t a
uto
pil
ot
cont
rol
sys
tem
con
sis
ts
of el
ect
ri
cal
,
me c
hani
cal
, a
nd h
ydr
aul
ic
de
vic
es
tha
t
move t
he
la
f ps,
ele
vato
rs,
f
uel
-lo
f w
cont
rol
le
rs,
a
nd ot
her
c
ompo n
ent
s t
hat
c
aus
e t
he
a
ir
cra
ft
t
o va
ry
i
ts
li
fght
.
Se
nso
rs
pr
ovi
de
i
nfo
rma
ti
on
on
v
elo
cit
y,
he
adi
ng,
r
at
e o
f r
ota
ti
on,
a
nd
ot
her
li
fght
dat
a.
Thi
s
inf
ormati
on
is
c
ombine
d wit
h t
h e
d
e s
ir
ed li
fght
c
har
acte
ris
ti
cs
(co
mma nds
)
e
lect
ron
ica
ll
y ava
il
abl
e t
o th
e aut
opi
lot
.
The a
utopi
lot
shoul
d be
abl
e t
o ly
f t
he a
irc
rat
f o
n a
h
ead
ing
a
nd
un
der
c
ond
it
io
ns
s
et
b
y t
he
p
il
ot
.
The
c
omma
nd
o
te
f n
c
ons
is
ts
o
f a
pr
ede
ter
min
ed
he
adi
ng.
De
sign
oft
en
foc
use
s on
af
orwa
rd-
movi
ng
air
cra
ft
th
at
moves
somewha
t u
p o
r down
wi
tho
ut
movi
ng r
ight
o
r l
et
f a
nd wi
tho
ut
rol
li
ng
(
rot
at
ing
the
win
gti
ps)
. S
uch as
tu勿is ca
ll
ed
pi
tch
axi
s des
ign.
The
a
irc
raf
ti
s
repr
ese
nted
by
a
pl
ant
145
K
G(
s)= (
5.53)
(
s+1
/z)
(S
I+2
杏,。,
S+。矛
)
wh
ere
Z
i
s
th
e t
i
me
c
ons
ta
nt
o
f t
he
a
ct
uat
or
.
Le
t:=1
/
4,
O
)l=2、a
nd杏,=I
/2.Th
en
t
he
s
-
pl
ant
p
lot
h
as
t
wo
c
ompl
ex
p
ole
s-a
p
ole
a
t
th
e o
rig
in
a
nd
a
po
le
a
t
s=-4,
a
s
sh
own
i
n
Fi
g5.
20.
Th
e c
omp
lex
p
ole
s,
r
epr
ese
nti
ng
t
he
a
irc
raf
t
dyn
ami
cs,
ca
n v
ary
wi
thi
n t
he
da
she
d-l
ine
bo
x s
hown
i
n
th
e f
igu
re.
We
th
en
c
hoo
se
t
he
z
ero
s o
f
th
e c
ont
rol
le
r
as
s=一1
.3士J2,
as
s
hown.
We
se
lec
t
th
e
g
ai
n
K。
。t
ha
t
t
he
ro
ot
s
r,
an
d }
z
a
re
co
mpl
e
x
wi
t
h。}
o
f
1
/
涯.T
he
o
t
her
ro
ot
s。a
nd
r
,
l
ie
ver
y near
t
he
z
eros
.
The
ref
ore
t
he
c
los
ed-
loop
t
rans
fer
f
unct
ion
i
s
appr
oxi
mat
ely
_、 C
)_2 5
I'
M — 二 一 一-
一一 一 一’ (
5.5
4)
s
`+2枷。
:+叫 s
`+3
.16
s+5-
s
i
nce
。。=万 ,
an
d=1
/万.
The
r
es
ul
t
in
g
re
sp
ons
e
t
o。s
t
ep
i
np
ut
h
as
a
n
ov
er
sh
oo
t
of
4.
5%
a
nd
a
s
ett
i
lng
t
i
me
(
2%
c
rit
eri
on)
o
f 2.
5
se
con
ds,
as
e
xpe
cte
d.
Fi
g.
5.
20
Root
l
ocu
s f
or
air
cra
ft
aut
opi
lot
Th
e c
omp
lex
pol
es
ca
n v
ay
r wi
thi
n t
he
dashe
d-l
ine
box
(
in
c
olo
r).
5.
9 THE DESIGN OF A SPACE TELESCOPE CONTROL SYSTEM
Sc
i
ent
is
ts
ha
ve
p
ropo
sed
t
he
ope
rat
io
n o
f a
s
pac
e v
ehi
cl
e
as
a
s
pac
e-b
ase
d r
ese
arch
l
abo
rat
oy
r
an
d t
est
bed f
or
equi
pment
to
be us
ed
on a
manne
d s
pace
s
tat
ion.
The
indust
ri
al
s
pace
fac
ili
ty
(I
S日
woul
d remain i
n as
pace,
an
d the
ast
ron
aut
s woul
d be
abl
e to us
e i
t onl
y when
the
shutt
le
is
at
tac
hed.
The I
SF wil
l
be t
he
fir
st
per
manent
,
human-o
per
ate
d comme r
cial
s
pace
fa
cil
it
y des
igne
d
f
or
R&D, t
est
ing,
a
nd ev
entual
ly
,
proc
essi
ng
in
t
he s
pace
e
nvir
o n
me nt
.
We wil
l
cons
ide
r a
n e
xper
iment
oper
ate
d i
n s
pace
but
con
tr
oll
ed
rom
f Ear
th.
The
goa
l i
s
to
ma
nipu
lat
e
and
posi
ti
on
a s
ma l
l
te
lesc
ope
to
acc
ura
tel
y po
int
at
a
pl
ane
t.
Th
e goal
i
s
to
have
a
s
tea
dy-
sta
te
e
rro
r e
qua
l t
o z
ero
, whi
l
e ma
int
ai
nin
g a
f
ast
r
esp
ons
e t
o a
s
tep
wi
th
a
n ov
ers
hoo
t of
146
l
ess
t
han
5%.
The
actu
ator
c
hose
n i
s a
l
ow power
act
uat
or,
and t
he
mo del
o
f t
he
combi
ned ac
tuat
or
a
nd t
ele
scope
i
s
shown i
n F
ig5.
21.
The c
omma nd
si
gnal
i
s r
ecei
ved
f r
om an
Eart
h s
tat
ion
wit
h a
d
ela
y of 7
r
/
1
6 sec
onds.
A s
ens
or
will
meas
u r
e t
he
poi
nti
ng
dir
ecti
on of
t
he t
el
escop
e ac
curat
ely
Ho
wev
er,
th
is
me
asu
reme
nt
i
s
re
lay
ed
b
ack
t
o
Ea
rt
h
wi
t
h a
d
ela
y
of
7
s/
16
s
eco
nds
.
.
斤oi
n Ear
th
s
t
rat
iou
Fi
g.
5.
21
Mo
del
o
f a
l
ow-
powe
r a
ctu
ato
r a
nd
t
ele
sco
pe
Th
us
t
he
t
ota
l
tr
ans
fer
f
unc
ti
on
o
f t
he
t
el
esc
ope
,
ac
tua
tor
,
se
nso
r,
a
nd
r
ound
-tr
ip
de
lay
(
Fig
5.2
2)
i
s
巴-s
ic
/
8
G(
s)= (
5.55)
(
s+1
)
2’
大(
s)+
Desi
red.
 ̄ ̄,
tele
1
1110l
e
co
pe
Fi
g.
5.
22
Fe
edb
ack
c
ont
rol
s
yst
em
f
or
t
he
t
ele
sco
pe
e
xpe
rime
nt
We
p
rop
ose
a
c
ont
rol
le
r t
hat
i
s
th
e t
hre
e-t
erm
c
ont
rol
ler
d
esc
rib
ed
i
n Se
cti
on
5.
6,
wh
ere
_ 、 _
_ K} __ K3
S2+K,
s+K2
U,
(
s)=K,+一
+K3 s= (
5.56)
s
The
u
se
o
f on
ly
t
he
p
rop
ort
ion
al
t
erm
wil
l
no
t b
e a
cce
pta
ble
s
inc
e we
r
equ
ire
a
st
ead
y-s
tat
e
e
rr
or
o
f z
ero
f
or
a
s
te
p i
npu
t.
Th
us
we
mu
st
us
e a
f
ini
te
v
alu
e o
f
K2,a
nd
he
nce
we
ma
y e
lec
t
to
us
e e
ith
er
p
ropo
rti
ona
l p
lus
i
nte
gra
l c
ont
rol
(
PI)
o
r p
rop
ort
i
ona
l p
lus
i
nte
gra
l pl
us
de
riv
ati
ve
cont
rol
(
PID)
.
We
wi
l
l f
ir
st仰 PI
c
ont
rol
,
so
t
hat
_
__ K,
K.
s+K,
G,
(s
)=K,+兰 =竺t
.竺. (
5.
57
)
o.- -- _
___ L ___ _ _ _
__ 」 ,_ --sT , 。
Si
nce
we
nave
a
pur
e oei
ay,
e ,we
us
e t
he
f
requenc
y r
espons
e me
tnoas
f
or
des
ignpr
oce
ss.
Thus
we
wi
ll
t
ran
sla
te
t
he
ov
ers
hoo
t s
pec
ifi
cat
ion
t
o t
he
f
requ
enc
y d
oma
in.
I
f we
h
avetwo
domi
nant
c
har
act
er
is
ti
c
ro
ots
,
th
e o
ver
sho
ot
t
o
a
s
te
p i
s
5%
wh
eng=0.
7,o
r
th
e p
has
e
ma
rgi
n
re
qui
re
men
t
i
s a
bout
700.
I
f
we
c
hoos
e Kt
=0.
022
and
KZ
=0
.22
,
we
h
ave
0.
22(
0.1
s+1
)e一
,8
/
n
G,
(
s)G(
s) (
5.58)
s
(s+1
)2
147
20
,、
!
} 日
日
、、 、
一 毛
一- ̄,‘
}
} -
,;
1
,
!
!
}
“、、
、 ,‘
餐,.
洲
可
之,
卜
只
!}
、
,,!;  ̄、、、 、
、
、、
-4() \
、
}!
,、
沙哎吧
、 ‘ 、.
.
卜
日}
、 、
、、
-60
}}\
丈
利
}
} 、
一25
7
.1
10-
1 1口0 11
J
I
Fi
g.5
.23
Bo
de
d
iag
ram
f
or
t
he
s
yst
em
wi
th
t
he
PIcontrol
ler
a
nd t
he
Bode
d
iag
ram i
s
sho
wn i
n F
ig5.
23.
The
loc
ati
on
o
f t
he
zer
o a
t s二1 0
wa
s chos
en
t
o a
dd
a
phas
e l
ead
a
ngl
e i
n or
der
to
a
tain
th
e de
sir
ed
pha
se
margi
n.
An it
era
ti
ve
p
roc
edure
yi
el
ds a
s
eri
es
of
t
ri
al
s f
or
K,
an
d K2
u
nti
l
th
e de
si
re
d ph
ase
ma
rgi
n i
s a
chi
eve
d.
No
te
t
hat
we
h
ave
a
chi
eve
d a
ph
se
a ma
rgi
n
of
a
bou
t
63'.
The
a
ctu
al
s
tep
r
espo
nse
wa
s p
lote
d,
a
nd
i
t
is
d
ete
rmi
ned
t
hat
t
he
o
ver
sho
ot
i
s 4.
7%
wit
h as
ett
li
ng
t
ime
(
2%
cri
te
rio
n)
o
f 1
6 s
eco
nds
, s
a r
eco
rde
d i
n
Tabl
e5.
5.
Th
e p
rop
ort
ion
al
p
lus
i
nte
gra
l
pl
us
de
riv
ati
ve
c
ont
rol
le
r
is
K,
s+K2+K3
s2
G,
(
s)= (
5.59)
s
We
now
h
ave
t
hre
e p
ara
met
ers
t
o va
y
r t
o a
chi
eve
t
he
d
esi
red
p
has
e ma
rgi
n.
I
f we
s
ele
ct,
a
te
f r
s
ome
i
t
era
ti
on
,
Ki=0.
8,
K2=0
.5
a
nd
K3=1
0"3
,we
o
bt
ai
n a
p
has
e ma
rg
in
o
f 6
40.T
he
pe
rce
nta
ge
ov
ers
hoo
t i
s 3
.7%,
a
nd
t
he
s
ett
i
lng
t
ime
(
2%
cit
reri
on)
i
s 5
.8
s
econ
ds.
Pe
rha
ps
t
he
e
asi
est
wa
y t
o
se
lec
t
th
e g
ain
c
ons
ta
nts
i
s
to
l
et
K3
be
a
s
mal
l,
but
no
nze
ro,
nu
mbe
r
in
it
ia
ll
y
an
d
Ta
ble
5
.5
S
tep
Re
spo
nse
o
f t
he
Sp
ace
Te
les
cope
f
or
Two
Co
ntr
oll
ers
S
tea
dy-
st
ate
Pe
rce
nt
S
ett
li
ng
Ti
me
Er
ror
Ove
rshoot
(
seconds)
PI
cont
rol
ler 0 4.
7 16.
0
PI
D cont
rol
ler 0 3.
7 5.
8
K,
=K2
=0,
th
en
p
lo
t
t
he
f
r
equ
en
cy
r
es
pon
se
.
I
n
th
is
ca
se
,
we
ch
oos
e K3=1
0-3
a
nd
o
bt
ai
n
a
Bo
de
p
lot
.
We
t
hen
u
se
K,“K2
a
nd
i
te
rat
e
to
o
bta
in
t
he
a
ppr
opr
ia
te
v
alu
es
o
f t
hes
e u
nsp
eci
fi
ed
148
gam s
The
perf
ormanc
e o
f t
he Pl
-an
d t
he
PI
D-compens
ate
d s
yst
ems
is
r
eco
rde
d i
n Ta
ble
5
.5.
The
PI
D
c
ontro
lle
r i
s
the
most
de
sir
abl
e,
sin
ce
i
t
pr
ovide
s a
s
hote
r r
se
tt
in
l g
t
ime.
5.
10
THE
DESIGN
OF
A ROBUST
BOBBI
N DRIVE
Monoil
f a
ment
nyl
on i
s
produ
c e
d by
an
ext
rusi
on
pr
oces
s
that
o
utp
uts
il
f a
me n
t a
t
a c
ons
tan
t r
ate
.
The
prod
uct
is
wound
ont
o a
bobbi
n t
hat
rot
at
es
at
a
maxi
mum spee
d of
2
000 r
pm. The
t
ens
ion
in
th
e f
il
ament
must
be
he
ld
between
0.2
an
d 0.
6 p
ound
to
e
nsure
t
hat
i
t
is
no
t s
tr
etch
ed.
The
wi
nding
di
amet
er
vari
es
bet
ween
2 t
o 4
i
nches.
The
fil
ament
is
l
aid
ont
o t
he bobbi
n b
y a
ball
s
crew-dri
ven
arm th
at
osci
ll
at
es b
ack a
nd f
oth
r a
t
con
stant
s
peed,
as
sho
wn i
n Fig.5.
24
a).
Th e
ar
m mus t
r
evers
e r
apidl
y at
t
he
end
of t
he
mov e
. The
r
equire
d bal
lsc
rew s
peed
is 60rpm.
The p
rime r
equire
me nt
of
the
bobbin
dri
ve i
s t
o pr
ovide a
con
trol
led
te
nsion.
Since
t
he win d
ing
di
ameter
va
rie
s by 2
to
1,
t
he ten
sio
n wil
l
fal
l
by 50
% ro
f m
st
art
t
o inis
f h.
Th
e c
ont
rol
s
yst
em
wil
l
ha
ve
a
s
yst
em
s
tru
ctur
e a
s s
hown
i
n Fi
g.5.
24
b)
,
fo
r wh
ich
we
s
ele
ct
a
PI
D c
ont
ro
ll
er
.
Th
e p
ara
met
er
v
ari
at
io
ns
a
re
1
.5‘Km‘2.
5,a
nd
3‘P‘5
wi
t
h t
he
n
omi
nal
c
ond
it
io
ns
K,
n=2,a
nd
p=4.F
urt
he
rmo
re
a
t
hi
rd
p
ole
a
t
s=
-50
h
as
b
een
o
mit
ed
ro
f m
t
he
mo
del
.
Th
e r
equ
ire
men
ts
a
re
a
n o
ver
sho
ot
l
ess
t
han
2.
5%
and
a
s
etl
ing
t
ime
(
2%
c
rit
eri
on)
l
ess
t
han
0.
4
se
con
d.
Th
e ma
gni
t
ude
o
f u(
t)
mus
t
be
l
ess
t
han
1
00.
召r
rbbr
u
”才
口介,r
,
Il
}+
+u+
jil
nrac
}r
u
H(
s)
Fi
g.5.
24
Mon
ofil
ame
nt
b
obbi
n
wi
nde
r
Us
ing
a
PI
D c
ont
rol
l
er,
t
he
I
TAE
d
esi
gn,
a
nd
t
he
n
omi
nal
p
ara
met
er
s,
we
d
ete
rmi
ne
0
),f
r
om
t
he
s
etl
i
ng
t
i
me
r
equ
ir
eme
nt.
Si
nce
we
e
xpe
ct
t
hat
杏二0.
8,we
u
se
149
4
T =一 — < 0.
4.
.8(
.
U,
0
We
s
el
ec
t
Cc
)=2
3
as
th
e
ma
xi
mum
a
ll
owa
bl
e
f
or
lu
l<1
00.
The
n
fo
r
G
。(
,)
=K,
+
K
z
+K
3s s
we
o
bta
in
K,=5
68.
68,Kz=6
083.
5
an
d K3=18.
13.
Usi
ng
t
he
a
ppr
opr
ia
te
p
ref
il
te
r,
t
he
r
espons
e i
s as
r
ecor
de nTable5.
d
i 6.
The syst
em dos
e not
ofe
r r
obus
t per
for
manc
e si
nce
t
he
o
ver
sho
ot
r
equ
ire
men
t
is
n
otsatisfi whent
ed he
wo
rst
-c
ase
pa
rame
ter
s
ar
e
co
nsi
der
ed.
自月...t』万..
了
‘
、
Set
tl
ing Ti
me
‘
矛
Paramet
ers
Per
cent
t、一 ‘
.了 } 、、 产 一训州
一 价‘
...
over (
Seconds
shoot )
1
.‘几r....月
了.、  ̄
』口r
.
、 . , 比以
- ̄
、.,  ̄ .,....几- , 护, .. , .. .‘  ̄
, ,‘. ‘
. , ..,.‘ 月,.. . .目目 诵  ̄ 甲 一 - 一 ‘ ̄ ,., ̄ 一 . 一 一 一 - .‘‘响 ..勺.. -
. 一 ,‘两 .... . .., .中‘声. ...., ,.勺 ..目.....
. 网 ...
. ‘..‘.
,.,.
....
...
...- ...
...
....
..‘  ̄ 一 ,‘. . ...‘..目‘.曰....
L洲
Nomi
nal 1.
96% 0.
318
Km=
=2
par
amet
ers
p=4
W ors
t-case 7.
48% 0.
375
Km二1
.5
pa
ramet
ers
p=3
Nominal
parameter
s 9.
82% 0.
732
and
a
dded thir
d pol
e
at
s=-
50
a
We l
so
exa
mine
the
per
for
manc
e of
t
he
sys
tem
wit
h t
he
no
min
al
p
ara
met
ers
b
ut
wi
th
t
he
u
nmo
del
ed
p
ole
adde
d,
s
o t
hat
t
he
ac
tua
l
pl
ant
is
2(
50)
G(
s)= (
5.60)
s(
s +
50)
(s
+
4)
(
Not
e
th
at
t
he
d
o g
ain
,
li
ms,
o s
G(s
),r
ema
ins
0
.5.
)
Th
e r
es
pon
se
o
f
th
e P
ID
c
ont
r
oll
er
wi
t
h t
he
a
dded
pol
e i
s
recorde
d i
n
Table
5.
6.
Aga
in
t
he s
yst
em f
ail
s
the
re
qui
rement
of
robu
st
per
for
mance
.
We
need t
o a d
just
t
he sy
stem s
o t
hat
t
he per
for
manc e
wit
h t
he wor
st-
cse
a pa
ramete
rs
i
s
a
cce
ptabl
e.
Examine
the
ro
ot
locu
s f
or
t
he
no
minal
par
ameters
s
hown
in
Fi
g.5.
25.
I
nse
rt
a
csc
a ad
e
g
ain
K p
ri
or
t
o
G.
(s
) s
o t
hat
we
h
ave
KG,
G(
s)・
The
n t
he
r
oot
s
fo
r K=
=1
a
nd
K=3
a
re
s
hown
o
n t
he
l
ocus
.
Si
nce
t
he
wo
rst
-c
ase
r
espo
nse
o
ccu
rs
wh
en
t
he
mo
tor
c
ons
ta
nt
Km
dr
ops
t
o
1
.5,
we
u
se
t
he
c
asc
ade
g
ain
K=3
t
o mo
ve
t
he
r
oot
s
to
t
he
l
ef
t
on
t
he
s-
p
lan
t.
Th
en
wh
en
t
he
g
ain
Km
d
rop
s t
o 1
.5
,
th
e r
oot
s
st
i
ll
a
re
i
n t
he
d
esi
re
d r
egi
on.
Th
e r
es
pon
se
o
f t
he
s
yst
em
wi
th
K=3
i
s
re
cor
ded
i
n Ta
ble
5
.7
f
or
t
he
no
min
al
a
nd
wo
rst
-ca
se
c
ond
iti
ons
,
an
d wi
th
t
he
a
dde
d p
ole
.
Th
is
s
yst
em
mee
ts
a
ll
t
he
s
pec
iic
fati
ons
.
Th
is
a
ppr
oac
h u
ses
a
c
asc
ade
g
ain
t
hat
,
wh
en
a
dju
ste
d
c
orr
ect
ly,
wil
l
dr
ive
t
he do
minan
t r
oot
s
nea
r t
he
compl
e x
z
eros
of
th
e PI
D cont
rol
le
r.
Then
wh
en
t
he
wor
st
pa
rame
ter
c
hange
oc
curs,
the
s
yst
em wi
ll
s
ti
ll
mai
ntai
n
th
e r
equi
red
per
for
ma nc
e.
“1
u
'
ni
r
n
.
ra
l邸
,
东
t
ft
I
+4
)
左二 1
---一“ 入一-
1
i
'
7}
‘.J 一
口‘. 口分
、 } }
I - 一 K - 1—
\
(6二3
峨址曰 \
/
一不
、J
二 考
沪
,撼‘
亡七飞馨七勺
目
月
八 』.‘
V V
乙r气 ̄
J、 --
一一洲
心D
ou
bl
e
r 一-阔 ̄ ̄子、 、尸
”‘
“
尸'
6 /
(、、、 卜吸. .一 人 _一 夕声沪口
产
洲
/
Wor
1
st
1
1
S
口c
as
e C
闪一s
(}
+万
-
30
-
25
-
:0
-
15 一
Ill
-
; 子 10
Ren了‘
“刀、
Fi
g.5
.25
Ro
ot
l
ocus
f
or
t
he
n
orma
l
ca
se
a
nd
t
he
wo
rst
ca
se
f
or
K =l
a
n
d K=3
Ta
ble
5.
7
Re
spo
nse
o
f t
he
Bo
bbi
n Dr
ive
S
yst
em
f
or
a
Un
it
St
ep
wi
t
h Ad
dit
iona
l
Cas
cade
Gai
n K-3
Pe
rce
nt
Se
tt
li
ng
Ti
me
Ove
rshoot
(s
econds
)
No
min
al
p
ara
met
ers 0.12% 0.
218
Wor
st-
ca
se
pa
rame
ter
s 0.
47% 0.
214
No
mina
l
pa
rame
ter
s
an
d t
hir
d po
le 0.
50% 0.
242
5.
11
THE
ROBUST
INTERNAL
M ODEL
CONTROL
SYSTEM
Th
e
i
nte
na
r l model
co
ntr
ol
sys
tem i
s s
hown
in
Fi
g5.26
. We
now r
econsi
der
t
he us
e of
t
he
i
nte
rnal
mode
l de
sig
n wi
th
spe
cia
l a
tte
nti
on
to
r
obus
t s
yst
em pe
rfo
rmanc
e .
The
i
nte
rnal
mod
el
p
ri
nci
pl
e
st
at
es
t
ha
t
if
G,
(
s)G(
s)
c
ont
ai
ns
R(
s),
the
n y
(t)
wi
ltr
ack
r
(t)
as
ymp
toi
t
cal
l
y (
i
n t
he
s
tea
dy
st
ate
),
a
n d
t
he
tra
cking
i
s
ro
bust
.
Exa
mi
ning
the
sys
tem of
Fi
gS.
26,
we
n
ote
t
hat
f
or
l
owe
r-o
rde
r p
lan
ts,
s
tat
e v
ari
abl
e f
eedb
ack
wi
l
l
no
t
be
r
equ
ir
ed,
an
d a
s
ui
ta
ble
G,
(
s)
c
an
b
e o
bta
ine
d.
Ho
wev
er,
wi
t
h h
igh
er-
or
der
s
yst
ems
,
t
he
f
eedb
ack
o
f a
ll
s
tat
e
va
ria
ble
s ma
y be
r
equ
ire
d.
Co
nsi
der
a
s
imp
le
s
yst
em
wi
th
G(
s)=1/s,f
or
wh
ich
we
s
eek
a
r
amp
r
esp
ons
e wi
t
h a
s
tea
dy-
st
ate
e
rro
r of
z
ero.
A
PI
c
ont
rol
ler
i
s s
ufi
ci
ent
, a
nd
we
l
et
K =0(
no
s
tat
e v
ari
abl
e
f
eedback)
.
Then
we
have
凡s+Kz
G
,G(
s
)・(
K,
+K2
)二- s2 s
s
(
5.61)
151
K61+  ̄ F
.0
(
今1s
1 而几-r
F
-
}
(s
-
)
K 卜J 1
.
Fi
g.5.
26
The
i
nter
nal
model
cont
rol
s
yst
em
No
te
t
ha
t
fo
r a
r
amp
,
R(
s)=1
/s2
,wh
ic
h i
s
co
nt
ai
ne
d a
s a
f
ac
to
r o
f E
gn.
(
5.6
1),
a
nd
t
he
cl
osed-
loop
t
rans
fer
f
unc
tion
i
s
(
K,
s
+K2) (
5.62)
T(
s)=
s
2+K,
s+K2
Us
ing
t
he
I
TAE
spe
ciic
fati
ons
f
or
a
r
amp
r
esp
ons
e,
we
r
equ
ir
e
T(
s)=
3.
2o
).
s+耐 (
5.63)
s
2
+3.
2o
on
s+。万
We
s
ele
ct
(
0n
t
o
sa
ti
sf
y
a
s
pec
iic
f a
ti
on
f
or
t
he
s
etl
i
ng
t
ime
.
Fo
r a
s
etl
in
g t
ime
(
2%
c
ri
te
ri
on)
o
f 1
s
eco
nd,
we
se
le
ct叭=
5.
Th
en
we
r
equ
ir
e Kl
=16
a
nd
K2=2
5.Th
e r
es
pon
se
o
f
th
is
s
yst
em
s
etl
es
i
n 1
s
eco
nd
a
nd
t
hen
t
rac
ks
t
he
r
amp
wi
th
z
ero
s
te
ady
-st
at
e
er
ror
.
If
t
his
s
yst
em
(
des
ign
ed
f
or
a
r
amp
i
npu
t)
r
ece
ive
s a
s
tep
i
npu
t,
t
he
es
r p
ons
e h
as
a
n o
ver
sho
ot
o
f 5
% a
nd
a
s
ett
i
lng
t
ime
of
1
.5s
econ
ds.
This
sys
tem i
s ver
y r
obust t
o c
hang
es in
t
he pl
ant
. F
or e
xampl
e, i
f
G(s)=K/s c
hang
es
gai
n s
o t
hat
K shi
ft
s ro
f m
K=1 b
y 15
0%,
t
he
cha
nge
i
n t
he
ramp
r
espo
nse
i
s
ins
ign
ifi
cant
.
FXAMPLES.
11
DESI
GNOFANI
NTERNALMODEL
CONTROLSYSTEM
Co
nsi
de
r
th
e s
yst
em
o
f F
ig.
5.
27
wi
t
h s
ta
te
v
ar
ia
bl
e
fe
edb
ack
a
nd
a
c
omp
ens
at
or
G,
(s
)・We
wi
sh
t
o
tr
ack
a
s
t
ep
i
npu
t
wi
t
h z
ero
s
te
ady
-st
at
e
er
ro
r.
He
re
we
s
el
ec
t
a
P
ID
c
ont
ro
ll
er
f
or
G,
(s
)・
W e
t
hen
have
_,、K3
s2+K,
s+K2
L
Y,
k
s)=— s
a
nd
G(
s)G,
(s
) wi
l
l
co
nta
in
R(
s)=1
/s,
the
i
npu
t
co
mma
nd.
No
te
t
hat
we
f
eed
b
ack
b
oth
s
ta
te
v
ari
abl
es
a
nd
a
dd
t
hes
e a
ddi
t
ion
al
s
ign
al
s
at
f
er
G,
(s
) i
n o
rde
r t
o r
et
ai
n t
he
i
nt
eg
rat
or
i
n
G,
(
s).
Pl
ant
G(
s)
R(
s]
G 仁夕 r】 厂了门 T
-
(S
)
C
}(s
)
些剑 t
i
},
j
S+?
)1X1
P
粼 L
-'
?
Fi
g.
5.
27
An
i
nt
ena
r l
mo
del
co
ntr
ol
wi
t
h s
ta
te
v
ari
abl
e
fe
edb
ack
a
nd
G,
(
s)
152
The
goa
l
is
t
o a
chie
ve a
s
ett
lin
g t
ime
t
o wi
thi
n 21
i(
t
)t
t
he
f
inal
v
alue
in
l
ess
t
han second
and a
d
eadbe
at
r
espo
nse whil
e r
eta
ini
ng ar
o
bust
res
p o
nse Her
e we ass
ume t
hat
tet
h wo pol
es of
G(
s)
c
an
c
han
ge
b
y士5
0%.
Th
en
t
he
wo
rst
-c
ase
c
ond
it
io
n i
s
G
(s
)=一
(
1一
50)
(s+1
)
.
s+0.
On
e d
esi
gn
a
ppr
oac
h i
s
to
de
sig
n t
he
c
ont
rol
f
or
t
his
wo
rst
-ca
se
c
ond
iti
on.
Anot
her
a
ppr
oac
h,
wh
ich
we
u
se
h
ere
, i
s
to
d
esi
gn
f
or
t
he
no
min
al
G(s
) a
nd
o
ne-
hal
f
th
e d
esi
re
d s
et
tl
ing
t
ime
.
Th
en
we
e
xpe
ct
t
o me
et
t
he
s
ett
in
l g
t
i
me
r
equ
ire
men
t a
nd
a
tt
ai
n a
ve
y
r f
st
a ,
hi
ghl
y r
obus
t
sys
tem.
No
te
t
ha
t
t
he
p
ref
i
lt
er
GP
(
s)
i
s
us
ed
t
o
at
t
ai
n
th
e
de
si
re
d
fo
rm
f
or
T(
s)・
Th
e r
esp
ons
e d
esi
red
i
s
de
adb
eat
,
so
us
e a
t
hir
d-o
rde
r
tr
ans
fer
f
unc
tio
n a
s
0
)
n
3
T(
s) 二二 ‘...州..,.,曰闷卜.晌.
(
5.64)
S
3+1
.
9(1
)s
2+2
.2
嵘s+叫’
a
nd
t
he
s
ett
i
lng
t
i 2% c
me
( ri
te
ri
on)
is
TS二4
.04/
o),
,
.Fo
r
a
s
et
ti
l
ng
t
i
me
o
f 1
/2second,
we
use
c
vn
=
¥,
o¥
.
TM'
e
fed
-l
oo
p
tr
ans
fe
r
fu
nct
i
on
o
f
th
e
sy
st
em
o
f
Fi
g.
5.
27
wi
t
h t
he
a
ppr
opr
ia
te
GP
(
s)
i
s
K,
T(
s)二二 . … ‘ (
5.65)
S
3+(
3+K3
+Kb
)S
2+(
2+K,
+Ka
+2Ke
)S
+`7
"
72
We
l
et
KQ=1
0,Kb
=
2,
K,=1
27.
6
K2
,二5
27.
5 a
nd
K3二1
0.3
5.No
te
t
ha
t T(
s)
c
oul
d
b
e a
chi
eve
d
wi
t
h o
the
r
ga
ins
,
i
ncl
udi
ng
Kb=0.
Th
e s
tep
r
espo
nse
o
f t
his
s
yst
em
hs
a a
d
ead
bea
t r
esp
ons
e wi
th
a
n o
ver
sho
ot
o
f 1
.65
% a
nd
a
s
et
tin
lg t
ime
o
f 0.
5 s
eco
nd.
Wh
en
t
he
pl
ant
po
les
of
G(s)
c
han
ge
b
y 15
0%,
t
he
o
ver
sho
ot
c
hang
es t
o 1.
86%,
and
th
e se
tt
in
l g t
ime
b
eco
mes
0.
95
s
eco
nd.
Thi
s
is
a
n o
uts
tan
din
g de
sig
n o
f a
v
ey
r r
obust
,
deadbe
at
r
espons
e s
yst
em.
5.
12 THE DESIGN OF AN ULTRA-
PRECISI
ON DI
AM OND
TURNI
NG
MACHI
NE
Th
e de
sig
n o
f a
n ul
tr
a-pr
eci
si
on
di
amon
d t
uni
r ng
mac
hine
h
as
b
een
s
tud
ied
a
t La
wrenc
e
Li
ver
mor
e Na
ti
ona
l La
bor
ato
ry.
Th
is
ma
chi
ned
s
hape
s o
pti
cal
d
evi
ces
s
uch
a
s mi
rr
ors
wi
t
h
u
ltr
a-h
igh
prec
isi
on
usi
ng
a
di
amond
to
ol
a
s t
he
cut
ti
ng
dev
ice.
In
th
is
di
scussi
on
we wil
l
co
nsi
der
o
nly t
he
z -
axi
s con
tro
l.
Us
ing
re
f que
ncy
resp
ons
e id
ent
iic
f a
ti
on wi
th
s i
nusoi
dal
in
put
to
the
a
ctua
tor
, i
t
was
det
ermi
ned
t
hat
4500
G(
s)二 (
5.66)
(
s+60)
Th
e s
yst
em
c
an
a
ccommo
dat
e h
igh
ga
ins
,
suc
h a
s 45
00,
s
inc
e t
he
i
npu
t c
omma
nd,
r(
t),i
s
a
s
eri
es
o
f s
te
p c
omma
nds
of
v
eys
r
mal
l
ma
gni
t
ude
(
a
ra
f c
ti
on
o
f
a
mi
cro
n).
Th
e s
yst
em
ha
sa
n o
ute
r
153
l
oop
f
or
p
osi
t
ion
f
ee
dba
ck
u
si
ng
a
l
as
er
i
nt
er
fe
rome
te
r
wi
t
h a
n a
ccu
rac
y o
f
0.
1
mi
cr
on(
10-
1m)
.
竺.『
An
i
nne
r f
eed
bac
k l
oop
i
s
al
so
u
sed
f
or
v
elo
cit
y f
eed
bac
k,
a
s s
hown
i
n
Fi
g5.
28.
V(
S)
『
十
朴 弓
t
石U
(S
) (
S
-
)
+ Ve
lo蕊o
〕『
1
7
s
'
1
Y
"
(
S
s
I
)
_
,r介夕
Z卜
刁少
C(
S)
d 争
.f 口 3I“ (矛几
Fi
g.5
.28
Tu
rni
ng
ma
chi
ne
c
ont
rol
s
yst
em
I
t
is
d
esi
re
d t
o
se
lec
t
th
e c
ont
rol
l
ers
,q (
s)
a
nd
Gz
(
s),
to
o
bta
in
a
n ov
erd
amp
ed,
hi
ghl
y
ro
bus
t,
h
igh-
ban
dwi
dt
h s
yst
em.
Th
e r
obu
st
s
yst
em
mus
t
ac
commod
ate
c
han
ges
i
n G(
s)
d
ue
t
o v
ary
ing
l
oad
s,
mat
eri
als,
a
nd cu
tti
ng
re
qui
eme
r n
ts.
Thus
we
see
k l
arge p
has
e margi
n a
nd ga
in
margi
n f
or
t
he
i
nner
a
nd out
er
lo
ops,
and
l
ow
root
sens
itiv
it
y.
Th
e s
pec
ific
ati
ons
ar
e s
umma r
ized
i
n Ta
ble
5.
8.
Ta
ble
5
.8
S
pec
iic
f a
tio
n f
or
Tu
ni
r n
g Ma
chi
ne
Co
ntr
ol
Sy
ste
m
Transf
er
Funct
ion
Spe
ciica
f tion: Ve
loc
it
y,
V(
s)/
U(S) Po
sit
ion
Y(
sYR(
s)
Mi
nimum
bandwi
dth 950rad/
s 95r
ad/
s
S
tea
dy-
st
at
e
er
ror
t
o a
s
tep 0 0
Mi
ni
mum
d
ampi
ng
r
at
io
,杏 0.
8 0.
9
,且
亡J
,
Ma
xi
mum
r
oo
t
s
en
si
t
iv
it
y,
I
S;
( 1.
0
”
O
甲碑 『
}
拼
勺
Mi
nimum
phas
e ma
rgi
n 900
B
们U
‘
钊
曰叼
Mi
ni
mum g
ain
mar
gin 40dB
Si
nce
we
d
esi
re
z
ero
s
tea
dy-
st
at
e
er
ror
f
or
t
he
v
eloc
it
y l
oop,
we
us
e a
v
elo
ciyl
t
oop
c
ont
rol
le
r
G2
(S
)=G3
(s
)G4
(s
),wh
ere
G3
(
S)
i
s a
P
I c
ont
r
oll
er
,
an
d G4
(s)
is
a
l
ead
c
ont
r
oll
er
.
We
ut
iize
l
(
1+K3
s)・—
(
1+K4
s) 八。,
G,
(s
)=认 (
s)G4
(
s)= K}
K3
S all+一 s)
a
a
nd
c
hoo
se
K3=0
.00
532,K4= 0.00272
a
nd
a =
2.95.
We
now
have
(s+18
8) (
s+368)
G2
(
s)=K2
(
s+1
085)
154
Th
e r
oot
l
ocus
f
or(
!2(
s)G(
s)
i
s
sh
own
i
n F
ig.
5.
29.
When
K2=2,
we
h
ave
f
or
t
he
v
elo
cit
y
c
los
ed-
loop
t
rans
fer
f
unc
tion
V
(
s) 9000(
s
+188
)(s+368) 1
04
T2
(
s)= (
5.67)
U
(
s) (
s+2
05)
(s+3
05)
(s+1
04)(
s+1
04)
whi
ch
i al
s a
rge
-ba
ndwi
dth
s
yst
em.
The
a
ctu
al
b
and
wid
th
a
nd
r
oot
s
ens
it
ivi
t
y a
re
s
umma
riz
ed
i
n
Tabl
e 9.No
5. te
t
hat
we
ha
ve
e
xce
ede
d t
he
s
pec
ifi
cat
ions
f
or
t
he
v
elo
ciy
t t
rans
fer
f
unc
ti
on.
61
10
400
1}
之越 卜
200 — 凡i
nc
re
asi
i
ig一
一
科一
户妞三警之闷
、 ,
尹、 入
乙7
・2口0
-400
一‘00
一
1200 一
1000 -
800
-600
-400
- 0
zoo
R翻 了
rt
e i
x
Fi
g.
5.
29
hoo
t
lo
cus
f
or
v
elo
ciy
t l
oo s K2
p
a v
ari
es
Ta
ble
5
.9
De
sign
Re
sul
ts
f
or
Tu
ni
r n
g Ma
chi
neCon
trol
Sy
ste
m
Vel
oci
y
t Pos
iti
on
Tra
nsf
er
Transfer Func
tion,
Achi
eve
d Res
ult Funct
ion Y(
s)/
R(s
)
Cl
ose
d-l
oop
ba
ndwi
dth 4000
rad/
s 1000rad/s
St
ead
y-s
tat
e
er
ror 0
Da
mpi
ng
r
ati
o,
S 1.
0 1.
0
R
oo
t } 0.
92 1.
2
Pha
se
ma
rgi
n 930 8
50
Gai
n mar
gin I
nfi
nit
e 76dB
155
脚动 细脚
‘
一
}
‘
一
}
口口明口、 一
厂
头
夕二屯
帐
护t
i 、- 了.甩二卜.
子
、厂,
7}一 、 厂 乒、
又
奢阳
侧
舀
.二翻曰户 ̄.
,200 护.瓜 ̄ 一 胃
一
・ ?(
JO
一4刀口
-
Ir
00
-
511
1
1 -
411
11
-
JOV
-20U 一
1x/
11
U
t
ort
j(
1
0
!t
eal
l
m i
s
L)
F
ig.
5.
30
Th
e
ro
ot
l
ocu
s
fo
r K,>0
Fo
r a
) o
ver
vie
w a
nd
b
) c
los
e-u
p n
ear
o
rigi
n o
f t
he
s
-pl
ane
We
wi
ll
u
se
a
l
ead
ne
two
rk
f
or
t
he
p
osi
ti
onl
oop
i
s
(
1+KS
s)
凡一a
G,
(s
)=K,
al
l十 s)
a
nd
we
c
hoo
se
a=2.
0
an
d KS=0.
018
5 s
o t
hat
K,
(s+5
4)
G,
(s
)=
(
s+1
08)
We
t
hen
pl
ot
t
he
r
oot
lo
cus
f
or
G,
(s
)・几(
s)・
156
I
f
we
us
e t
he
a
ppr
oxi
mat
e爪(
、)of
Eq
n.
(
5 7
',
),
we
h
ave
t
he
r
oot
l
ocus
o
f F
ig.
5.
30
a)
.
Us
ing
t
he
a
ct
ual
几(s),we
g
et
t
he
cl
os
e-u
p
of
th
e r
oot
l
oc
us
sh
own i
n
Fi
g.
5.
30
b)
.
We
sel
ec
t K2”1000
a
nd a
chi
eve
t
he a
ct
ual
r
es
ult
s
fo
r t
he
t
ot
al
s
yst
em
t
ra
nsf
er
f
unc
ti
on
a
s
r
eco
rde
d i
n Ta
ble
5.
9. The t
ota
l s
yst
em has
a
hi
gh phas
e mar
gin
, a
l
ow
sens
it
ivi
ty,
a
nd
i
s
o
ver
dampe
d wi
th
a
l
arge
ban
dwidth.
This
s
yst
em
is
v
ey
r ro
bust.
5.
13
THE
PSEUDO-
QUANTI
TATI
VE
FEEDBACK
SYSTEM
Qua
nti
ta
ti
ve
f
eed
bac
k t
heo
ry
(
QFT)
us
es
a
c
ont
rol
le
r,
a
s s
hown
i
n F
ig.
5.
31,
t
o a
chi
eve
r
obus
t
p
erf
orma
nce
.
Th
e go
al
i
s
to
a
chi
eve
a
wi
de
b
and
wid
th
f
or
t
he
c
los
e-l
oop
t
rans
fer
f
unc
ti
on
wi
th
a
h
igh
l
oop
g
ain
K.Ty
pic
al
QF
T de
sig
n me
tho
ds
u
se
g
rap
hic
al
a
nd
nu
mer
ic
al
me
tho
ds
i
n
co
njunc
ti
on wit
h t
he Nich
ols
ch
art
.
Gene
ral
ly
QFT
d
esi
gn
s
eeks
h
igh
l
oop
g
ain
a
nd
l
arg
e ph
ase
marg
in
so
th
a t
robust
p e
rfor
mance
i
s
ac
hieve
d.
R
(
s)+ Y(
s)
K 乓了
砂 G付
Fi
g.5.
31
Fe
edba
ck
s
yst
em
I
n t
his
s
ect
i
on
we
p
urs
ue
a
s
imp
le
me
tho
d o
f a
chi
evi
ng
t
he
g
oal
s
of
QFT
wi
th
a
n s-
pla
ne,
roo
t
l
ocu
s
ap
pro
ach
t
o
th
e
se
le
ct
io
n o
f
th
e
ga
in
K
a
nd
t
he
c
omp
ens
at
or
GC
(s
),
Th
is
a
ppr
oac
h,
du
bbe
d ps
eudo
-QFT,
folo
ws
t
hes
e s
teps
:
1
. Pl
ace
t
he
n po
les
a
nd
mz
er
os
o
f G(
s)
o
n t
he
s-
p
la
ne
f
or
t
he
n-
t
h
or
der
G(
s)
.
Al
so
a
dd
a
ny
p
ol
es
o
f G,
(s
).
2
.
S
tar
t
ing
n
ear
t
he
o
ri
gi
n,
p
lac
e t
he
z
ero
s o
f G,
(s
)
imme
dia
tel
y
to
t
he
l
eto
f
f
ea
ch
o
f t
he
(
n-1
) p
ole
s o
n t
he
l
eft
-ha
nd
s
-pl
ane
.
Th
is
l
eav
es
o
ne
po
le
f
ar
t
o t
he
l
eft
of
t
he
l
eft
-ha
nd
s
ide
o
f
t
he
s
-pl
ane.
3.
I
ncr
eas
e
th
e g
ain
K
s
o t
hat
t
he
r
oot
s
of
t
he
c
har
act
er
is
ti
c
equ
ati
on
(
pol
es
o
f
th
e c
los
ed-
lo
op
t
ra
nsf
er
f
unc
ti
on)
ar
e
cl
os
e
to
t
he
z
ero
s
of
G
js
)G(
s)
Thi
s met
hod
i
ntroduces
z
eros
s
o t
hat
al
l
but
one
of
t
he
r
oot
l
oci
e
nd
on
f
ini
te
ze
ros.
I
f t
he
gai
n
K
i
s
su
fic
ie
nt
ly
l
ar
ge,
t
he
n t
he
p
ole
s o
f T
(
s)
a
re
a
lmo
st
e
qua
l
to
t
he
z
ero
s o
f Gc
(
s)G(
s)・
Thi
s
le
ave
s o
ne
po
le
o
f T(
s)
wi
th
a
s
igni
fi
can
t pa
rt
ia
l f
ra
cti
on
r
esi
due
a
nd
t
he
s
yst
em
wi
th
a
ph
ase
ma
rgi
n
of
a
ppr
oxi
mat
ely
9
00
(
act
ual
ly
a
bout
850
).
EXAMPLE
5
.12
DESI
GN
USI
NG
THE
PS
EUDO-
QFT
METHOD
Co
nsi
der
t
he
s
yst
em
o
f Fi
g.5
.31
wi
th
157
G(
s)=
P}
)(s+P2
)
wh
ere
t
he
n
omi
nal
c
ase
i
s p,= a p2= 2
nd k
ith
+50% va
ria
tion.
The
wor
st
ca
se
i
s wi
th
Pt=0
.5
a
nd
p2=1
.We
wi
sh
t
odes
ignt
he
s
yst
em
f
or
z
ero
s
tea
dy-
st
at
e e
rro
r f
or
a
s
tep
i
nput
,
s
o we
us
e t
he
PI
l cont
rol
ler
。,
_、(
s+:
,)(
S+Z
2)
。。t
、曰S,=—
, 夕 一— 、
S
We
t he
n i
nvoke
t he e, wi
int
enal
r model pr
inci
pl th
R(
s)=1
/s
i
nco
rpo
rat
ed
wi
t
hin
Gc
(
s)G(
s).
Usi
ng
s
te
pl,
we
pl
ac
e t
he
p
ole
s
of
Gc
(
s)G(
s)
o
n t
he
s
-p
lan
e,
a
s
sh
own
i
n
Fi
g.
5.
32.
Th
ere
a
re
t
hr
ee
p
ole
s,
a
t
s=0,
一1,
an
d一2,
as
s
hown.
St
ep
2
c
als
f
or
p
lac
ing
a
z
ero
t
o
t
he
l
eto
f
f
t
he
po
le
a
t
th
e o
ri
gin
a
nd
a
t
th
e po
le
a
t
s=一1,
as
s
hown
i
n
Fi
g.
5.
32.
Fi
g.
5.
32
Ro
ot
l
ocu
s f
or
KG,
G(s)
Th
e c
omp
ens
ator
i
s
thu
s
。 ,、 (
s+0.
8)(
s+1.
8) (
5.68)
vc
l
s)
_— .
s
We
s
el
ect
K
=100,s
o t
hat
t
he
r
oot
s
of
t
he
c
har
act
er
is
ti
c
equ
ati
on
a
re
c
los
e t
o t
he
z
ero
s.
Th
e
cl
osed-
loop
t
rans
fer
f
unct
ion
i
s
100
(s+0.80)(s+1.80) 100
T
(
s) (5.
69)
(s+0.798)(s+1.797)(s+1 00.4) ( s+1 00)
Th
i
s clo
sed-
loo
p sy
ste
m provi
desa f
st
a resp
onse and p
oss
esse
s a p
hase
mar
gin of
a
ppr
oxi
ma t
el
y 8
50 .
The
per
formance
issummarized
in
Tabl
e 5.
10.
When
the
worst
-ca
se
condit
i
on s
a
re r
eali
zed(PI=0. 5 and
p “1) ,t
he
pe
rfo
rman
ce
r
emains
158
e
ssent
ial
ly
uncha
nged,
as
s
hown
i
n Tabl
e 5.
10.
Ps
eudo-
QFT
dei
gn
r
esul
ts
i
n ver
y r
obus
t s
yst
ems
.
Tabl
e 5.
10
Per
for
manc
e of
Ps
eudo-
QFT
Des
ign
Percent
overshoot 一旦e
tt
li
ng
T
ime
Nomi
nal
G(
s) 0.
0!% 40m s
Wor
st-
cas
e G(
s) 0.
97% 40ms
5.
14
ROBUST
CONTROL
SYSTEMS
USI
NG
MATLAB
I
n
th
is
s
ect
ion
we
wi
ll
in
vest
iga
te
rob
ust
cont
rol
sy
stems
usi
ng
MATLAB. I
n p
arti
cul
ar
we
will
c
onsi
der
t
he
commonl
y use
d PID c
ontr
oll
er
in t
he f
eedba
ck co
ntr
ol
s
yst
em shown i
n Fi
g.5
.1.
No
ti
ce
t
ha
t
t
he
s
ys
te
m h
sa
a
pr
ef
il
t
er
Gp
(S
)・
The
PI
D c
ont
rol
ler
ha
s t
he
f
orm
K,
s2+K,
s+K,
G,
(S
)二一— s .
No
tic
e t
hat
t
he
PI
D c
ont
rol
l
er
i
s n
ot
a
p
rope
r r
ati
ona
l f
unc
ti
on
(
i.
e.
,
th
e d
egr
ee
o
f t
he
n
ume
rat
or
p
oly
nomi
al
i
s
gr
eat
er
t
han
t
he
de
gre
e o
f t
he
d
eno
min
ato
r po
lyno
mia
l).
The
o
bje
cti
ve
i
s
to
c
hoo
se
t
he
p
ara
met
er
s
K,
,K2
,an
d K3
t
o me
et
t
he
p
erf
or
man
ce
s
pec
if
ic
at
io
ns
a
nd
h
ave
d
esi
r
abl
e
r
obu
stn
ess
p
rop
ert
i
es.
Unfo
rtu
nat
ely
i
t
is
no
t i
mme
dia
tel
y c
lea
r ho
w t
o c
hoo
se
p
ara
met
ers
i
n t
he
PI
D c
ont
rol
l
er
t
o
obt
ai
n c
eta
r i
n
rob
ust
nes
s c
har
act
eri
st
ics
.
An
i
ll
ust
rat
ive
e
xampl
e
wi
ll
s
how
t
hat
i
t
i
s po
ssi
ble
t
o c
hoos
e t
he
pa
rame
ter
s i
te
rat
ive
ly
a
nd
v
eri
fy
t
he
r
obus
tne
ss
b
y s
imul
at
ion.
Us
ing
MATELAB
hel
ps
i
n t
his
p
roc
ess
, s
inc
e t
he
e
nti
re
d
esi
gn
a
nd
s
imu
lat
ion
c
an
b
e me
cha
niz
ed
u
ti
iz
l i
ng
s
cri
pts
a
nd
c
an
e
si
a l
y
be
e
xec
ute
d r
epe
ate
dly
.
EXAMPLE反13
ROBUST
CONTROL
OF
TEMPERATURE
Cons
ider
t
he
f
eedbac
k cont
rol
s
yst
em
i
n Fi
g.5.
1,
whe
re
1
G(
s)=
(
S+c
(
3)2
t
he
n
omi
na
l
va
lu
e
is
co=I
,
an
d GP
(
s)=I
.We
wi
l
l
de
si
gn
a
c
omp
ens
at
or
b
se
a d
o
n c
o=I
an
d
c
hec
k r
obus
tne
ss
b
y s
imu
lat
i
on.
Our
d
esi
gn
s
pec
ifi
cat
ion
s a
re
s
a f
oll
ows
:
1
.
Se
ti
ln
g t
i
me
(
2%
c
ri
te
ri
on)
T,
:
!
-
}0.
5
se
con
d,
a
nd
2.
Op
ti
mum
ITAE
per
for
man
ce
f
or
a
s
te
p i
npu
t.
Fo
r
t
his
des
ign
we
wil
l
not
ut
il
iz
e a
pre
fil
te
r t
o me
et
s
p e
cif
ica
ti
on (
2)
but
wi
ll
i
nst
ead
sh
ow
t
hat
a
cce
pta
ble
per
for
mance
(
i.
e.
,
low
overs
hoot)
ca
n be
o
bta
ined
by
inc
reas
ing
a
csc
a ade
ga
in.
Th
e
cl
ose
d-l
oop
tr
ans
fer
f
unct
ion
is
T(
s)=
K3
s2+K,
s+K2
(
5.70)
5
s
3+(
2+K3
)S
2+(
1+K,
)s
+K2’
1
59
Th
e a
sso
cia
ted
r
oot
l
ocus
e
qua
ti
on
i
s
1
+
‘
K(
-
s-
2
-
+'
s
a
s+
b,
一0
、
where
K=K3+2,a二
1+KI b=一K2.
2+K3’ 2十K3
T
he
s
et
tl
i
ng
t
i
me
r
eq
ui
re
men
t T<0
.5
s
ec
ond
l
e
ads
us
to
c
hoo
se
t
he
r
oo
ts
o
f (
S2
+
a
s
+
b
)
t
o
th
e l
ef
t
of
t
he
s=一枷 ,=
-8
i
l
ne
i
n
th
e s
-pl
ane
,
as
s
hown
i
n
Fi
g.
5.
33,
t
o
en
sur
e
th
at
t
he
l
ocus
t
rav
els
int
o
the
r
equ
ire
d s
-pl
ane
r
egi
on.
We
have
c
h o
sen a=16
and b=70
t
o
ensu
re
th
e
l
ocus
t
rav
els
pas
t
th
e s=-8 l
ine
. We
s
el
ect
a
p
oint
on
the
ro
ot
lo
cus
i
n
the
pe
rf
or
manc
e r
egi
on,
a
nd
u
sin
g t
he
r
loe
fi
nd
f
unc
ti
on,
we
f
in
d t
he
a
ss
oci
at
ed
g
ain
Ka
nd
t
he
a
sso
cia
te
d v
alu
e o
f叭.
F
or
o
ur
c
hos
en
po
int
we
f
ind
t
hat
K =118.
Th
en
wi
th
K,
a,
and
b,
we
c
an
s
olv
e f
or
t
he
PI
D c
oeic
f i
ent
s
as
f
oll
ows
:
K3=充一2=1
16,
K,=a(
2+K3
)一1
-11
87,
K2=b(
2+
K3
)二8
260.
叩 1.’
.
I
r ,一 一 -一 一牛—
一 一 一一,
6
-口 ̄一- 口一  ̄-,尸 ̄肠 ̄
1t)
I
、
、、、、
进
共州
s= -,
8 \
,
,
j
响定之
} 办、
口
沪一
知 下p
o
le
-10 Se
lect
ed
r
oom 洲/ 一
. .曰.........- 一
洲洲尸一-
-15 I 一 ,
少0
{
一
-15
4
x
_
i;
s
了0
Real
> a二16;
b=70:
num=[
1a :而不石万Q
b] 可三
sys
=tf
(
num,
den
):
>r l
ocus(sys)
卜a
r1
oc
fi
nd
(s
ys) 一
一一一一一一一一曰
Fi
g.
5.
33
Ro
ot
l
ocu
s f
or
t
he
PI
D-c
omp
ens
ate
d t
emp
era
tur
e
co
ntr
ole
r
as
Kv
a
ri
es
To
me
et
t
he
ov
ers
hoo
t p
erf
orma
nce
r
equ
ire
ment
s
fo
r a
s
tep
i
nput
,
we
wi
l
l u
tiiz
le a
c
asc
adega
in
K
t
hat
wi
l
l
be
c
hos
en
b
y i
te
rat
iv
e me
thod
s u
si
ng
t
he
s
tep
f
unc
ti
on,
as
i
ll
ust
ra
ted
i
n
Fi
g.
5.
34.The
160
s
te
p r
esp
ons
e
co
rre
spo
ndi
ng
t
o
K二
5
h
as
a
n a
cce
pta
ble
ov
ers
hoo
t
of
?
'/
( Wi
th
t
he
a
ddi
ti
on
o
f
the
ga
in
K=5
.
th
e in
f a
l
PI
D c
ont
rol
le
r i
s
(
K3s
2+K,
s )二 5(
+K2 116
x2+1
187
,+8
260
) (
5.71)
G,
(
s)=K s
s
We do
not
use
t
he
prefi
l
ter
,
as
i
n Exa
mple
5.
8.
I
nst
ead
we
in
crea
set
he
cas
cade
gain
K
toobtain
s
ati
sf
act
ory
t
rans
ient
r
espons
e.
Now we
ca
n c
onsi
der
t
he
qu
esti
on
of
r
obu
stnes
s t
o c
hang
es
in the
pl
ant
p
ara
met
er
c
o・
Th
e i
nve
sti
gat
ion
i
nto
r
obus
tne
ss
o
f t
he
de
sign
c
ons
ist
s o
f a
s
tep
r
espo
nse
a
nal
ysi
s u
sin
g t
he
PI
D cont
rol
l
er gi
ven
in Eqn.
(
5 .
71)
f
or
a r
ange of
pla
nt
para
me t
er
v
ari
at
ion
s of
0.
1
_c
< o
_1
<
0.
The
re
sul
ts
are d
ispl
aye
d in
Fig.
5.35
.
The
scri
pt i
s wri
tt
en
to co
mput
e t
he s
te
p res
pons
e f
or
a
gi
ven
c・ I
t
might
be
a
go o
d i
dea
to
pl
ace
t
he
input
of
co
at
t
he
command
pr
ompt
leve
l
to
make
t
he
s
cri
pt
mo
re
i
nte
rac
ti
ve.
一
矛滋三
『,
卜,,曰‘二二二
. . . 甘 一 ;,, - ,
Y‘ }
O
习
一 J
.
[
犷
弓
州’
‘一,{
道
O
6
息
愁
七侧
!
}
{
I
a
口乎
!
E
!
} !
} i
一
0 伙05
0.
1 0.15
0
2
0.
25
0.3
Ti
me
(
sec
)
・
I
ncr
ea
se
s
ys
t
em
g
ai
n
t
o
r
ed
uc
e
ov
er
sh
oo
t}
Fi
g.5.
34
St
ep
r
esp
ons
e of
t
he
PI
D t
emp
era
tur
e c
ont
rol
ler
161
/1
!
(
! ! }
-一-
一 1
{
} }
,
一
尸
巴里口..闰
- 丁
}
I
}
老、
0s !
}
} 里 l
!
」“
,荟‘
・
”“ {
仁11乌、
0 6 1’ .
命,
{
一一 一
0了 I
{
I
}
0少
0 口夕 0 1 015 0口 0少5 0
.
3
Twi
t,(
‘,‘
)
・
Sp
eei
f
y
pl
an
t
j
vra
net
er
.!
Fi
g.5
.35
Ro
bus
t
PI
D c
ont
rol
l
er
a
nal
ysi
s
wi
t
h v
aia
r t
ion
s i
n
c0
The
s
imul
at
ion
r
esu
lts
i
ndi
cat
e t
hat
t
he
MI
D d
esi
gn
i
s
ro
bus
t
wi
th
r
esp
ect
t
o c
han
ges
i
n
co.
The
d
ife
ren
ces
i
n
th
e s
t
ep
es
rpon
ses
f
or
0
.1
_c
<
o
<
1
0 a
re
b
are
ly
d
is
ceni
rbl
e
on
t
he
p
lot
.
I
f
th
e r
esu
lt
s
sh
owe
d o
the
rwi
se,
i
t
would
be p
ossi
ble
to
i
ter
ate
on
th
e des
ign
u
nti
l
an
acc
ept
abl
e p
erf
o r
mance
wa
s a
chi
eve
d.
The i
nte
rac
tiv
e capa
bil
it
y of
MA TL
AB a
llo
ws u
s t
o ch
eck t
he
ob
r us
tness
b
y
s
imul
ati
on.
5.
15
SEQUENTI
AL
DESI
GN
EXAMPLE:
DISK
DRIVE
READ
SYSTEM
In
thi
s
chapt
er
we wi
ll
d
esi
gn a
PI
D c
ont
rol
le
r t
o a
chie
ve t
he
des
ir
ed
sy
ste
m r
espo
nse.
Man
y
a
ctu
al
dis
k dr
iv
e he
ad c
ont
rol
s
yst
ems
us
e a
PI
D cont
role
r a
nd u
se
a c
ommand
s
igna
l r(
t)
t
hat
u
til
iz
es
a
n i
dea
l v
elo
cit
y p
rof
il
e a
t
th
e ma
ximu
m a
ll
owa
ble
v
elo
cit
y u
nti
l
the
he
ad
a
rri
ves
n
ear
t
he
d
esi
re
d t
ra
ck,
wh
en
r(
t)
i
s
swi
t
che
d t
o a
s
tep
-t
ype
i
nput
.
Th
us
we
d
esi
re
z
ero
s
tea
dy-
st
at
e e
rro
r
f
or
a
r
amp
(
vel
oci
ty)
s
ign
al
a
nd
a
s
tep
s
ign
al.
Ex
ami
nin
g t
he
s
yst
em
s
hown
i
n Fi
g.536
we
no
te
1
62
t
hat
t
he
f
orwa
rd
pa
th
p
oss
ess
est
wo
pu
re
i
nte
gra
ti
on,‘
,i
d
we
e
xpe
ct
z
ero
s
tea
dy-
st
at
e e
rro
r f
or
a
v
elo
cit
y i
npu
t r(
t)=At
,
t>0
7
I(砂
I
TO
xf)
人州 G
气(
似=
t)
I
t
3(
s
+
}1
)(
s全盯
7
(
1
_
(T
)
} GZ
(.
r
I
) NO +20)
Fi
g.5.
36
Di
sk
d
riv
e f
eed
bac
k s
yst
em
wi
th
a
PI
D c
ont
rol
l
er
The
PID
cont
rol
ler
i
s
K3
(s
+Z,
)(
s+Z
,)
。
。(
:)
=K,
+K2
+、3
,= s
The
mot
or
f
iel
d t
rans
fer
f
unct
ion
i
s
50(
洲)
G,
(S
)= 二 5.
(
S+1
000)
Fi
g.5.
37
A
sk
etc
h o
f a
r
oot
loc
us
a
s凡 i
ncea
r s
es
f
or
e
sti
mat
ed
r
oot
lo
cat
ions
wi
th
a
d
esi
rab
le
s
yst
em
re
spons
e
T
he
s
eco
nd-
or
der
mod
el
u
se
G,
(s
)=5,
and
t
he
d
esi
gn
i
s
de
te
rmi
ned
f
or
t
hi
s
mo
del
.
We
us
e t
he
s
econ
d-o
rde
r mo
del
a
nd
t
he
PI
D c
ont
rol
l
er
f
or
t
he
s
-pl
ane
d
esi
gn
t
ech
niqu
e
i
ll
ust
rat
ed
i
n Se
cti
on
5.
6.
Th
e p
ole
s a
nd
z
ero
s o
f t
he
s
yst
em
a
e
r s
hown
i
n t
he
s
-pl
ane
i
n l
ig.
5.
37
163
f
or
t
he
s
econ
d-or
der
mo
del
a
nd
G,
(s
)=t
i Th
en
we
ha
ve
5K,
(‘
、十z
,
)(
s+:
,)
G,
G,
GZ
(
s)
s
`(
s+2
0)
We
s
ele
ct
z
,=一1
20+
j40
a
nd
d
ete
rmi
ne
5K3
s
o t
hat
t
he
r
oot
, a
re
t
o t
he
l
ef
t o
f t
he
l
ine
s=一1
00.
If
we
a
chi
eve
t
hat
re
qui
reme
nt,
th
en
I l 1
} {
/ \\ !
一 _
_
__
_一
} \、
/ }
I
;
孑
_
_
__7 I }-------.一
门
1
t
』
代n‘
、__ _____ j
一’一
不 !
一
_一
_
__
一_
{
产
I
‘、)
!
I
卜1
,」I
!奢之11
.
、 1_ _ _ _ _ _
,以 b}
} i_
___
___
__
__
__
鱼
“1
V D
o
ub
l
e
)‘,
’
“
一’
一
r---一 -一--
--一 「
‘ 、
I
t
r 一
}
\ '
)_
1
\只 、、.一 r 洲
// }
d0) -300 -}
00 -100 -0 100 少1
I(/
Real
axi
s
Fi
g.5
.38
Ac
tua
l
ro
ot
l
ocu
s f
or
t
he
s
eco
nd-
ord
er
mod
el
a
nd
t
he
o
ver
sho
ot
t
o a
s
te
p i
npu
t i
s
(i
dea
ll
y)
l
es
s t
han
2
% s
in
ce杏o
f t
he
c
omp
lex
r
oot
s
is
a
ppr
oxi
mat
el
y
0.
8.
Of
c
our
se,
thi
s
sk
etc
h i
s
onl
y
a
f
ir
st
s
tep
.
As
a
s
eco
nd
s
tep
we
u
se
MAT
LAB
i
t
era
ti
vel
y t
o d
ete
rmi
ne
K3
.
We
t
hen
o
bta
in
t
he
a
ct
ual
r
oot
l
ocu
s a
s s
hown
i
n F
ig.
5.
38
wi
t
h K3=8
00.T
he
s
yst
em
r
es
pon
se
i
s r
eco
rde
d i
n T
abl
e 5
.11
.
Th
e s
yst
em
mee
ts
a
ll
t
he
s
pec
iic
f a
ti
ons.
Th
e
desi
gn
of
h
igh
ly
a
ccu
rat
e c
ont
rol
s
yst
ems
i
n t
he
p
res
enc
e o
f s
ign
if
ica
nt
p
lan
t u
nce
rta
int
y
r
equ
ir
es
t
he
d
es
ig
ner
t
o
se
ek
a
r
obu
st
c
ont
r
ol
s
yst
em.
A
r
obu
st
c
ont
ro
l
sy
st
em
e
xhi
bi
ts
l
ow
s
ens
it
ivi
t
ies
t
o
pa
rame
ter
c
han
ge
a
nd
i
s
st
abl
e
ov
er
a
wi
de
r
ange
of
pa
rame
ter
v
ari
at
ions
.
164
Ta
ble
5
.11
Di
sk
Dr
iv
e Co
ntr
ol
S
yst
em
S
pec
if
ica
io
t ns
a
nd
Ac
tua
l
Pe
rfo
rma
nce
Per
for
mance
Meas
ure Desi
red
Val
ue Res
pons
e f
or
Second-
Order
Model
Percent
overshoot <5%
4.
5%
Se
tt
in
l g
t
ime <50ms 6ms
f
or
s
tep
i
nput
Ma
ximu
m r
esp
ons
e f
or <5x1
0"3 7.
7x
1
0-'
a
uni
t
st
ep
di
stur
bance
Th
e t
hre
e-mo
de,
o
r PI
D,
c
ont
rol
ler
wa
s c
ons
ide
red
a
s a
c
omp
ens
ato
r t
o a
id
i
n t
he
d
esi
gn
o
f
r
obu
st
c
ont
rol
s
yst
em.
The
d
esi
gn
i
ssu
e f
or
a
PI
I)
c
ont
rol
le
r i
s
th
e s
ele
cti
on
o
f t
he
g
ain
a
nd
t
wo
z
eros
o
f t
he
c
ont
rol
ler
t
rans
fer
f
unc
tio
n.
We
u
tiiz
l e
d t
hre
e de
sig
n me
tho
ds
f
or
t
he
s
ele
cti
on
o
f t
he
c
ont
rol
le
r:
t
he
r
oot
l
ocus
me
thod
,
th
e f
req
uen
cy
r
espo
nse
me
tho
d,
a
nd
t
he
I
TAE
per
for
man
ce
i
nde
x me
tho
d.
An
o
per
ati
ona
l a
mpl
if
ier
c
irc
uit
us
ed
f
or
a
,
PI
D c
ont
rol
l
er
i
s
sho
wn
i
n Fi
g.
5.3
9.玩
ge
ner
al,
the
u
se
o
f a
PI
D c
ont
rol
le
r
wi
ll
e
nab
le
t
he
d
esi
gne
r t
o
at
ta
in
a
r
obus
t
co
ntr
ol
s
yst
em.
T
he
i
nt
ena
r l
mo
del
c
ont
ro
l
sy
st
em
wi
th
s
ta
te
v
ari
abl
e
fe
edb
ack
a
nd
a
c
ont
r
oll
er
G,
(s
)
wa
s
u
til
iz
ed
t
o o
bta
in
ar
o
bus
t c
ont
rol
s
yst
em.
Fi
nal
ly
t
he
r
obu
st
n
atu
re
o
f a
ps
eudo
-QFT
c
ont
rol
s
yst
em
wa
s de
mons
trat
ed.
〔’
1 了’,
R,
R1
R3
Rl
1
、
l
}
Fi
g.
5.3
9 Op
era
tio
nal
a
mpl
if
ier
c
irc
uit
us
ed
f
or
PI
D c
ont
rol
le
r
A ro
bust
cont
rol
s
yst
e m
prov
ides
st
able,
cons
ist
ent
perf
ormance
as
spec
ifi
ed
by t
he
des
ign
e r
i
n
s
pit
e o
f wide v
ari
at
ion of
pla
n t
par
a met
ers a
nd di
s t
urbanc
es.
It
al
so provi
des
hig
hly
robus
t
r
esp
onse
to
command i
nput
s
and a
st
eady-s
tat
e t
rac
kin
g err
or e
qual
t
o z
ero.
Fo
r sy
stems wit
h uncer
tai
n paramete
rs, t
he need for ro
bust s
ystems wil
l r
equ
ire t
he
i
nco
rpo
rat
io
n o
f
ad
van
ced
ma
chi
ne
i
nt
el
li
genc
e.
a
s s
hown
i
n
Fi
g.
5.
40.
165
P孙1
Hi
gh
,nb孙、灿勺u勺斗之勺、
-
1da
pti
re,
,%I
elt
t
加tl灿u、 Robust
/
srst
ein
/
魂勺二次
_
t1der
ate
Fe
edbac
k s
yst
em
/p
a
r
a
me
t
e
r
sa
n
d
Low印 e n-
loop
syst
em
(wi
thot
d
feedback)
Loin
Lo
w
Mo
der
at
e
H心h
Unc
ert
aint
y o
f par
ame
ter
s and
di
st
urbanc
es
Fi
g.
5.
40
I
nt
el
i
lge
nce
r
equ
ied
r v
ers
us
u
nce
rt
ai
nt
yf
or
mo
den
r c
ont
ro
l
sy
ste
ms
166
CHAPTER
6 DI
GITAL
CONTROL
SYSTEM S
A
dig i
tal
computer
may ser
ve as
a c
ompensat
or or
cont
rol
ler
in
af
eed bac
k cont
rol
sys
tem.
Si
nce
the co
mp ute
r r
ecei
ves
data
only
at
s
peci
fi
c i
nte
rval
s,
it
i
s
necess
ay
r to dev
elop
a
method
for
d
escr
ibing
an d
anal
yzi
ng t
he
perfor
mance
of
compute
r c
ontro
l s
yst
ems.
Th
e comp ute
r s
yst
em uses
data
sampl
ed a
t pr
espe
cif
ied i
nte
rval
s,
r
esul
ting i
n a
ti
me se
ries
of
s
igna
ls.
These t
ime
seri
es,
ca
ll
ed sa
mpled
dat
a,
can
be
tra
nsfor
me d
t
o t
he s
-do mai
n a
nd,
ult
i
ma t
ely
,
t
o t
he
Z
-
doma
in
b
y t
he
r
el
at
io
n
Z=e
sr.Th
e c
omp
le
x re
fque
ncy
d
oma
in
i
n t
er
ms
o
f Z
h
as
prope
rti
es
si
mi l
ar
to
those of
the
Laplac
e s
-domain.
We
ma y use t
he z-
t
ra ns
form of a t
ransf
er fu
nction
to analyz
e the
sta
bil
ity
and t
ransi
ent
res
ponse o
f a s
y s
tem. Thu s
we may r
eadily
deter
min e
the
re s
ponse
of a
clos
ed-l
oop f
eedback
syst
em wit
h ad i
git
al comp ute
r ser
vin
g as th
e compe ns
ator
( o
r con
trol
le
r)
block.
We can
also
uti
li
ze
the
root
locus
me t
ho d to
dete
rmine
the l
ocat
ion of
th
e roots
of
the
cha
ract
eri
sti
c e
quati
on.
These
computer c
o n
trol
systems ar
e wi
dely us
ed in
indust
y.
r They al
so
play
an
import
ant
rol
e in
t
h e
cont
rol
o f
i
nd us
tria
l processe
s,
wh i
ch uti
li
ze a c
o mbi
nation of
a c
ompu t
er
and an
actu
ator
wo
rki
ng
t
oge
the
r
to
pe
rfo
rm
as
e
ri
es
o
f t
as
ks.
Th
is
c
hapt
er
c
onc
lud
es
wi
th
t
he
de
si
gn
o
f a
d
igi
ta
l
c
ont
rol
le
r f
or
t
he
Se
que
nti
al
De
sign
Exa
mpl
e:
Di
sk
Dr
ive
Re
ad
S
yst
em.
6.
1 INTRODUCTION
Th
e u
se
o
f a
d
igi
ta
l c
ompu
ter
a
s a
c
omp
ens
ato
r (
con
trol
l
er)
de
vic
e h
as
g
rown
d
uri
ng
t
he
p
st
a
t
wo d
eca
des
a
s
the
pr
ice
a
nd
re
li
abil
it
y
of
di
gi
ta
l
comput
er
s
have
i
mpro
ved
dra
mati
cal
ly
.
A
bloc
k
d
iag
ram
of
a
si
ngl
e-l
oop
di
gi
ta
l
contr
ol
s
yst
em
is
shown
i
n Fi
g.
6.
1.
The
di
gi
tal
c
omput
er
i
n
this
s
yst
em
con
fig
ura
ti
on
r
ece
ive
s t
he
e
rro
r i
n di
git
al
f
orm
and
pe
rfo
rms
c
alc
ula
ti
ons
i
n o
rde
r t
o
p
rov
ide
a
n o
utp
ut
i
n d
igi
ta
l
fo
rm.
Th
e c
ompu
ter
ma
y b
e pr
ogr
amme
d t
o
pr
ovi
de
a
n o
utp
ut
s
o t
hat
t
he
p
erf
orma
nce
o
f t
he
pr
oce
ss
i
s
ne
ar
o
r e
qua
l
to
t
he
d
esi
red
p
erf
orma
nce
. Ma
ny
c
ompu
ter
s a
re
a
ble
t
o
re
cei
ve
a
nd
ma
nipu
lat
e
se
ver
al
i
npu
ts,
s
o a
d
igi
ta
l
comp
ute
r
co
ntr
ol
s
yst
em
c
an
o
te
f n
b
e a
mul
ti
var
iabl
e s
yst
em.
A
d
igi
t
al
c
omp
ute
r r
ece
ive
s a
nd
o
per
ate
s on
s
ign
als
i
n d
igi
ta
l
(nu
mer
ica
l)
f
orm,
as
co
ntr
ast
ed
t
o
c
ont
inuo
us
s
ign
als
.
A di
gi
tal
c
ont
rol
s
yst
em u
ses
di
gi
tal
s
ign
als
a
nd a
d
igi
tal
c
omput
er
t
o
con
tr
ol
a
p
roc
ess.
The
meas
ure
me n
t d
ata
a
re
conve
rt
ed ro
f m
anal
og
for
m to
di
git
al
fo
rm by
me
ans
of
t
he
a
nal
og-t
o-di
git
al
conver
ter
shown i
n Fig.
6.1
. Af
ter
pr
oces
sin
g t
he i
npu
ts,
t
he
dig
ita
l c
omput
er
p
rovi
des a
n outp
u t
in di
git
al f
orm. Thi
s outpu
t i
s t
hen conver
ted
to
anal
og f
orm by t
he
d
igi
tal
-t
o-a
nal
og c
o n
vert
er
shown i
n Fi
g.6.
1.
Re
fe
re
nce _
厂下i
,
;i
r
a
l刁(thgatrl) Di
git
al (
uual
og) ()r
r7nr
t t
f
i
l
p“一  ̄}c
oa
mat
e
r r 匆一伽诫 0
con
g
ref六"
r
tct
aator 了h,cess
(
analog)
(
d
i
g
i
t
a)
l上遍 .
七,
t
.6'
o-t
g-
hgac
rl
convert
er
(
rnur
lo}) He
usurenr
SeJ
I
e
SOr
nt
Fi
g.6
.1
A bl
ock
dia
gra
m of
a
c
ompute
r c
ontr
ol
s
yst
em,
i
ncl
udi
ng
t
hesi
gna
l c
onver
ts
.
The
si
gna
l
is
i
ndi
cated
sd
a
igi
ta
l
or
a
nal
og
167
6.
2 DI
GITAL
COMPUTER
CONTROL
SYSTEM
APPLICATIONS
The
tot
al
number
of
compute
r cont
rol
sys
tems
ins
tal
le
d in
indust
ry
hs
a g
r own over
the
past
t
hree
decad
es.
Curr
entl
y t
her
e a
re appr
oxi
ma t
ely
100
mill
ion
contr
ol s
yst
ems us
ing comput
ers,
a
lth
o u
gh th
e comput
er s
ize
and p
owe r
ma y
vary
si
gni
ic
f antl
y. I
f we co
nsider o
n l
y compute
r
c
ontrol
s
yst
ems of
a
r
ela
tiv
ely
co
mp l
e x
nat
ure,
such
s
a ch
emical
proc
ess
cont
rol or
a
ircr
a f
t
con
trol
,
t
he
n
umbe
r
of
c
omp
ute
r
co
ntr
ol
s
yst
ems
i
s
ap
pro
xima
tel
y
20
mi
li
l
on.
s
外e ize
of comp uter
s and
th
e cost
for
the
activ
e logi
c dev i
ces
u s
e d
to
constr
uct t
hem have
both
decli
ned expon enti
ally.
The ac
tive c
omp onents per cu
bic c e
nti
me ter h
a ve
incre
se
a d so
that
the
actual
compu t
e r
can be r
educed i
n si
ze to
the
po i
nt where r
e l
ati
vely i
nexpensive,
powe rfu
l l
aptop
compu t
e r
s a
re pr
ov i
ding mo bi
le high-per
formanc e compu tat
ional capabi
it
l y to s
tud e
nts and
profes
sionals
a l
ike,
a nd i
s,
in
ma ny inst
ances,
replaci
ng the tr
ad i
tio
nal deskt
op mi c
ro computer
.
Th e
speed of comp uters
hs
a a
lso in
cr e
ased exponenti
ally.
Th e
transi
stor
de n
sit
y (a me asur
e of
compu t
a t
ional performa nce
) on NTEL
I mic r
op r
ocessor i
ntegrat
ed cir
c u
it
s hs i
a ncr
ese
a d
exponen t
ial
ly over t
h e l
ast
30 years
.
I n
fac
t,
ac c
ording to "Mo ore's
Law" the t
ransistor
dens
ity
doubles ev
e r
y year,
and will
be
mo e
r fst
a er
for
the next
twe nty
years.
A si
mp le
calculaio
t n s
hows
that b
y 2012 microprocess
ors will
c ont
ain over a bilio
l n t
ransis
tors with o
pe ra
ting speeds
approaching 10GHz . Since
the popular
8086 central
proce s
sin
g unit
s were i
ntroduced in 1976
contai
ning only 29,000 tr
ansis
tors a
nd ope
rain
t g
a t
10MHz ,
signi
f i
cant
process i
n co mputat
ion
capabil
ity has b e
en and will
con t
i
nu e t
o be ma de.
Cl eay,
r imp r
oveme nts i
n c ompu t
aio
t nal
capabil
ity
h ave revolut
ioni
zed the a
pp l
icat
ion of cont
rol theoy
r a
nd desi
gn in t
he mo d
e r
n-e
ra.
Wi
t
h t
he
a
vai
la
bil
it
y
of
f
st
a ,
l
ow-
pic
red,
a
nd
s
mal
l-
si
zed
mi
cr
opr
oce
sso
rs,
muc
h o
f t
he
c
ont
ol
r o
f
i
ndu
str
ial
a
nd
c
omme
rci
al
p
roc
ess
es
i
s mo
ving
t
owa
rd
t
he
us
e o
f c
ompu
ter
s wi
thi
n t
he
c
ont
rol
s
yst
em.
Digit
al con t
rol sy
stems ar
e used
in many appic
l ati
ons: f
or machine t
ools,
metal
worki
ng
p
roce ss
es, chemi ca
l proces
ses,
ai
rcr
at
f c
ontol
r , and aut
o mobi
le t
raff
ic contr
ol.
Au t
omaic
t
c
ompu t
er-c
ont r
olled sy
stems ar
e used f
or pur
p o
se s
s
a dive
rse s
a mesu
a in
r g th
e o
bjec
tiv
e
r
efr
ac io
t n of
the human ey
e and
con
tro
li
lng
the
engi
ne spar
k imi
t ng
or
air-
fuel
ra
tio
of
aut
omobil
e
e
ngine s
. The latt
er
innovaio
t ns
ar
e ne
cessa
y
r to re
du ce
aut
o mobi
le
emi s
sio
ns and i
ncea
r se
fuel
econom y.
The a
dvant
age
s of
us
ing
di
git
al
c
ontr
ol
i
ncl
ude
i
mproved
mesu
a r
ement
s
ens
iiv
t i
ty;
t
he
use
of
di
gi
ta
ll
y co
ded s
ign
als
,
dig
it
al
se
nsor
s
and
tr
ans
duc
ers
, a
nd mi
cr
opr
oce
sso
rs;
r
educ
ed s
ens
iiv
t i
ty
t
o s
ign
al
no
ise
; a
nd
t
he
c
apa
bil
i
ty
t
o e
asi
l
y r
eco
nfi
gur
e t
he
c
ont
rol
a
lgo
it
r h
m i
n s
otwa
f re.
Imp
roved
s
ens
it
iv
it
y r
esul
ts
ro
f m t
he
low-
ene
rgy
si
gnal
s
re
qui
red
by
di
git
al
sens
ors
an
d dev
ice
s.
The u
se o
f di
gi
tal
ly co
ded s
ignal
s pe
rmi
ts t
he wi
de ap
pl
ica
ti
on of dig
it
al de
vic
es and
c
ommu
nic
aio
t ns
. Di
git
al
s
ens
ors
a
nd
t
ran
sdu
cer
s c
an
e
fec
tiv
ely
mesu
a r
e,
t
rans
mit
,
and
c
oupl
e
s
ign
als
a
nd
d
evi
ces
.
In
a
ddi
t
ion
ma
ny
s
yst
ems
a
re
i
nhe
ren
tl
y d
igi
ta
l
be
caus
e t
hey
s
end
o
ut
p
uls
e
s
ign
als
.
Exa
mpl
es
o
f s
uch
a
di
git
al
s
yst
em
a
re
a
r
ada
r t
ra
cki
ng
s
yst
em
a
nd
a
s
pac
e s
ate
ll
it
e.
6.
3 SAM PLED-
DATA
SYSTEM S
Co
mpu
ter
s u
sed
i
n c
ont
rol
s
yst
ems
a
re
i
nte
rco
nne
cte
d t
o t
he
a
ctu
ato
r a
nd
t
he
pr
oce
ss
b
y me
ans
1
68
of
s
igna
l
co
nve
rte
rs.
The
o
utp
ut
o
f t
he
c
omp
ute
r i
s
pr
oce
sse
d by
a
d
igi
tal
-t
o-a
nal
og
c
onv
ete
r r
.
We
wil
l
ss
a u
me tha
t
al
l
the n
umb er
s t
hat
e
nte
r or
l
eav
e t
he comput
er
do
so
at
t
he sa
me f
ixed
pe
ri
od T,
ca
lle
d t
he s
ampli
ng pe
riod.
Thus,
f
or
exampl
e,
th
e re
ferenc
e i
nput
s
hown i
n
Fig.6.
2
is
a
seq
uence
of
sampl
e val
ues
r
(kT ).The va
ria
ble
s
r(kT),m(kT ),and
u
(kT)
ar
e
disc
rete
s
ignal
s
in
con
tra
st
t
o m(t)
an
d y(t
)
which a
re
conti
nuous
f
uncti
ons
of
t
ime.
Da
ta
obt
aine
d f
or t
he syst
em var
ia
bles
only
at
disc
ret
e i
nter
val
s,
an
d denot
ed sx
a
(kT
),r e
c
all
ed
s
amp
led
da
ta
o
r a
d
isc
ret
e s
ign
al.
A
s
amp
ler
i
s
ba
si
cal
ly
a
s
wit
ch
t
hat
c
los
es
e
ver
y Ts
econ
ds
f
or
o
ne
i
nst
ant
o
f
ti
me.
Co
nsi
der
a
n
i
de
al
s
amp
le
r
ss
a
ho
wn
i
n
Fi
g.
6.
3.
Th
e i
np
ut
i
s
r(
t)
,an
d t
he
o
utp
ut
i
s
r*
(t
),wh
er
e n
T i
s
th
e
c
ur
re
nt
s
amp
le
t
i
me,
an
d
th
e
cu
rr
en
t
va
l
ue
o
f r
*
(
t
)
is
r(
nT)
。We
th
en
h
ave
r
(
t
)=r
(
n7At
-0)
,
wh
ere
8i t
s
he
i
mpu
lse
f
unc
ti
on.
r
(O)
了钊
日夕
r
a(kT) Digi
tal
- Act
uat
or
Rc
fe
renc
e Di
git
al 多
,)
,
(
口
.
comput
er
八,1
a1al
og and
conrerter Out
put
p代= SS
乙引
j少
m(
kT) Ana城 -
t
o d
igi
tal
e口1
1 P曰 了er
Fi
g.
6.
2
A
di
git
al
c
ont
rol
sy
ste
m
r
*
(t
)
C卜-一 一 .一
Cont
inuous Sampled
S
烤It
al si
glt
td
F i
g.6.
3
An id
eal
samp
ler
wi
th
a
ninput
r(t
)
Le
t u
s a
ssume
t
hat
we
s
ample
a s
ignal
r
(t),a
s s
hown
inFig.
6.3,
ando
bta
in
r
*
(t
)
.
The
n we
p
ort
r
ay
t
he
s
er
ie
s
fo
r r
'
(t
)
as
a
s
t
ri
ng
o
f
imp
ul
se
s
st
ar
ti
ng
a
t
t=0,
spa
ced
a
t T
s
ec
ond
s,
a
nd
o
f
a
mpit
l u
de
r(
kT).Fo
r ex
ampl
e,
co
nsi
der
t
he
i
nput
sig
nal r(
t)
s
hown
in
Fi
g.
6.4
a)
.
The
s
ampl
ed
s
ign
al
i
s
sh
own i
n
Fig
.6.
4
b)
wi
th
an
i
mpul
se
r
epre
sente
d b
y a
ve
rt
ic
al
a
rr
ow o
f
magni
tu
de
r(
kT
).
r(
t)
0 T 2T 了T 4T , 口
(
1)
169
厂防孙 r(
2T)
r(
3T)
r(T) r(
4T)
0 T 2T 了T -
f
T
Ti
me(
kT
b)
F
ig
.
6.
4
a
)
An
in
pu
t
s
ig
na
l
r
(t
)
a
nd
b
)
t
he
sa
mpl
e
d
si
g
nal
r*
(t
)=艺乙r
(k
T)S
(t
-k
T)
Th
e c
olo
red
v
ert
ica
l
ar
row
r
epr
ese
nts
a
n i
mpu
lse
.
Sampl
er
r
(
t)一一f
Fi
g.
6.
5
A
sampl
er
a
nd
zer
o-or
der
hol
d c
ircui
t.
^d
ig
it
al
-
to
-a
na
lo
g c
onv
er
te
r
se
rv
es
a
s
a
d
evi
ce
t
ha
t
co
nve
rt
s
t
he
s
amp
le
d
si
gn
al
r
*
(t
)
t
o
a
c
oninu
t ous
s
ign
al
p(
t).
The
di
gi
ta
l-
to-
ana
log
c
onv
ert
er
c
an
u
sua
ll
y
be
r
epr
ese
nte
d by
a
z
ero
-or
der
ho
ld
c
ir
cui
t,
as
s
hown
i
n
Fi
g.
6.
5.
Th
e z
ero
-or
der
ho
ld
t
ake
s t
he
v
alue
r
(kT)
a
nd
h
old
s i
t
co
nst
ant
f
or
k
T S
t<(
k+1
)T,
as
s
hown
i
n
Fi
g.
6.
6
fo
r k=0
.
Th
us
we
u
se
r
(kT)
du
ri
ng
t
he
s
amp
li
ng
pe
riod.
P
(
r)
1
0 1'
Ti
me
Fi
g.
6.
6
The
r
esp
ons
e
of
a
z
ero
-or
der
ho
ld
t
o
an
i
mpul
se
i
nput
r
(
kT),
whi
ch
e
qua
ls
1
wh
en
k=0
a
nd
e
qua
ls
z
er
o
wh
en
k
#。,
so
t
ha
t r
(t
)=r
`
(0
)8(
t)
.
A
s
ampl
er
a
nd
z
ero
-or
der
h
old
c
an
a
ccu
rat
ely
f
oll
ow
t
he
i
npu
t s
ign
al
i
f Ti
s
sma
ll
c
omp
are
d t
o
t
he
t
ran
sie
nt
c
han
ges
i
n t
he
s
igna
l.
Th
e r
esp
ons
e o
f a
s
ampl
er
a
nd
z
ero-
ord
er
h
old
f
or
a
r
amp
i
npu
t
i
s
sh
own
i
n Fi
g.
6.
7.
Fi
nal
ly
t
he
r
espo
nse
o
f a
s
ampl
er
a
nd
z
ero-
ord
er
h
old
f
or
a
n e
xpo
nen
ti
al
ly
170
d
eca
yin
g s
igna
l i
s s
hown
i
n Fi
g.6
.8.
f
or
t
wo
v
alu
es
o
f s
amp
lin
g pe
rio
d.
Cl
ear
ly
t
he
o
utp
ut
p(
t)
wi
ll
app
roa
ch
t
he
i
npu
t r(
t)
a
s Ta
ppr
oac
hes
z
ero
(
as
we
s
ampl
e
ve
y
r o
fte
n).
Th
e i
mpu
lse
r
esp
ons
e o
f a
z
ero
-or
der
h
old
i
s s
hown
i
n F
ig.
6.
6.
Th
e t
ran
sfe
r f
unc
tio
n o
f t
he
ze
ro-
orde
r hol
d i
s
1
1
一 一 e一
,: (
1一e
-S
T
)
尸大
J
‘.、
G,
,
(s)=一s
s
它礼飞注
Z
_ 5
厅 Tune
Fi
g.
6.
7
The
r
esp
ons
e
of
a
s
ampl
er
a
nd
z
ero-
ord
er
h
old
f
or
a
r
amp
i
npu
t r(
t)=t
.
了、
妇西
The preci
si
o n
of t
he di
gital
computer
an
d t
he as
soc
iat
ed s
ignal
conve
rter
s is
li
mite
d
『.L
「J
F
ig.
6.2)
.
Preci
sion
is
t
he
degree
of e
xact
ness
or
di
scri
mina
tio
n wi
th
which a
q
ua n
tiy
t i
s
st
ated.
p
rec
isio
n of
the
compute
r i
s l
imit
ed by
a
fi
nit
e word
le
ngt
h.
The
p r
eci
si
o n
of
t
he anal
og-t
o-di
git
al
c
onvert
er i
s l
imit
ed b
y an abil
i
y
t t
o st
ore
it
s out
put
onl
y i
n di
git
al
logi
c composed o
f a f
ini
te
nu
mber
of b
inay
r d
igi
ts
. Th
e c
onver
te
d si
gnal
,
m(
U) ,i
s t
hen
sa
id
to
inc
lude
an
ampl
it
ude
qu
ant
iz
ati
on
err
or.
When t
he
qua
nti
zat
io
n e
rro
r a
nd
t
he
e
rro
r d
ue
to
a
co
mpute
r's
f
in
it
e
word
si
ze
ar
e s
mall
r
ela
tiv
e t
o t
he ampli
tude
of
t
he
s
igna
l,
t
he
s
yst
em
i
s s
ufi
ci
ent
ly
p
ec
r i
se,
a
nd
t
he
pr
eci
si
on
i
lmit
ati
ons
ca
n b
e ne
glect
ed.
3tl、毛 它、
口
}l
e
f
t
)
0,
5
r
(
t)
/'
1
2 子
Ti
na
(
sec
ondv
)
川 T二 0.
5 Yeconf
ls
171
它气1,、
个lrl\
,1,仑』
1.
夕
/P
(r
)
产了声了(l)
扶s
尸
n 3
4
Ti
me
(
see
undq
h
) T
=
0.
J
}
ecf
nji
A
Fi
g.
6.
8
The
r
espo
nse
of
a
s
amp
ler
a
nd
z
ero
-or
der
h
old
t
o a
n i
nput
r
(t)=e一
‘fo
r t
wo
v
alu
es
o
f
sa
mpin
lg p
eri
od
T.
6.
4 THE
Z-
TRANSFORM
Be
cau
se
t
he
o
utp
ut
o
f
th
e
id
eal
sa
mpl
er
,
r*
(t
),i
s
a
s
eie
rs
of
imp
uls
es
wi
t
h v
al
ue
s r
(kT
)
,we
have
r
.(
;)
=艺r
(k
T)8
(
k月
t
-
k
T
), (
6.2)
f
or
a
s
ign
al
f
or
t>0.
Usi
ng
t
he
La
pla
ce
t
ra
nsf
orm
we
h
ave
L
j
r
'
(
t)
}
=分(
k
T
)
e-
k
s
T. k=0
(
6.3)
We
n
ow
h
ave
a
n
in
i
fni
t
e s
ei
re
s
th
at
i
nvo
lv
es
f
ac
to
rs
o
f e
'T
a
nd
i
t
s
po
wer
s.
We
d
efi
ne
(
6.4)
Z二e
sT,
t
his
r
ela
ti
ons
hip
i
nvol
ves
a
c
onf
orma
l ma
ppi
ng
f
rom
t
he
s
-
pla
ne
t
o t
he
Z
-
pla
ne.W e
then
a
new
t
rans
for
m,
ca
lled
t
he
Z
-t
rans
for
m,
s
o t
hat
Z
jr
(
t)
j=
Z}
`(
t)
}
=乞;
(
、:)
:一
“。 k
=O
(
6.5)
e
xamp
le,
l
et
us
d
ete
rmi
ne
t
he 7
-
tr
ans
for
m o
f t
he
u
nit
s
te
p f
unc
ti
on
u(
t
)
[n
ot
t
o b
e
c
onf
use
d wi
th
t
he
c
ont
rol
si
gna
l
u
(
t)
l.
f W e
obt
ain
Z{
r(
t
)}一l
u
(k
T
)4
.
-k=I
Z
-k
k=0 k=0
(
6.6)
sin ce u
(
kT
)=1
f
or
k?
0.
Thi
s
se
ri
es
c
anbe
wri
tten
i
n osedf
cl orm
a
s (
Rec
all
th
at
t
he
i
nfni
i t
e
g
eome
tr
ic
s
er
ie
s ma
y
b
e
wr
it
t
en
(
1一b
x)一
,=1
+
b
x+(
bx)
2+(
bx)
3+二
。,i
f(
b
x)
2<1
.
之
U(
z)= (
6.7)
1一z z一1
I
n
ge
ner
al
we
wi
ll
d
efi
ne
t
he
z
-
tr
ans
for
m of
a
f
unc
ti
on
f
(
t)
a
s
172
Z
仃(:
)}
=F(
z)=艺f
(
kT)
z-
k k=0
(
6.8)
EXAM PLE
6.1
TRANSFORM
OF
EXPONENTI
AL
Letu
s d
ete
rmi
ne
t
he
z
-
tr
ans
for
m of
f
(
t)=e
-
a`
f
ort
?
0.
The
n
Z
}一
I
-F(
z)
=E。
一:一i
(
z
e'
aT
)一
‘ (
6.9)
Aga
in
t
his
s
eri
es
can
be
wr
itt
en
i
n c
los
ed
f
orm
as
F(
z)= z一e
-
aT
(
6.1
0)
1
一(z
eal
)一,
I
n ge
ner
al
we
ma
y s
how
t
hat
Z
苍一,
(,
)}
一F(
e
aT
z)
EXAM F生石 6.
2刀?
ANSFOR材 OF
SINUSOI
D
Le
t u
s d
ete
rmi
ne
t
he
z
-
tr
ans
for
m o
f f(
t
)=s
in
c
ot
f
ort
?
0.We
c
an
wr
it
e s
in
c
e
n a
s
s
i
n。二e
兰立翌.
2j
Ther
efore
r
. 7
I
st
I
e
n
j0
(
T一—。
ax
t
=
一
S—
]
‘I》 (
6.11
)
l
2J
2j
Then
厂
-一
一三
.-
一_一止匕一
F(
z)=
1/
2jI,
__v
i
.
、‘一 “
_,_I
,
WT _一,
WT、
一。
-JC
OT
、
) (
6.12)
=二 }
一一止竺二一二生二卫_ 】
二 zs
ino)T
2
j(z
’一z
(e
,W
T
+
e
-'
0'
T
)
+‘)z2一2z
co
s
coT+1’
A
t
abl
e o
f z
-
tr
ans
for
ms
is gi
ven
i
n Tab
le
6.1
an
d a
t t
he
MCS we
bsi
t
e.
A t
abl
e o
f p
roper
ti
es
of
t
he
Z
-
tra
nsf
orm i
s g
ive
n i
n Tabl
e 6
.2.
As i
n t
he
cas
e of
La
pla
ce
t
ran
sfo
rms,
we a
re u
lti
mate
ly
i
nt
er
est
ed
i
n t
he
o
utpu
t
y(
t)
o
f t
he
s
yst
em.
Th
ere
for
e we
mus
t
us
e a
n i
nve
rse
t
ra
nsf
orm
t
o o
bta
in
y(
t)
f
rom
Y(
Z).We
ma
y o
bt
ai
n t
he
o
utp
ut
b
y (
1)
e
xpa
ndi
ng
Y(z
)
in
ap
o
wer
s
eri
es
, (
2)
e
xpa
ndi
ng
Y(Z
)in
to
pa
rt
ia
l
fr
act
io
ns
a
nd
us
ing
Ta
ble
6.1
t
o
obt
ai
n
th
e i
nve
rse
o
f
ea
ch
t
erm,
or
(
3)
o
bta
ini
ng
t
he
i
nve
rse
Z
-
tr
ans
for
m b
y a
n i
nve
rsi
on
i
nte
gra
l.
We
wi
ll
l
imi
t
ou
r me
thod
s t
o (
1)
a
nd
(
2)
i
n t
his
l
imi
ted
di
scus
sion.
1
73
Tabl
e 6.
1 2-
Trans
for
ms
x(
t) X
(
S) X
(
z)
_ (
1
t=0 一
— 一
6(
t)=弋_ _
__ _ 1
1
t=
S(
t一kT)= 丁
‘ e-
kTs Z-
k
t
o
钊灯.
之
S
U
(
t),
uni
t
st
ep
之一1
钊灯
Tz
2
S
t
(
z一1
)2
1
之
,-
at
S十 a
z一e-
aT
1一e
-a
t
1
(1
一e -
aT
)
z
S(
S+a) (
z一1)
(z一e-a
T)
CO
z
s
in
c
OT
sin
OW
S2+口2 z2一2
z c
os
a
x+1
Cos
Cf
)t
s
z(
z一c
os
(
oT
)
S2+田2 2
z一
2 2z
c
os
c
ot+1
CU
z
e-a
T
si
n
of
e
-a
t s
in
c
ot
(
S+a)
2+0
)2 22 2z
z一 e-a
T
co
s
co
T+e
-2a
T
s十 a 22一z
z e-a
T
cos
o
f
e
-a
t
cos以
(
s+a)
2
+c
l
)2
z2一2z
e-a
T
co
s
of+
e
-
2aT
x(
t) X(z
)
1
.
kx(
t) kX
(
z)
2
.
X
I(t
)+x
,
(t
) X,
(z
)+X2
(z
)
3
.
x(t+T) z
X (
z)一z
X(0)
4
.
tx(
t) T_
dX
(
z)
- 1乙 —
dz
Ca
t
X
(t) X
(
zea
T)
6. x(
0),
in
iti
al
val
ue l
imn
、一X
(
z)
i
f
t
he
i
l
mit
ex
is
ts
x(
o),
f
inal
v a
lue l
im_
},
(
z一1
)X(
z)
i
f
th
e l
i
mit
s
ex
is
ts
a
nd
t
he
s
yst
em
is
s
tabl
e;
t
hat
i
s,
i
f a
ll
po
les
of
(
z一1
)X
(
z)
ar
e
in
si
de
u
ni
t
c
ir
cl
e
IZ
I=1
on
『
L
.
,.
J
,
…『.
z
-pl
ane
.__
_,_,
「..
「」二 ‘ __
174
EXAMPLE
63
TRANSFER
FUNCTI
ON
OF
OPEN-
LOOP
SYSTEM
Le
t
us
c
ons
ide
r t
he
s
yst
em
s
hown
i
n F
ig.
6.
9
fo
rT=
=1.
Th
e t
ra
nsf
er
f
unc
ti
on
o
f t
he
z
ero-
ord
er
hol
d (
Egn.
6.1
) i
s
六 ,、 (
1一e
fo
L l
s)= —
s
T
her
ef
or
e
t
he
t
r
ans
fe
r
f
unc
t
ion
Y(
s)
/R*
(s
) i
s
Y(
s) 。,
_、。,
_、 。,
_、_1一CS
T
二;丁下,= 。。l
s)Y ,
L ls)= Vl
y)=
. (
6.1
3)
找 Ls) s
2(
s+1
)’
Ex
pand
ing
i
nto
pa
ti
r a
l
ra
f c
ti
ons
,
we
h
ave
G
(
s)
一
“一‘
传一香
・s
+
1
1 (
6.1
4)
G
(
z
,二
‘,
“,
,
=(
1
一‘
)
Z(
s
?一
香・
s
+
1
1
). (
6.1
5)
r
(
t
)一丫/
T=1
Fi
g.6.
9
An
o
pen-
loo
p,
s
ampl
ed-
dat
a
sy
ste
m(wi
tho
ut
f
eedb
ack)
Us
ing
t
he
ent
rie
s of
Tabl
e 6.
1 t
o obt
ain
t
he
Z -
tra
nsf
orm
of
e
ach
t
erm,
we
ha
ve
_,、 ,
, _
,、
「TZ
Z :1( Z
e-T一Z+T
z)+(
1一e
-T一T
e-T
)
i
1LZ)二 L
1一Z 一
刀份 ̄
一,二二 一—
十一一一一
二}
L(
z一1
)0
Z-1
Z一e
-`
j (
Z一1
)(
Z一e-
T)
W hen T =1
we
obt
ain
z
e-,
+1
一2e
-' 0.
3678z+0.
2644
0.3678z+0.
2644
G(
z) (
6.16)
(
z一1
)(
z一e
一,)(
z一1
)(
z一0.
367
8)
za一1
.36
78z+0
.36
78’
Th
e
re
spo
nse
o
f
th
is
s
yst
em
t
o
a
u
nit
imp
uls
e
is
o
bta
ine
d f
or
R(
z)=I
so
t
hat
Y(
z)=G(
z)・
1.
We
ma
y o
bta
in
Y(z
) b
y d
ivi
din
g t
he
d
eno
mina
tor
i
nto
t
he
n
ume
rat
or
a
s:
0
.36
78z
一,+0
.76
75z
-2+0.
914
5z一+…=Y(
z)
z
,一1
.3
678
z+0
.3
678
)0.
36
78z+0
.2
644 ” (
6.
17
)
洲-
洲
0.
3678 z一 0.
5031 + 0.1
353
+ 0.
7675 一 0 .
1353
175
心 -
+0.
767
5一1.
0497
z一
I+0.2823
z
刁
0.
9145
z-1一0.
2823
z
Th
is
c
alc
ula
tio
n外e
lds
t
he
r
esp
ons
e a
t
th
e s
amp
lin
g i
nst
ant
s
an
d c
an
b
e c
arr
ied
a
s f
ar
a
s i
s
ne
ede
d
f
or
Y(
z).
Fro
m Eqn
.
(6
.5)
,
we
ha
ve
Y
(z)
=艺y
(kT
):
一‘
k=0
I
n
th
is
c
se
a we
h
ave
o
bta
in
ed
y
(U)
as
f
ol
lo
ws:
y(
0)=0
,
y(
T)=0
.36
78,
y(
2T)=0.
767
5,
a
nd
y
(3T)
=0 .
914
5.Not
e
that
y(U )
p
rov
ides
t
he
va
lue
s o
f y(t
)
at
t=kT.
We
hav
e d
ete
rmin
edY(z),t
he Z-
t
r
ans
for
m of
t
he
out
put
sa
mple
d s
ign
al.
Th
e z
-
tr
ans
for
m o
f
t
he
i
npu
t
sa
mpl
ed
s
igna
l
is
R(
z).
The
t
ra
nsf
er
f
unc
ti
on
i
n
th
e z-
d
oma
in
i
s
Y(
z)
二G(
z). (
6.18)
R(
z)
Si
nce
we det
e r
mined t
he
sampl
ed
output
,
we
can
use
a
n out
put
s
ample
r t
o depi
ct
thi
s
c o
ndit
ion,
a
s shown in F
ig.6.
10, whi
ch r
epre
sents
th
e s
yst
em of
Fig.
6.9
wit
h t
he add i
t
ional
(ic
f t
it
ious)
s
amp l
es
(shown in c
olor)
.
We ss
a u
me t
hat
bo
th
sampl
ers
have
t
he sa
me s a
mpl i
ng
pe r
iod
and
o
perate
s
ynchronousl
y.
Then
Y(
z)二G(
z)R(
z), (
6.1
9)
a
s r
equi
ed
r .
We ma
y epr
r ese
nt
Eg
n.(
6.1
9),
whi
ch
i
s
a
z-t
ran
sfo
rm
e
qua
ti
on,
by
t
he
bl
ock
di
agr
am
of
Fi
g.6.
11.
F啥.
6.1
0 Sy
ste
m wi
t
h s
amp
led
o
utp
ut
Fi
g.
6
.11
Th
e z
-
tr
ans
for
m t
ran
sfe
r f
unc
tio
n i
n b
loc
k d
iag
ram
f
orm
6.
5 OPEN-
LOOP
DIGITAL
CONTROL
Co
mmerci
all
y avai
la
ble
digi
ta
l a
ctu
ator
s i
ncl
ude e
lec
tri
ca
l and
el
ect
ro-
hydr
aul
ic s
te
ppin
g
mot
ors,
di
git
al
valv
es,
and
di
git
al
pi
st
on/cy
li
nder
s.
Bef
ore
des
cri
bi
ng
the
act
uat
ors
i
n det
ai
l
we
f
ir
st
pre
s e
nt
the
thumbwhee
l s
wit
ch,
a
simpl
e d
evic
e,
cos
ti
ng a
bout
¥
5,
for
ent
eri
ng
numeri
cal
c
omma nds
int
o a
digi
tal
s
ystem.
In
an a
nal
og el
ect
ri
cal
system the
r
efer
enc
e in
pu t
i
s usual
ly
a
c
ont
inuousl
y va
ria
ble
do vol
tage
, s
ay 0
to
10V,
whi
ch we c ou
ld ent
er
manual
ly wi
th a1 0
-tun
r
p
ote
nti
o met
er e
xcit
ed wit
h ap r
eci
se 10
-V power
suppl
y. To ent
er
commands in
digit
al f
orm
r
equ
ire
s th
at
the
digi
tal
"b
its
" us
ed t
o de
ine
f t
he
nume r
ica
l val
ue be
tur
ned
"on" or
"
of f
" i
n t
he
176
prope
r patt
en.
r In
a ty p
ical
dig
ital
elec
tro
nic s
ystem a vol
tag
e a nywhere
betwe en 0a nd 4.
5V
cor
responds
to
"off'
a nd o
ne anywhere
between
7.5 t
o 12V co
r r
espo n
ds t
o "on". Of
the va
r i
ous
codi
ng s
chemes i
n us e,
I wi
ll
de s
cri
be ab i
nay-
r c
oded-dec
ima l (
BCD) method i
n wh i
ch each
deci
ma l
digi
t of
a numbe r i
s s
pe c
ifi
ed by t
he on-of
f pa
tte
n
r of f
ou r
digi
tal s
ignals
. The leas
t
si
gnif
icant
bit
(
LS B) has a
weight
of 1,
t
he nex
t 2,
the
next
4, a
n d
the l
st
a (mo st
sig
n i
fi
cant bi
t
MS B) 8.
To rep
resent
d ec
imal
zero,
none
of
the
four
bit
s
are
HI (on).
To r
epr
ese n
t deci
ma l
3,
both
th
e 1
bit
a n
d t
he 2 b
it are
HI;
to
repre
sent
deci
ma l
9,
the
LSB and the
MSB are
HI ,
and so
fort
h.
Thus ald e
cimal
numb e
rs
between 0a
nd 9 c
an be r
epres
ente
d by some combi n
a ti
on of
the f
our
b
it
s.
F
or
n
ume
ri
ca
l
va
lu
es
g
rea
te
r
th
an
9
we
n
eed
a
n呻e
r
se
t
of
f
our
BCD
i
l
nes
t
o
re
pre
sen
t
th
e
10'
s
di
git
,
an
othe
r s
et
of
fo
ur f
or
th
e 100'
s,
a
nd
s
o f
ort
h.
Thus
,
fo
r e
xamp
le,
t
o r
epr
ese
nt
d
eci
mal
numbe
rs
f
rom 0t
99
o 9 r
equi
res
12
l
ine
s of
BCD
d
igi
tal
data
.
Fi
gur
e 6.
12
s
hows
s
che
mat
i
cal
l
y a
t
humb
whe
el
s
wit
ch
u
sin
g o
ur
BCD
c
ode
t
o e
nte
r ma
nua
ll
y
deci
ma l
numbers
bet
we e
n 0a nd 9i n
to ad
ig
ita
l con
tro
l syst
em. The s
witch i
s sh
own in t
he
posi
ti
o n
neede
d t
o re
prese
nt dec
imal 1;
t
hat
is
, l
ine
Q, i
s HI whil
e Q2,伍, an
d Q8 are
LO. By
sid
l i
ng the movi
ng con
tact
s succe
ssivel
y t
hro
ugh the 1
0 d e
tent
ed posi
ti
ons we se
lec
t al
l t
he
combinat
ions
of Ql,
Q2 ,
Q4, and Q8 nee
ded t
o repr
esent
deci
ma ls
0 th
rough 9.
For l
arge
r
numbe r
s
we re
quir
e o
ne addi
ti
onal
such s
wit
ch f
or
each
decimal
digi
t.
Si
nce
codes
other
t
han ou
r
BCD
ar
e i
n us
e,
s
wit
ches
ar
e a
vai
labl
e i
n va
rie
ty
of
pat
tens.
r
-
NI'
,
r}
}
,p
}}
wel
t妙 t
o好r
l
ei议
Fi
g-6
.12
Th
umb
whe
el
s
wit
ch
(
BCD)
fo
r d
igi
ta
l
da
ta
e
nty
r
Re
tuni
r n
g no
w t
o d
igi
ta
l a
ctu
ato
rs,
Fi
g.6.
13
s
hows
a
f
lui
d p
ist
on/
cyl
ind
er
de
vic
e u
sin
g t
hre
e
b
ina
ry-
wei
ght
ed
s
tag
es
t
o p
rov
ide
l
oad
p
osi
t
ion
ing
ov
er
t
he
r
ang
e 0
t
o
7
u
nit
s
wi
th
a
r
esol
ut
io
n o
f
1
/7
of
f
ull
s
cal
e.
Fo
r e
xamp
le
i
f
th
e LSB
i
s a
0.
01-
in.
mot
ion
, we
c
an
p
rov
ide
p
osi
ti
ons
be
twe
en
177
0.
00
a
nd
0.
07
i
n.
i
n 0.
01-
in.
i
ncr
eme
nts
.
Fi
g.6
.13
s
hows
t
he
a
ctu
ato
r wi
th
t
he
l
oad
po
sit
ion
ed
a
t
6
/7
o
f f
ull
s
cal
e.
Not
e t
hat
e
ach
c
yli
nde
r r
equ
ire
s i
ts
o
wn
v
alv
e,
ho
wev
er
t
hes
e v
alv
es
a
re
i
nex
pens
ive
a
nd
r
eli
abl
e o
n-o
f s
ole
noi
d v
alv
es,
no
t p
rec
isi
on
s
erv
ova
lve
s.
Al
so
t
he
a
ccu
rac
y o
f
l
oad
po
sit
ion
ing
de
pen
ds
o
nly
o
n t
he
c
yli
nde
r f
ull
s
tr
oke
,
whi
ch
i
s e
sse
nti
all
y f
ixe
d a
t
th
e t
ime
o
f
ma
nuf
act
ure
a
nd
c
an
be
h
eld
t
o c
los
e t
ole
ran
ces
. No
po
sit
io
n-me
asu
ring
t
ran
sdu
cer
of
f
eedb
ack
l
oop
i
s
ne
ces
say;
r t
hi
s
is
a
n o
pen
-lo
op
s
yst
em.
So
l
ong
a
s
lui
f d
su
ppl
y
pr
ess
ure
i
s
su
ffi
ci
ent
l
y h
igh,
e
xte
rna
l d
ist
urb
ing
f
orc
es
h
ave
no
e
fec
t o
n pos
it
ioni
ng
a
ccu
rac
y.
We
c
oul
d u
se
l
ine
s Q1
,
Q2
,
an
d
Q4
o
f t
he
t
humbwh
eel
s
wit
ch
o
f F
ig.
6.1
2 t
o t
urn
t
he
s
ole
noi
d v
alv
es
o
n a
nd
of,
the
reby
c
rea
ti
ng
a
s
imp
le
o
pen-
loo
p d
igi
tal
po
sit
ion
ing
s
yst
em
wit
h ma
nua
l
co
mma
nd
i
npu
t.
Fi
g.
6.
13
Di
gi
ta
lactuator
An
oth
er
c
omme
rci
al
ly
a
vai
lab
le
di
git
al
"
act
uat
or"i
s t
he
di
git
al
low-
f cont
rol
val
ve.
Al
thou
gh
i
t
c
an
b
e u
sed
f
or
o
pen-
loo
p c
ont
rol
i
t
is
mor
e o
fte
n us
ed
a
s a
n a
lt
ena
r t
ive
t
o t
he
p
neu
mat
ic
all
y
pos
it
io
ned
pr
opor
ti
onal
va
lve
in
pr
oce
ss
c
ontr
ol
fee
dbac
k sys
tems
usi
ng di
re
ct
d
igi
ta
l c
ont
ro
l
(DDC).
Fi
g.
6.1
4 a
)
compar
es
th
ese
t
wo
met
hods
of
mani
pul
at
ing
proc
ess
lu
fid
low
f i
n r
espo
nse
t
o
di
git
al
c
omma
nds
f
rom
a c
ont
rol
c
ompu
ter
; wh
ere
as
6
.14
b)
s
hows
s
ome
de
tai
ls
of
v
alv
e
c
ons
tru
cti
on
a
nd
o
per
ati
on.
Si
mpl
e o
n-o
f s
ole
noi
d v
alv
es
a
re
ma
nif
old
ed
t
o pr
ovi
de
t
he
d
esi
red
c
omb
ina
ti
on
o
f bi
nar
y-we
igh
ted
lo
f w
ar
eas
, t
he
p
ict
ure
d e
xampl
e u
sin
g s
ix
b
its
t
o p
rov
ide
a
r
eso
lut
ion
o
f 1
/63
o
f f
ull
s
cal
e.
Di
git
al
v
alv
es
h
ave
t
hei
r own
pr
obl
ems
, i
ncl
udi
ng
po
ssi
ble
wat
er-
hammer
t
rans
ient
s due
t
o s
udden
val
ve
cl
osur
es
a
nd
low
f coef
fi
cient
`
uncer
tai
nti
es
whe
n
lo
fw i
s
ne
ar
t
he
l
ami
nar
/t
urb
ule
nt
ra
tnsi
ti
on.
Pr
ope
r
de
si
gn
a
nd
a
ppl
i
cat
ion
c
an
u
sua
ll
y
av
oid
s
uch
d
iff
icu
lti
es.
F
igu
re
6.
15s
.
hows
t
wo
l
ar
ge
d
igi
tal
v
alv
es
(
12
b
its
) t
hat
a
re
u
sed
t
o c
ont
rol
lo
f w
179
(
5-5
00
g
pm)
o
f s
odi
um
hy
poc
hlo
ri
te
i
n a
he
avy-
wat
er
p
lan
t.
Th
e c
orr
osi
ve
lu
fid
r
equ
ire
d u
se
o
f a
s
pec
ial
t
it
ani
um
a
s v
alv
e c
ons
tru
cti
on
ma
ter
ial
.
Some
i
nte
rna
l c
ons
tru
cti
on
d
eta
ils
o
f t
hes
e v
alv
es
a
re
s
hown
i
n F
ig.
6.1
6.
Sol
id
s
tat
e re
hn:
.
O>
nve
nti
onnl
me功 od
撒
Co
ny}
nit
er
c
ont
rol
of
a
1
1i
nd
f
low Ne
w DDI
n
eet
li
od
(
one
pe
r bi
t)
一撇一
撇
矛'
o ces
r s
一
徽
control
一
撒
di
g
igal
一SSR
comput
er
」i
r
source
.
. 呀  ̄ ̄ 明硬 ̄一 自
Pr
oce
ss户f
il
Fr
act
ion
g
fto
tal
Cv
Fi
g.
6.
14
Di
gi
tal
v
alv
e (
Dig
ita
l
Dy
nami
cs
I
nc.
)
By
f
ar
t
he
mos
t
co
mmo
n d
igi
tal
a
ctu
ato
r i
s
th
e e
lec
trome
cha
nic
al
s
tep
ping
mo
tor
,
wi
del
y u
sed
i
n o
pen-
loo
p a
nd
c
los
ed-
loo
p pos
it
ion
a
nd
s
pee
d c
ont
rol
s
yst
ems
.
On
e mi
ght
a
rgu
e t
hat
i
t
is
no
t
r
eal
ly
a
di
git
al
a
ctu
ato
r s
inc
e i
t
do
es
n
ot
r
esp
ond
d
ire
ctl
y t
o a
s
et
o
f bi
nay-
r we
igh
ted
b
its
,
howe
ver
i
t
is
n
ot
u
sua
lly
d
iff
icul
t
to
pr
ogr
am
t
he
c
ompu
ter
t
o p
rod
uce
a
pu
lse
t
rai
n wi
th
a
c
ert
ai
n pu
lse
r
ate
a
nd
t
ota
l nu
mbe
r of
pu
lse
, wh
ich
i
s t
he
t
ype
of
i
npu
t t
he
s
tep
ping
mot
or
r
equ
ire
s.
Ba
sic
all
y,
a
s
tepp
ing
mot
or
pl
us
i
ts
a
sso
cia
ted
dr
ive
r e
lec
tro
nic
s a
cce
pts
pul
se
t
rai
n i
npu
t a
nd
pr
odu
ces
a
n
179
i
mpr
over
n-
ent
of
r
ota
ry
di
spl
ace
ment
,
sa
y 1
.8"
(
200
pl
uses
per
r
evol
uti
on),
for
e
ach
p
uls
e.
Thus
we
c
an
con
tr
ol mot
or
aver
age
spe
ed
by
mani
pula
ti
ng p
uls
e ra
te,
a
nd moto
r p
osi
ti
on
by
cont
rol
in
l g
t
ota
l
pu
lse
c
oun
t.
We
s
hou
ld
i
mme
dia
tel
y ma
ke
c
lea
r t
hat
:
Fi
g.6.
15
12-
bit
di
git
al
va
lve
Fi
g.
6.
16
Co
nst
ruc
ti
on
de
tai
ls
o
f d
igi
ta
l
va
lve
190
a
. I
f
th
e i
npu
t pu
lse
r
ate
i
s t
oo
f
ast
,
th
e mo
tor
wi
ll
"
mis
s"
s
teps
,
ma
king
s
pee
d a
nd
po
sit
ion
i
naccur
ate.
b.
I
f t
he motor
i
s at
res
t,
an
ext
ena
r l
di
stur
bing
t
orqu
e grea
ter
t
han t
he
mo t
or'
s
hol
din
g t
o r
que
c
an t
wi s
t t
he
motor s
haft
a
wa y
fr
om its
commande
d posi
ti
on by
any
number
of
s
tep
s.
Wh en
t
he
dist
urbanc
e is
r
emo v
ed,
th
e moto
r wil
l
not
r
ecov
er i
ts
cor
rec
t p
osi
ti
on.
Two
t
ypes of
s
teppi
ng moto
rs
are
in
commo n
use:
varia
ble
rel
ucta
nce
and
per
manent
magnet
.
Th
e p
erma
nen
t ma
gne
t t
ype
i
s
some
wha
t e
asi
er
t
o u
nde
rst
and
s
o i
t
wi
ll
b
e d
esc
rib
e i
t
br
iely
f .
In
Fi
g.6.
17
t
he
mo
tor
ha
s a
pe
rma
nent
ma
gne
t r
oto
r a
nd
t
he
s
tat
or
ha
s f
our
s
ai
le
nt
e
lec
tro
mag
net
ic
po
les
t
hat
a
re
e
ner
giz
ed
ro
f m
a
do
s
uppl
y.
I
f S3
i
s c
onne
ct
t
o po
sit
ive
po
lar
it
y,
S
l
t
o ne
gat
ive
po
lar
it
y,
a
nd
S
2,
S4
a
re
o
pen,
an
el
ect
roma
gne
tic
f
iel
d i
s c
rea
ted
b
etwe
en
S3
a
nd
S1
wi
th
a
n
ort
h
po
le
a
t S3
a
nd
a
so
uth
po
le
a
t S1
, a
tt
rac
ti
ng
t
he
r
oto
r i
nto
t
he
po
sit
ion
s
hown.
Wi
th
t
his
f
iel
d
ma
int
ai
ned,
i
f we
t
ry
t
o t
wis
t t
he
s
haf
t a
way
rom
f i
ts
p
ref
err
ed
p
osi
ti
on
we
f
eel
a
"
mag
net
ic
s
pri
ng"
r
est
ori
ng
t
orq
ue;
h
owe
ver
a
s
uff
ic
ien
tl
y
la
rge
e
xte
na
r l
t
or
que
c
an
o
ver
come
t
he
ma
gne
ti
c
s
pri
ng
a
nd
we
c
an
t
un
r s
haf
t
an
gle
ea
s
f
ar
a
s we
wi
sh.
Whe
n we
l
et
g
o,
t
he
r
oto
r wi
ll
"
sna
p
b
ack"
a
fte
r s
ome
o
sci
ll
ati
on
t
o t
he
p
osi
ti
on
of
Fi
g.6
.17
, h
owe
ver
we
h
ave
"
los
t"
a
n i
nte
ger
n
umber
o
f
re
vol
ut
io
ns.
To
co
mmand
t
he
r
ot
or
t
o
tu
n9
r 0
(
one
s
te
p)
c
oun
te
rc
lo
ckwi
se
we
c
onne
ct
p
osi
t
ive
t
o
S2,
nega
ti
ve
t
o
S4
(
S
1,
S3
ope
n)
.
A
fur
t
her
C
CW 90
r
equ
ir
es
+S
1,-
S3,
(S2,
S4
op
en)
a
nd
t
he
f
ina
l po
sit
ion
r
equ
ire
s +S
4,-
S2,
(
S1,
S3
o
pen
).
Obv
iou
sly
, c
loc
kwi
se
r
ota
ti
on
mer
ely
r
equ
ir
es
t
he
p
rop
er
s
equ
enc
e of
s
ta
tor
e
xci
ta
ti
ons
,
an
d r
oto
r a
ver
age
s
pee
d c
ont
ro
l
is
o
bta
ine
d by
c
han
ging
t
he
r
ate
a
t wh
ich
s
tat
or
e
xci
tat
ion
i
s s
wit
che
d.
Al
thou
gh
a而c
rop
roc
ess
or
c
an
b
e
pr
ogr
amme
d t
o p
erf
orm
t
he
r
equi
red
s
tat
or
e
xci
ta
ti
on
s
wit
chi
ng,
an
oth
er
a
ppr
oac
h us
es
a
vai
lab
le
Fi
g.6
.17
Pe
rma
nen
t-ma
gne
t
st
epp
ing
mo
tor
i
nte
gra
ted
c
irc
uit
c
hip
s (
st
epp
er
mo
tor
d
riv
ers
)
ins
tea
d.
Th
ese
c
hips
r
equi
re
a
s i
npu
t o
nly
a
pu
lse
t
rai
n a
t t
he
s
tep
pin
g re
f qu
enc
y a
nd
a
l
ogi
c s
igna
l
to
s
pec
ify
CW
or
CCW
rot
ati
on.
An a
dju
sta
ble
181
re
f q
uen
cy
pul
se
t
rai
n i
s r
ead
ily
o
bta
ine
d rom
f ano
the
r I
C c
hip,
a
v
olt
age
-co
ntr
oll
ed
o
sci
ll
at
or
(
VCO)
.
Th
e ou
tpu
t re
f qu
enc
y of
s
uch
c
hip
s d
epe
nds
t
ypi
cal
l
y o
n t
he
v
alu
e of
t
wo
r
esi
sto
rs,
o
ne
c
apa
cit
or,
a
nd
a
n i
nput
v
olt
age
(
dc)
,
al
lowi
ng
e
sy
a re
f q
uen
cy
c
ont
rol
rom
f manu
al
i
npu
ts
o
r d
o
cont
rol
vol
tages
.
A
s
impl
e
op
en-
loo
p s
pee
d c
ont
rol
c
an
b
e ss
a e
mbl
ed
ro
f m
aV
CO,
s
te
ppe
r
mot
or
d
ri
ver
,
po
wer
s
upp
ly,
a
nd
s
tep
per
mot
or.
Fo
r ma
nua
l s
pee
d c
omma
nd
we
c
oul
d u
se
a
1
0-t
un
r po
t s
a av
ari
abl
e
r
esi
st
or
i
n t
he
VCO.
As
lo
ng
s
a s
tepp
ing
r
ate
s a
nd
d
ist
urbi
ng
t
orq
ues
a
re
l
ow
e
nou
gh
t
hat
n
o s
tep
s
a
re
mi
sse
d,
s
pee
d c
ont
rol
wi
ll
be
a
ccu
rat
e,
a
t
ypi
cal
a
ppl
i
cat
ion
b
eing
a
ch
art
-
pap
er
d
riv
e f
or
a
s
tr
ip-
cha
t
r r
eco
rde
r.
By
dr
ivi
ng
s
eve
ral
mo
tor
s ro
f m
t
he
s
ame
f
req
uenc
y s
our
ce,
s
ync
hro
niz
ed
mot
i
ons
a
t d
iff
ere
nt
po
int
s i
n am
ac
hine
a
re
e
si
a l
y o
bta
ine
d.
Us
ing
s
tan
dar
d re
f q
uen
cy-
div
ide
r
c
hip
s we
c
an
d
riv
e a
s
econ
d mo
tor
a
t a
p
rec
ise
ra
f c
ti
on
o
f a
not
her
mo
tor
's
s
pee
d,
g
ivi
ng
a
n
"
ele
ctr
oni
c g
ear
t
rai
n."
Fo
r o
pen-
loo
p po
sit
ion
c
ont
rol
t
he
a
rra
nge
men
t of
F
ig.
6.1
8 c
an
be
us
ed.
Ad
dit
iona
l c
hip
s s
hown
t
her
e a
re
t
he
c
ount
er
, c
omp
ara
tor
, a
nd
f
i
lp
f
lop.
We
s
et
t
he
d
esi
red
po
sit
ion
a
s a
c
eta
r i
n nu
mbe
r o
f mo
tor
s
tep
s o
n t
he
t
humb
whe
el
s
wit
che
s,
a
nd
t
he
va
ria
ble
r
esi
st
or
o
n t
he
VCO
s
ele
cts
t
he
p
uls
e r
ate
(
spe
ed)
wi
th
wh
ich
we
wi
sh
t
o ma
ke
t
he
mo
ve.
As
pu
lse
s a
e
r f
ed
ro
f m
t
he
VCO t
o t
he
mo
tor
,
th
e c
oun
ter
c
ount
s
th
em
a
nd
s
ends
t
his
n
umb
er
t
o t
he
c
omp
ara
tor
,
S
c
t
I
)l
t
l
se
rate
厂一 一I
el
l
/CCT
V
l
ogi
c
in
put
了、rhe
train
」人ttor
了(.
'O dri
ver
E;abl
e/di
subl
e
岭i
s
in
put
BCD Co
unt
u
p-do
,二,
cor
n er
砷i
s i
nput
Bl
I
B:
I
B_
41
Bi
_
d>B
BCD
」 <B
Comparat
or 」 二B
JLnr
dul
el
l
/
l
l
介〕’
r
nl
_
4
,
f
i
.
t
4
m
:3
l
9
B
r
b
1
.
C
'
I
4
}
t
r
.
D
e
l
r
t
eA
l
4
Fi
g.6
.18
Ope
n-l
oop
p
osi
ti
on
c
ont
rol
us
ing
s
tep
mo
tor
182
wh
ich
a
lso
r
ece
ive
s t
he
t
humb
whe
el
s
wit
ch
da
ta
a
nd
c
ompa
res
t
he
t
wo.
When
B
(p
uls
es
c
oun
ted
)
e
qua
ls
A
(pu
lse
s wa
nte
d)
t
he
c
ompa
rat
or
A=B
l
ogi
c s
igna
l go
es
HI
a
nd
d
isa
ble
s t
he
VCO,
s
top
ping
t
he
p
uls
es.
It
we
no
w c
han
ge
t
he
t
humb
whe
el
s
wit
ch
s
ett
i
ng
i
n e
it
her
d
ire
cti
on,
th
e A<B,
A>B
c
omp
ara
tor
o
utpu
ts
(
thr
ough
t
he
f
i
lp-
lop
f ) s
wit
ch
t
he
mo
tor
d
riv
er
t
o t
he
p
rop
er
(
CW/
CCW)
d
ir
ect
io
n a
nd
s
inc
e
A
#
B.Th
e A二B
in
le g
oes
LO,
ena
ble
t
he
VCO
a
nd
pr
odu
cin
g p
uls
es
un
ti
l
t
he
c
omp
ara
tor
i
s a
gai
n s
ati
sf
ied.
Not
e t
hat
mot
or
mo
tio
n i
s n
ot
s
ens
ed,
t
hus
i
f
st
eps
a
re而s
sed
d
ue
t
o e
xce
ssi
ve
pu
lse
r
ate
a
nd/
or
di
st
urbi
ng
t
orq
ues
,
th
e s
yst
em
doe
s no
t k
now
t
his
a
nd
e
rro
r
r
esul
ts.
Si
nce
t
he
c
ost
a
nd
s
impl
i
cit
y a
dva
nta
ges
o
f s
tepp
er-
mot
or
mo
tio
n c
ont
rol
s
yst
ems
e
rode
whe
n
mo
tio
n s
ens
ors
a
nd
f
eed
bac
k l
oops
a
re
a
dde
d,
muc
h e
ffo
rt
ha
s、
gone
i
nto
i
mpr
ovi
ng
t
he
p
erf
orma
nce
of
ope
n-l
oop
sys
tems
.
A pa
rti
cul
ar
mot
or/
loa
d i
ner
ti
a wi
ll
exh
ibi
t a
n
e
rro
r-f
ree
-st
art
-s
top
r
ate
(
EFSS
R)
t
hat
d
epe
nds
o
n l
oad
t
orq
ue
s
a i
n Fi
g.6
.19
. As
l
ong
a
s o
ur
c
omma
nds
d
o n
ot
r
equ
ire
t
he
s
yst
em
t
o e
xce
ed
i
ts
EFSS
R,
t
her
e wi
ll
b
e o
ne
mo
tio
n p
uls
e f
or
e
ach
e
lec
tri
cal
c
omma
nd
pu
lse
a
nd
a
ccu
rac
y wi
ll
b
e p
res
erv
ed.
By
a
cce
ler
ati
ng
a
nd
d
ece
ler
ati
ng
t
he
l
oad
g
rad
ual
ly,
we
ca
n e
nsu
re
s
uch
o
per
ati
on.
Thi
s
is
c
all
ed
r
ampi
ng
a
nd
i
s
esi
a l
y a
chi
eve
d whe
n
we
u
se
a
VCO
s
a t
he
pu
lse
t
rai
n s
our
ce,
s
inc
e t
he
re
f q
uen
cy
c
an
b
e s
che
dul
ed
wi
t
h a
t
ime
-va
ryi
ng
d
o i
nput
vo
lta
ge.
Bot
h e
xpo
nent
ia
l
an
d s
tra
ight
-l
ine
"
ramp
s"
a
re
u
sed,
ex
pon
ent
ial
be
ing
p
ref
err
ed
f
or
a
cce
ler
ati
on
a
nd
s
tra
igh
t-l
ine
be
ing
a
dequ
ate
f
or
d
ece
ler
ati
on.
Th
e c
irc
uit
y
r f
or
c
onv
ert
i
ng
a
c
omma
nd
p
uls
e r
ate
t
o t
he
pr
ope
r s
eque
nce
o
f mo
tor
p
hase
s
wit
chi
ngs
i
s c
all
ed
at
r
ans
lat
or;
wh
ere
as
t
he
me
ans
f
or
g
ene
rat
i
ng
a
pr
ope
r pu
lse
r
ate
a
nd
c
ount
i
s c
all
ed
a
n i
nde
xer
.
Fo
r t
hos
e
u
ser
s who
mus
t t
rad
e d
esi
gn
t
ime
a
nd
s
kil
l f
or
c
ost
,
re
ady-
mad
e s
yst
ems
o
f g
rea
t v
ers
ati
li
ty
i
ncl
udi
ng
mot
or,
i
ndexe
r,
a
nd
t
rans
lat
or
a
re
avai
labl
e.
之l
b、足 Pno7
100 200
(
Pil
l
seb
/S)
Fi
g.6.
19
Er
ror
-re
f e
-st
at-
r s
top-
rat
e
fo
r s
tep
The
s
ingl
e-s
tep
r
espons
e of
s mot
ors
i
s of
ten
qui
te
os
cil
lat
ory
and
va
rious
s
che
mes
f
or
i
mpr
ovi
ng
d
amp
ing
a
re
a
vai
la
ble
. Th
e Lan
che
ste
r t
ype
o
f me
cha
nic
al
d
amp
er,
whi
ch
was
tes
n
o
193
e
ner
gy
i
n f
ri
cti
on
f
or
c
ons
tant
v
elo
cit
i
es,
i
s a
s
imp
le
s
olu
ti
on
i
n s
ome
c
se
a s
. A
mor
e c
ompl
ex
a
ppr
oac
h u
ses
t
he
l
st
a f
ew
p
uls
es
o
f a
mo
ve
t
o a
ppl
y ma
gne
tic
c
oun
ter
-t
orq
ue
(
"br
aki
ng
pu
lse
s")
t
o r
edu
ce
o
sci
ll
at
ion.
Even
mor
e c
omp
lex,
b
ut
q
uit
e s
ucc
ess
ful
,
is
t
he
c
onc
ept
o
f mi
cro
ste
ppi
ng.
Th
e mo
tor
of
F
ig.
6.1
7 h
as
f
our
s
teps
p
er
r
evo
lut
ion
,
but
it
i
s
mo
re
c
ommo
n t
o
us
e a
ddi
ti
ona
l s
tat
or
a
nd
r
oto
r ma
gne
tic
p
ole
s t
o g
et,
s
ay,
2
00
s
teps
p
er
r
evol
ut
ion
. Mi
cro
ste
ppi
ng
p
rov
ide
s 2
0,00
0 t
o
5
0,0
00
s
teps
p
er
r
evo
lut
ion
a
nd
i
s
not
a
cco
mpis
lhed
by
mo
tor
r
ede
sign
bu
t b
y d
riv
ing
a
s
tan
dar
d
2
00
s
tep
s pe
r r
evo
lut
ion
mot
or
wi
th
"
sma
t"
r e
lec
tro
nic
s.
The
b
asi
c c
onc
ept
c
an
b
e u
nde
rst
ood
i
n
t
erms
o
f Fi
g.6.
17.
By
e
ner
giz
ing
b
oth
S1
/S3
a
nd
S2
/S4
s
imu
lta
neo
usl
y,
t
he
r
oto
r ma
gne
ti
c f
iel
d
c
an
b
e or
ie
nte
d a
t 45',c
aus
ing
t
he
r
oto
r t
o mo
ve
45',r
at
her
t
han
9
0*.Us
in
g t
he
p
rop
er
c
omb
ina
ti
ons
o
f i
ndi
vidu
al
a
nd
s
imu
lta
neou
s s
tat
or
p
ole
e
xci
ta
ti
ons
, we
no
w h
ave
do
ubl
ed
t
he
n
umb
er
o
f s
teps
pe
r r
evo
lut
ion.
If
we
n
ow
c
an
a
lso
a
djus
t
th
e r
ela
tiv
e s
tr
engt
hs
o
f t
he
s
tat
or
p
ole
e
xci
ta
ti
on
(
cur
rent
) f
or
t
wo
a
dja
cen
t po
les
, t
he
a
ngu
lar
or
ien
tat
ion
of
t
he
s
tat
or
f
iel
d (
and
t
hus
r
oto
r p
osi
ti
on)
c
an
be
i
nde
xed
i
n e
ven
f
ine
r s
teps
,
cu
rre
nt
p
rac
tic
e p
rov
idi
ng
t
he
2
0,0
00
t
o 5
0,0
00
s
tep
s pe
r r
evol
uti
on
me
nti
one
d e
arl
ie
r.
Thi
s
te
chn
iqu
e n
ot
o
nly
i
mpr
ove
s da
mpi
ng
b
ut
g
ive
s f
ine
r
a
ngu
lar
mo
ti
on
r
eso
lut
ion
t
han
e
xpe
nsi
ve
s
haf
t
en
cod
ers
i
n c
omp
eti
t
ive
c
los
ed-
loo
p s
yst
ems
.
Al
tho
ugh
t
he
a
dva
nta
ges
of
s
tep
per
dr
ive
s i
n o
pen
-lo
op
s
yst
ems
a
re
mo
st
o
bvi
ous
,
cl
ose
d-l
oop
a
ppl
ic
ati
ons
a
lso
e
xis
t.
So
meo
ne
c
omp
are
s
se
ver
al
a
lt
er
nat
iv
e s
olut
io
ns
t
o
a
l
ine
pr
int
er
p
ape
r-f
eed
d
riv
e p
robl
em
and
s
how
how
a c
los
ed-
loo
p s
tepp
er
d
riv
e c
an
b
e a
nal
yze
d u
sing
c
las
sic
al
t
ech
niq
ues
(
roo
t l
ocu
s,
e
tc.
)
empl
oye
d f
or
c
ont
i
nuo
us-
mot
io
n s
yst
ems
.
Mo
st
s
te
ppe
r-
mot
or
d
riv
es
(
eit
her
o
pen
l
oop
o
r c
los
ed
l
oop
) a
re
r
ela
ti
vel
y
low
po
wer
(
a f
ew
h
und
red
wa
tts
o
r l
ess
),
t
he
l
arg
est
a
vai
la
ble
e
lec
tri
cal
mot
ors
c
urr
ent
ly
be
ing
a
bou
t 3
hp.
Whe
n hi
gh-
powe
r d
riv
es
u
sin
g t
he
s
tep
-mo
tor
p
rin
cip
le
a
re
r
equ
ire
d,
e
lec
tro
-hy
dra
uli
c s
tep
mot
ors
a
re
a
vai
la
ble
. Th
ese
u
se
a
s
mal
l
e
lec
tro
mec
han
ica
l
st
ep
mot
or
t
o p
osi
ti
on
t
he
s
poo
l o
f a
s
erv
o-v
alve
i
n a
n o
pen
-lo
op
f
shi
a on.
Us
ing
me
cha
nic
al
f
eed
bac
k i
n a
c
los
ed-
loo
p f
ash
ion
, t
he
s
poo
l p
osi
t
ion
i
s
re
pro
duc
ed
a
t a
h
igh
po
wer
l
eve
l (
10
hp
o
r mo
re)
b
y ar
ota
ry
h
ydr
aul
ic
mo
tor
.
Fi
gur
e 6.
20
s
hows
t
wo
s
che
mes
us
ed
t
o
i
mpl
eme
nt
t
his
c
onc
ept
. I
n 6
.20
(
a)
a
l
ow-
powe
r e
lec
tri
cal
s
tep
mot
or
r
ota
tes
t
he
s
pool
o
f a
s
erv
o-v
alv
e,
wh
ich
a
lso
t
hen
t
rans
lat
es
(
ope
nin
g t
he
v
alv
e)
be
cau
se
t
he
l
eft
end
of
t
he
s
poo
l i
s
th
e
s
cre
w o
f a
ba
ll
s
cre
w/nu
t u
nit
.
Th
e nu
t o
f t
his
u
nit
r
ota
tes
wi
th
t
he
h
igh
-po
wer
o
utpu
t s
haf
t
of
a
r
ota
ry
h
ydr
aul
i
c mo
tor
d
riv
en
b
y t
he
v
alv
e.
Ro
tat
i
on
o
f t
his
s
haf
t
te
nds
t
o t
ra
nsl
at
e
th
e v
alv
e s
poo
l
ba
ck
t
owa
rd
i
ts
c
los
ed
p
osi
ti
on,
t
hus
h
ydr
aul
i
c mo
tor
r
ota
ti
on
c
ont
inu
es
u
nti
l
it
j
ust
e
qua
ls
t
he
r
ota
tio
n c
omma
nde
d b
y t
he
e
lec
tri
cal
s
tep
mot
or,
wh
ere
upo
n t
he
h
ydr
aul
i
c mot
or
s
tops
,
si
nce
t
he
v
alve
h
as
n
ow
r
etu
ne
r d
t
o n
ull
.
Thi
s
hy
dro
-me
cha
nic
al
a
rra
nge
men
t i
s a
c
omp
let
e c
los
ed-
loo
p
s
erv
o s
yst
em,
thu
s hy
dra
uli
c
mot
or
pu
lse
s d
upic
l a
te
t
hos
e o
f t
he
e
lec
tri
cal
mo
tor
.
Howe
ver
,
if
t
he
o
ver
all
mo
tio
n c
ont
rol
s
yst
em
doe
s n
ot
me
asu
re
l
oad
mot
ion,
we
c
an
s
til
l
ha
ve
e
rror
s d
ue
t
o t
he
e
lec
tri
cal
mo
tor
s
haf
t
no
t du
pic
l a
ti
ng
e
ach
e
lec
tri
cal
i
nput
p
uls
e.
Fi
gur
e 6
.20
(
b)
s
hows
a
s
imi
lar
194
a
rrangeme
nt
f
or
t
rans
lat
iona
l l
oad
mot
ion.
Rot
ay
r mot
orsup
t
o a
bout
20
hp
a
re
ava
ilabl
e,
a
t
ypi
cal
2
0 h
p mo
tor
ha
viga
n s
tep
s
ize
o
f 0.
720 m axim u m pul
se
r
ate
of
1
3,000/
s,
EFSSR
of
2000
pul
ses
/s,
wi
th
t
orqueof
1
150
i
n.-
Ibf
a
t t
his
pul
ser
ate
.
A
t
ypi
cal
t
ran
sla
tio
nal
a
ctu
ato
r mi
ght
h
ave
a
s
tep
s
ize
of
0.
001
or
0.
0005
i
n.
wi
th
a
ma
ximum s
peed
of
300
i
n/mi
n.
功"
dra
rdr
r
pi
st
on
r
uot
or El
ect
ric
s
teppi
rr
g mot
or
Ball
,
screw-
nut
uni
t
,
s
a
w
-
r
u
o
F hi
-,
lrnr
rl
ic
ral
re
Fi
g.6
.20
El
ect
ro-
hyd
raul
i
c s
tep
pin
g a
ctua
tor
s
6.
6 CLOSED-
LOOP
FEEDBACK
SAMPLED-
DATA
SYSTEM S
In
t
his
s
ect
ion
we con
sid
er
cl
ose
d-l
oop,
sample
d-da
ta
cont
rol
sy
s t
ems.
Consi
der
th
e s
yst
em
s
hown
in
F
ig.
6.21
(
a).
The s
ampl
ed-
dat
a z-
tr
ansf
onn mode
l o
f t
his
f
igure
wi
th a
sampl
ed-
out
put
s
ign
al
Y(
z)
i
s s
hown
i
n Fi
g.
6.
21
(
b).
Th
e c
los
ed-
loo
p t
ra
nsf
er
f
unc
ti
on
T(
z)
(
usi
ng
bl
ock
di
agr
am
r
educ
tion)
i
s
Y(
z) G(
z)
=
T(z
) (
6.20)
R(
z) 1
+G( z
)
195
,
、
:)
一 ̄/产
(a)
十 ̄ 月Z
)
G(
Z) Y(
Z)
(
b)
Feedbac
k cont
rol
s
yst
em
wit
h uni
ty
f
eedbac
k
He
re
we
a
ssume
t
hat
t
he
G(z)
i
s t
he
z
-t
ra
nsf
orm
of
G(s
)=玩(
s)
G
p
(
s),
wh
er
e GO
(s)
is
t
he
z
er
o-
or
de
r
ho
ld,
an
d GP
(
s)
i
s
t
he
p
la
nt
t
r
ans
fe
r
fu
nct
i
on.
exa
A n mpl
e
of
a
d
igi
tal
c
ont
rol
s
yst
em
i
s
th
e r
obo
t d
og
Ai
bo
s
hown
i
n Fi
g.
6.
22.
Th
e f
eed
bac
k
cont sy
rol ste
m o
f o
ne
j
oint
wi
th
a
d
igi
tal
co
ntr
oll
er
i
s
sh
own
i
n F
ig.
6.
22
(
a).
Th
e z
-t
ran
sfo
rm
b
loc
k
d
iag
ram
mod
el
i
s
sho
wn
i
n
Fi
g.
6.
22
(
b).
The
c
los
ed-
lo
op
t
ra
nsf
er
f
unc
ti
on
i
s
Y(
z) G(
z
)D( z)
=T(
z) (
6-21)
R(
z) 1+G(z)
D( z)
DI
S;W
cnnv川 l
et
r
(
r) ̄/户 口(
a) 。闷卜Y
(
Z)
几(L)
186
Fi
g.
6.
22
Ai
bo
i
s a
s
ophi
st
ic
ate
d e
nte
rta
inme
nt
r
obo
t.
Ai
bo
l
ook
s l
ike
a
Ch
ihu
ahu
a a
nd
wa
gs
i
ts
t
ail
,
doe
s t
ri
cks
a
nd
go
es
f
or
wa
lks
.
Ai
bo
d
epe
nds
o
n a
wi
de
r
ang
e of
s
ens
ors
, i
ncl
udi
ng
t
ouc
h,
c
olor
CCD
c
ame
ra,
ra
nge
f
ind
er
a
nd
v
elo
cit
y s
ens
ors
.
A
6
4-b
it
RI
SC
mi
cro
pro
ces
sor
a
nd
1
6MB
o
f
me
mor
y a
re
bu
ilt
-i
n.
I
t
ha
s 1
8 j
oin
ts
p
owe
red
b
y 1
8 mot
ors
. (
a)
F
eed
bac
k c
ont
rol
s
yst
em
wit
h a
d
ui
o
6i
二 t
几a
“l
‘ 二c
.o
Un
l
ltr
o
“Ul
几l
e
 ̄ r
.
・(
b
。)
、 Bl
o
分c
,户1 k
“d
协i
a
“舀r
几am
‘m
 ̄二 二
1 o
de
 ̄协‘l
玉.
.No
Ut
人’ e
 ̄ t
h
“1at
““G
甘(
、z)=Z
‘,一  ̄ 七。 (
卜.0、s
。、
jG _
(
\:
"、}.
户
P ,
Ij.
EXAM PLE
6.4 OF
CLOSED-
LOOP
SYSTEM
No
w l
et
us
c
ons
ide
r t
he
cl
ose
d-l
oop
sys
tem
as
s
hown
i
n Fi
g6.
23.
We
ha
ve
obt
ai
ned
t
he
z一t
rans
for
m model
of
t
his
s
yst
em
as
s
hown
i
n Fi
g.6.
21.
The
ref
ore
we
ha
ve
Y(
z)
G(习
(
6.22)
R(
z) 1+G(z)
I
n Ex
amp
le
6.
3 we
o
bta
ine
d G(
z)
a
s Eg
n(6
.16
) wh
en
T=1
s
econ
d.
Su
bst
it
ut
ing
G(
z)
i
nt
o
Eg
n(6.
22)
we
o
bta
in
1
'() 0.
Z 36782+0.
2644
(
6.23)
R(
z) z
2一z+0.
632
2’
S
inc
e t
he
i
npu
t
is
a
u
nit
st
ep,
之
R(
z)= (
6.24)
之一I
t
hen
z
(0.
3678z+0.
2644) 0.
3678
22+0.
264
42
Y(
z) :
二二 一 二二一
(
z一1
)(
22一z+0
.63
22)
z3一2
22
+1
.63
222一0
.63
22
Co
mpl
et
in
g t
he
d
ivi
si
on,
we
ha
ve
Y(
z)=0.
3678
2一,
+z-
2
+1
.4z
-3
+1
.4z
一+1
.14
7z-
5.… (
6.25)
e
(
r) e"h)
r(
i) t(
Go
(S
) s(
s+ I)
T= 1
Fi
g.
6
.23
A
cl
ose
d-l
oop,
sa
mpl
ed-
dat
a
sy
ste
m
Th
e v
alue
s o
f
y(
kT)
a
re
s
hown
i
n Fi
g.
6.
24,
u
sin
g t
he
s
ymb
ol口。
The
c
omp
let
e
re
spo
nse
o
f t
he
s
ampl
ed-
dat
a,
c
los
ed-
loo
p s
yst
em
is
s
hown
(
obt
ain
ed
by
MAT
LAB;
s
ee
Sec
tio
n 6.
13)
a
nd
c
ont
ras
te
d t
o t
he
r
esp
ons
e o
f a
c
ont
in
uou
s s
yst
em
(
whe
n T=0)
.Th
e o
ver
sho
ot
o
f t
he
s
amp
led
s
yst
em i
s
45%,
i
n c
ont
ras
t
to 1
7% for
t
he
co
nti
nuous
sy
stem.
Fu
rth
ermo
re,
th
e s
ett
li
ng
t
ime
o
f t
he
s
ample
d s
yst
em
i
s t
wic
e a
s l
ong a
s t
hat
of
t
he
cont
inu
ous s
yst
em.
I只7
(
-(
t1
16
了
.刃‘
:
/ 尸声尸韶 ̄入 {
,‘
.声
t.
0 // 卜闷入一 睁 ,‘ ....侧,
\ 〕、.
E
t).
。甘
名E
t
f
(
(
)1
t
tl
X
t
l
IX
t
,
一。,
毛
、
“。
一扮
:
‘
Il.
‘
办
君
l).
J
r !
获
,曲
一
匕_
______
_
/
6
T
il
tt
e(verot
tds
)
Fi
g
.
6.
24
Th
e r
espons
e o
f a
s
econd
-or
der
s
yst
em:
a
) c
ont
i
nuo
us
(T=0),not
sa
mpled;
b)
s
ampl
ed
sys
tem,
T=I
s
eco
nd
6.
7 STABILI
TY
ANALYSIS
IN
THE z
-
PLANE
A
in
l e
ar
c
ont
i
nuou
s f
eed
bac
k c
ont
rol
s
yst
em
i
s s
tab
le
i
f
alp
ole
s of
t
he
c
los
ed-
loo
p t
ran
sfe
r
f
unc
ti
on
T(
s
)
i
le
i
n t
he
l
eft
h
alf
o
f t
he
s-
p
la
ne.
Th
e z-
p
l
ane
i
s
re
lat
ed
t
o
the
s-
p
la
ne
b
y t
he
t
rans
for
mat
ion
z=e
ST=e
`a
+1
XU
)
T (
6.
26
)
We
ma
y a
lso
wr
it
e
th
is
r
ela
ti
ons
hip
a
s
'
z
}一e
d
r
and
Lz=(
OT. (
6.2
7)
I
n
the
l
et-
fhan
d s
-
pl
ane
,
o'
<0,
and
t
her
ef
ore
t
he
r
el
at
ed
ma
gni
tu
de
o
f zv
ari
es
b
etwe
en
0
a
nd
1
.
Th
ere
for
e t
he
i
mag
ina
ry
a
xis
o
f t
he
s
-
pla
ne
c
orr
espo
nds
t
o t
he
u
nit
c
irc
le
i
n t
he
z
-pl
ane
,
an
d t
he
i
nsi
de
of
t
he
u
nit
ci
rcl
e c
orr
espo
nds
t
o t
he
l
eft
h
alf
of
t
he
s
-
pla
ne.
Th
ere
for
e we
c
an
s
ta
te
t
hat
a
s
amp
led
s
yst
em
i
s
st
abl
e
if
a
ll
t
he
po
les
of
t
he
c
los
ed-
loo
p t
ra
nsf
er
f
unc
ti
on
T(z
)
li
e
wi
th
in
t
he
u
nit
ci
rc
le
of
t
he
z-
p
la
ne.
EXAMPLE
6.5
STABILITY
OF
A CLOSED-
LOOP
SYSTEM
Le
t
us
c
ons
ide
r
the
s
yst
em
s
hown
i
n
Fi
g-
6.
25,
whe
n T=1
a
nd
K
G_
尸
(
s)二
(
6.28)
s(
s+1
)
Re
cal
i
lng
Eq
n.
(
6.1
6),
we
no
te
t
hat
!88
G
(
z)
・z
K
2
(-
0
.
1
3
.6
3
7
6
8
7z
8
+
2+
0
0
.
2
.6
3
4
6
4
7)
8
:
“一(
1+a
)z+a
K(
az+b)
(
6.29)
whe
re
a=0.
3678,
and
b=0.
2644.
r
(t)
r
(t)十 e(
t) e击
)
,
(
Gr
}
(
s) GI
)
(
V
I)
Fi
g.
6.
25
A
cl
ose
d-l
oop
s
ampl
ed
s
yst
em
Th
e po
les
o
f t
he
c
los
ed-
loo
p t
ran
sfe
r f
unc
tio
n T(
z)
a
re
t
he
r
oot
s
of
t
he
e
qua
tio
n
[
1+G(
z)]=0.
We
c
all
q(
z)=1+G(
z)=0
t
he
c
har
act
er
is
ti
c
eq
uat
io
n.
The
ref
ore
we
o
bta
in
q
(z)
=1
+
G(z
)
=
z
2一(
l+a
)z+a+Ka
z+Kb二0.
(6
.3
0)
When
K =1,we
have
q(
z)=z
“一z+0.
632
2=(
z一0.
50+
j
0.61
82)
(z一0.
5一j
0.61
82)=0.
(
6.
31
)
Th
ere
for
e t
he
s
yst
em
i
s
st
abl
e
be
cau
se
t
he
r
oot
s
li
e
wi
t
hin
t
he
u
nit
ci
rcl
e.
Wh
en
K=1
0,we
ha
ve
q(
z)=z+
2
22 .
3102+3
.01
2=(
z+1
.15
5+j
l.
295
)(z+1
.15
5一j
1.2
95)
,
(
6.
32
)
a
nd
t
he
s
yst
em
i
s u
nst
abl
e b
eca
use
b
oth
r
oot
s
i
le
o
uts
ide
t
he
u
nit
c
irc
le.
Thi
s
sy
ste
m i
s s
tab
le
f
or
0< K < 2.
39.The
l
ocus
of
t
he
r
oot
s as
K var
ies
i
s di
scus
sed
i
n Sect
ion
6.
11.
We
no
ti
ce
t
hat
a
s
eco
nd-
orde
r s
amp
led
s
yst
em
c
an
be
u
nst
abl
e wi
th
i
ncr
esi
a n
g ga
in
whe
re
a
s
eco
nd-
ord
er
c
ont
inuo
us
s
yst
em
i
s s
tab
le
f
or
a
ll
v
alu
es
o
f g
ain
(
ass
umi
ng
bo
th
t
he
p
ole
s o
f t
he
o
pen
-lo
op
s
yst
em
l
ie
i
n t
he
l
eft
h
alf
s
-
pla
ne)
.
6.
8 PERFORMANCE
OF
A
SAMPLED-
DATA,
SECOND-
ORDER
SYSTEM
Le
t
u
s co
nsi
der
the
p
erfo
rmance
of
a
s
amp
led
s
eco
nd-
ord
er
s
yst
em
wit
h a
z
ero-
orde
r ho
ld,
a
s
s
hown
i
n Fi
g.6
.25
,
when
the
pl
ant
i
s
K
Gp
(s
)= (
6.33)
s
(
TS+1
)
We
t
hen
o
bta
in
G(z
) f
or
t
he
u
nspe
cif
ie
d s
amp
li
ng
p
eri
od
Ta
s
号
G(
z)
Ki
z一E)
[T一,
t(z一1
)]+i(
z一1
)
(
6.34)
(
z一1
)(z一E)
r
_一了h
m , ,.
1. o 1
wnere a = e ・ r
he
staor
nty
or
the
sys
tem i
s anat
yzea
oy
cons
taenng
the
cnar
act
enst
nc
equat
ion
q()=
Z
z
+
z z
{K[
T一;
(1一E)
]
-
(
1+E)
}+
K[
z
(
1一E)
一TE]
+E=0. (
6.35)
Because
t
hep
oly
nomi
al
q(
z)
i
s a
q
uad
rat
ic
a
nd
ha
s r
eal
c
oefi
ci
ent
s,
t
he
ne
ces
sar
y a
ndsufficient
199
c
ond
it
io
ns
f
or
q(
z)
t
o
ha
ve
a
ll
i
ts
r
oot
s
wi
th
in
t
he
u
nit
c
ir
cl
e
ar
e
j
q(
0)
j<1
,
q(
1)
>0,、
(一1
)>0・
Tab
le
6.
3
Max
imu
m Ga
in
f
or
a
S
eco
nd-
Orde
r S
amp
led
Sy
ste
m
T/丁 0 0.
1 0.
5 1 2
M axi
mum Kz CO 20.
4 4.
0 2.
32 1
,45
Th
ese
s
tabi
it
l y
c
ond
iti
ons
f
or
a
s
econ
d-o
rde
r s
yst
em
ca
n b
e e
sta
bli
she
d b
y ma
ppi
ng
t
he
Z-
p
lan
e
c
har
act
er
is
ti
c
equ
ati
on
i
nt
o t
he
s-
p
la
ne
a
nd
c
hec
kin
g f
or
p
osi
t
ive
c
ef
o ic
fien
ts
o
f q(
s).Us
in
g
t
hes
e c
ond
iti
ons
,
we
e
sta
bl
is
h t
he
n
ece
ssa
ry
c
ond
iti
ons
f
rom
Eqn
.
(6.
35)
a
s
1一E
K2< (
6.36)
1
一E一(
T
/
z)
E’
2(
1+E)
K,
r< (
6.37)
(
T
/
z)
(1
+E)
一2(
1一E)’
a
nd
K>0,
T>0.Fo
r t
hi
s
sy
ste
m we
c
an
c
alc
ula
te
t
he
ma
ximu
m g
ain
pe
rmi
ss
ibl
e f
or
a
s
tabl
e
s
yst
em.
Th
e maxi
mum gai
n al
l
owabl
e i
s g
iven
in
Tabl
e 6.
3 f
or
s
eve
ral
va
lue
s of
T /
Z.I
f t
he
c
omput
er
sys
tem
ha
s s
ufic
f i
ent
sp
eed
of compu
tat
ion
and
dat
a h
and
lin
g,
it
i
s po
ssi
ble
t
o s
et
TJz=0.
1 a
nd
o
bta
in
s
yst
em
c
har
act
er
is
ti
cs
a
ppr
oac
hin
g t
hos
e of
a
c
ont
in
uou
s (
nons
ampl
ed)
s
yst
em.
Th
e ma
ximu
m ov
ers
hoo
t
of
t
he
s
eco
nd-
ord
er
s
yst
em
f
or
a
u
nit
s
te
p i
npu
t
is
s
hown
i
n
Fi
g.
6.
26.
The
p
erf
orma
nce
c
rit
eri
on,
int
egr
al
s
qua
red
e
rro
r,
c
an
b
e wr
it
te
n s
a
1
I
丁 J
o
e
2
(
t
)d
t’ (
6.38)
2‘
夕 }
、
\\ 火
0
,口
-
\ \ \ SO恢岌反
\
、
St
abi
li珍l
i
nni
l
弓}
、
矛通
.
、 、、
\
4
_
}
K
-
r
\ \、 又众 0.
?
、、、
1.
夕
卜
、、
}、、
 ̄,月..巨
\、
L
,'
I=0
.1 }( 、气、、 、七
 ̄ . 卜匆.‘ . .一 .一
a
.
,
卜,妇卜一 耳 }
1
.
 ̄,..叫巴巴匕 .阅.1门.门 ......
}
,....如.口 .....曰..
口...匕..晌 ...
0
0.
25
0.
SO .水
)
( 1.
0 1.
25
150
1万 2.
0
r.
"公
F
ig.
6.
26
Th
e
ma
xi movershoot }
mu ,}
f
or
a
s
econ
d-o
rde
r s
ampl
ed
s
yst
emfor
a
uni
tst
ep
i
nput
190
Th
e l
oci
o
f
th
is
c
ri
te
ri
on
a
re
gi
ven
i
n
Fi
g.
6.
27
f
or
c
ons
ta
nt
v
alu
es
of
I.
Fo
r
a
gi
ven
v
alu
e o
f
T/T,one
can
det
ermi
ne
t
he
mi
nimum
vaueo
l f
1
a
nd
t
he
r
equ
ir
ed
v
alu
e o
f
KT.
The
o
pti
mal
c
urv
e s
hown
i
n F
ig.
6.
27
i
ndi
cat
es
t
he
r
equ
ird
e KT
f
or
a
s
pec
ifi
ed
Tl
}
r t
hat
mi
nimi
zes
I.Fo
r
e
xamp
le,
whe
n T
/
T“0.
75,we
r
equ
ir
e KT=I
i
n o
rde
r t
o mi
ni
miz
e t
he
p
erf
orma
nce
c
rit
eri
on
1.
,从
亏
\ 双\
\
,
。
,朴
月仔
I
‘ 钾 _ 一  ̄ 一 一
众 \汉一
{
1.
5
一
又“
} \
}
灯 i
v
i
n
t
i
r
l
c
1u
=
r
v
1
e
.
5
左公
0 m
a
1. r尸
// 一 ,一 一 ,毛
匕、、
0.5
I= 王5一
一 ̄份尸一 ̄ f
{
__
、
伪;
_
_
__! 乏- 斧 、、,、
{
}
‘ ̄ ̄, ̄曰川‘闷,一 丫 圈
一 ! 2口
1
l .
l 一 口叭 .曰....臼曰阳卜.. 口晚目.口.
} { ! - 2.
4
0
0.
2s d
l
.
SO 从水 1.
0
1
.21 1
,51
1
1.
万
ir
f
r
Fi
g.
6.
27
Th
e l
oci
o
f i
nte
gra
l
sq
uar
ed
e
rro
r
f
or
a
s
eco
nd-
orde
r
sa
mpl
ed
s
yst
em
f
or
c
ons
ta
nt
va
lue
s
of
I
The
s
tea
dy-
st
ate
er
ror
f
or a
unit
r
amp i
nput r(
t)=t i
s sho
wn inFig For
a
given
T
/
T
we
can
r
educ
e t
he
st
eady
-st
at
e e
rro
r,
b
ut t
he s
yst
em
yi
elds
gre
ate
rovershootand
se
ttl
ing
t
ime
f
or
a
s
tep
i
nput
.
,时
.
j
一
\
{
}
淮
}
l
!
一
之
户
\一、又 卜 \ 、火 饰
'r
f”
i
+对l
a助l
u 犷 :r夕
丫
1.
5 }r
a
j 卜\眨 、
I
’
一、汉 、、
广一一一‘
-
万r
.
、、
交姿卜 、 、、 ̄
戈芝吮  ̄ 、御.
1口
、、、勿
物%断、‘ 门月... ‘一 一
、 ̄ 泛 舜三
「-
一 ;  ̄
(
).5  ̄了。=户(
 ̄ 一
J l
一 .. . .
一
处
叫卜 ,.  ̄. .. . 目曰目门勺.口.卜
, ̄ 吸 、
‘勺.一 一 .自自.阅口.匕  ̄ .
廿
}
口. .,门 .. 阳比曰 匕. . 目曰 ... 曰.... 味 曰. ..曰 臼. 曰口自 .
!
{
口:
'
{ 以乍
1.
U
1.2
l
}i(
1
1_
^,
稼 ?
R
r :
Fi
g.
6.
28
Th
e s
tea
dy-
st
ateerorof
a
second-
orders
amp
le
s
yst
em
for
a
uni
tramp
i
nput
r
(t)
=t
,
v0
191
EXAM PLE
6.6
DESIGN
OF
A SAMPLED
SYSTEM
Le
t u
s c
ons
ide
r a
c
los
ed-
loo
p s
ampl
ed
s
yst
em
a
s s
hown
i
n Fi
g.
6.
25
wh
en
K
G,
(
s) s( (
6.39)
0.1
s+1
)(0.
005s+1
)
a
nd
we
n
eed
t
o
se
le
ct
T
a
nd
K f
or
s
ui
ta
bl
e.
pe
rf
or
man
ce.
As
a
n
ap
pro
xi
mat
i
on
we
n
egl
ec
t
t
he
e
fec
ts
o
f
th
e t
ime
c
ons
ta
nt
22=0.
005
s
eco
nd
b
eca
use
i
t
is
o
nly
5%
o
f
th
e pr
ima
ry
t
ime
c
ons
ta
nt
Z
l=0.
1.Th
en
we
c
an
u
se
F
igs
.6.
26,
6.
27,
an
d 6
.28
t
o
se
le
ct
T
a
nd
K.
Limi
t
ing
t
he
o
ver
sho
ot
t
o 30% for
the
st
ep
in
put,
we s
ele
ct
T/t=025 yie
ldi
ng K'
r=1 .
4. For
t
hes
e v
alue
s,
the
s
tea
d y
-st
at
e
erro
r
fo
r
a
uni
t
ramp
in
put
i
s
appr
oxi
mat
ely
0.
6(s
ee
Fi
g.6
.28
).
Be
c
ause T=0.
l,
we t
hen s
et T=0.025 s
eco
nd and
K==14.The sa
mpl
ing
ra
te i
s t
hen
r
eq u
ire
d t
o b
e 40
s
amples
pe
r s
econ
d.
The
overs
hoot
t
o s
tep
i
nput
and
t
he
st
ead
y-s
tat
e
er
ror
f
or
a
r
amp
i
npu
t
ma
y b
e r
educ
ed
i
f
we
s
et
T
1'
rt
o
e
qual
0.
1.
The
over
shoot
t
o a
s
tep
in
put
wil
l
be
25%
for
KT“1
.6.Us
ing
Fi
g.
6.
28,
we
e
st
ima
te
t
hat
the
st
eady
-st
at
e
e r
ror
f
or
a
uni
t
ra
mp
input
i
s
0.
55
f
or KT=1.
6.
6.
9 CLOSED-
LOOP
SYSTEMS
W ITH
DIGITAL
COMPUTER
COMPENSATION
A c
lose
d-l
o o
p,
sa
mpl
ed
sys
tem wi
th a
di
git
al
compu
ter
u
sed
t
o i
mpr
ovet
he
per
for
mance
i
s
s
hown
in
Fi
g.6.
22.
The
c
los
ed-
loo
p t
ran
sfer
f
unc
tio
n i
s
Y(
s) 。,、 G(
z) D(z
)
花二叮丫卜= 1L之) (
6.40)
R(
s) 1
+G( z)D(z)
The
t
rans
fer
f
unct
ion
of
t
he
comput
er
i
s
re
pres
ent
ed
by
U(
z)
D(
z). (
6.41
)
E(
z)
I
n o
ur
p
rio
r c
alc
ula
ti
ons
D(z)
wa
s r
epr
ese
nte
d s
impl
y b
y a
ga
in
K.As
a
n i
ll
ust
ra
ti
on
o
f t
he
p
owe
r of
t
he
c
ompu
ter
a
s a
c
omp
ens
ato
r,
we
wi
ll
r
eco
nsi
der
t
he
s
eco
nd-
ord
er
s
yst
em
wit
h a
z
ero-
orde
r hol
d and
a
pl
ant
I
when
T =I.
G,
(
s)=s(
s+1
)
The
n (
see
Eg
n.6.
16)
0.
3678(
2+0.
718
9)
G(
z)二 (
6.42)
(
:一I
)(
,:一0.
3678)
If
we
sel
ect
K(
z一0.
3678
)
4
内j
、月了
D(
z)= (
6
(
7,
+r)
、
幼
e
we
c
anc
el
t
he
p
ole
o
f G(
z)
a
t z=0.
3678
a
nd
h
ave
t
wo
p
ara
met
er
s,
r
a
nd
K,t
o s
et.
I
f
!92
sel
ect
1
.359(
2一0.
3678)
D(
z) (
6.44)
(
z+0.
240)
we
have
0.50(
2+0.
7189)
G(
z)D(
z) (
6.45)
(z一1)(
z+0.
240)
I
f
we c
alc
u l
at
e t
he r
espons
e o
f t
he
sys
tem t
o a
uni
t
st
ep,
we
fi
nd t
hat
t
he
out
put
is
e
qua
l t
o t
he
i
nput
a
t t
he
fourt
h s
amplin
g i
nst
ant
an
d t
here
aft
er.
The,
r
espo
nse
s f
or
bot
h t
he
unco
mpensa
ted
and
t
he
compens
ated
s
yst
em
are
s
hown
i
n Fi
g.6.
29.
The
over
shoot
of
t
he
compens
ated
s
yst
em
is
4%,
wh
ere
as
t
he
ov
ers
hoot
o
f t
he
u
nco
mpe
nsa
ted
s
yst
em
i
s 4
5%.
I
t
is
be
yon
d t
he
o
bje
cti
ve
of
t
his
b
ook
t
o
di
sc
uss
a
ll
t
he
e
xte
nsi
ve
me
tho
ds
f
or
t
he
a
nal
yt
ic
al
s
ele
cti
on
o
f t
he
pa
rame
ter
s
of
D(z),
a
nd
t
he
r
ead
er
i
s
re
fer
red
t
o o
the
r t
ext
s.
Ho
wev
er,
we
wi
ll
c
ons
ide
r t
wo
me
tho
ds
o
f c
omp
ens
ato
r
d
esi
gn:
(
1)
t
he
G,
(
s)-
t
o-D(
z)
c
onv
ers
ion
me
tho
d,
i
n t
he
f
ol
lo
win
g p
ara
gra
phs
,
an
d (
2)
t
he
r
oot
l
ocus
z
-pl
ane
met
hod,
i
n Sect
ion
6.
11.
攀1‘
I子
日
1
1.
2 / ‘
火-"
01
7
1
*一?
,
・
、
“
1.
0 一 一 份 二 甲 一 ,
J
户 , .一.
臼 . .卜 乙一 . \ 一
卜
一‘
一
了
{
} 火 、卜、 ̄
l
l
{
Cong
)r1
1sat
ed
O'
s
0.
6 / 一’‘
1‘
! {
一
/V
0.
4 c
!
l
{
L
从2
r一 -
/ 落
}
{ 功
T 2T 3T 孑T ST
67'
7T
Thne(seconds
)
Fi
g.
6
.29
Th
e r
esp
ons
e o
f a
s
ampl
ed-
dat
a s
eco
nd-
ord
er
s
yst
em
t
o a
u
nit
s
tep
i
npu
t
On
e me
tho
d f
or
d
ete
rmi
ni
ng
D(z
) i
f
rst
d
ete
rmi
nes
a
co
ntr
ol
le
r G,
(s
) f
or
a
gi
ven
p
la
nt
GP
(
s)
f
or
t
he
s
ys
te
m s
hown
i
n
Fi
g.
6.
30
.
Th
en
t
he
c
ont
ro
ll
er
i
s
:
,o
nve
rt
ed
t
o D(
z)
f
or
t
he
g
ive
n s
amp
li
ng
p
eri
od
T.
Thi
s
de
si
gn
me
tho
d i
s
ca
ll
ed
t
he
G,
(s
)
-t
o-
D(
z)
c
onv
ers
io
n me
tho
d.
I
t
c
onv
ert
s
th
e
G,
,
(s)
of
Fi
g.
6.
30
t
o D(
z)
o
f
Fi
g.
6.
22.
(
s) c
:(s) G
(
s
) 3'
(s
)
Fi
g.6.
30
Th
e c
ont
inuous
s
yst
em
model
of
a
s
ampl
ed
s
yst
em
193
We
cons
ider
a
f
irs
t-or
der
c
ompe
nsa
tor
G,
(
s)二K竺旦 (
6.46)
s+b
and
a
di
git
al
c
ont
rol
ler
_ 之一A
D(
z)= (.'— . (
6.47)
之一B
We
det
ermi
ne
t
he
z
-t
rans
for
m of Gc
(s
)
an
d s
et
i
t
eq
ual
to
D(
z)
a
s
Z,
G,
(
s)
}=
D(
z). (
6.48)
d
an
-aT
Th
en
t
he
r
ela
tio
nshi
p b
etwe
en
t
he
t
wo
t
ran
sfe
r f
unc
ti
ons
i
s
A=e B=e-
tT when
s=0
we
r
equi
re
C
(
1-A
)一K
-
a
-. (
6.49)
(
1一B)
b
EXAMPLE
6.7
DESIGN
TO
M EET
A PHASE
MARGI
N
SPECIFICATION
Co
nsi
der
a
s
yst
em
wi
th
a
p
lant
1740
G_(
s)=
Y 、
(
6.50)
s
(O.
25s+1
)
.
wl
th
We
wi
l
l a
tt
emp
t t
o d
esi
gn
G,
(s
)
so
t
hat
we
a
chi
eve
a
p
has
e ma
rgi
n o
f 4
5
0 a crossover
re
fque
ncy
0
),=1
25
r
ad
/s
.
Us
i
ng
t
he
Bo
de
d
ia
gr
am
o
f
GP
(s
),we
f
i
nd
t
ha
t
th
e
ph
ase
ma
rg
in
i
s
20
.We
f
in
d t
hat
th
e r
equ
ir
ed
p
ole
-z
ero
r
at
io
i
s a=6
.25.
It
i
s
sp
eci
fi
ed
t
ha
t
(
0,=1
25,
and
s
o
112
we
n
ote
t
hat
叭 =(
ab) .Ther
efo
re
a=50
and
b=31
2.The
l
ead
c
omp
ens
ato
r i
s
the
n
K(
s+50)
.5
l
G,
(s
)= (
6
(
s+31
2)
fi
ll
W e
sel
ect
K i
n or
der
to
y
ie
ld
I
GG,
(j
(0
)I
=1when
0)二 叭 =1
25
r
ad/
s.
Th
en
we
K 二5.
6.Thec
omp
ens
at
or
G,
(s
)
is
t
o
be
r
eal
i
zed
b
y
D(
z),
and
s
o we
s
olve
t
he
r
ela
tions
hips
wi
th
as
el eds
ect ampl
i
ng
pe
ri
od.
Se
tt
ing
T=0.
001
s
eco
nd,
wehave
A=e
00
- =0
s .95
,
B=e
03
-
1
2=0
.73
, a
nd
C二4.
85。
Then
we
have
4.
85(
2一0.95)
D(
z) (
6.52)
(z一0.
73)
Of
c
our
se
i
f we
s
ele
cte
d a
not
her
v
alu
e f
or
t
he
s
amp
lin
g p
eri
od,
t
hen
t
he
coe
ffi
cie
nts
o
f
D(z)
woul
d
di
fer.
I
n
ge
ner
al
we
sel
ect
a
s
mal
l
sa
mpl
in
g p
eri
od
s
o t
hat
t
he
d
esi
gn
b
ase
d o
n t
he
c
ont
inu
ous
s
yst
em
wi
ll
a
ccu
rat
el
y c
arr
y ov
er
t
o t
he
z
-
pla
ne.
Howe
ver
,
we
s
hou
ld
n
ot
s
ele
ct
t
oc
s
mal
l
a
T,o
r t
he
194
c
omp
uta
tio
n r
equ
ire
men
ts
ma
y be
mo
re
t
han
n
ece
ssa
y.
r I
n ge
ner
al,
us
e a
s
ampl
i
ng
pe
rio
d
T二1
/10
几,wh
ere几 =C
UB
l
21
G,a
nd
C
UB
i
s
th
e b
and
wid
th
o
f t
he
c
los
ed-
l
oop
c
ont
i
nuo
us
s
yst
em.
T
he
b
and
wid
th
o
f
th
e s
yst
em
d
esi
gne
d i
n
Ex
amp
le
6
.7
i
s C
UB=1
80
r
ad/
s
or几 =28
.6
Hz
.
Th
en
we
s
el
ect
a
pe
ri
od
T=0.
003
s
econ
d.
No
te
t
hat
T=0.
001
s
econ
d wa
s
us
ed
i
n
Ex
amp
le
6.
7.
6.
10
TIDE
DESI
GN OF
A W ORKTABLE
MOTION CONTROL
SYSTEM
A n
impo r
tant
po s
iti
oning sys
tem in
ma nuf
act
urin
g syst
ems is a workta
ble moti
o n contr
ol
s
yst
em. The s
ystem contr
o l
s t
he
mo t
ion
of
a workt
able
at
a
cert
ai
n locat
ion.
We a
ssume that
the
t
abl
e i
s
acti
vate
d in
each axi
s by
a mot
or
and l
ead
scre
w,
as s
hown in
Fig.
6 .
31 a).
We consider
the
x
-axi
s
a n
d examine
the motio
n cont
rol
f
or
feedba
ck s
yst
em, as
s
hown i
n Fig.
6.31
b).
The g o
al
is
t
o o
btai
n a fa
st r
esponse wit
h ar
apid
is
r e
time a
nd set
ling
ti
me t
o a s
tep comma n
d while
not
e
xceedi
ng an
overs
ho ot
of
5%.
Th
e spec
iic
f ati
ons
are
then (
1)
a pe
rcen
t o
ve r
shoot
e
qual
to
5% and (
2)
a minimum setl
ing ti
me
(
2% cri
teri
on) and r
ise t
ime .
Ris
e t
ime is
dein
f ed as
t
he ti
me to re
a c
h the magnit
ud e of
the
command.
To c
onfi
gur
e t
he
s y
stem,
we choos
e a
po
wer
ampl
iie
f r
and
mot
or
s
o t
hat
t
he s
yst
em
is
de
scr
ibe
d
b
y F
ig.
6.32
.
Obtai
ning t
he
t
rans
fer
fu
nct
ion
of
t
he
moto
r a
nd p
ower
a
mplif
ier
,
we h
ave
_ 、 1
G。(
S)二一丁一一::二:一一二二二. (
6.53)
S(
S+1U)
LS+LU)
Senae
rr
二 偏.、
.
困
t
f
i
r
l
r
m
二 4df
口
.
llr
alor .
ftr
rkahl
e
X(J
) pac+
2cr
o-or介r
t
oir
r
pur
er 穿 叫州口脚
r
(..
..
 ̄问 hol
d
Lead
se
res
t-
u 心成l
c ag
八 r沉 ed)
rous
nror:d
。
砂
7
,er
o-or
der )
人(
+)
双t
s) )tol
d
、
6 ,)(
))(
u十2的
h)
Fi
g.
6
.31
A
t
abl
e
mo
tio
n c
ont
rol
sy
ste
m:
a
) a
ctu
ato
r a
nd
t
abl
e;
b
) b
loc
k di
agr
am
We
wil
l
in
it
ia
ll
y
us
e
a
c
ont
in
uous
s
yst
em
an
d de
si
gn G,
(
s)
a
s
de
scr
i
bed
in
S
ect
i
on
6.9.
We t
he
n
ob
tai
n D(z)
fr
om
G,
(
z)・
Fir
st
we
s
el
ect
th
e c
ont
ro
ll
er
a
s
a
si
mpl
e
gai
n,
K,i
n
or
der
t
o
det
er
mine
t
he
res
pon
se
t
hat
ca
n be
ach
iev
ed wi
thout
a
c
ompe
nsat
or.
Pl
oting th
e r
oot
lo
cus
, we f
ind
t
hat
whe
n K=700,t
he
domin
ant
compl
ex r
oot
s
ha
ve
a
dampi
ng r
ati
o o
f 0.
707,
a
nd
we
ex p
ect
a
5%
195
I
,
+ I
I
了
万一eel
R仁) -
(
枯仁) 仁十20) V(
V干了0
post
八ID
/!
Fi
g.6.
32
Model
of
t
he
whe
el
c
ont
rol
f
or
a
wor
k t
abl
e
o
ver
shoo
t.
Then
usi
ng
a s
imula
ti
on,
we
fin
d t
hat
t
he
ov
ers
hoot
i
s 5%,
t
he
r
ise
t
ime
i
s 0.
48
seco
nd,
a
nd
the
s
etli
ng
time
(2
% c
rit
eri
on)
i
s 1
.12
se
conds.
Thes
e v
alu
es a
re
r
eco
rded
as
i
te
m 1
i
n Ta
ble
6.
4.
Tabl
e 6.
4 Per
for
manc
e f
or
Two
Cont
rol
ler
s
Co
mpe
nsat
or
Per
cen
t
Se
ttl
i
ng
Time Ri
se
Time
G(
s)
K
o
ver
sho
ot
(s
econ
ds) (
seconds)
1.
K 700 5.
0
2.
K(s
+ l
l
)1(
s+62) 8000 5.
0
The
next
s
tep
i
s t
o i
ntr
oduce
a
l
ead
Compens
ato
r,
s
o t
hat
K(
s+a)
G,
(s
) (
6.54)
(
s+b)
We
wi
l
l s
ele
ct
t
he
z
ero
a
t s=一11
s
o t
hat
t
he
c
ompl
ex
r
oot
s
ne
ar
t
he
o
rig
in
do
min
ate.
We
fi
nd
t
ha
t
we
r
eq
ui
re
t
he
p
ol
e
at
s=-
62・
Eva
lu
at
i
ng‘
夕e
g
ai
n
at
th
e
ro
ot
s,
we
f
i
nd
t
ha
t
K=8
000
・
Then t
he
st
ep
re
spons
e ha
s a
ris
e t
ime of
0.
25 s
e con
d and
a s
ett
li
ng t
ime (2
% cri
ter
ion)
o
f 0.
60
sec
ond.
Thi
s
is
a
n i
mprove
d re
sponse,
and we
fi
na l
iz
e t
his
s
yste
m as
accept
abl
e.
I
t now
remai
ns t
o s
elect
t
he sampli
ng per
iod and t
hen us
e the
me t
hod o
f Secti
on
6.9
to
obt
ain
D(z
).The
ri
se
ti
me of
t
he co
mp ens
ate
d conti
nuous
sy
stem i
s
0.2
5 s
eco
nd.
Then we
r
equi
re
T<<几 i
n o
rde
r t
o o
bta
ini
ng
a
sy
stmr
e e
spo
nse
p
red
ict
ed
b
y t
he
de
sign
o
f t
he
c
ont
inu
ous
s
yst
em.
Le
t u
s s
ele
ct
T=0.
01
s
eco
nd.
Wehave
8
000(
s+11
)
G,
(
s)
(
s+62
)
Then,
D(
z)一C
-
z--
A,
7
.一B
wher
e
A二e
一ri
r二0.
8958,
and
B
=
e-
62
7'二0.
537
9.
W e
now
have
。a
l
l一B)
8000(
11)
(0.
462)=
t 二二二八 — 二二— 0471
)。
b(1一A) 62(
0.1
042)
106
Us
ing
t
hi
s
D(
z),we
e
xpe
ct
a
r
esp
ons
e v
ey
r s
imi
la
r t
o t
hat
o
bta
ne
i d
f
or
t
he
c
ont
in
uous
s
yst
em
model
.
6.
11
THE
ROOT
LOCUS
OF
DIGITAL
CONTROL
SYSTEM S
Let
us
cons
ide
r t
he
t
rans
fer
f
unc
tion
of
t
he
s
yst
em
shown
i
n Fi
g.6.
33
Re
cal
l
tha
t
G(
s)=Go
(s
)G
p
(
s)・
The
c
lo
sed
-l
oop
t
r
ans
fe
r
fu
nct
i
ons
is
Y(
z) KG(
z)D(
z)
洲
口
办
勺
(
6
R(z) 1
+KG(
z)D(
z)
R了
s)+ 1了,
)
I
i
G
(
s)
Fi
g.
6.
33
Cl
ose
d-l
oop
s
yst
em
wi
th
a
d
igi
tal
co
ntr
oll
er
The
c
har
act
eri
sti
c equat
ion
i
s
1+KG(
z)D(
z)=0,
已
U
whi
ch
is
ana
log
ous
t
o t
he
char
act
eri
st
ic
eq
uat
ion
for
t
he
s-
pla
ne
ana
lys
is
of
KG(s).
Thus
we
pl
ot
t
he
root
loc
us
fo
r
the
c
h a
rac
ter
is
ti
c
equat
io
n o
f
the
sa
mpled
s
yst
em a
s Kv a
ri
es.
The
rul
es
for
obt
aini
ng
t
he
r
oot
l
ocus
a
re
s
umma
riz
ed
i
n Ta
ble
6.
5.
EXAMPLE
6.8
ROOT
LOCUS
OF
SECOND-
ORDER
SYSTEM
Co
nsi
der
th
e
sy
st
em
s
hown
一i
n
Fi
g.
6.
33
wi
t
h D(
z)=l
andG_
(:)=1
/s`.
The
n we
 ̄ o
bta
in
、 产 Y 、 产 ,
K
G(
z)
一Tz
K(z+1
)
Z
(
z一1
)
2
L
et
T=扼 a
nd
p
lo
t
t
he
r
oo
t
l
oc
us.
We
now
h
ave
K(
z+1
)
KG(
z)=
(
z一1
)z
Ta
ble
6.
5
Roo
t
Lo
cus
i
n t
he
z
-pl
ane
1
.
The
r
oot
lo
cus
s
tar
ts
a
t
th
e po
les
a
nd
p
rog
res
ses
t
o t
he
z
ero
s.
2.
The
r
oot
lo
cus
l
ies
o
n a
s
ect
ion
o
f t
he
r
eal
ax
is
t
o t
he
l
eto
f
f
an
o
dd
nu
mbe
r of
p
ole
s a
nd
z
eros
.
3.
The
r
oot
lo
cus
i
s
sy
mme
tri
cal
wi
th
r
esp
ect
t
o t
he
ho
riz
ont
al
r
eal
ax
is.
4
.
Th
e r
oot
l
ocus
ma
y b
rea
k a
way
f
rom
the
r
eal
a
xis
a
nd
may
r
een
ter
t
he
r
eal
a
xis
. Th
e
b
rea
kawa
y a
nd
e
ntr
y po
int
s
ar
e d
ete
rmi
ned
f
rom
t
he
e
qua
ti
on
197
K二一N(
z)二F(
z),
D(
z)
dF(
a)=0
wi
th
z= Q .
Then
obt
ain
t
he
s
olut
ion
of
d6
5.
Plot
t
he
l
ocus
of
r
oot
s t
hat
s
ati
sfy
1
+KG(
z)D(
z)=0
or !
KG(
z)
D(z
)}
=1
and LG(
z)D(
z)=18
0’士k3600 k二0,
1,
2,.
..
a
nd
t
he
p
ole
s a
nd
z
ero
s a
re
s
hown
o
n t
he
z-
pl
ane
i
n Fi
g.6
.34.
The
c
har
act
er
ist
i
c e
qua
ti
on
i
s
1
、KG
(z
)=;
+耳 斗琴=
(z一1)-
0
Fi
g.
6
.34
Ro
ot
l
ocu
s f
or
Exa
mpl
e 6
.8
Let
z=a
and
s
olve
f
or
K t
o obt
ain
K __(
a一
一1
)
l
)‘
2=F(C),
(
口+1
)
T
hen
o
bt
ai
n
th
e
de
ri
va
ti
ve
d
F(6)
/d6=0
a
nd
c
al
cu
la
te
t
he
r
oot
s
as
6,
=-3
a
nd
a
'2=1
.Th
e
l
ocu
s l
eav
es
t
he
t
wo
po
les
a
t o'
2二1
a
nd
r
eent
er
s a
t
6,=-3,a
s s
hown
i
n Fi
g.
6.
34.
Th
e u
nit
c
irc
le
i
s a
lso
s
hown
i
n F
ig.
6.3
4.
Th
e s
yst
em
al
way
s h
as
t
wo
r
oot
s o
uts
ide
t
he
u
nit
c
irc
le
a
nd
i
s
a
lwa
ys
u
nst
abl
e
fo
r a
ll
K>0.
198
We
no
w t
un
r t
o
th
e de
si
gn
o
f
a
d
igi
ta
l
con
tr
oll
er
D(z
)
to
a
chi
eve
a
s
pec
if
ie
d r
espo
nse
ut
il
i
zin
g
a
root
l
ocus
met
hod.
W e
wil
l sel
ect
a
cont
rol
ler
(Z一a)
D(
z)=
(
Z一b)
Us
e (
z一a)
t
o c
anc
el
o
ne
po
le
a
t G(
z)
t
hat
l
ie
s o
n t
he
p
osi
ti
ve
r
eal
a
xis
o
f t
he
z
-pl
ane
.
Th
en
s
ele
ct
(
z一b)
s
o t
hat
t
he
l
ocus
o
f t
he
c
ompe
nsa
ted
s
yst
em
wil
l
gi
ve
a
s
et
o
f c
ompl
ex
r
oot
s a
t a
de
sir
ed
po
int
wi
thi
n t
he
u
nit
c
irc
le
o
n t
he
z
-
pla
ne.
EXAM PLE
6.9
DESIGN
OF
A DIGITAL
COMPENSATOR
Le
t us
des
ign
 ̄
a comDens
ator
D(z)t
hat
wi
ll
r
esul
t i
n a
s
t
‘
abl
、
e s
ys
产
t
em
when
G.(
s)
i
s s
a 一 Y
de
scr
ibe
d i
n
Ex
ampl
e
6.
8.
Wi
th
D(
z)=1,
we
ha
ve
a
n u
nst
abl
e
sy
ste
m.
Se
lec
t
a
D(z
)
as
之一 a
D(
z)=
之一b
so
t
hat
K(
z+1
)(z一a)
KG(
z)D(
z)
(
Z一1
)
2
(
Z一b
)’
I
f
we
s
et
a=1
a
nd
b=0.
2.we
have
K(
z+1)
KG(
z)D(
z)
(
Z一1)
(Z一0.
2)
Us
ing
t
he
e
qua
ti
on
f
or
F(
a),
we
o
bta
in
t
he
e
ntr
y p
oin
t a
s
z=-2.
56,a
s s
hown
i
n
Fi
g.
6.
35.
Th
e
r
oot
lo
cus
i
s
on
t
he
un
it
ci
rc
le
a
t
K=0.8.Thus
t
he
sy
ste
m i
s s
ta
ble
f
or
K<0.
8.I
f
we s
el
ect
K=0.25,we
fi
nd
tha
t t
he
st
ep
r
esp
onse
has
an
ove
rsh
oot
of 2
0% a
nd
a
se
tt
li
ng
ti
me (
2%
c
rit
eri
on)
eq
ual
t
o 8.
5
seco
nds
.
I
f t
he
syst
em per
for
mance
we
re
i
nade
qua
te,
we
woul
d i
mpr
ove
t
he
r
oot
l
ocu
s b
y s
ele
cti
ng
a= 1
a
nd
b= -0.
98
so
t
hat
K(
z+l
) K
KG(
z)D(
z)
(
z一1
)(z+0.
98) (z一1)
Th
en
t
he
r
oot
l
ocu
s wou
ld
l
ie
o
n t
he
r
eal
a
xis
o
f t
he
z
-pl
ane.
Whe
n K=1,t
he
r
oot
of
t
he
c
har
act
eri
st
ic
e
qua
ti
oni
s
at
th
e o
ri
gi
n,
a
nd
T(
z)=1
/z=z
一,.
Th
en
t
he
r
es
pon
se
o
f
th
e
sa
mpl
ed
s
yst
em
(
at
t
he
s
ampl
i
ngi
nst
ant
s)
i
s
th
e i
npu
t s
tep
de
lay
ed
by
o
ne
s
amp
in
l g
p
eri
od.
We
c
an
d
raw
l
i
ne
o
f c
ons
ta
ntgo
n t
he
z
-p
la
ne.
T
he
ma
ppi
ng
b
etwe
en
t
he
s
-pl
an
e a
nd
t
he
z-
pla so
ne
i b
ta
in
ed
b
y t
he
r
el
at
i
on
z=e
sr.
The
i
l
nes
o
f
co
nst
an
t杏 o
n t
he
s
-p
la
ne
a
re
r
ad
ia
l
l
ines
wit
h
6
Si
nce
s=6+j
o),
we
h
ave
O
一
‘
an
”一‘
a
n
(s
‘
一“
,一扁 ・
199
l
n I
t
, "
-1 它
飞
、
、
、
、
Fi
g.
6
.35
Ro
ot
l
ocu
s f
or
Ex
amp
le
6.
9
劲叮.1 、 口 {
) 1
{ 。
一
I
一 . 1
{ I
{
}
l
哥
、
I
一
l_
1
I)
户一
I
1
}
0.
s /
I
吐
}
i
{
!
____,
厂/令仁
-
-
_-_
__a2
0,6
丫/ 一 一0
Oa
/一 / }_‘_艺‘_灸尸尸
0.
了
/
一
_
一
__
_
_
了一l厅:
一, 布万
冷了二 1
2尸一 1 石
、
下、
.
3
1氢
i
口2
干
一’
一琳热
r I
-10
一
-08
以
}
-0 6
-0.
4 _刀2
、 一
6夕咋
}
}
C
,O
4
0
1
1 02 0a 0.6 4.x
LO
Re{二声
F
ig
6
.36
Cu
rve
s
of
co
nst
ant杏o
n t
he
z
-pl
ane
z=e
0
7'
e
j
c
uT
wher
e
J 二二 — 一
杏 田
1
-V
T
he
p
lot
of
t
hes
e
li
nes
f
or
c
ons
ta
nt杏i
s
sh
own
i
n
Fi
g.
6.
36
f
or
a
r
ang
e o
f
T.
A c
ommo
n v
al
ue
o
f
200
杏f
or
ma
ny
d
es
ig
n
sp
ec
if
i
cat
i
on
i
s=1
/五.
The
n
we
ha
ve。=一
。an
d
z=e-
(Q
T
e
;
(
OT=e
'T
L
6
whe
re
B二 (
OT 。
6.
12
IMPLEM
E NTATION
OF
DIGITAL
CONTROLLERS
W e
wil
t consi
der
t
he
PID
cont
rol
ler
wit
h an
s-
domai
n t
ransf
er
funct
ion
U(
s
X(
)
S)
=
。。(
:
)=、
、十K
z+K
3s
. s
(
6.56)
We
c
an
de
ter
min
e a
d
igi
tal
i
mpl
eme
nta
ti
on
o
f t
his
c
ont
rol
l
er
by
us
ing
a
di
scr
ete
a
ppr
oxi
mat
ion
f
or
t
he
de
riva
tive
a
nd
i
ntegr
ati
on.
For
t
he
t
ime
de
riva
tive,
we
us
e t
he
bac
kwar
d di
fer
ence
r
ule
1 ._、 _. ___
‘
T
(x(
k!'
)一x[
(k一I)
T1). (
6.57)
The
z
-t
rans
for
m o
f Eq
n.
(
6.5
7)
i
s
tu
(
k
T)
・
he
nd
d
x
I
f
_k
:
二 1
(
1一z
一,)、
,,_
、(z一1
)
U(
z)=一一二二一目一 人 气z) X
(
Z).
1
Tz
Th
e i
nte
gra
ti
on
of
x(
t)
c
an
b
e r
epr
ese
nte
d b
y t
he
f
orwa
rd-
rec
tan
gul
ar
i
nt
egr
at
io
n a
t
t=kT
s
a
u(
kT)=u[
(k一1
)T+Tx(
kT
)
,
(
6.
58)
wh
ere
u
(
kT)
i
s
the
o
utp
ut
of
t
he
i
nt
egr
at
or
a
t
t=k
T.Th
e z
-t
ran
sfo
rm
of
Eqn
.
(6
.58
)
is
U(
z)=z
一,U(
z)+T
X(z
),
and
t
he
t
rans
fer
f
unct
ion
i
s t
hen
U(
z)
Tz
X
(
z)
(
z一1
)
He
nce,
t
he
z-
domai
n t
rans
fer
f
unct
ion
of
t
he
PI
D cont
rol
ler
i
s
。
。(
之)
一二1
t
+K
/2T
z +
二,(
z -
认一1) 一 It
1)
(
6.59)
Th
e c
omp
let
e di
fer
enc
e e
qua
ti
on
a
lgo
ri
th
m t
hat
pr
ovi
des
t
he
PI
D c
ont
rol
l
er
i
s o
bta
ine
d b
y
a
ddi
ng
t
he
t
hr
ee
t
er
ms
t
o
ob
tai
n
[we
u
se
x
(kT)二x(
k)
I
u(
k)=K,
x(k
)+K2
[
u(k一1)+Tx(
k)]
+(K31
T)[
x(k)
一x(
k一1)
]
(
6.60)
=[
K1+K2T+(
K3
IT))
x(k)
+K3 T
x(k一1
)+K2U(
k一1
).
Egn.
(
6.6
0)
c
an
b
e i
mpl
eme
nte
d u
sing
a
d
igi
ta
l
co
mpu
ter
o
r mi
cr
opr
oce
sso
r.Of
c
our
se,w e can
o
bta
in
a
PI
o
r PD
c
ont
role
r b
y s
ett
i
ng
a
n a
ppr
opr
iat
e
ga
in
e
qua
l
to
z
ero.
6.
13
DIGI
TAL
CONTROL
SYSTEMS
USI
NG
M ATLAB
Th
e pr
oce
ss
o
f d
esi
gni
ng
a
nd
a
nal
yzi
ng
s
ampl
ed-
dat
a s
yst
ems
i
s e
nha
nce
wi
th
t
he
u
se
o
f
int
er
act
ive
compu
ter
too
ls.
Ma
ny
of
t
he MATLAB f
unc
tio
ns
for
cont
inuous
-ti
me
con
tro
l de
sig
n
have
equi
vale
nt
cou
nter
pats
r f
or
s
ample
d-d
a t
a s
yst
ems.
Dis
cre
te-
ti
me t
ran
sfer
f
unc
ti
on model
201
o
bje
cts
a
re
o
bta
ine
d wi
th
t
he
t
f
fu
nct
i
on
s
imi
lar
t
o c
ont
i
nuo
us
t
ime
mo
del
s.
Fi
g.
6.3
7 i
ll
ust
rat
es
t
he
T
h
J
7
,
O
t
k
r
N
T
s
e
'
J
Yr
e
T
t
F
e
"
-
.
/
t
r
i
N
rU
r
"
7
i
n
〕
r
t
'
{
"
}
_
I
d
R
r
"
t
rr
匕.
几 1
c ¥
3,r
=扩(
n i
nn,
rl
r"
u,乃)
41
)
C
'
(
`
)]1
s
,
!
拜〕}
-
平
-
-}
困匣z
-
-- e
r
o
-
o
r
de
r
h
o
ld
Com "
rrwtor
tia
r"reta-hir
ra
口仁夕=
s
'1
"
sd sr
rur
pli
rr
8 t
ivu" us
s7r
rnl
rr
a
G
{
砂二+
;
se
j
}M
fr=一
s,'
s
,d
j二e:
r
l(
sys
e,
T
:
'
,oh'
)
GI
巨亘二颐困仁
巫醉〕
[遮班习
('
ourr
"rt
t
d emrt
inrr
rnrs-t
irr
re
G
y
(
s
,l
”st
’‘
’ C
i{)二
一、r
J 从川塑I
’啥 t
l e
m
i 瓜“
as
sr,x
1
v功i
u烤 a
,v
ro-
orr
lr
"rkr
rl
rl
向 ,了一r
心
Jy
e
(s
厉一三一习
7
"r
r1
.
.T.
';ni
}`
)
Fi
g.6
.37
a
)
Th
e t
f
fu
nct
i
on
b)
The
c
2d
f
unc
ti
on
c
) Th
e d
2e
f
unc
ti
on
u
se
o
f t
f.
Mo
del
c
onv
ers
ion
c
an
be
a
cco
mpl
i
she
d wi
th
t
he
f
unc
tio
ns
c
2d
a
nd
d2
c,
s
hown
i
n
Fi
g.
6.
37.
The
f
unc
ti
on
e
2d
c
onv
ert
s
co
nti
nuo
us-
ti
me
s
yst
ems
t
o d
isc
ret
e-
ti
me
s
yst
ems
;
th
e f
unc
ti
on
d2
c c
onv
ert
s d
isc
ret
e-t
ime
s
yst
ems
t
o c
ont
i
nuou
s-t
ime
s
yst
ems
.
Fo
r e
xampl
e,
c
ons
ide
r t
he
p
lan
t
t
rans
fer
f
unct
ion
1
G _
r
(s)=
"
s
(s+1
)
ss
a
pawn
i
n
Fi
g.
6.
23.
Fo
r
a
sa
mpli
ng
pe
rio
d of
T=l
s
econd
,
we
know
f
rom
Eg
n.(
6.1
6)
t
hat
.6l)
0.
367¥(
2+0.7189)
0.
36792+0.
2644
G(
&) (
6
一(
z-.
1
)
(z-
-
0.
36
80
)一z
2一1
8;
6
3 +
0.
36
80“
The.unlt
we
c
an
u
se
MA
TI,
AB
t
o
obt
ai
n
th
e G(
z),
as
s
hav
en
i
n
Fi
g.
6.
38.
The f
unct
ions
ste
p, impulse,
an
d Isi
m are u
sed
for
si
mulat
ion
of
sampled-
data
syst
ems.
u
nit
st
ep
res
ponse
is gene
rated
by
step.
The
step
fu
ncti
on f
ormat
is
s
hown in
Fig.
6.39.
The
i
mpuls
e r
espon
se i
s gener
a t
ed by
t
h e
func
tio
n impu
lse
, a
nd t
he r
espo
nse t
o an
arbi
tra
y
r i
nput
i
s
o
bta
ined
by
the
ls
im funct
ion.
The
impulse
an
lsi
m f
unct
ions
ar
e s
hown i
n F
igs.
6.40
and 6.
41,
202
%Thi
s scr
ipt
conver
ts
t
he
t
ransf
er
f
unc
tion
%Gp
(s)
=11
s(s
+ f
)to
a
di
scr
ete-
ti
me
s
yst
em
%wi
th
a
sa
mpl
ing
per
iod
of
T=1
sec.
%
num=[
1];
den=[
'l
1
0]
.sy
sc=t
f(
num,
den)
;
T=飞;
[
sysd]
=c2d(
sysc,
T.'
zoh
Zer
o-or
der
hol
d
>>
,
0
,
ut
y
}u
t 0.
3679z+0.
2642
Transf
er
functi
on:
z^2-
1.368z+0.
3679
Sampl
ing
t
ime:
1
F
ig.
6.
38
Us
in
g
th
e
c2
d f
unc
ti
on
t
o
co
nve
tG
r (
s)=Go
(s
)G,
(s
)
to
G(
z)
r
esp
ect
ive
ly.
Th
ese
s
amp
led
-da
ta
s
yst
em
si
mul
at
ion
f
unc
ti
ons
o
per
ate
i
n e
sse
nti
all
y t
he
s
ame
ma
nne
r
as
t
hei
r
co
unt
er
par
ts
f
or
c
ont
in
uous
-t
ime
(
uns
ampl
ed)
s
yst
ems
.
Th
e ou
tpu
t
is
y(
kT)
a
nd
i
s
s
hown
a
s
y(
kT)
h
eld
c
ons
ta
nt
f
or
t
he
p
eri
od
T.
・
一
(=)
R
一
-
仁一对
`巨
-
-t
D.
T
s
.
T
f
t
.
i
m
-
h
e
in
h
T
e
s
f
i
t
s
〕=nu夺r
ut
r
espn+
tse Tshoul
d be 功。.
八)I`
7
11
T
= s
rnt
ulunon
t
ir
tte <
*( =s
Ps 乃.几.茸 nhere几 ,
、rh e
rect
or
s
ample
[
Y.T
]=s
tr
h(s
ys,
T)
Fi
g.
6.
39
The
s
tep
f
unc
ti
on
g
ene
rat
es
t
he
o
utp
ut
y(
kT)
f
or
a
s
te
p i
npu
t
n
We ow
re
consi
der
Exa
mpl
e 6.
4a
nd
a
ppr
oac
h t
he
p
rob
lem
o
f o
bta
ini
ng
a
s
tep
r
espo
nse
wi
t
hout
u
til
i
zin
g l
ong
div
isi
on.
20
I
7
Z
)
嵘:
I
nipul
ce
R(
,
) 二I
Fi
g.
6.
4
T
I
"
=
s
r
o
e
u
n
c
t
p
r
o
u
l
a
t
r
0
i
o
e
s
n
p
o
t
i
r
n
s
e
Th
G
(
;
)
=
e
s
t
'
i
T
0
t
h
s
;
e
T
c
.
o
a
u
T
m
l
f
d
p
b
w
l
e
ht
m
r
m
e
f
i
r
T
.
e
s
f
i
o
'
r
n
mpul
s
e
f
unc
ti
o
I
YI
n
T
I=i
ge
ne
nr
r
a
pul
t
e
s
s
t
h
e(
s
e
o
u
r
t
s
,
p
T)
ut
y(
kT)
f
or
a
n i
mpu
lse
i
npu
t
!_
1
-
r
}
):
Ar
brt
rarr
r
npur ig
f s
?em
一 卿巨肠;
GO
.
、
’=out
jnr
t
rc
匀,01
25盯 u,
u
rpu
tsl
rvr
dd
b
e
r
r
e
r
es
p
o
n
s
e旱
I
F
u
s
(
I
a
N
.m
w
p
l
u
e
d
t
s
I
a
h
t
o
u
h
l
e
db
e
T
= snnul
aNon
trr
rrc G(
)
=s
l'
s 5
存),
I刀的7ar功廿s
; ame
rot
e
1,
己ctor 口.
’s
t '
i's
h'
.
TI=l
s'
i
m(s
t'
s,
u)
Fi
g.
6.
41
The
I
si
m f
unc
ti
on
g
ene
rat
es
t
he
o
utp
ut
y(
kT)
f
or
a
n a
rbi
t
rayi
r
npu
t
E叉通掬仔生石疾I
EXAMPLE
6.10
OU入了TSTEP
RESPONSE
In
Example
6.
4,
we
cons
idere
d the
pro
ble
m of
comp ut
ing
the
ste
p r
espons
e of
a
cl
ose
d-l
oo p
s
ample
d-da
ta
sys
tem.
I
n t
hat
example
, t
he
re
spon s
e,
y(kT),wa s c
omputed
usi
ng l
ong
di
vis
ion.
We
can
use
MATLAB
to
compute t
he
res
pons
e y(kT) usi
ng t
he st
ep
fun
cti
on,
shown
in
Fi
g.
b.39.
Wit
h
the
cl
ose
d-l
oop
t
ran
sfe
r f
uncti
on
give
n b
y
Y(
z)
0.3678z+0. 2644
R(
z)
z
2一z+0
.63
22’
Th
e s
ubj
ect
o
f d
igi
tal
co
mpu
ter
c
omp
ens
ati
on
wa
s d
isc
uss
ed
i
n Se
cti
on
6.
9.1
n t
he
ne
xt
e
xampl
e
we
r
econs
ider
t
he
s
ubj
ect
ut
iizi
l ng
MATLAB
204
%仆 i
s
¥o
ri仍卯 n
era
tes
t
he
u
nit
鱿印 r
espo
nse,
只kT)
,
%t
ort
samp
led
da
ta
s
yst
em
gi
ven
i
n Exampl
e 6
.4
%
num=[1[
;
den=[1
1句;
SYs
t-峨n um,den)
;
sysd
,c
w 2
d(s
ys c
,l,
'
zol
Y);
sy
s=t
eebac
k(s
ysd,
jl
))
;
T
-[0
:1:
20
,s
1
t
ep(
sy
s,
T)
1
4
12
令、
扣
掀
乏.奢}
口
〕
九
扮J i
t
rz 二n
j a卫:
+
o
.
r44
I
t
W :+口h
,;=
:
以2
了U
IS 20
'
S
'u
of
Ru
n洲 .
}
r
Fi
g.
6.
42
Th
e di
sc
ret
e
re
spo
nse
,
y(
kT
),
of
a
s
amp
led
s
eco
nd-
ord
er
s
yst
em
t
o
a
u
nit
s
te
p
%%%
This
s
crip
t comput
es
t
he
cont
inuous-
ti
me
unit
st
ep r
esponse f
or
t
he
sys
tem
i
n Example
6
.4
numg=[
i];
deng=(
1 1
0
];sys
g=t
f(
numg,
deng)
;
%
[nd,
dd]-pade(1,
2);
sysp
=tf
(nd,
dd)
;
sysi
=tf
([1
],[
1,0
]);
s
ysi
=se
ri
es(
i.
sys
p,s
ysi
)
;'
}'
4 一
 ̄
%
sys o=seri
es(sysl,
sysg);
sys =fe
edback( sys
o,[
1])
;
t
= [0:0.
1:201;
step(sys,t
)
1.
5
食之
之冬食,
0.5
0 5 l0 1 20
Ti
me
Fi
g.
6.
43
Th
e c
ont
in
uous
r
esp
ons
e,
y(
t),
to
a
u
nit
s
te
p f
or
t
he
s
yst
em
o
f
Fi
g6.
23
205
EXAMPLE
6.11
ROOT
LOCUS
OF
A DIGITAL
CONTROL
SYSTEM
Re
cal
l
ro
f m
Eqn.
(6.
16)
t
hat
a
pl
ant
i
s
gi
ven
by
0.
3678(z+0.
7189)
G(
z)
(z一1)(
z一0.3680)
K(
z一0.
3678
)
The
c
omp
ens
ato
r
is
s
ele
cte
d t
o
be
D(
z)
z+0.
2400
wi
t
h t
he
pa
rame
ter
Ka
s
a
v
ari
abl
e
ye
t t
o
be
d
ete
rmi
ned.
When
_ 、_ 、 _
,0.
3678(
z-+
-0.
7189)
行(z)
口(z)=式 一 — (
6.62)
(
z一1
)(z+0.
2400)
%Thi
s
sc
rip
t ge
ner
ate
r
the
r
oot
locu
s f
or
%t
he
sampl
ed
dat
a syst
em
%
% K(
0.3678)
(z+0.
7189)
%
% (
z-1)
(z+0.
2400)
%
num=[
0.
367
8 0.
264
4];
den=
[1.
000
0-0
.76
00
-
0.24
00]
:sy
s=t
f(
num,
den
):
rl
ocus(sys)
;hold
on
x=[-
1:0.
1:1]
;y=sgrt
(l-
x.^2)
;
pl
ot(
x,y
a'
--
',
x,
一,’
一’
) 一 巨u
r
un
it
4
'
t司
an3=
=
4.
6390
Fi
g.6.
44
The
r
1oc
us
f
unc
tio
n f
or
s
ampl
ed
d
ata
s
yst
ems
h
We ave
the
pr
obl
em in
af
orm for
whic
h the r
oot
lo
cus meth
od is
dir
ectl
y a
ppli
cable.
The
r1
oeus
f
unct
i
o n
wor
ks
for
di
scr
ete
-ti
me sy
stems i
n t
he
same way
as f
or
conti
nuous
-ti
me syste
ms .
Usi
ng
a
MAT LAB
s
cri
pt,
t
he
r
oot l
ocus
ass
oci
a t
ed
with
Egn.(
6.62
) i
s
easi
ly
gener
ate
d,
as
sh own i
n
Fi
g.
6.
44.
Re
memb
er
t
hat
the
s
ta
bil
i
ty
r
egi
on
i
s
de
fi
ned
b
y t
he
u
nit
c
ir
cl
e
in
t
he
c
omp
lex
pl
ane
.
Th
e
206
MATLAB
fu
nct
ion
r
1od<
i nd
can
b
e us
ed wi
th
th
e di
scr
ete
-ti
me sys
tem
roo
t l
ocu
s i
n e
xact
ly
the
s
ame way
as
f
or
co
n t
inuous
-t
ime
sys
tems
t
o de
ter
mine
the va
lue
of
th
e s
yst
em ga
in
ass
ocia
ted
wi
th
a
ny
p
oin
t o
n t
he
l
ocu
s.
Us
ing
r
1oc
fi
nd
we
d
ete
rmi
ne
t
hat
K=4.
639
pl
ace
s t
he
r
oot
s
on
t
he
un
it
ci
rcl
e.
6.
14
SEQUENTI
AL
DESI
GN
EXAMPLE:
DI
SK
DRIVE
READ
SYSTEM
I
n
thi
s
chapt
er
we wil
l
des
ign
.
a di
git
al
contr
oll
er
for
t
he
disk
driv
e sys
tem. As
the
di
sk r
otat
es,
th
e sens
or h
ead r
eads t
he
patt
ens
r used t
o pr
ovide
the
ref
ere
nce error
i
nformati
on.
This e
rro
r
in
formati
on
pat
ter
n i
s
read
i
nte
rmitt
entl
y
as t
he
head
read
s s
tor
ed d
a t
a,
and
then t
he
pat
ten
r i
n t
un.
r
Sinc
e th
e di
sk
is
rota
tin
g a
t a
const
ant
speed,
th
e t
ime,
T,between posi
t
i o
n-err
or
rea
dings
is
a
cons
tant
. Thi
s sa
mpling p
eri
od i
s t
ypica
lly 100脚 to 1 ms. Thus we hav
e sampl
ed e
rro
r
i
nfor
mat
ion.
We ma
y a
lso
us
e a
d
igi
ta
l c
ont
rol
le
r,
a
s s
hown
i
n F
ig.
6.
45,
t
o a
chi
eve
s
ati
sfa
cto
y
r
s
yst
em
r
esp
ons
e.
I
n t
hi
s
ch
apt
er
we
wi
l
l
de
sig
n D(
z)
L
.
(z
)
D(
Z) G() 3讹)
+
45
.
Fe
edb
ack
Co
ntr
ol
s
yst
em
wi
th
a
di
git
al
c
ont
rol
ler
.
No
te
t
ha
tG(
z)
=
ZI
Go
(s
)G,
(s
)]・
F
irs
t
we
de
ter
mine
G(
z),
whe
re
G(
z)“Z[
Go
(
s)G,
(s
)]・
Si
nce
5
G_
(s)=
夕
(
6.63)
s
(s+20)
w e
have
,
_ ,‘
7
5
Go
(s
)G,
(s
)二(
竺二一一
)一
s(
s+20)
We
n
ot
e
th
at
f
or
a=2
0
an
d T=l
ms
, e-
0r
i
s
eq
ual
t
o
0.
98
.
Th
en
we
s
ee
t
ha
t
th
e p
ole
a
t
‘=-
-20
i
n
Eqn
.
(6.
63)
ha
s
an
i
nsi
gni
ic
f a
nt
e
fe
一
c
t.
Th
ere
for
e we
c
oul
d
ao
oro
xima
te
G_
(
s)
a
s:
‘击 Y
0.
25
G(
s
)“
Then
we
need
207
G
(z
)=
Z{
1-
e-
s
T{
0.
2
5
L
月一
(,
一;
一)
(
0.
2
5)
Z干
粤)
s
l
s 刀 LS
J
、
,__
_了 T
z 、0
.25
T
0
.25
x
1
0-3
二1
1一Z)
(U.
LJ)
I —
,
_
:
}
I二万一.
,井二一一万-一代刃一一・
以Z一1
)一) (
z一1
)
(
Z一1
)
We
n
eed
t
o s
el
ect
t
he
d
igi
ta
l
con
tro
ll
er
D(
z)
s
o t
hat
t
he
de
si
re
d r
esp
ons
e i
s
ac
hie
ved
f
or
a
s
tep
i
nput
.
I
f
we
s
et
D(
z)二K,
the
n we
h
ave
K(
0.
25
x
1
0-3
)
D(
z)G(
z)=
(
z一1
)
Th
e r
oot
l
ocus
f
or
t
his
s
yst
em
i
s
sho
wn
i
n Fi
g.6
.46
.
Wh
en
K =4000,
the
n
1
D(
z)G(
z)=
(
z一1
)
Th
ere
for
e
th
e c
los
ed-
loo
p t
rans
fer
f
unc
tio
n i
s
D(
z)G(
z) 1
T(
z)=
1+D(
z)G(
z) 之
We
e
xpe
ct
a
r
api
d r
esp
ons
e f
or
t
he
s
yst
em.
Usi
ng
MAT
LAB
we
d
emo
nst
rat
e t
hat
t
he
r
esp
ons
e i
s
s
tab
le
a
nd
r
api
d.
Th
e pe
rce
nt
ov
ers
hoo
t
to
a
s
tep
i
nput
is
0
%,
a
nd
t
he
s
ett
li
ng
t
i
me
i
s
2
ms
.
Fi
g.6.
46.
Root
l
ocu
s
Th
e u
se
o
f a
d
igi
tal
c
omp
ute
r a
s t
he
c
omp
ens
ati
on
d
evi
ce
f
or
a
c
los
ed-
loo
p c
ont
rol
s
yst
em
has
g
rown
d
uri
ng
t
he
p
ast
t
wo
d
eca
des
a
s t
he
p
ric
e a
nd
r
eli
abi
li
ty
o
f c
ompu
ter
s h
ave
i
mpr
ove
d
d
rama
ti
cal
ly.
A
comp
ute
r c
an
b
e us
ed
t
o c
ompl
et
e ma
ny
c
alc
ula
tio
ns
d
uri
ng
t
he
s
amp
in
l g
i
nte
rva
l
T
a
nd
t
o
pr
ovi
de
a
n ou
tpu
t
si
gna
l
th
at
i
s
us
ed
t
o
dr
iv
e a
n a
ctu
ato
r
of
a
p
roc
ess
.
Co
mpu
ter
c
ont
rol
is
u
sed
t
oda
y f
or
c
hemi
cal
pr
oce
sse
s,
a
irc
raf
t
con
trol
,
ma
chi
ne
t
ool
s,
a
nd
ma
ny
c
ommo
n pr
oce
sse
s.
208
The
z
-tr
ans
for
m ca
n be
us
ed
t
o anal
yzet
he
s
tabi
li
tya
nd
r
espo
nse
of
a
s
amp
led
s
yst
em
and
t
o
de
sign
a
ppr
opr
iat
e s
yst
ems
i
nco
rpo
rat
inga
comput
er. cont
rol
s
yst
ems
have
be
come
i
ncr
esi
a n
gly
c
ommo
n a
s l
ow-
cos
t
compu
ters
have
become
r avai
labl
e.
209
REFERENCES
1
.
T.
J.
Vie
rsma
,
inv
est
i
gat
io
n i
nto中e
a
ccu
rac
y of
h
ydr
aul
i
c s
erv
omot
or
s,
Th
esi
s
De
lft
Un
iver
si
ty
o
f Techno
log
y,
1961;
Ph i
i
lps
Re
searc
h Rep
ort,
1
6(1
961)
507;
17(
196
2)2
0.
2
.
G.
S.
Brown
a n
d A.
C.
Hal
l,
Dynamic
b
ehavi
or
and
desi
gn
of
s
erv
ome c
hani
sms,
Trans
.
ASME,
(
1946)
503.
3.
H.
E.
Me r
it
r ,
Hydraul
ic Cont
rol
Sy s
tems,
Wil
ey,
Ne w
Yo r
k, 1
967.
4.
M.
Gui
ll
on,
Etude
at
deter
minat
ion
des
sys
tems
hyd r
aul
iques,
Dunod,
Pari
s,
1961.
5.
A.
C.
Mo r
se,
El
ect
ronhyd r
aul
ic
ser
vomechani
sms,
Mc Graw-Hill
,
New Yo
rk,
1963.
6.
R.
Wa l
ter
s,
Hydr
aulic
and El
ect
ro-
hydraul
i
c se
rvosyst
ems,
Iif
l f
e,
Lond
on,
1967
7.
L.
H.
Ge y
er,
Cont
roll
ed dampi
ng t
hroug
h dyna
mi c
pres
sure
feed
back,
Dowt y
Equi
pme
nt,
Che
ltenha
m,
June,
1
958.
8.
J .
J.
'
t
Ma
nnet
je,
St
abi
i
lzi
ng ne
twor
ks
f
or
hydr
aul
i
c mot
ors
,
Contr
ol
Eng.
,
Ju
ne (
1974)
55.
9.
W.
J
.
Thaye
r,
Tr
ansf
er
f
unc
tio
ns f
or
Moo
g s
erv
oval
ves,
Mo o
g Te
chnic
al
Bul
le
tin
103
,
Ne w
Yor
k,
1
965.
1
0.
P
Lensse
n,
The
infl
u e
nce
of d
y
r fri
ct
ion
and
me c
hanical
pa r
ame t
er
s on
th
e s
tabi
li
ty
a
nd
a
ccura
cy of
a
h
yd r
auli
c c
opying
sys
tem,
Proc,
1
0th
M. T.
D. R.
Confer
ence,
1
969.
1
1.
PL
ensse
n,
Analog
simula
ti
o n
of
hydra
uli
c s
erv
o sy
stems wit
h s
tabi
li
zing
i
nf
luence
of
d
y
r
f
ri
cti
on,
(i
n
Dutch),
Thes
is,
Univ
ersi
ty
of
Leuve
n,
Be l
gium,
Ap r
il
, 1
970.
1
2.
R.
Ch
iapp
uli
ni
,
Fr
ic
ti
on
a
nd
i
ts
e
ffe
cts
on
l
i
nea
r
pos
it
ion
al
h
ydr
aul
ic
a
nd
e
lec
tr
o-h
州ra
uic
l
s
ervos
yst
ems
,
Anna
ls
CI
RP,
19(
1971
) 125.
1
3.
T.
J.
Vi
ersma, I
nves
ti
gat
ions
int
o
th
e ac
cur
acy
of
hy
dra
uli
c
ser
vomotor
s,
Thes
is
De
lft
Un
i
ver
sit
y o
f Tech
nolog
y, 1
961;Phi
li
ps
Rep.
,
16
(1961
)
507;]7(
196
2) 20.
1
4.
T.
J.
Vie
rsma
,
De
sig
nin
g l
oad-
compe
nsa
tes
f
ast
re
spo
nse
h
ydr
aul
ic
s
erv
os,
Co
ntr
ol
En
g.,
May
(
1962)
111.
1
5.
PB lok
and
T.
J.
Vie
rsrr
t
a ,
A
new constr
uct
ion
f
or
a l
oad-c
o mp e
nsat
ed
fa
st
r
esp
onse
hy
dra
uli
c
f
ol ow-u
p s
yst
em, Proc,
Thi
rd I
FAC Congr
,
196
6,Vol.
1No.2,Report
25D.
1
6.
PB lok,
Linea
r e
lect
rohydrau
lic
st
epp i
ng
motor
f
or
NC machines,
Hydr
aul.
Pneum.
Vol
.21
,
No.
10,
(
1968
) 1
17.
1
7.
R.
v
on
Mi
ses
,
Be
rec
hnu
ng
v
on
Au
sfl
uss
-un
d Ue
ber
fal
lza
hle
n,
Z.
ve
r.
De
uts
ch.
Ing.
61
(
1917)
447
1
8.
T.
J.
Vi
ersma,
Inve
sti
gat
ions
in
to
the
accur
acy
of
hydr
au l
i
c s
ervomo
tor
s,
Thesi
s
Del
ft
Un
i
ver
sit
y of
Technolog
y, 1
961;Phil
ips
Rep.
,
16(1
961)507;
17
(196
2)20.
1
9.
W.
Wuest
, St
rot
nung
durch
Schli
tz
-
und Loch
ble
n d
en
be i
kl
ei
ner
Reyn
o l
ds-
Zahl
en,
IngAr
ch.
22(1
954)
357
2
0.
R.
C.
Dor
f,
Th
e Enc
ycl
ope
dia
o
f
Ro
bot
ic
s,
J
ohn
Wi
le
y S
ons
,
Ne
w Yo
rk.
1
988.
21
.R,
C.
Do
rf,
El
ect
ri
cal
En
gin
eer
ing
Ha
ndbo
ok,
2n
ded.
CRC
Pr
ess
,
Bo
ca
Ra
ton,
Fl
a.
,
19
98.
2
2.
R.
S.
Sa
nch
ez-
Pen
a a
nd
M.
Sz
nar
er
,
Ro
bus
t
Sy
ste
ms
Th
eor
y
an
d Ap
pli
cat
io
n,
J
ohn
Wi
le
y
Sons,
N丫 1998.
23
. G.
Zames,
"
Inpu
t-Ou
tpu
t
Fe
edba
ck
St
abi
li
ty
a
nd
Ro
bus
tne
ss,
"
IEEE
c
ont
rol
Sy
ste
m,
J
une
1
996,
p
p.61
-66.
210
2
4.
K.
Zh
ou
a
nd
J
.
C.
Doyl
e,
Es
sen
tia
ls
o
f Rob
ust
Co
ntr
ol,
Pr
ent
ice
Ha
ll
,
Up
per
Sa
ddl
e
Ri
ver
,
NJ
,
1998
2
5.
C.
M.
Clo
se
a
nd
D.
K.
Fr
ed
eri
ck
,
Mo
del
i
ng
a
nd
An
al
ys
is
of
Dyn
ami
c
Sy
st
ems
,
2"
d
ed
.,
Houghto
n Mifl
in,
Bost
on,
1993.
2
6.
A.
Charar
a,
"Nonl
ine
ar
Con t
rol
of
a
Ma gn
e t
ic
Levi
ta
ti
on S
yste
m,"
IEEE Tr
ans
act
ion
s on
Co
nt
rol
Syst
em
Technol
ogy,
Septe
mbe r
1996,
pp.
513-
523.
2
7.
J.
Ye n,
Fuzzy
Logi
c:
Int
ell
ige
n c
e a
nd Contro
l,
Pre
nti
ce
Hall
,
Uppe
r Sad
dle
Riv
er,
NJ, 19
98.
2
8.
X.
G>Wa
ng,
"Es
ti
mat
ion
i
n
Pa
per
Ma
chi
ne
Co
nrol
t ,
" I
EEE
Co
nro
t l
Sy
ste
ms,
Aug
ust
1
993
,
pp.
34一3.
2
9.
D.
S
bar
bar
o-Ho
fer
,
"Co
ntr
ol
o
f a
S
tee
l
Ro
lli
ng
Mi
ll
,"
I
EEE
Co
nro
t l
Sy
ste
ms,
Ju
ne
1
993
,
pp.
69-
75.
30.
N. Mo
han,
Power
El e
ctr
oni
cs,
J
ohn
Wil
ey
Sons,
New York,
1
995.
31.
J.
M.
Wesis
,
"The
TGV Comes
to
Te
xas
,"
Europ
e ,
March
199
3,
pp.
1
8-2
0.
32.
S>Lee,
"
Int
ell
ige
nt Sens
ing
an
d Con
ro
t l
fo
r Advance
d Tel
epo
rta
ti
on,"
I
EEE
Con
rol
t S
yst
ems
,
J
une
1
993,
pp.
1
9-28.
3
3.
J
.
V
Wa
it
a
nd
L.
P
Hue
ls
man
,
OA
T,
2
0d
e
d.,
Mc
Gra
w-Hi
l
l,
Ne
w Yo
rk,
19
92.
3
4.
R
G.
Mart
in,
The Art
o
f Rob
otic
s,
Pre
ntic
e Hal
l,
Up
per
Sa
ddl
e
River
,
NJ,
19
9 9.
3
5.
R.
Sh
our
eshi
,
"I
ntel
l
igent
Conro
t l
Sy
stems,"
J
ouna
r l
of
Dy
namic
Sy
stems
,
June
199
3,
pp.
392-
400.
36
. A.
But
ara
nd
R.
Sal
es,
"Con
ro
t l
fo
r Ma
gLev
Ve
hic
les
,"正EE
Con
rol
t Sy
ste
ms,
Augus
t
19
98,
pp.
18-
25
3
7.
L.
W.
Cou
ch,
Di
gi
ta
l
an
d
An
al
og
Co
mmu
ni
ca
ti
on
S
yst
ems
,
5`
h“.
,Ma
cmi
l
la
n,
Ne
w Y
ork
,
1997.
3
8.
H.
Pa
raci
a
nd
M. J
amshi
di,
De s
ign
an
d I
mplement
ati
on
of
I
nte
ll
igent
Manuf
act
uri
ng
Sys
tems
,
Pr
ent
ice
Hal
l,
Upper
Sad
dle
Rive
r,
N.
J.,
1
997.
3
9.
C.
T
Chen,
Ana
log
and
Digi
tal
Conro
t l
Syst
ems
Desi
gn,
HarcoutB
r r
ace
Jo
vanovi
ch,
Or
lando,
Fl
a.,
1993.
40.
B.
J
ohns
ton
e,
"J
apan'
s
Fr
ien
dly
Rob
ots
,"
Tec
hnol
ogy
Revi
ew,
June
1999,
pp.
66-6
9.
41.
W.
J.
Gr
ant
ham
and
T.
L.
Vince
nt,
Mo d
en
r Conro
t l
Syst
ems
Anal
ysi
s
and
Des
ign,
Jo
hn
Wi
le
y
Sons
,
Ne
w Yor
k,
1
993.
4
2.
K.
Capek,
Rossu
m' Un
ive
rsa
l
Robots
,
Engl
ish
e
dit
ion妙 P
Sel
ver
a
nd
N.
Pla
yfai
r,
Do
ubl
eday,
Page
,
New Yo
rk,
19
23.
4
3.
L.
L. Cone
,
"Sky
cam:
An
Aeri
al
Robot
ic
Camer
a S
yste
m,"
By
te,
Oc
tob
er
1985,
PP 12
2-1
28.
4
4.
H.
Ka
zer
ooni
,
"Hu
man
Ex
tend
ers
,"
J
ouna
r l
of
Dy
nami
c
Sy
ste
ms,
ASME,
Jun
e 1
993,
pp.
281-
290.
45.
C.
La
pis
ka,
"Fli
ght
Simul
at
ion,
"
Aero
space
Amer
ica
,
Au gus
t
19
93,
pp.
14
-17.
46.
D.
E.
Bosse
rt,
A Root
-
Locus
An
an
alysi
s
of
Qua
nti
ta
tiv
e Fee
dbac
k The
ory,
"
Proc
eed
ing
s o
f
t
he
Ame
ri
can
Co
nro
t l
Co
nfe
ren
ce,
Ju
ne
1
993,
pp.
1
698
-17
05.
4
7.
J
.
A.
Gut
ie
r e
z a
nd
M.
Ra
bin,
"
A
Comp
ute
r Loo
p s
hapi
ng
Al
gor
it
hm
f
or
Con
ro
t l
le
rs.
"
Pr
o c
eedin
gs
of
the
Ameic
r an
Conrol
t Conf
er
ence
,
June
1993,
pp.
1
711一
1715.
4
8.
J.
W. Song,
"
Syn t
hes
is
of
Co
mp e
nsat
ors
i
n
Line
ar
Unceta
r i
n Plan
ts,
"
Pro
ceed
ing
s o
f
th
e
Co
nfe
rence
o
n Decis
ion
a
nd Con
ro
t l
,
Dec
ember
1992
,
pp.
2882-288
3.
211
49.
M.
Got
ts
chal
k,
"
Part
S
urgeon-
Part
Robot
,"
Desi
gn
News,
Ju
ne 7,
1
993
, p
p.6
8-7
5.
50.
D.
E.
Wh i
tne
y,
"
From
Ro b
o t
s
to
Dei
gn,
"
Jouna
r l
of
Dy
nami
c S
yst
ems,
ASME,
Jun
e 1
993
,
pp.
262-
270.
51.
S.
Ja
yasuri
ya,
"F
requen
cy
Doma i
n
De s
ign fo
r Robust
Per
for
mance
Under
Unce
rtai
nti
es,
"
J
o
urnal
of
Dynamic
Syst
ems,
June 1
993p p.
439-
450 .
52.
L.
S.
Shie
h,
"Contr
ol
of
Unce
ta
r in
Syst
e ms,
"IEEE Proc
eedi
ngs
, Mar
ch
1993
,
pp .
99-1
10.
53.
M.
van de
Panne,
"
A Cont
rol
l
er
for
the
Dy namic
Wa l
k of
a
Bi
ped,"
I
EEE,
December
1992,
pp.
2668-
2673.
5
4.
S.
Bennet
t,
"
The
Devel
opment
of
t
h e
PI
D Cont
rol
le
r.
"
IEEE Con
trol
Sy
s t
ems,
Dece
mbe
r
1
993
, p
p.58-
64.
5
5.
J.
C.
Doyl
e,
Fee
dbac
k Cont
rol
Th e
o y,
r Macmil
la
n,
New Yor
k,
1992.
5
6.
R.
C.
Do r
f,
t
he
Enc
yclop
edi
a of
Roboti
cs,
Joh
n Wil
ey&S ons,
New York,
198
8.
5
7.
C.
L.
Phi
lip
ls a
nd
H.
T.
Nag
le,
Di
git
al
Co
ntr
ol
Sy
ste
ms,
pr
ent
ic
e Ha
ll
,
En
gle
woo
d Cl
ifs
,
N.
J.
,
1995.
58
. G.
E
Fra
nkl
i
n,
e
t
al
.,
Di
git
al
Co
ntr
ol
o
f Dyn
ami
c
Sy
ste
ms,
2n
d e
d.,
Pr
ent
ice
Ha
ll,
Uppe
r
Saddl
e Ri
ver
,
NJ.
1998.
59
. S.
H.
Za
k,
"
Ripp
le-
Fre
e De
adb
eat
Cont
rol
,
"I
EEE
Con
trol
Sy
ste
ms,
Augu
st
1
993,
pp.
51-
56.
6
0.
C.
La
pis
ka,
"Fl
ig
ht
Si
mul
at
io
n,"
Ae
ros
pac
e Ame
ri
ca,
Aug
ust
1
993
,
pp
.
14
-17
.
61
. R
G.
Ma
rti
n,
Th
e Ar
t
of
Ro
bot
i
cs,
Pr
ent
ic
e Ha
ll,
Uppe
r Sa
ddl
e
Ri
ver
,
NJ
, 1
999
.
62
. D.
Ra
viv
a
nd
E.
W.
Dj
aj
a,
"
Dis
cre
ti
zed
Con
rol
t l
ers
,"
I
EEE
Co
nt
ro
l
Sy
ste
ms,
Ju
ne
1
999
,
pp.
52-
58.
63
. R.
C.
Dorf
,
Ele
ctr
ica
l
Engi
nee
rin
g Han
db o
ok,
2nd
e
d.,
CRC Pr
ess,
Bo
ca Rat
on,
Fla.
,
1998
.
64
. T.
M.
Fol
ey,
"Engi
neeri
ng
t
he
Space
St
at
ion,"
Ae
ros
pac
e Amer
ica
,
Octob
e r
1
996
,
pp.26
-32.
6
5.
A.
G.
Uls
oy,
"Co
nro
t l
o
f
Ma
chi
ni
ng
P
roc
ess
es,
"
ASME
J
ouna
r l
of
Dy
nami
c
Sy
ste
ms,
Ju
ne
1
993
, p
p.301
-310.
6
6.
K.
J.
Asro
t m,
Comput
er-
Conro
t ll
ed
Sy
ste
ms,
Pr
ent
ice
Ha
ll,
Upper
Sa
dd l
e
Rive
r,
N.J.
,
19
97.
6
7.
R.
C.
Drof
and
A.
Kusi
ak,
Han
dbook
of
Manuf
act
uri
ng
a
nd Au
tomat
ion,
Jo
hn
Wi l
ey&So n
s,
Ne
w Yor
k,
1994.
6
8.
L.
W.
Co
uch
, Di
git
al
a
nd
An
alog
Co
mmu
nic
ati
on
S
yst
ems
,
5t
h
ed.
,
Ma
cmi
ll
an,
New
Yor
k
1995.
6
9.
M. v
an de
Pa
nne,
"A Con
ro
t l
le
r f
or
t
he Dy
nami
c Wa
lk
of
a
Bi
ped
,'
Pr
oceed
ings
o
f t
he
Co
nfer
ence
on
Dec
isi
on
and
Conro
t l
,
IEEE,
De
cemb
er
199
2,
pp.
2668
-26
73.
70
. K.
S.
Yeun
g a
nd
H.
M.
La
i,” Re
for
mat
i
on
o
f
th
e Ny
qui
st
Cr
it
er
ion
f
or
Di
scr
et
e
Sys
te
ms,
"
IEEE Tran
sact
ions
on
Educat
ion,
Februa
y
r 1
988,
pp.
32-3
4.
71
. M. Esla
mi,
Theoy
r of
Sens
iti
vit
y
in Dynamic
Sys
tems,
Spr
inger-
Ver
lag
,
New
Yor
k,
199
4.
72
. T.
R.
Ku rf
ess,
"
Pre
d i
cti
ve
Cont
rol of
a
Roboti
c
Grind
ing
Sy s
tem."
J
ouna
r l
of
Eng
ine
eri
ng
f
or
I
ndus
ty
r AS ME,
No v
emb e
r 199
2, pp.
412-4
20.
7
3.
T
St
udt,
"Wh y
You
Real
ly
Nee
d a
DSP
"R&D
Magazi
ne,
June
1993
,
pp.3
2-3
4.
7
4.
D.
M. Aus
lan
der,
Me c
haro
t ni
cs
,
Pren
ti
ce
Hal
l,
Engl
ewood
Cli
fs,
N.
J.,
1
996.
7
5.
R.
Sho
ures
hi,
"
Int
el
li
gen
t Co
nro
t l
S
yst
ems
,"
J
ouna
r l
o
f Dy
nami
c
Sy
ste
ms,
Ju
ne
1
993,
p
p.
392-
400.
212
,.
}
0
D.
J.
Le o,
"Co ntrol
of
a
Flexi
ble F
r a
me i
n Sl
ewing,
" Procee
dings of
Ame ri
can
Cont
rol
Confer
e nce, 1
99 2,
pp.2
535 -
25 4
0.
,了
7
M.
Al pert,
"A Bul l
et
Trai
n f
or t
he U.S.
," Fort
une,
May 1993,
pp.32.
月
只
2
J.
L. Jones and A. M.
Flynn,
Mo bil
e Robo t
s,
A.
K.
Pe t
ers
Publis
hing,
New York,
1
993.
1
n,
1
V.
Skormi n,
"On -Line
Diagnosti
cs o
f a
Se l
f-
Co n
tai
ned Fl
ight
Ac t
uator
,"
IEEE Tr
ansa
cti
ons
on
Ae r
o space and El
ectr
onic Syst
ems ,
J a
n uay
r 1
994,
pp.130-141.
只)
0
H.
H. Ot t
esen,
"F ut
ure
Servo Technolo
g i
e s
f
or
Ha r
d Disk
Drives,
" J
.
of
the Magn
eti
cs
Soci
ety of J
a p
an ,
vol.
18,
1994,
pp.31-36.
213
APPENDI
X 1.
HYDRAULI
C SYM BOLE
PUMPS,
MOTORS,
AND
DRI
VES
Fi
xed
Vari
abl
e
S
ing
le
d
ire
cti
on
p
ump
测 二巴)
Doubl
e di
rec
tion
pump
拭勃铡
Si
ngl
e di
rect
ion
mot
or
城全书多
Doubl
e di
rect
ion
mot
or
拭 〕铡
Si
ngl
e
di
rec
tio
n p
ump
加ot
or
二
二
城兰)书多
wi
t
Si
wi
t
h
ng
r
ever
l
e
h
s
d
i
n
i
sal
r
e
gl
c
e
t
of
i
o
lo
f
low
f
n
p
ump
w
di
re
di
c
t
/
r
ect
mo
io
n
i
t
o
on
r
升
蓄。
二泣)
-
7
f}
Z
Do
ubl
e
di
rec
ti
on
pu
mp/
mot
or
wi
th
t
wo
di
rec
tions
of
low
f
Hydr
ost
ati
c dr
ive,
s
pl
it
sys
tem
t
ype
Hydr
ost
at
ic
dri
ve,
co
mpa
ct,
re
ver
si
ble
ou
tput
Semi
r
ota
ry
act
uat
or
LINEAR
ACTUATORS
S
i
ng
le
ac
t
in
g
r
am
(
l
oa
d
r
et
u
n
rs
t
h
e
r
am
)
}T
l
S
in
gl
e
a
ct
i
ng
a
ct
ua
to
r
(
lo
ad
r
et
uns
r。。p
is
t
on) 二
吕那
S
in
gl
e
a
ct
i
ng
a
ct
ua
t
or
(
s
pri
ng
r
et
uns
r t
he
pi
s
to
n) 二
赓铆
Do
ubl
e
a
ct
i
ng
a
ct
ua
t
or 幸到)
Di
fe
re
nt
ia
l
a
ct
ua
to
r
wi
t
h
ov
er
si
ze
r
od
214
Do
ubl
e
a
ct
in
g
ac
t
uat
or
wi
t
h
do
ubl
e
e
nde
d
ro
d 二
扫玲
P
is
t
on
w
it
h
a
dj
us
t
ab
le
en
d
cu
sh
i
oni
n
g 二
乒卿
P
is
to
n
wi
t
h
f
ix
ed
e
nd
c
us
hi
on
in
g 幸豆习
T
el
es
co
pi
c,
si
ng
le
a
ct
i
ng
a
ct
ua
to
r
T
ele
sco
pic
,
do
ubl
e
ac
ti
ng
a
ctu
at
or
P
re
ss
ur
e
i
nt
en
si
f
ie
r
仁粗心仁咽侧仲片归屯一户归归
VALVE
CONTROL
M ECHANISM S
Undef
ined
cont
rol
Ha
nd
l
eve
r (
rot
ary
o
r i
l
nea
r)
Pus
h but
ton
Foot
l
eve
r
Cam
rol
ler
Pl
ung
er
(
pis
ton
o
r ba
ll
)
Sp
ring
De
tent
mechani
sm
Pr
essure
r
eli
ef
Pr
ess
ure
a
ppl
ie
d
Pne
uma
ti
c
pi
lot
’
Hyd
raul
i
c p
ilo
t
Sol
enoi
d
So
len
oid
/hy
dra
uli
c
pi
lot
215
Pn
euma
ti
c/h
ydr
aul
i
c pi
l
ot M〔二}
S
pri
ng
c
ent
red 叮丁丁  ̄
种
DIRECTIONAL
CONTROL
VALVES
Di
rect
iona
l c
ont
rol
val
ve
wit
h
} I
I
朴中。
t
wo
d
isc
et
r e
po
sit
i
ons
Di
rect
ional
cont
rol
val
ve
wit
h
t
hr
ee
d
isc
ret
e
po
siio
t n
s
Di
rect
ional
cont
rol
val
ve
wi
th
s
ign
iic
f a
nt
c
ros
s-o
ver
po
siio
t ns
Val
i
nf
ve
i
ni
wi
t
t
e
h
t
wo
di
s
numbe
cr
r
ete
of
i
and
nte
r
an
medi
ate
-
t
hro
tti
l
ng
p
osi
io
t ns
Va
lve
wi
th
t
hree
di
scr
ete
a
nd
a
n
助吻聊卿咖
i
nf
ini
te
number
of
i
nter
medi
ate }{ } }
t
hro
tli
ng
p
osi
io
t ns
Two
po
sii
t
on,
two
p
ort
va
lve
Two
po
sit
ion
,
th
ree
p
ort
v
alv
e
Two
p
osi
io
t n
,
fo
ur
p
ort
v
alv
e
Two
p
osi
ti
on,
fi
ve
p
ort
va
lve
Th
ree
p
osi
io
t n
,
fo
ur
p
ort
v
alve
wi
th
fu
ll
y c
los
ed
c
ent
er
co
nigu
f r
ati
on
人B
Po
rt
l
abe
in
l g:
Wo
rki
ng
li
nes
A,
B X
--蜘
-
-
Y PT
Pi
lot
l
ine
s
X,
Y
Pr
ess
ure
ine
l P
Tank
ine
l
T 翩A
=j
.
L
_
L
1
-L
1
B
LL
i
W PT
Check
val
ve
一 ̄-考卜一一.
216
}
号.
Sp
rin
g l
oad
ed
c
hec
k va
lve
髯命恤
Pi
l
ot
o
per
ate
d c
hec
k v
alv
e
OR
funct
ion
val
ve
AND
funct
ion
val
ve
Decel
erat
ion
val
ve
Decel
erat
ion
val
ve
SERVO
Pr
o
el
r
p
ie
or
i
t
f
v
a
AND
o
l
na
v
PROPORTIONAL
l
e
(
c
o
wi
nt
t
r
o
h
i
n
l
t
p
e
g
r
r
e
s
a
su
l
r
e
ma
VALVE
x.
幕嘶 卿
pr
ess
ure
l
imi
tat
ion)
Pi
l
ot
o
per
ate
d d
ire
cti
ona
l
pr
opor
iona
t l val
ve
4-
way
s
erv
o v
alv
e wi
th
mechanical
feedback,
st
andard overl
apping
and
hydr
aul
ic
zer
o
咖
PRESSURE
CONTROLS
伽
Thro
tti
l
ng or
if
ice
no
rmal
ly
cl
osed
or
nor
ma l
l
y o
pen
颐
(o
pti
onal)
Pr
ess
ure
r
eli
ef
v
alv
e (
fi
xed
)
Pr
ess
ure
r
eie
lf
va
lve
(
adj
ust
abl
e)
Det
ai
led
sy
mbol
of
pi
lot
op
erat
ed
pr
ess
ure
r
eie
l f
va
lve
(
compou
nd r
eli
ef
val
ve)
217
啄啊矛尽琳怀啄---ty潮t---the甸可
Si
mpif
l i
ed
s
ymb o
l o
f
c
ompound
r
eli
ef
val
ve
(
Pil
ot
lo
f w
e
xtena
r ll
y
dr
ained)
(
Pll
ot
lo
f w
i
nte
na
r l
l
y
dr
aine
d)
BRAKE
VALVE
Brake
val
ve
Unl
oad
ing
v
alv
e
(
accumul
ator
cha
rgi
ng
val
ve)
Count
er
bal
ance
va
lve
(
bac
k
pr
ess
ure
val
ve)
S
equen
ce v
alv
e wi
th
r
emo
te
c
ont
rol
(
ext
ena
r l
pi
lot
)
Sequenc
e v
alve
wi
th
d
ire
ct
c
ont
rol
(i
nte
na
r l
pi
lot
)
Of
lo
f a
din
g v
alv
e
Pr
ess
ure
r
edu
cin
g v
alv
e (
fi
xed
)
Pr
ess
ure
r
educ
ing
v
alv
e (
adj
ust
abl
e)
Pi
lot
ope
rat
ed
pr
ess
ure
r
educi
ng
val
ve
Pr
ess
ure
r
edu
cin
g v
alv
e wi
th
s
econda
y
r s
yst
em
r
eli
ef
FLOW
CONTROLS
Th
rot
tl
e
va
lve
no
t
afe
cte
d by
v
isc
osi
ty \才
一 - 艺欢尸--
218
子曰圈甲回
Thr
ott
le
val
ve
(
fixed)
Thr
ott
le
v
alve
(
adjus
tabl
e)
Fl
ow
c
ont
rol
v
alv
e,
p
res
sur
e a
nd
t
empe
rat
ure
compensa
ted
Fl
ow
cont
rol
val
ve
wit
h r
ever
se
re
f e
low
f
check
By
-pa
ss
low
f c
ont
rol
v
alve
Fl
ow
divi
der
FLUID
PLUM BING
AND
STORAGE
Pr
essure
sour
ce
Wor
king
ine,
l ret
ur
n l
ine,
f
eed
ine
l
Pi
lot
cont
rol
ine
l
Drai
n l
ine
Encl
osur
e l
ine
Fl
exi
ble
ine
l 奋一 -,
El
ect
ric
l
ine 爷
十上山山
Pi
pel
ine
conne
cti
ons +未
Cr
oss
p
ipe
lin
e (
not
co
nne
cte
d)
Ai
r vent
Res
ervoi
r wi
th
i
nle
t bel
ow
lui
f d l
evel
Reservoi
r wi
th
i
nlet
above
lui
f d l
evel
219
MISCELLANEOUS
SYMBOLS
赶
亡
El
ect
ric
mot
or
衅二 }拳。个
匠
二
Hea
t e
ngi
ne
Ele
ctri
c
moto
r wi
th
p
ump
a
nd
d
riv
e
coupin
l g.
Pl
ugg
ed
l
i
ne
Pl
ugg
ed
l
i
ne
wi
th
t
ake
-of
f
in
l e
Qu
ick
c
onn
ect
co
upl
i
ng
Ro
tar
y c
onn
ect
ion
Accumul
ator
含令勿
Fi
lt
er,
s
trai
ner
Cool
er
wi
th
cool
ant
l
ines
Heat
er
令今
Pr
ess
ure
g
auge
,
pr
ess
ure
i
ndi
cat
or
州一
R ow
met
er
Ther
momet
er
Pr
ess
ure
s
wit
ch
(
elect
ri
cal
)
Shut
-of
val
ve
220
APPENDI
X 2.
CONTROL
TERM S
AND
CONCEPTS
Act
uat
or
Th
e d
evi
ce
t
hat
c
aus
es
t
he
p
roc
ess
t
o p
rov
ide
t
he
o
utp
ut.
Th
e d
evi
ce
t
hat
pr
ovi
des
t
he
mo
tiv
e
power
t
o t
he
pr
oces
s.
Ampl
it
ude
quant
iz
ati
on
er
ror
Th
e
sa
mp l
ed
sig
nal
a
vai
la
ble
o
nly
wi
th
a
l
imi
te
d pr
eci
si
on.
Th
e e
rro
r be
twe
en
t
he
a
ctu
al
s
ignal
and
th
e s
ample
d si
gna
l.
Angl
e of
depart
ur
e
Th
e
angl
e
at
whic
h a
lo
cus
l
eave
s a
c
omp
lex
po
le
i
n t
he
s
-pl
ane
.
Asy
mpt ote
ce
ntoi
r d
Th
e c
ent
er
o
f t
he
l
ine
ar
sy
a mp
tot
es,
6A・
As
ymptot
es
The
pat
h t
he
root
l
ocu
s f
oll
ows
s
a t
he
par
amete
r becomes
ve
ry
l
arg
e an
d a
ppr
oaches
i
nini
f t
y.
Th
e
n
umb er
o
f a
symptot
es
i
s
eq
ual
t
o t
he
number o
f p
ole
s mi
nus
t
he
number
o
f z
eros
.
Aut
om at
ion
The
con
trol
of a
p
roce
ss
by
aut
omati
c
means
.
Aux
ili
ary
pol
y no
mi al
Th
e
equ
ation
that
i
mme d
iat
el
y
pre
cedes
t
he
zer
o e
nty
r i
n t
he
Ro
uth
a
rra
y.
Bandwi
dth
The
re
f q
uenc
y a
t wh
ich
t
he
re
f qu
enc
y r
esp
ons
e h
as
d
ecl
ine
d 3
dB
ro
f m
i
ts
l
ow-
re
f qu
enc
y
val
ue.
Bl
ock
di
agrams
Un
idi
rec
ti
ona
l,
o
per
ati
ona
l
bl
ocks
t
hat
r
epr
ese
nt
t
he
t
ran
sfe
r f
unc
tio
ns
o
f t
he
e
leme
nts
o
f t
he
s
yst
em.
Bode
pl
ot
The
lo
gait
rhm o
f t
he mag
nit
ude
of
t
he t
rans
fer
f
unct
ion
is
pl
ott
ed ve
rsu
s t
he
log
ari
thm
of
co,
t
he
re
f quen
cy .
The pha
se,
0,of
t
h e
t
ran
sfer
f
unct
ion
i
s s
epara
tel
y p
lot
ted
ver
sus
t
he
loga
rit
hm
o
f
t
he
re
f quency
.
Br
eak
f
equenc
r y
T he
frequenc
y at
which t
he sy
a mp t
oti
c a
pproxi
ma t
ion
of
th
e re
f que
ncy r
espo
nse
f
or
a
po
le
(
o r
z
ero)
change
s s
lope
Br
eaka way point
Th
e
po i
nt
on
the
rea
l
axis
wh er
e t
he
locus
de
parts
ro
f m t
he
rea
l
axis
o
f t
he
s-
pla
ne.
Casc
a de
compe nsat
ion
network
A
co
mp ensat
or
net
wo r
k plac
ed i
n
cascade
or
se
ie
r s wi
th
the
s
yste
m proc
ess.
Cauchy's
theoem
r
I
f
a
c
ont
our
e
nci
rc
les
Z
z
eros
a
nd
P
po
les
o
f F(
s)
t
ra
ver
si
ng
c
loc
kwi
se,
t
he
c
orr
espo
ndi
ng
c
ont
our
i
n t
he
F(s
)-pl
ane
e
nci
rc
les
t
he
o
ig
r i
n o
f t
he
F(s
)-p
lan
e N=Z一P
ti
mes
cl
ockwi
se.
Charac
ter
ist
ic
equat
ion
221
The
re
lat
ion
fo
r me
d by
equa t
ion
to
zer
o t
he
deno
minat
or
of
a
tr
ans
fer
f
unc
tio
n.
Cl
ose
d-l
oop f
eedback
contol
r syste
m
A
s
yst
em t
hat
uses
a
meas u
r e
me nt
o
f t
he
out
put
and
c
omp a
res
i
t
wit
h t
he
des
ir
ed
o
utp
ut.
Cl
ose
d-l
oop f
eque
r ncy
res
po nse
Th
e f
re
que
ncy
r
es
pon
se
o
f
th
e
cl
os
ed-
l
oop
t
r
ans
fe
r
fu
nct
i
on
T(
可).
Cl
osed-
loop
s
yst
em
A syste
m wit
h a
me as
urement
of
t
he out
put s
ign
al
and
a co
mp ar
iso
n wi
th
t
he des
ir
ed out
put
t
o generat
e an
error
s
ignal
t
hat
i
s
appl
ied
to
the
act
uat
or.
Co
mpe nsation
Th
e alt
era
tion
or
adjus
tment
of
a
cont
rol
sys
tem i
n or
der
t
o pr
ovide
a
sui
t
abl
e
p e
rfo
rma n
ce.
Co
mpe n s
ator
An
addi
ti
onal component
or
cir
cui
t tha
t i
s ins
ert
ed i
nto
the s
yst
em to
compensa
te f
or a
p
erformance
de f
ici
ency.
Co
mple xit
y of
desig
n
Th
e i
ntri
cate
patt
en
r of
int
erwov
en p
arts
and k
nowledge
r
equi
red.
Co
nformal ma pping
A
c
ont
our
ma
ppi
ng
t
hat
r
et
ai
ns
t
he
a
ngl
es
o
n t
he
s-
p
lan
e o
n t
he
F(
s)-
pl
ane
.
Cont
our
map
A
c
ont
our
o
r t
ra
jec
tor
y i
n o
ne
pl
ane
i
s
ma
ppe
d i
nto
a
not
her
pl
ane
b
y a
r
ela
tio
n F(
s).
Cont
ol
r s
yst
em
An
i
nte
rco
nne
cti
on
o
f c
ompo
nen
ts
f
ormi
ng
a
s
yst
em
c
onigu
f rat
ion
t
hat
wi
ll
p
rov
ide
a
d
esi
red
response.
Cont
olabl
r e s
yst
em
A
s
yst
em
i
s c
ont
ro
ll
ab
le
o
n t
he
i
nt
er
val
[
to,
t
f
]
i
f t
her
e e
xis
t
s a
c
ont
i
nuo
us
i
npu
t
u(
t)
s
uch
t
ha
t
an
y i
ni
ti
al
s
ta
te
x(
to
)ca
n b
e d
ri
ve
n t
o
an
y a
rbi
t
rar
y
tr
ia
l
st
at
e x
(t
f
)
in
a
i
f
ni
te
ti
me
i
nt
er
va
l
t
,一t
o>0.
Cor
ner
f
eque
r ncy
Se
e Brea
k re
f que
nc y
.
Cr
it
ica
l damping
Th
e c
ase
where
dampin
g i
s
on
t
he
b
oun
dar
y be
twe
en
u
nde
rda
mpe
d a
nd
o
ver
damp
ed.
Damped
os
cil
lat
ion
An
os
cil
lat
ion
i
n wh
ich
t
he
a
mpl
it
ude
decr
eas
es
wit
h
time.
Da
mp i
ng
rat
io
A
measu
re
of
dampi
ng.
A
di
men
sion
les
s n
umber
o
f t
he s
eco
nd-
ord
er
c
har
act
eri
st
ic
e
qua
ti
on.
Dc
mot
or
An
e
lec
tri
c
ac
tua
tor
t
hat
u
ses
a
n i
npu
t
vo
lta
ge
a
s a
c
ont
rol
v
ari
abl
e.
De
adbeat
es
r ponse
A
s
yst
em
wi
th
a
r
api
d r
esp
ons
e,
mi
ni
mal
o
ver
sho
ot,
a
nd
z
ero
s
tea
dy-
st
at
e
er
ror
f
or
a
s
te
p
inpu
t.
De
cibel口b)
222
The
u
nit
s
of
t
he
l
oga
it
r h
mic
g
ain
.
Des
ign
The
process
of c
oncei
vin
g or i
nvent
ing
the
forms.
par
ts,
and
deta
il
s of
a
sys
tem t
o ac
hieve
a
s
pec
ifi
ed purp
ose.
De
sign
gap
A
gap
between t
he
comple
x physi
cal
syst
em and
the
desi
gn
model
in
tri
nsi
c
to t
he
progr
essi
on
f
r
om the i
ni
tia
l
concept
t
o
the f
inal
p
roduct.
De
sign
of
a contro
l s
yst
em
The
arrangement
or
t
he pl
an of
th
e sys
tem s
truct
ure
and
the
s
ele
cti
on of
s
uit
able
component
s
and
pa
ramet
ers
.
De
sig
n spe
cif
ica
tions
A
s
et
of
pr
escri
bed
p
erf
orma
nce
c
rit
eri
a.
Di
gi
ta
l
co
mput
er
c
omp
ens
ator
A
s
yst
em
t
hat
u
ses
a
di
gi
ta
l
co
mput
er
a
s t
he
c
omp
ens
ato
r e
leme
nt.
Di
gi
ta
l
co
ntol
r s
yst
em
A
c
ont
rol
s
yst
em
u
sin
g d
igi
ta
l
si
gna
ls
a
nd
a
d
igi
tal
c
ompu
ter
t
o
co
ntr
ol
a
p
roc
ess
.
Di
ect
r s
yst
em
Se
e
Open-
loop
s
yste
m.
Di
scet
r e-
ti
me appr
oximati
on
An
a
pprox
imati
on
used
to
o
bta
in
t
he
t
ime
r
espo
nse
of
a
s
yst
em
bas
ed
o
n t
he
d
ivi
si
on
o
f t
he
t
ime
i
nt
o
sma
ll
i
ncr
eme
nts
,△,
.
Di
stur
bance
s
ignal
An
u
nwa
nte
d i
nput
si
gna
l
tha
t
afe
cts
t
he
s
yst
em'
s
out
put
s
ign
al.
Domi
nant
r
oot
s
The
r
oot
s o
f t
he
c
har
act
eri
st
ic
e
qua
ti
on
t
hat
c
aus
e t
he
do
min
ant
t
ran
sie
nt
r
esp
ons
e o
f t
he
s
yst
em.
Eng
inee
ring
de
sign
The
pro
ces
s of
d
esi
gni
ng
a
t
ech
nic
al
s
yst
em.
Err
or
si
gnal
Th
e d
if
fe
ren
ce
b
etwe
en
t
he
d
esi
re
d o
utpu
t,
R(
s),a
nd
t
he
a
ctu
al
o
utp
ut,
Y
(
s)
.
The
ref
ore
E(
s)=R(
s)一Y(
s)
Fe
edback
si
gnal
A
mea
sure
o
f t
he
o
utpu
t
of
t
he
s
yst
em
u
sed
f
or
f
eed
bac
k t
o c
ont
rol
the
s
yst
em.
Fouri
er
t
rans
form
Th
e t
ra
nsf
orma
ti
on
o
f
a
f
unc
ti
on
o
f
ti
me,
f
(
t),
int
o
th
e re
f q
uen
cy
d
oma
in.
Fundament
al
mat
rix
Se
e Tr
ans
iti
on
mat
rix.
Ga
in
mar
gin
Th
e i
ncr
eas
e i
n t
he
s
yst
em
ga
in
whe
n p
has
e=一1
800
th
at
wi
ll
r
esul
t
in
a
ma
rgi
nal
ly
s
ta
ble
s
yst
em
wit
h
in
ter
sec
tio
n o
f t
he
-1+j
0 po
int
o
n t
he
Ny
qui
st
d
iag
ram.
I
ntegrat
ion
ne
twor
k
223
A
n
etwo
rk
t
hat
ac
ts,
in
pa
rt,
l
i
ke
a
n mt
egr
at
(,
Lag
net
wor
k
See
Phas
e-l
ag n
etwor
k.
Lapl
ace
t
ransfor
m
A
tr
ansf
ormati
on of
a
fu
nct
ion
f
(
t)
f
rom
the
t
ime
d
oma
in
i
nto
t
he
c
omp
lex
f
req
uen
cy
do
mai
n外e
ldi
ng
F(
s),
Lead
net
work
See
Phse
a -l
ag
networ
k.
Le
ad-l
ag net
work
A
n
e t
work
wit
h t
he
char
act
eri
st
ic
of
bo
th
a
lea
d net
work
and
a
l
ag
net
wor
k.
Li
near
appr
oximati
on
An
a pp
roxi
mate
model
t
hat
r
esul
ts
i
n a
l
ine
ar
rel
at
ions
hip
be
twe
en t
he
ou
tpu
t a
nd
t
he
i
npu
t o
f
t
he
devi
ce.
Li
near
s
yst
em
A
s
yst
em
t
hat
s
ati
sie
f s
t
he
pr
ope
rti
es
o
f s
upe
rpo
sit
ion
a
nd
ho
mog
ene
iy.
t
Locus
A
path
or
t
ra
jec
tor
y t
hat
is
t
rac
ed
o
ut
a
s a
p
ara
met
er
i
s
cha
nge
d.
Lo
gari
thmic
magni
tude
T
he
l
og
ari
t
hmi
c
of
a
ma
gni
t
ude
o
f
th
e
tr
ans
fe
r
fu
nct
i
on,
20
l
ogl
of
GI
.
Lo
ga r
ithmi
c plot
S
ee Bo
de plot
.
Mason
loop
rule
A
r
ule
t
hat
e
nab
les
&e
us
er
t
o o
bta
in
a
t
ra
nsf
er
f
unc
ti
on
b
y t
ra
cin
g pa
ths
a
nd
l
oops
wi
t
hin
a
s
yst
em.
M at
he mati
cal
mode
ls
De
scr
ipt
io
ns
o
f
th
e b
eha
vio
t
of
a
s
yst
em
u
sin
g ma
the
mat
i
cs.
Max
imum
val
ue
o
f t
he
f
eque
r ncy
es
r po
nse
A
pa
ir
o
f
co
mpl
ex
po
les
wi
ll
re
sul
t
i
n
a
ma
ximu
m v
alu
e f
or
t
he
f
req
uen
cy
r
esp
ons
e
oc
cur
ri
ng
a t
the
re
sonant
re
f quen
cy.
Mi
nimum phase
Al
l
t
he
zer
os o
f a
tr
ans
fer
f
unc
tio
n l
ie
i
n t
he
l
eft
-ha
nd
s
ide
o
f t
he
s-
p
lan
e.
Mul
tivari
abl
e cont
ol
r s
yst
em
A
s
yst
em
wi
th
mo
re
t
han
o
ne
i
npu
t v
ari
abl
e
or
mor
e t
han
o
ne
o
utp
ut
v
ari
abl
e.
Nat
ural
f
equency
r
The
re
fque
ncy
o
f n
atu
ral
os
ci
ll
at
io
n t
hat
wo
uld
oc
cur
f
or
t
wo
c
omp
lex
po
les
i
f
th
e da
mpi
ng
were
equal
t
o z
ero.
Ne
gati
ve
feedback
The
output
s
ignal
i
s
fe
d b
ack
s
o t
hat
it
sub
tra
cts
ro
f m
t
he
i
npu
t
si
gna
l.
Ni
chol
s c
hart
A c
hart
di
spl
ayi
ng
the
c
urv
es
f
or
t
he
r
el
at
io
nsh
ip
be
twe
en
t
he
o
pen
-lo
op
a
nd
c
los
ed-
lo
op
re
f
quency
r
e s
pon
se.
224
No
nminimum pha
s e
Tr
ansf
er
f
unct
ions
wi
t
h z
ero
s i
n t
he
r
ight
-ha
nd
s一p
lan
e.
Number
of
separat
e l
oci
Equ
al
t
o the
number
of
pole
s o
f t
he
tr
ansf
er
func
ti
on
ass
uming
tha
t t
he
nu
mbe
r of
p
ops
i
s
g
re
ate
r t
han
or
equal
t
o t
he
number
of
z
ero
s of
t
he
tr
ans
fer
f
unc
tio
n.
Nyqui
st
st
abil
ity
cr
ite
rion
A
f
eed
bac
k s
yst
em
i
s
st
abl
e
if
,
an
d o
nly
i
f,
t
he
c
ont
our
i
n t
he
G(
s)-
pla
ne
d
oes
no
t
en
cir
cl
e
t
he(
一1,
0)
p
oint
wh
en
t
he
n
umb
er
o
f p
ole
s o
f
G(
s)
i
n t
he
r
ig
ht-
han
d s-
p
l
ane
i
s
ze
ro.
I
f
G(
s)
h
s
a p
ole
s
P
i
n
th
e r
igh
t-h
and
pl
ane
,
th
en
t
he
n
umb
er
of
c
ount
er
cl
ockwi
se
e
nci
rc
leme
nts
of
t
he(
一1,
0)
poi
nt
mu
st
b
e e
qua
l
to
P
f
or
a
s
ta
ble
s
yst
em.
Obs
ervabl
e s
yst
em
A
sy
st
em
i
s
ob
ser
va
ble
on
t
he
i
nt
erv
al
[
to
,t
f
]
if
an
y
in
it
ia
l
st
at
e
x(
to)
is
u
ni
qu
el
y
de
te
rmi
ne
d
b
y
obs
er
vi
ng
t
he
o
utp
ut
y(t
)
on
t
he
i
nt
erv
al
[
to
,tf]・
Ope
n-l
oop c
o ntol
r s
yste
m
A
syst
em t
hat
uti
li
zes
a
dev
ice
to
cont
rol
t
he
proce
ss
wi
t
hout
usi
ng
fee
dback.
Thus
th
e o
utp
ut
ha
s n
o efe c
t
upon t
he
si
gnal
t
o t
he
proces
s.
Ope
n-l
oop s
y ste
m
A
syst
em wit
hout
fe
edba
ck t
hat
di
rec
tly
gene
rat
es
t
he
ou
tpu
t i
n
es
r p
onse
t
o an
i
npu
t s
ign
al.
Opt
imi
zat
io
n
The
a
djus
tme
nt
of
t
he
p
ara
met
ers
t
o a
chi
eve
t
he
mos
t
fa
vor
abl
e
or
a
dva
nta
geo
us
d
esi
gn.
Opt
imum
c
ont
ol
r s
yst
em
A
s
yst
em
whos
e p
ara
met
ers
a
re
a
dju
ste
d s
o t
hat
th
e pe
rfo
rma
nce
i
nde
x r
eac
hes
a
n e
xtr
emu
m
val
ue.
Opt
iona
l
co
ntol
r s
yst
em
A
s
yst
em
whos
e
pa
rame
ter
s
ar
e
ad
jus
ted
s
o t
hat
t
he
pe
rf
orma
nce
i
nde
x r
eac
hes
a
n e
xtemu
r m
val
ue.
Over
shoot
Th
e amount
by whi
ch t
he
sys
tem
ou
tput
r
esponse
pr
oce
eds
beyon
d t
he
des
ied
r es
r po
nse
.
Pa
ramete
r de
sig
n
A
method
of s
elec
tin
g one
or
t
wo
par
ameter
s u
sing
t
he
roo
t
locus
met
hod
.
Pe
ak
t
ime
Th
e t
ime
f
or
a
s
yst
em
t
o
re
spo
nd
t
o a
s
tep
i
npu
t a
nd
r
is
e t
o a
p
eak
r
esp
ons
e.
Pe
rformance
i
nde
x
A
qu
anti
t
ati
ve
me
asu
re
o
f t
he
p
erf
orma
nce
o
f a
s
yst
em.
Pha
se
mar
gin
Th
e a
moun
t
of
p
has
e s
hif
t
of
t
he
GH
(
j叻 a
t
un
it
y
ma
gni
tu
de
t
hat
wi
ll
re
sul
t
in
a
ma
rgi
nll
ay
s
tab
le
s
yst
em
wi
th
i
nte
rse
cti
ons
o
f t
he一1
+j0
poi
nt
o
n t
he
Ny
qui
st
d
iag
ram.
Pha
se-
lag
net
wor
k
A
net
work
t
hat
p
rov
ide
s an
ega
tiv
e p
has
e a
ngl
e a
nd
as
ign
ifi
can
t a
tte
nua
tio
n o
ver
t
he
f
reque
ncy
r
ange
of
i
nte
rest
.
225
Phase-
lead
network
A n
etwork
t
hat
pr
ovide
s a
pos
it
ive
phas
e a
ngle
ove
r t
he
fre
quenc
y r
ange
of
i
nte
res
t.
Th
us
p
hase
le
ad
can
be
use
d t
o c
ause
a
sy
stem
to
hav
e a
n a
dequa
te
pha
se
margi
n.
PI
c
ont
rol
ler
Co
ntr
oll
er
wi
th
a
p
rop
oti
rona
l
te
rm
a
nd
a
n i
nte
gra
l
te
rm
(
Pro
por
ti
ona
l-I
nte
gra
l).
PI
D c
ont
ol
r l
er
A c
ontr
oll
er
wit
h t
hre
e t
erms
in wh
ich t
he
out
put
is
t
he sum of
a
pro
poti
r o
nal
t
erm,
a
n
i
nt
egra
ti
ng t
erm,
and
a
dif
fer
ent
iat
i
ng
ter
m, wi
th
a
n a
djus
tabl
e g
ain
f
or
e
ach
te
rm.
Pl
ant
See
Pr
oces
s.
Pol
ar
pl
ot
A
p
lot
of
t
he
r
eal
pa
rt
of
) v
f
i
(
G ers
us
t
he
i
mag
ina
ry
p
art
of
G(
j(o
).
Pos
iti
ve
f
eedback
Th
e o
utpu
t s
ign
al
i
s
fe
d b
ack
s
o t
hat
is
a
dds
t
o t
he
i
npu
t s
ign
al.
Preg
ilter
A
t
ra
nsf
er
f
unc
ti
on,
G,
(
s),t
ha
t
i
fl
t
ers
t
he
i
np
ut
s
ig
nal
R(
s)
p
ri
or
t
o
th
e c
al
cu
la
ti
on
o
f
th
e
er
ror
s
ignal
.
Pr
inci
ple
o
f t
he ar
gument
Se
e
Cauc
hy's
the
ore
m.
Process
Th
e de
vic
e,
p
lant
,
or
s
yst
em
u
nde
r
co
ntr
ol.
Process
cont
ol
r l
er
Se
e PI
D cont
rol
ler
.
Produc
tivi
ty
Th
e r
ati
o
of
p
hys
ica
l
ou
tput
to
p
hys
ica
l
in
put
of
a
n i
ndu
str
ial
pr
oce
ss.
Re
lat
ive
s
tabi
li
ty
The
prop
ert
y t
hat
i
s meas
ure
d by
t
he
r
ela
ti
ve
r
eal
pa
rt
o
f e
ach
r
oot
o
r p
air
o
f r
oot
s o
f t
he
c
harac
ter
ist
i
c e
quat
ion.
Re
sonant
f
equenc
r y
The
f
req
uenc
y,件,
at
wh
ich
t
he
ma
ximum
v
alu
e o
f
th
e f
re
que
ncy
r
esp
ons
e
of
a
c
omp
lex
pa
ir
of
pol
es
i
s a
tai
ned.
Ri
se
t
ime
The
ti
me f
or
a
s
yst
em t
o r
espon
d t
o a
st
ep
i
nput
a
n d
a
tta
in
a
r
espo
nse
eq
ua l
to
a
p
erce
nta
ge
of
t
he magni
t
ude
of
the
i
nput.
The
0-100%
ris
e t
ime,
T,,meas
ures
t
he
time
to
100% o
f t
he
ma
gni
t
ude
o
f t
he
i
npu
t.
Al
t
ern
at
iv
el
y,爪me
as
ur
es
t
he
t
i
me
f
ro
m 1
0%
t
o 9
0%
o
f t
he
r
esp
ons
e t
o t
he
s
tep
i
nput
.
Ri
sk
Un
ceta
r i
nti
es
e
mbo
die
d i
n t
he
u
nin
ten
ded
c
ons
equ
enc
es
o
f a
de
sign
.
Robot
Pr
ogr
amma bl
e c
ompute
rs
int
egr
ate
d wit
h a
ma
nip
ula
tor
.
A
r
epr
ogr
amma
ble
, mu
lti
func
tio
nal
ma
ni
pul
ato
r u
sed
fo
r a
var
iet
y
of
ta
sks.
226
Robus
t c
ont
rol
s
yst
em
A
sys
tem
tha
t e
xhi
bit
s t
he
des
ir
ed
per
for
man
ce
ui
the
pr
ese
nce
of
s
ign
iic
f a
nt
pl
ant
unce
rta
nt
i y.
Root
l
ocus
Th
e l
ocu
s o
r p
ath
o
f t
he
r
oot
s
tr
ace
d o
ut
o
n t
he
s
-pl
ane
a
, a
pa
rame
ter
i
s
ch
ange
d.
Root
l
ocus
met
hod
Th
e me
tho
d f
or
d
ete
rmi
ni
ng
t
he
l
ocu
s of
r
oot
s
of
t
he
c
har
act
er
is
ti
c
eq
uat
i
on
1+KP(
s)二0
a
s
K
v
ari
es
ro
f m
0
t
o
in
fin
it
y.
Roo
t l
ocus
se
gme nt
s
on t
he
r
eal
ax
is
Th
e r
oot
l
ocus
lyi
ng
in
a
se
cti
on
of
t
he
r
eal
a
xis
t
o t
he
l
eto
f
f a
n o
dd
nu
mbe
r o
f p
ole
s a
nd
zeros.
Ro
ot
s
ens
it
iv
iy
t
The
s
ensi
ti
vi
ty
of
t
he
roo
ts
as
a
p
ara
meter
ch
ang
es ro
f m i
ts
nor
ma l
va
lue
.
The
ro
ot
se
nsit
ivi
ty
i
s
the
i
ncr
ement
al
c
hange
in
t
he
r
oot
di
vid
ed
by
the
pro
port
iona
l c
hange
of
t
he
par
amete
r.
Rout
h-Hurwi
tz
c
rit
eri
on
A
c
rit
er
ion
f
or
d
ete
rmi
nin
g t
he
s
tabi
l
ity
o
f a
s
yst
em
by
e
xami
nin
g t
he
c
har
act
eri
st
ic
e
qua
ti
on
of
t
he
t
rans
fer
f
unc
tion.
The
c
rit
eri
on
st
ates
t
hat
t
he
number
of
r
oot
s of
t
he
c
har
act
eri
sti
c
equat
ion
wit
h pos
it
ive
r
ea l
par
ts
is
equal
to
t
he
number
of
changes
o
f s
ign
of
t
he
c
oeff
ic
ient
s
i
n
t
he f
ir
st
col
umn of
t
he Rout
h a
rra
y.
Sa
mple dat
a
Da
ta
obt
aine
d for
t
he s
ystem
vari
able
s only
at
d
isc
ret
e i
nte
rv a
ls.
Data
obt
ain
ed
onc
e eve
ry
s
a
mp l
ing
per
iod.
Sa
mple-dat
a s
yst
em
A
s
yst
em
wh
ere
pa
to
r
f
th
e s
yst
em
a
cts
o
n s
ampl
ed
d
ata
(
samp
led
v
ari
abl
es)
.
Sa
mpl
i
ng
p
eri
od
Th
e p
eri
od
wh
en
a
ll
t
he
nu
mbe
rs
l
eav
e o
r e
nte
r t
he
c
ompu
ter
.
Th
e p
eri
od
f
or
wh
ich
t
he
s
amp
led
va
ria
ble
i
s
he
ld
c
ons
tan
t.
Set
ting
l t
ime
Th
e t
ime
r
equ
ire
d f
or
t
he
s
yst
em
out
put
t
o s
ett
l
e wi
thi
n ac
ert
ai
n p
erc
ent
age
o
f t
he
i
npu
t
ampl
it
ude.
S
ign
al-
flow gra
ph
A
dia
gram t
hat
consi
st
s of
nod
es conn
ecte
d by
se
ver
al
d
ire
cte
d b
ran
che
s a
nd
t
hat
i
s a
gr
aphica
l
repres
ent
ati
on o
f a
s
et
o
f in
l e
ar
rel
at
ion
s.
Si
mul
ati
on
A mod el
of
a sys
tem t
hat
is
used
to
i
nves
ti
gate
t
he beh
avio
r o
f a
s
yst
em by
ut
il
iz
ing
ac
tual
i
nput
signals
.
Specif
ica
tions
St
a
temen t
s t
hat e
xpic
l it
ly s
tat
e what
t
he dev
ice
or pr
oduct
i
s t
o be
and
to
do.
A se
t of
p
r
escrib
e d
perf
ormance
cri
te
ria.
Sta
bili
y
t
A
pe
rforma n
ce measu
re
of
a sy
stem.
A
sys
tem
is
st
able
i
f
all
the
pol
es
of
t
he
t
ran
sfe
r f
unc
tio
n
h
ave
negati
ve r
eal
par
ts.
227
S
tabi
li
ty
o
f a
s
ampl
ed-
dat
a s
yst
em
Th
e s
tab
le
c
ond
it
io
n e
xis
ts
wh
en
a
ll
t
he
p
ole
s of
t
he
c
los
ed-
lo
op
t
ra
nsf
er
f
unc
ti
on
T(z
)
ar
e
wit
hin
t
he
u
nit
c
irc
le
o
n t
he
z
-pl
ane
.
S
tabl
e
syst
em
A
d
yna
mic
s
yst
em
wi
t
h a
b
oun
ded
s
yst
em
r
espo
nse
t
o a
b
oun
ded
i
npu
t.
St
ate
di
ff
erent
ial
equat
ion
Th
e
difer
ent
ial
eq
uati
on
f
or
t
he
s
tat
e
ve
cto
r:
z二Ax+Bu.
St
ate
of
a
s
yst
em
A s
et
of
nu
mber
s s
uch
th
at
th
e knowle
d g
e o
f t
hes
e nu
mb e
rs
and
t
he
i
npu
t f
unc
ti
on
wi
ll
,
wi
th
t
he
equa
ti
ons
d
escr
ibi
ng
the
dyn
a mi
cs,
pr
ovide
t
he
fut
ure
st
at
e
of
t
he
s
yst
em.
St
ate
var
iabl
e f
eedback
Oc
cur
s
wh
en
t
he
c
ont
rol
s
ign
al
u,
for
t
he
pr
oce
ss
i
s
a
d
ire
ct
f
unc
ti
on
o
f al
l
t
he
s
tat
e
va
ria
ble
s.
St
ate
var
iabl
es
The
s
et
o
f v
ari
abl
es
t
hat
d
esc
rib
e t
he
s
yst
em.
St
ate
vect
or
The
v
ect
or
c
ont
ai
nin
g a
ll
n
s
tat
e
va
ria
ble
s,
x,
,x2……,
xR・
St
eady-
state
eror
r
Th
e er
ror
wh en th
e ti
me per
iod i
s
large
and th
e tr
ansi
ent
res
po ns
e has
de
caye
d,
le
avi
ng t
he
c
ont
inuous
response.
Sy
nthesi
s
The
p r
ocess
by which n
ew physi
cal
conf
igurat
ions
are
cre
ate
d .
The
combin
ing
of
s
epa
rate
el
eme nts
or
devic
es t
o f
orm ac
oherent
whole.
Sy
stem
An
i
n t
erco
nn e
ction of
e
lements
and
device
s for
a
desi
red
pur
p ose.
Sy
stem s
e n
sit
ivi
ty
Th
e rat
io
of t
he c
h a
nge i
n t
he sys
tem tr
an s
fer
funcio
t n
to
the chan
ge of
a
pr
oces
s t
ran
sfe
r
f
unc
io
t n
(or
parame t
er)
f
or
a s
ma li c
n re
me nt
al
change.
Te
st
i
nput
s
igna
l
An
i
npu
t s
igna
l
us
ed
a
s a
s
tan
dar
d t
est
of
a
s
yst
em'
s
abi
l
ity
t
o
re
spo
nd
a
deq
uat
el
y.
Thr
ee-
mode
c
ont
ol
r l
er
See
PI
D cont
rol
ler
.
Thr
ee-
ter
m c
ont
ol
r l
er
See
P1
D cont
rol
ler.
Ti
me
de
lay
A
pu
re
ime
t d
ela
y,
T,
so
t
hat
ev
ent
s
oc
cur
in
r g
a
t
ti
me
ta
t
on
e po
int
i
n
th
e s
yst
em
o
ccur
a
t
a
not
her
p
oin
t i
n t
he
s
yst
em
a
t a
l
at
er
t
ime
,
t+T.
Ti
me
domai
n
The
ma
the
mat
ica
l
doma
in
t
hat
i
nco
rpo
rat
es
t
he
t
ime
r
esp
ons
e a
nd
t
he
d
esc
ip
r t
ion
o
f a
s
yst
em
i
n t
erms
of
t
ime
t
.
Ti
me-v
aryi
ng
s
yst
em
A
syst
em
f
or
whi
ch
on
e o
r mo
re
p
ara
met
ers
ma
y v
ary
wi
th
t
ime
.
228
Trade-
off
Th
e r
esu
lt
o
f ma
kin
g aj
udg
men
t a
bou
t h
ow
muc
h c
omp
romi
se
mus
t b
e ma
de
be
twe
en
c
onf
li
cti
ng
c
rit
eri
a.
Transf
er
F
uncti
on
Th
e r
ati
o o
f t
he
La
pla
ce
t
rans
for
m o
f t
he
o
utpu
t v
ari
abl
e
to
t
he
La
pla
ce
t
ran
sfo
rm
o
f t
he
i
npu
t
vari
abl
e.
Tr
ansf
er
f
unctio
n i
n t
he
feque
r ncy
domai
n
Th
e
r
ati
o of t
he
out
put
to
th
e in
put
si
gna
l wh
ere
t
he
i
nput
i
s a
s
nu
i s
oid.
I
t
is
e
xpr
ess
ed
a
sG(
jco).
Trans
ient
es
r pons
e
The
re
spo
nse
of
a
s
yst
em
a
s a
f
unct
i
on
of
t
ime
.
Tr
ans
iti
on
matri
x,
D(
( t
)
T he
ma
tri
x e
xpo
nen
tia
l
fu
nct
ion
t
hat
d
esc
rib
es
t
he
u
nfo
rce
d r
esp
ons
e o
f t
he
s
yst
em.
升p
e number
The
n
umb
er
N
of
p
ole
s
of
t
he
t
ra
nsf
er
f
unc
ti
on,
G(
s),
at
t
he
o
ri
gi
n.
G(
s)
i
s
th
e f
orwa
rd
pa
th
t
ransfer
funct
ion.
V
elo
cit
y
er
or
r c
ons
tan
t,凡
Th
e c
ons
tan
t e
val
uat
ed
a
s l
im[
sG(
s
'-,o
)].
The
s
tea
dy-
st
ate
e
rro
r f
or
a
r
amp
i
npu
t (
of
s
lop
e A)
f
or
a
s
yst
em
i
s
eq
ual
to
A
/
K,,
.
Z-
plane
The
pl
ane
wit
h t
he
ver
ti
cal
a
xis
e
qua
l
to
t
he
i
magi
nay
r pa
rt
o
f z
a
nd
t
he
ho
riz
ont
al
a
xis
e
qua
l
t
o
t
he
r
eal
pa
rt
of
z.
Z-
trans
for
m
A
c
onf
or
mal
ma
ppi
ng
f
or
m t
he
s
-p
la
ne
t
o
th
e
z-
pl
an
e b
y t
he
r
el
at
io
n
z=e
sr.
^tr
ans
fo
rm
f
rom
t
he
s
-doma
in
t
o t
he
z-
domai
n.
229
图书在版编目(CI
P)数据
液压控制系统设计=Hy
dra
ul
ic
Con
tr
ol
S
yst
ems
De
-
s
ig
]训汤逢瑜编著 一兰州:
兰州大学出版社,
200
4.2
I
SBN 7-311一02346-7
I.
液… I.
杨二 I
Q.液压控制一系统设计一教
材一英文 I
V.TH
13
7
中国版本图书馆CI
P数据核字(2
004)
第00
-
967
9号
液压控制系统设计
HYDRAULI
C CONTROL
SYSTEMS
DESI
GN
杨逢瑜
兰州市大学出版社出版发行
兰州市天水路3
08号 电话:
861
715
6 邮编:
730
000
E-ma
il,
pre
ss@onbook.
com.
cn
h
ttp
-
//www
o
nboo
k.co
m e
n
兰州残联福利印刷厂印刷
开本 :
787X1
092
1
/16 印张 :
15
20
04年2月第 1版 20
04年2月第1次印刷
字数:37
8千字 印数 :
110
00册
I
SBN7-311-02346-7 定价:
250
0元