You are on page 1of 21

Linkpings universitet

TMHP51
IEI / Fluid and Mechanical Engineering Systems
____________________________________________________________________________________

Feedbacks in Hydraulic Servo Systems


Karl-Erik Rydberg
2008-10-15

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

FEEDBACKS IN ELECTRO-HYDRAULIC SERVO SYSTEMS


1.

Linear valve controlled position servo

A linear valve controlled position servo is shown in Figure 1. Leakage flow over the
piston with the flow-pressure coefficient Cp and a viscous friction coefficient Bp are
included in the model. The servo amplifier (controller) is proportional with the gain Ksa.
Fix reference

Cp

be

Ap

P1 V1

xp

Bp

be

Ap

Mt

V2 P2

FL

qL2

qL1

Position
transducer

xv
Servo

uc + amplifier i
Ksa
u f

Kf
Ps = const.

Figure 1: Valve controlled position servo

The transfer functions (in the frequency domain) of the components in the position
servo are illustrated in Figure 2. Threshold and saturation in the servo valve are
included.
FL

Vt
Kce

+
s
1

Ap2 4 be K ce
Threshold

uc

Ksa

Saturation

iv imax

ir
ein

Kqi
1
Ap 1 + s +
wv

Au(s)

1
xp 1 xp
s 2 + 2 dh s +
s
1
wh2 wh

Kf
Figure 2: Block-diagram of a linear position servo including valve dynamics and non-linearitys

The transfer function of the valve is G ( s) =


v

and damping is expressed as: h =

4 e A p2

1
1+

. The hydraulic resonance frequency

and h =

K ce
Ap

eMt

M tVt
Vt
The parameter values of the system are as follows:
Ap = 2,5.10-3 m2
e = 1,0109 Pa
Kf = 25 V/m
Bp = 0
-11
5
Kqi = 0,02 m3/As
Kce = 1,010 m /Ns
Mt = 1500 kg
Ksa = 0,1 A/V
-3
3
v = 1/v = 0,005 s
Vt = 1,010 m

Bp
4 Ap

Vt
.
eM t

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

These parameter values gives h = 129 rad/s and h = 0.155.


The open loop gain (Au(s)) of the position servo with Kv = hh = 20 1/s (Am = 6 dB) is
shown in Figure 3. Observe that the bandwidth of the valve v = 1/v = 200 rad/s is
higher than the hydraulic resonance frequency h.

Phase shift [degrees]

Amplitude [dB]

Kv
= dh = 0,155
wh

Am

Frequency [w/wh]
Figure 3: Bode-diagram of the open loop gain of the position servo depicted in Figure 2
when the servo valve is assumed to be very fast

Influence of valve dynamics


To really make use of the actuator capability of controlling the load it is very important
that the servo valve is fast enough. Normally the selected valve will have a bandwidth
(v) of at least twice as high as the hydraulic resonance frequency (h). Figure 4 shows
the open loop gain of the position servo depicted in Figure 2, with an ordinary valve
(v=200 rad/s) and a valve with slow response (v = 20 rad/s).
2

10

Amplitude

Amplitude

10

10

10

10

10

10

10

10

10

10

50

50
100

10

150

Phase

150

Phase

10
Frequency [rad/s]

100

200
250

200
250
300

300
350
0
10

10

Frequency [rad/s]

350
1

10

10
Frequency [rad/s]

a) Normal valve bandwidth, v = 200 rad/s

10

400
0
10

10

10

10

Frequency [rad/s]

b) Valve with low bandwidth, v = 20 rad/s

Figure 4: Bode-diagram of the open loop gain of a position servo with a) fast valve and b) slow valve

From Figure 4 it can be recognised that the open loop gain and thereby the amplitude
margin will be change because of the valve dynamics. For a slow valve (v < h) the
open loop gain can be approximated as
Kv
Au
, which gives Kvmax = v for a reasonable stability margin.
(1 + s / v )s

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

Closed loop stiffness


The most important characteristic of the servo system is the closed loop stiffness. The
stiffness of the closed loop system describes the controlled signal deflection Xp due to
variations in the disturbance force FL. By setting Uc = 0 in the block-diagram in Figure
2 the new block-diagram becomes as in Figure 5.

Vt
DFL Kce

1 + 4 b K s 2
Ap
e ce

1
s 2 + 2 dh s +
1
wh2 wh

Saturation

Kqi
1
Ap 1 + s
wv

xp

1
s

DXp

Threshold

imax

ein

Ksa

Kf

Figure 5: Block-diagram describing the stiffness of a closed loop position servo

The stiffness of the closed loop servo is defined as S c =

FL
. If the valve dynamics
X p

and the threshold are neglected the stiffness becomes

s
s 2 2 h
2 h 2
s

+
1

+
s
+
1
+
s
+
+
1
2
A p2 K v h2 K v h
A p2 K v
Kv
h
h

Sc = K v

Kv

V
K ce
K ce

s
t
1+
s
1 +

4 e K ce
2 h h
s3

where the steady state loop gain Kv = KsaKqiKf/Ap. The closed loop stiffness including
valve dynamics is shown in Figure 6. The amplitude curve is normalised as
A p2
Sc
, where K s = K v

Ks
K ce
2

10

Amplitude, (Sc/Ks)

Amplitude, (Sc/Ks)

10

10

10

10

10

10

10

10

10

10

10

10

10

200
150

Phase

150

Phase

10
Frequency [rad/s]

200

100

50

0
0
10

10

Frequency [rad/s]

100
50
0

10

10
Frequency [rad/s]

a) Normal valve bandwidth, v = 200 rad/s

10

50
0
10

10

10

10

Frequency [rad/s]

b) Valve with low bandwidth, v = 20 rad/s

Figure 6: Bode-diagram of the closed loop stiffness with a) fast valve and b) slow valve

In Figure 6b) it can be seen that the valve dynamics reduce the stiffness just at
frequencies around the bandwidth of the valve (v = 20 rad/s).

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

The threshold of the servo valve will also cause a position error Xp. If the threshold is
in
in the position error is X p =
, where in is nominal valve input current.
K sa K f

2.

Valve controlled position servo with load pressure feedback

The load pressure feedback is used to increase the hydraulic damping in the system. A
negative load pressure signal acts in the same way as a Kc-value (flow-pressure
coefficient) of the servo valve. Load pressure feedback can be of proportional or
dynamic type. Proportional pressure feedback is shown in Figure 7.
Proportional pressure feedback

Kpf
-

uc +

Ksa

Gv Kqi

+
-

1
Kce + Vt s / 4be

PL

Ap

FL
-

1
Mt s

xp 1_ xp
s

Ap
Kf
Figure 7: Block-diagram of a linear position servo with proportional pressure feedback (Bp = 0)

Load pressure feedback will mainly increase the hydraulic damping. It works just as a
Kc-value. In the above block diagram the proportional pressure feedback will increase
the effective Kc-value as follows, K ce' = K ce + K pf K sa Gv K qi . The resulting bode
diagram of the open loop gain (Au(s)) and the closed loop stiffness (Sc(s)) for a
hydraulic damping of h = 0,46 is shown in Figure 8. One negative effect of
proportional pressure feedback is that the steady state stiffness will be reduced.
2

10

Amplitude, (Sc/Ks)

Amplitude

10

10

10

10

10

10

10

10

10

10

10

10

Frequency [rad/s]

10

10

Frequency [rad/s]

50

200

100
150

200

Phase

Phase

150

250
300

100

50

350
400
0
10

10

10
Frequency [rad/s]

10

0
0
10

10

10

10

Frequency [rad/s]

Figure 8: Open loop gain (to the left) and closed loop stiffness of a position servo
with load pressure feedback

Dynamic pressure feedback is shown in Figure 9. The idea of using dynamic pressure
feedback is that the feedback signal shall reach its maximum value at a frequency,
which has to be damped (the hydraulic frequency h). Therefore, the pressure signal

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

will be high-pass filtered. At low frequencies the pressure feedback signal is low and
the reduction of the steady state stiffness will be very low compared to proportional
pressure feedback.
Dynamic pressure feedback

s/wf
s/wf + 1
uc

Ksa

Kpf

Gv Kqi +

1
Kce +

Vt
s
4be

PL

Ap

FL
-

1
Mt s

1
_ xp
s

xp

Ap
Kf
Figure 9: Block-diagram of a linear position servo with dynamic pressure feedback (Bp = 0)

3.

Valve controlled angular position servo with acc. feedback

Acceleration feedback works in principal as dynamic pressure feedback. When the load
starts oscillate there will be a feedback signal, which increase the hydraulic damping
just at the resonance frequency. The good thing with acceleration feedback is that the
steady state stiffness will not be affected. An angular position servo with acceleration
feedback is shown in Figure 10 and the corresponding block-diagram is expressed in
Figure 11.
Kf
Uc

Kac
Greg

V1
ps

..
qm

qm
Jt

pL

TL

V2

Figure 10: An angular position servo with acceleration feedback (Bm = 0)

From Figure 11 the effect of the acceleration feedback can be expressed as a change in
1
the second order transfer function of the hydraulic system, Gh ( s ) = 2
.
2 h
s
+
s +1
2

This transfer function will now change to Gh ( s ) =

1
K qi
2 h

+
+
K
K
Gv ( s ) s + 1
ac
sa
2

Dm
h h

s2

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

TL

Vt
Kce

+
s
1
Dm2 4 be K ce

uc

Ksa

iv

Kqi
1
Dm 1 + s +
wv

.
1
qm
s 2 + 2 dh s +
1
wh2 wh

Kac.s

Acceleration feedback

Kf

Position feedback

1 qm
s

Figure 11a: Block-diagram of an angular position servo with acceleration feedback (Bm = 0)

With h =

4 e Dm2
, Gv(s) = 1,0 and Bm = 0 the effective hydraulic damping (including
J tVt

acceleration feedback) will follow the equation: h* =

K ce
Dm

eJ t
Vt

+ K ac K sa K qi

e
Vt J t

Constant acceleration feedback gain (Kac) means that the total damping ( h* ) varies
according to variations in the inertia load Jt, as shown in Figure 11b.

Figure 11b: Damping in an angular position servo with acceleration feedback (Bm = 0)

4.

Velocity feedback in position control servos

Pressure and acceleration feedback is used to increase the hydraulic damping and this
makes it possible to increase the steady state loop gain Kv and the closed loop stiffness
will increase. Another way to increase the stiffness of a position servo is to introduce a
velocity feedback. A block-diagram of a linear position servo with velocity feedback is
shown in Figure 12.

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

FL

Vt
Kce

+
s
1

Ap2 4 be K ce
Threshold

uc

Ksav

Saturation

Kqi
1
Ap 1 + s +
wv

iv imax

ir
ein

1
xp 1 xp
s 2 + 2 dh s +
s
1
wh2 wh

Kfv

Velocity feedback

Kf

Position feedback

Figure 12: A linear valve controlled position servo with velocity feedback

If the bandwidth of the valve is relatively high and threshold and saturation is neglected
the velocity feedback will give the effect on the hydraulic resonance frequency and
damping as shown in Figure 13.
FL

Vt
Kce

1 + 4 b K s
2
Ap
e ce

uc

Ksav

Kqi
Ap

iv

Kf

Kvfv = 1 + Kfv Ksav

Kqi
Ap

1/ Kvfv
xp
2
2
s
d
h
+
s +1
Kvfv wh2 Kvfv wh

1 xp
s

Position feedback

Figure 13: A linear position servo with velocity feedback included

From Figure 13 the new resonance frequency and damping (hv and hv) caused by the
velocity feedback can be evaluated as
hv = h K vfv , hv = h

1
K vfv

, where the velocity loop gain is K vfv = 1 + K fv K sav

K qi
Ap

Designing the position control loop for the same amplitude margin as without velocity
feedback gives the following relations:
Steady state loop gain without velocity feedback: K v = K sa
Steady state loop gain with velocity feedback: K vv = K sav

K qi
Ap

Kf

K qi
A p K vfv

Kf

A certain amplitude margin means that K v h h . In this case h h = hv hv , which


implies that K v = K vv and thereby the servo amplifier gain K sav = K sa K vfv . With
velocity feedback, the servo amplifier gain (Ksav) can be increased in proportion to the
velocity loop gain Kvfv and the servo amplifier gain without velocity feedback, Ksa.
The open loop gain (Au(s)) for a position servo without (Kv = 20) and with velocity
feefback (Kvfv = 10 and Kvv =20) is shown in Figure 14.

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

Amplitude

10

10

10

10

10

10

10

Frequency [rad/s]
50

Phase

100
150
200
250
300
0
10

10

10

10

Frequency [rad/s]

Figure 14: Open loop gain for a position servo without and with velocity feefback (Kv = Kvv)

5.

Valve controlled velocity servo

If an integrating amplifier is used in a velocity servo the loop gain Au(s) will be in
principle the same as for a position servo with proportional control. Such a velocity
servo is shown in Figure 15.
Ap

vp

Ap

Mt

FL

Velocity
transducer

uc +

Integrating
servo ampl.

uf -

Ksa
____
s

Kf
i

Ps = const.

Figure 15: A linear valve controlled velocity servo

A block diagram of the velocity servo is shown in Figure 16.


FL
Integrating
amplifier

uc

Ksa

1 ir
s

Threshold

Kce
Gl
Ap2
-

Saturation

i imax

Kqi
G
Ap v +

ein

Gh

Au(s)
Kf
Figure 16: Block-diagram of a linear valve controlled velocity servo

xp

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

The transfer functions in the above block-diagram are:


Vt
1
1
Gv ( s ) =
, G1 ( s ) = 1 +
s , Gh ( s) = 2
s
2 h
4 e K ce
s
+1
+
s +1
2

An integrating amplifier means that the control error will be integrated and the steady
state control error becomes zero.

6.

Proportional valves with integrated position and pressure


transducers

In all fluid power applications a load has to be controlled by an actuator in respect of


speeds and forces. A new dimension of the ways to look upon these control aspects is to
use a control valve (proportional or servo valve), which is capable of controlling both
flow and pressure in the actuator ports (two ports for a double cylinder or motor). Such
a proportional valve has been developed by Ultronics. The principle design of the valve
is shown in Figure 17.
U

Load

Uc
+

Pressure
and
position
signals

Valve
controller

Output signals

Supply pressure

Figure 17: Application with Ultronics proportional valve

From Figure 17 it can be seen the valve has two spools, which make it possible to
control meter-in and meter-out flow of any actuator independently. This facility gives
the opportunity of smooth acceleration and deceleration control of the load by
individual pressure control in each cylinder chamber. The pressure transducers can also
be used for load pressure feedback to increase the hydraulic damping. By measurement
of the pressure drop (p) over a spool the load flow (qL) can be controlled by
calculation of the spool displacement (xv) from the flow equation of the valve, which
gives
qL
xv =
2
Cq w
p

K-E Rydberg

7.

Feedbacks in Electro-Hydraulic Servo Systems

10

Electro-hydraulic servo actuators

Today electro-hydraulic actuators are normally manufactured as integrated units. The


servo valve is connected to the actuator (cylinder or motor) and all the transducers
needed for close loop control are integrated in the valve and actuator. An industrial
actuator for linear position control is depicted in Figure 18. The control card for this
actuator includes connectors for all feedback signals and the controller is implemented
in a micro-processor. The input signals to the control card are electric power supply and
a set point signal and than the card deliver a current signal (i) to the servo valve. The
hydraulic part of the actuator system has two connectors, one hydraulic supply line and
one return flow line.
In many industrial applications there is a need for multiple degrees of freedom control
of the load. One application, which requires advanced control, is motion simulator
platforms. This type of platform is often used for dynamic simulation of air-crafts and
cars. A common way to design a platform, which can be moved in a 3D-space, is to use
6 electro-hydraulic linear actuators as shown in Figure 19.

Figure 18: Industrial electro-hydraulic linear position control actuator, MOOG

K-E Rydberg

11

Feedbacks in Electro-Hydraulic Servo Systems

Figure 19: Electro-hydraulic motion platform with 6 degrees of freedom, Rexroth

For low power applications (low load weights) the platform shown in Figure 19 is often
realised by using electro-mechanical actuators (electric motor and a ball screw).
A similar control strategy as for the 6 DOF platform can be used for crane (or industrial
robot) tip control. Electro-hydraulic control of a lorry crane is shown in Figure 20.
Mechanical System
Z3
Optronic
Sensor

Hydraulic
System

Computer Based
Measurement and
Control System
X3

Figure 20: Crane tip control with optronic sensor for vertical position measurement

K-E Rydberg

12

Feedbacks in Electro-Hydraulic Servo Systems

The strategy for 2 DOF crane tip control is shown in Figure 21. A range camera
(optronic sensor) is used to measure the vertical distance (h) between the camera and
the object. Z3 is the vertical co-ordinate from the base line of the crane to the crane tip.
The reference value for the vertical crane tip position is calculated as Z3ref = Z3+hrefh.
The kinematics of the crane structure is calculated by using the signals from position
transducers in the hydraulic cylinders and a geometric description of the crane structure.
However, this will not give the true tip position of the crane tip because of the weakness
in the mechanical structure. By using a range camera it is possible to compensate the
vertical position control according to the mechanical weakness.

X3ref

href

Xp1r
+

Inverse
Kinematics Xp2r

Z3ref
+

Leadfilter

PI-

xv1r

PI-

xv2r

contr

contr

Leadfilter

Gcyl1

Xp1

Kinematics

Xp2

Gcyl2

X3
Z3

q3

h Range

Camera

Figure 21: Control strategy for crane tip positioning

8.

Design examples

Hydraulically operated boom with lumped masses


The figure shows a valve controlled cylinder used for operation of a mechanical arm.
The total mass of the moving arm is ML. The distance from the gravity centre of the
mass to the joint (0) is L. The lever length for the hydraulic cylinder is e, which will
vary according to xp. The piston area is Ap and its pressurised volume is VL and this
volume varies according to the piston position. The effective bulk modulus is e. The
pressure on the piston rod side is assumed as constant, pR = constant. The mass of the
cylinder housing is M0 and the mechanical spring coefficient for the connection is KL.
L

ML

e
xp

Ap
V L pL
KL

pR = constant

M0

xv

Figure 22: Application with variable mechanical gearing between cylinder and load

K-E Rydberg

13

Feedbacks in Electro-Hydraulic Servo Systems

Equivalent cylinder mass


The equivalent mass loading the piston rod is found from the torque equation for the
joint (0).
..

Inertia : J t = M L L2
..

Torque : T = M L L2 = p L A p e

Xp

L ..
With =
p L Ap = M L X p .
e
e
..

Introducing the mechanical gear U =

L
, the equivalent cylinder mass can be expressed
e

as, M t = M LU 2
Hydraulic resonance frequency and dampning
Assuming ML as the dominant mass the resonance frequency can be calculated as,
M0 << ML gives h =

Ke
1
=
Mt U

2
Ke
where 1 = V L 2 + 1 K e = K L e A p 2
ML
K e e Ap K L
K LV L + e A p

Low mechanical friction gives the hydraulic damping: h =

e M L
K ce
U
2 Ap
VL

According to the gear U it can be observed that an increase of the gear gives a reduced
h but an increased h. Studying the product hh the expression is,
2
e
K
K ce
e M L 1 e Ap
U
h h
h h ce
, for KL >> Kh
VL U VL M L
2 VL0 + Ap x p
2 Ap

Cylinder design according to max pressure level

This example is aimed to demonstrate how the cylinder design will influence the
hydraulic frequency and damping. Figure 23 shows a system with a stiff mechanical
structure and the cylinder is assumed to be loaded by one mass (ML).
L
ML
e

xp

Ap
V0 pL

Figure 23: Cylinder controlled mass with mechanical gear

xL

K-E Rydberg

14

Feedbacks in Electro-Hydraulic Servo Systems

As in Fig. 22 the mechanical gear U = L/e. The piston area is selected as, Ap = U
The cylinder volume depends of the load displacement (XL) as, V0 = Ap
hydraulic resonance frequency the basic equation is, h =

M Lg
.
pL

XL
. For the
U

e Ap2

. If the cylinder is
V0 M LU 2
designed for some maximum load pressure (pLmax), with Ap and V0 as described above,
the hydraulic frequency will follow the expression:

h =

e g
X L p L max

The hydraulic damping is described as, h = K ce

2 Ap

e M LU 2 or = K ce U
h
V0

e p L max

2 Ap

XL g

where the flow/pressure coefficient (Kce) is assumed to be constant.


The product hh is expressed as, hh =

K ce e K ce e p L max
=
.
2 V0
2 Mt g XL

Figure 24 shows how the frequency, damping and the product varies according to the
design parameter max load pressure, pLmax.

Figure 24: Hydraulic resonance frequency and damping versus max load pressure

From the equations it can be noticed that the hydraulic damping will be proportional to
2
p L3 /max
and the product hh p L max . This indicates that the cylinder-load response will
show less oscillations when the max load pressure is increased. The system response for
different pLmax is illustrated in Figure 25.

K-E Rydberg

Feedbacks in Electro-Hydraulic Servo Systems

15

Figure 25: Response of the cylinder-load dynamics with cylinder design for max load pressure of 100,
200 and 300 bar respectively

K-E Rydberg

Controller design

______________________________________________________________________

Controller Design for Hydraulic Servo Systems


General structure of the controller
The most general controller of conventional type is the PID-controller. However, even
with this controller there can still be a need of more dynamic compensations in the
control loop. In a hydraulic system the relative damping is often quite low. A
stabilisation feedback (load pressure or acceleration feedback) can be used to increase
the damping. Depending of the variation of the command signal there will be a delay
between the derivative of the command signal and the output signal. This delay can be
reduced to a minimum by use of a feed forward gain.
The action of the PID-controller means that the derivative gain increases proportionally
to the frequency. In spite of this behaviour it is important to reduce the gain of the Daction at high frequencies. Otherwise, the high frequency disturbances on the signals
will be amplified to a level which can mainly influence the function of the system. A
forward loop filter is used to reduce the derivative gain at high frequencies.
From the above discussion the general structure of the controller will be as shown in
Figure 1.

Figure 1: Structure of a PID controller with feed forward gain and stabilisation feedback.

K-E Rydberg

Controller design

______________________________________________________________________

Feed forward gain for reduction of velocity error in pos. servo


Assume a linear position servo with a valve controlled piston. In this case a plain
proportional controller is suitable to use and easy to adjust for stability. However, if the
command signal is changed there will be a phase lag from input to output signal in the
servo. In the position servo the phase lag cause a position error proportional to time
derivative of the command signal (velocity).
If the feed forward gain introduces a derivative of the command signal it will be
possible to more or less eliminate the phase lag. This feed forward gain helps the servo
control loop (servo valve) to react quickly to a change in the command signal.
Implementation of a feed forward gain in a position servo is shown in the simulinkmodel in Figure 2. The feed forward gain is represented by the transfer function
Gff(s) = s/Kv, where Kv is the steady state gain in the control loop from feed forward
input to system output signal. In this case Kv = 20 sec-1 and 1/Kv = 0.05 sec. The feed
forward gain also includes a low-pass filter with a break frequency of 1000 rad/s
(compare with the forward loop filter in Figure 1).

Figure 2: Simulink-model of a valve controlled cylinder with position feedback and feed forward gain.

The command signal in Figure 2 is a sine wave. The simulation results in Figure 3
shows that the output signal can follow the command signal with a very small phase lag.
The oscillations at start depends on the relatively low hydraulic damping (h = 0.155) in
the system.

Figure 3: Command and output signal with feed forward gain.

The effect of the feed forward gain can preferable be studied by plotting the output
signal (Y) versus command signal (X), as illustrated in Figure 4.

K-E Rydberg

Controller design

______________________________________________________________________

Figure 4: Output versus command signal without (to the left) and with feed forward gain in
a position servo with proportional control.

A notable behaviour of the feed forward gain is that its action is like a pre-filter, which
not affect the control loop gain and the stability margins.

PID Controller
The ProportionalIntegralDerivative controller (PID controller) is a control loop
feedback mechanism widely used in industrial control system. A PID controller
attempts to correct the error between a measured system variable and a desired
command signal by calculating and then outputting a corrective action that can adjust
the process accordingly. A PID controller and its control algorithm are shown in Fig. 5.
P_action

k={1}
Output_Y
Input_U

I_action

I
startTime={0.2}

Sum
+1
+1

Saturation

+1

k={3}

uMax={2}

D_action
DT1

k={0}

Y (t ) = K P U (t ) +

1
dU (t )
U ( )d + TD

TI t0
dt

Figure 5: PID Controller.

Proportional gain
Proportional gain is used for all tuning situations. It introduces a control signal that is
proportional to the error signal. As proportional gain increases, the error decreases and
the feedback signal tracks the command signal more closely. Proportional gain increases

K-E Rydberg

Controller design

______________________________________________________________________
system response by boosting the effect of the error signal. However, too much
proportional gain can cause the system to become unstable.
Command signal

Too low gain

Output signal

Optimal gain

Too high gain

Figure 6: Effects of proportional gain.

Integral gain
With an integral control mode the error signal will be integrated over time, which
improves mean level response during dynamic operation. Integral gain increases system
response during steady state or low-frequency operation and maintain the mean value at
high-frequency operation. The I-gain adjustment determines how much time it takes to
improve the mean level accuracy. Higher integral gain settings increase system
response, but too much gain can cause slow oscillations, as shown in Figure 7.
Command signal

Output signal

Optimal gain

High gain

Too high gain

Figure 7: Effects of integral gain.

The integrator output signal depends upon the I-gain and the input signal level, see
Figure 8. It is very important to set a limit for the output signal, as shown in Figure 8,
to prevent the integrator for windup.

Figure 8: Integrator action with different input signals.

An Anti-windup implementation for a PID controller is shown in Figure 9.

K-E Rydberg

Controller design

______________________________________________________________________

Figure 9: Anti-windup implementation for I-action in a PID controller.

Derivative gain
With a derivative control mode the feedback signal means it anticipates the rate of
change of the feedback and slows the system response at high rates of change.
Derivative gain provides stability and reduces noise at higher proportional gain settings.
The D-gain tends to amplify noise from sensors and to decrease system response when
set is too high. Too much derivative gain can create instability at high frequencies.
Overshoot
Ringing

Low rate

Optimum rate
Figure 10: Effects of derivative gain

Too much rate

You might also like