Chapter 10

PID Control
Based on a survey of over eleven thousand controllers in the refining, chemicals and pulp and paper industries, 97% of regulatory controllers utilize PID
feedback.
Desborough Honeywell, 2000, see [DM02].

PID control is by far the most common way of using feedback in natural
and man-made systems. PID controllers are commonly used in industry
and a large factory may have thousands of them, in instruments and laboratory equipment. In engineering applications the controllers appear in
many different forms: as a stand alone controller, as part of hierarchical,
distributed control systems, or built into embedded components. Most controllers do not use derivative action. In this chapter we discuss the basic
ideas of PID control and the methods for choosing the parameters of the
controllers. Many aspects of control can be understood based on linear
analysis. However, there is one nonlinear effect, that has to be considered in
most control systems namely that actuators saturate. In combinations with
controllers having integral actions saturations give rise to an effect called
integral windup. This phenomenon that occurs in practically all control
systems will be discussed in depth for PID controllers. Methods to avoid
windup will also be presented. Finally we will also discuss implementation
of PID controllers, similar methods can be used to implement many other
controllers.
301

302

CHAPTER 10. PID CONTROL

Error
Present
Past

00000
11111
00000
11111
000000000
111
00000
11111111111
t

Future

t + Td

Time

Figure 10.1: A PID controller takes control action based on past, present and
prediction of future control errors.

10.1

The Controller

The ideal version of the PID controller is given by the formula

Z t
de
e( ) d + kd ,
u(t) = kp e(t) + ki
dt
0

(10.1)

where u is the control signal and e is the control error (e = r y). The
reference value, r, is also called the setpoint. The control signal is thus a
sum of three terms: a proportional term that is proportional to the error, an
integral term that is proportional to the integral of the error, and a derivative
term that is proportional to the derivative of the error. The controller
parameters are proportional gain kp , integral gain ki and derivative gain kd .
The controller can also be parameterized as


Zt
1
de(t)
u(t) = kp e(t) +
e( )d + Td
,
Ti
dt

(10.2)

where Ti is the integral time constant and Td the derivative time constant.
The proportional part acts on the present value of the error, the integral
represents an average of past errors and the derivative can be interpreted as
a prediction of future errors based on linear extrapolation, as illustrated in
Figure 10.1. Note that the control signal u is formed entirely from the error
e, there is no feedforward term (which would correspond to kr r in the state
feedback case). In Section 10.5 we will introduce a modification which also
uses feedforward.
We begin our analysis by considering pure proportional feedback. Figure 10.2a shows the responses of the output to a unit step in the command

303

10.1. THE CONTROLLER

1.5
1.5

y
0.5

0.5

10

20

10

20

10

20

10

20

0.5

1.5
1

1
0
2

0
0

10

20

10

20

(a)

(b)

t
(c)

Figure 10.2: Responses to step changes in the command signal for a proportional
controller (left), PI controller (center) and PID controller (right). The process
has the transfer function P (s) = 1/(s + 1)3 , the proportional controller (left) had
parameters kp = 1 (dashed), 2 and 5 (dash-dotted), the PI controller has parameters
kp = 1, ki = 0 (dashed), 0.2, 0.5 and 1 (dash-dotted), and the PID controller has
parameters are kp = 2.5, ki = 1.5 and kd = 0 (dashed), 1, 2, 3 and 4 (dash-dotted).

signal for a system with pure proportional control at different gain settings.
In the absence of a feedforward term, the output never reaches the reference
and hence we are left with non-zero steady state error. Letting the process and the controller have transfer functions P (s) and C(s), the transfer
function from reference to output is
Gyr =

PC
.
1 + PC

(10.3)

The zero frequency gain with proportional control C(s) = kp is

Gyr (0) =

P (0)kp
1 + P (0)kp

and thus the steady state error for a unit step is 1Gyr (0) = 1/(1+kp P (0)).
For the system in Figure 10.2a with gains kp = 1, 2 and 5, the steady state
error is 0.5, 0.33 and 0.17. The error decreases with increasing gain, but the
system also becomes more oscillatory. Notice in the figure that the initial
value of the control signal equals the controller gain.

304
e

CHAPTER 10. PID CONTROL

u
1
1 + sTi

1
1+sT

(a)

(b)

Figure 10.3: Implementation of integral action (left) and derivative action (right)
by combining simple blocks.

To avoid having a steady state error, the proportional controller can be

changed to
u(t) = kp e(t) + ud ,
(10.4)
where ud is a feedforward term that is adjusted to give the desired steady
state value. If we choose ud = r/P (0) = kr r, then the output will be exactly
equal to the reference value, as it was in the state space case. However,
this requires exact knowledge of the process dynamics, which is usually not
available. In early controllers the term ud , which was also called the reset,
was adjusted manually to obtain the desired steady state.
As we saw in Sections 1.5 and 6.4, integral action guarantees that the
process output agrees with the reference in steady state and provides an
alternative to including a feedforward term. To see the effect of integral
action in the frequency domain, we consider a proportional-integral (PI)
controller, which has a transfer function
C(s) = kp e +

ki
.
s

We see that the controller has infinite zero frequency gain (C(0) = ) and
it then follows from equation (10.3) that Gyr (0) = 1, which implies that
there is no steady-state error.
Integral action can also be viewed as a method for generating the feedforward term ud in the proportional controller (10.4) automatically. An
alternative way to represent this action is shown in Figure 10.3a, where the
low pass part of the control action of a proportional controller is filtered and
feed back with positive gain. This implementation, called automatic reset,
was one of the early inventions of integral control. Integral action is often
realized in this way in biological systems.

305

10.1. THE CONTROLLER

The transfer function of the system in Figure 10.3 is obtained by loop

tracing: assuming exponential signals, we have
u = kp e +

1
u,
1 + sT

and solving for u gives

u = kp

kp
1 + sT
e = kp +
,
sT
sT

which is the transfer function for a PI controller.

The properties of integral action are illustrated in Figure 10.2b. The
proportional gain is constant, kp = 1, and the integral gains are ki = 0, 0.2,
0.5 and 1. The case ki = 0 corresponds to pure proportional control, with a
steady state error of 50%. The steady state error is removed when integral
gain action is used. The response creeps slowly towards the reference for
small values of ki , but faster for larger integral gains, but the system also
becomes more oscillatory.
We now return to the general PID controller and consider the use of
the derivative term, kd . Recall that the original motivation for derivative
feedback was to provide predictive action. The input-output relation of a
controller with proportional and derivative action is
u = kp e + kd

de

de
,
= k e + Td
dt
dt

where Td = kd /kp is the derivative time constant. The action of a controller

with proportional and derivative action can be interpreted as if the control
is made proportional to the predicted process output, where the prediction
is made by extrapolating the error Td time units into the future using the
tangent to the error curve (see Figure 10.1).
Derivative action can also be implemented by taking the difference between the signal and its low-pass filtered version as shown in Figure 10.3a.
The transfer function for the system is

1 
sT
C(s) = 1
=
U (s).
1 + sT
1 + sT
The system thus has the transfer function G(s) = sT /(1 + sT ), which approximates a derivative for low frequencies. Notice that this implementation
gives filtering automatically.
Figure 10.2c illustrates the behavior of a system with a PID controller:
the system is oscillatory when no derivative action is used and it becomes

306

CHAPTER 10. PID CONTROL

more damped as derivative gain is increased. A comparison of the systems

with P, PI and PID control in Figure 10.2 shows that the steady-state error
is removed by introducing integral action and that the response speed can
be improved by introducing derivative action.

10.2

Tuning

Users of control systems are frequently faced with the task of adjusting
the controller parameters to obtain a desired behavior. There are many
different ways to do this. One way to do this is to go through the steps of
modeling and control design. Since the PID controller has so few parameters
a number of special empirical methods have also been developed. A simple
idea is to connect a controller, increase the gain until the the system starts
to oscillate, and then reduce the gains by an appropriate factor. Another
is to measure some features of the open loop response and to determine
controller parameters based on these features. We will present the ZieglerNichols methods which are the most celebrated tuning rules.

Ziegler-Nichols Tuning
Ziegler and Nichols develped two techniques for controller tuning in the
1940s. The idea was to tune the controller based on the following idea: Make
a simple experiment, extract some features of process dynamics from the
experimental data, and determine controller parameters from the features.
One method is based on direct adjustment of the controller parameters.
A controller is connected to the process, integral and derivative gain are
set to zero and the proportional gain is increased until the system starts
to oscillate. The critical value of the proportional gain kc is observed together with the period of oscillation Tc . The controller parameters are then
given by Table 10.1. The values in the table were obtained based on many
simulations and experiments on processes that are normally encountered in
process industry. There are many variations of the method which are widely
used in industry.
Another method proposed by Ziegler and Nichols is based on determination of the open loop step response of the process, as shown Figure 10.4a.
The step response is measured by applying a step input to the process and
recording the response. The response is scaled to correspond to a unit step
input and characterized by parameters a and Tdel , which are the intercepts
of the steepest tangent of the step response with the coordinate axes. The
parameter Tdel is an approximation of the time delay of the system and

307

10.2. TUNING

Table 10.1: Controller parameters for the Ziegler-Nichols frequency response

method which gives controller parameters in terms of critical gain kc and critical period Tc . Parameter Tp is an estimate of the period of damped oscillations of
the closed loop system.

Controller
P
PI
PID

kp /kc
0.5
0.4
0.6

Ti /Tc

Td /Tc

0.8
0.5

0.125

Tp /Tc
1.0
1.4
0.85

a/Tdel is the steepest slope of the step response. Notice that it is not necessary to wait until steady state to find the parameters, it suffices to wait
until the response has had an inflection point. The controller parameters are
given in Table 10.2. The parameters were obtained by extensive simulation
of a range of representative processes.

Improved Ziegler-Nichols Rules

There are two drawbacks with the Ziegler-Nichols rules: too little process
information is used and the closed loop systems that are obtained lack robustness. Substantially better tuning is obtained by fitting the model
P (s) =

K
esTdel
1 + sT

(10.5)

to the step response. A simple way to do this is illustrated in Figure 10.4b.

The zero frequency gain of the process K is determined from the steady
state value of the step response. The time delay Tdel is determined from the
y

K
0.63K

Tdel

t
(a)

Tdel

T63

(b)

Figure 10.4: Characterization of the unit step response by two (left) and three
parameters (right). The point where the tangent is steepest is marked with a small
circle.

308

CHAPTER 10. PID CONTROL

Table 10.2: Controller parameters for the Ziegler-Nichols step response method.
Parameter Tp is an estimate of the period of damped oscillations of the closed loop
system.

Controller
P
PI
PID

akp
1
0.9
1.2

Ti /Tdel

Td /Tdel

3
2

Tdel /2

Tp /Tdel
4
5.7
3.4

intercept of the steepest tangent to the step response and the time T63 is
the time where the output has reached 63% of its steady state value. The
parameter T is then given by T = T63 Tdel . Notice that the experiment
takes longer time than the experiment in Figure 10.4a because it is necessary to wait until the steady state has been reached. The following tuning
formulas have been obtained by tuning controllers to a large set of processes
typically encountered in process control


kp K = min 0.4 T /L, 0.25
(10.6)
Ti = max (T, 0.5Tdel ).
Notice that the improved formulas typically give lower controller gain than
the Ziegler-Nichols method, and that integral gain is higher, particularly for
systems with dynamics that are delay dominated, i.e. Tdel > 2T .

Relay Feedback
The experiment used in the Ziegler-Nichols frequency response method,
where the gain of a proportional controller is increased until the system
reaches instability, gives the frequency 180 where the process has a phase
lag of 180 and the process gain K180 at that frequency. Another way to
obtain this information is to connect the process in a feedback loop with
a relay as shown in Figure 10.5a. For many systems there will then be an
oscillation, as shown in Figure 10.5b, where the relay output u is a square
wave and the process output y is close to a sinusoid. Notice that the process
input and output have opposite phase and that an oscillation with constant
period is established quickly.
To provide some analysis, we assume that the relay output is expanded in
a Fourier series and that the process attenuates higher harmonics effectively.
It is then sufficient to consider only the first harmonic component of the

309

10.3. MODELING AND CONTROL DESIGN

ysp

y
G(s)
y

0.5
0
0.5
1

10

20

30

(a)

(b)

Figure 10.5: Block diagram of a process with relay feedback (left) and typical signals
(right). The process output y is solid and the relay output u is dashed. Notice that
the signals u and y are out of phase.

input. The input and the output then have opposite phase, which means
that the frequency of the oscillation 180 is such that the process has a phase
lag of 180 . If d is the relay amplitude, the first harmonic of the square wave
input has amplitude 4d/. Let a be the amplitude of the process output.
The process gain at 180 is then given by
K180 = |P (i180 | =

a
.
4d

(10.7)

The relay experiment is easily automated. Since the amplitude of the oscillation is proportional to the relay output, it is easy to control it by adjusting
the relay output. The amplitude and the period can be determined after
about 20 s, in spite of the fact that the system is started so far from the
equilibrium that it takes about 8 s to reach the correct level. The settling
time for the step response of the system is 40 s.
Automatic tuning based on relay feedback is used in many commercial
PID controllers. Tuning is accomplished simply by pushing a button which
activates relay feedback. The relay amplitude is automatically adjusted to
keep the oscillations sufficiently small, the relay feedback is switched to a
PID controller as soon as the tuning is accomplished.

10.3

Modeling and Control Design

Parameters of PID controllers can also be determined by modeling process

dynamics and applying some method for control design. Since the complexity of the controller is directly related to the complexity of the model it is
necessary to have models of low order.
To illustrate the ideas we will consider the case where a process dynamics

310

CHAPTER 10. PID CONTROL

is approximated by a first order transfer function

P (s) =

b
.
s+a

The approximation is reasonable for systems where storage of mass, momentum and energy can be captured by one state variable. Typical examples
are the velocity of a car on the road, control of the velocity of a rotating
system, electric systems where energy is essentially stored in one component, incompressible fluid flow in a pipe, level control of a tank, pressure
control in a gas tank and temperature in a body with uniform temperature
distribution.
A PI controller has the transfer function
C(s) = kp +

ki
,
s

and the transfer function of the closed loop system from reference to output
is
b(kp s + ki )
PC
= 2
.
Gyr =
1 + PC
s + (a + bkp )s + bki
The closed loop system has the characteristic polynomial
s2 + (a + bkp )s + bki .
Assuming that the desired characteristic polynomial is
s2 + 20 s + 02 ,

(10.8)

we find that the controller parameters are given by

20 T 1
20 a
=
b
K
02
02 T
ki =
=
.
b
K

kp =

(10.9)

The parameter 0 determines the response speed and determines the

damping. Notice that controller gain becomes negative if 0 < 0.5, which
gives a closed loop system with bad properties, as will be discussed in Section 12.5.
The same approach can be used to find the parameters of a PID controller
for a process with dynamics of second order (Exercise 1).

10.3. MODELING AND CONTROL DESIGN

311

Example 10.1 (Cruise control design). Consider the problem of maintaining the speed of a car as it goes up a hill. In Example 5.14we found that
there was very little difference between the linear and nonlinear models when
investigating PI control provided that the throttle does not reach the saturation limits. We will now use a linear model to design a controller and to
investigate the effects of design parameters. A simple linear model of a car
was given in Example 5.10:
d(v ve )
= a(v ve ) + b(u ue ) g,
dt

(10.10)

where v is the velocity of the car, u is the input from the engine and
is the slope of the hill. The parameters were a = 0.101, b = 1.3203,
g = 9.8, ve = 20, and ue = 0.1616. This model will be used to find suitable
parameters of a vehicle speed controller. To investigate the behavior of the
closed loop system we start with the linearized process model (10.10) and
we assume that the cruise controller is PI controller is described by
Z t
(ve v( ))d.
(10.11)
u = kp (ve v) + ki
0

To compute the error dynamics, introduce the error e = ve v, differentiate

equations (10.10) and (10.11), and eliminate u. The error is then given by
the differential equation
de
d
d2 e
+ (a + bkp ) + bki e = 9.8 .
2
dt
dt
dt

(10.12)

We notice that the steady state error is zero if and e are constant, which
is no surprise since the controller has integral action.
To understand the effects of the controller parameters K and ki we can
compare equation (10.12) with the standard second order system
q + 20 q + 02 q = ku.
This gives

2
a + 20
ki = 0
b
b
where denotes relative damping and 0 is the undamped natural frequency.
The parameter 0 gives response speed and determines the shape of the
response. Since it is desirable that a cruise control system should respond to
changes smoothly without oscillations we choose = 1, which corresponds
to critical damping.
kp =

312

CHAPTER 10. PID CONTROL

v [m/s]

0.5

0.5
1

throttle

1.5

10

20

30

0
0.5
1

40

0.8

0.8

0.6

0.6
throttle

v [m/s]

0.4
0.2
0

10

10

20

30

40

20

30

40

0.4
0.2

10

20

t [s]

30

40

t [s]

Figure 10.6: Illustrates the effect of parameters 0 (left) and 0 (right) on the
response of a car with cruise control.

The consequences of different choices of 0 and 0 are illustrated in

Figure 10.6. The figure shows the velocity and throttle for a car that first
moves on a horizontal road and encounters a hill with slope 4 at time 6.
Between time 5 and 6 the slope increases linearly. The choice of 0 is a
compromise between response speed and control actions. The compromise
is illustrated in Figure 10.6, which shows the velocity error and the control
signal for a simulation where the slope of the road suddenly changes by 4 .
The largest velocity error decreases with increasing 0 , but the control signal
also changes more rapidly. In the simple model (10.10) it was assumed that
the force responds instantaneously to throttle commands. For rapid changes
there may be additional dynamics that has to be accounted for. There are
also physical limitations to the rate of change of the force, which also restrict
the admissible value of 0 . A reasonable choice of 0 is in the range of 0.2
to 1.0. Notice in Figure 10.6 that even with 0 = 0.1 the largest velocity
error is only 1 m/s.
Another interpretation of the effect of the integral action can be given
by returning to the basic force balance model of the car
m

dv
= F Fd ,
dt

where m is the mass of the car, F is the applied force (from the engine)
and Fd is the disturbance force (aerodynamic drag and force of gravity).
Since zero steady state error implies that v is constant, we see that the PI
controller generates an output force F that in steady state is equal to the

10.4. INTEGRATOR WINDUP

313

drag force Fd . Since the error is zero in steady state the controller output
equals the output of the integrator of the PI controller. The output of the
integrator in the PI controller can thus be interpreted as an estimator of the
drag force.

10.4

Integrator Windup

Many aspects of a control system can be understood from linear models.

There are, however, some nonlinear phenomena that must be taken into account. These are typically limitations in the actuators: a motor has limited
speed, a valve cannot be more than fully opened or fully closed, etc. For
a system which operates over a wide range of conditions, it may happen
that the control variable reaches the actuator limits. When this happens
the feedback loop is broken and the system runs in open loop because the
actuator will remain at its limit independently of the process output as long
as the actuator remains saturated. The integral term will also build up since
the error typicallys zero. The integral term and the controller output may
then become very large. The control signal will then remain saturated even
when the error changes and it may take a long time before the integrator and
the controller output comes inside the saturation range. The consequence is
that there are large transients. This situation is colloquially referred to as
integrator windup which is illustrated by the following example.
Example 10.2 (Cruise control). The windup effect is illustrated in Figure 10.7, which shows what happens when a car encounters a hill that is
so steep (6 ) that the throttle saturates when the cruise controller attempts
to maintain speed. When encountering the slope at time t = 5 the velocity
decreases and the throttle increases to generate more torque. The torque
required is however so large that the throttle saturates. The error decreases
slowly because the torque generated by the engine is just a little larger than
the torque required to compensate for the gravity. The error is large and
the integral continues to build up until the error reaches zero at time 30, but
but the controller output is still much larger than the saturation limit and
the actuator remains saturated. The integral term starts to decrease and at
time 45 and the velocity settles to quickly to the desired value. Notice that
it takes considerable time before the controller output comes in the range
where it does not saturate resulting in a large overshoot.

314

Velocity [m/s]

Velocity [m/s]

CHAPTER 10. PID CONTROL

21
20
19
18

20

40

60

1.5
1
0.5
0

20
19
18

20

20

40

60

40

60

2.5

Throttle

Throttle

2.5

21

20

40

60

2
1.5
1
0.5
0

Time [s]

Time [s]

(a)

(b)

Figure 10.7: Simulation of windup (left) and anti-windup (right). The figure shows
the speed v and the throttle u for a car that encounters a slope that is so steep that
the throttle saturates. The controller output is dashed. The controller parameters
are kp = 0.5 and ki = 0.1.

Avoiding Windup
There are many ways to avoid windup. One method is illustrated in Figure 10.8: the system has an extra feedback path that is generated by measuring the actual actuator output, or the output of a mathematical model
of the saturating actuator, and forming an error signal (es ) as the difference
between the output of the controller (v) and the actuator output (u). The
signal es is fed to the input of the integrator through gain kt . The signal
es is zero when there is no saturation and the extra feedback loop has no
effect on the system. When the actuator saturates, the signal es is feedback
to the integrator in such a way that es goes towards zero. This implies that
controller output is kept close to the saturation limit. The controller output
will then change as soon as the error changes sign and integral windup is
avoided.
The rate at which the controller output is reset is governed by the feedback gain, kt , a large value of kt give a short reset time. The parameter kt
can, however, not be too large because measurement error can then cause
an undesirable reset. Areasonable compromise is to choose kt 1/Ti for PI
control and as kt 1/ Ti Td for PID control. We illustrate how integral
windup can be avoided by investigating the cruise control system.
Example 10.3 (Cruise control). Figure 10.7b shows what happens when

315

10.5. IMPLEMENTATION
y

kd
Actuator
model

e = ysp y

ki

1
s

Actuator

es
kt

Figure 10.8: PID controller with anti-windup.

a controller with anti-windup is applied to the system simulated in Figure 10.7a. Because of the feedback from the actuator model the output of
the integrator is quickly reset to a value such that the controller output is
at the saturation limit. The behavior is drastically different from that in
Figure 10.7a and the large overshoot is avoided. The tracking gain kt = 2
in the simulation.

10.5

Implementation

There are many practical issues that have to be considered when implementing PID controllers. They have been developed over time based on practical
experiences. In this section we consider some of the most common. Similar
considerations also apply to other types of controllers.

Filtering the Derivative

A drawback with derivative action is that an ideal derivative has very high
gain for high frequency signals. This means that high frequency measurement noise will generate large variations of the control signal. The effect of
measurement noise be reduced by replacing the term kd s by
Da =

kd s
.
1 + sTf

(10.13)

This can be interpreted as an ideal derivative that is filtered using a firstorder system with the time constant Tf . For small s the transfer function is

316

CHAPTER 10. PID CONTROL

approximately kd s and for large s it is equal to kd /Tf . The approximation

acts as a derivative for low-frequency signals and as a constant gain for
the high frequency signals. The filtering time is chosen as Tf = (kd /k)/N ,
with N in the range of 2 to 20. Filtering is obtained automatically if the
derivtive is implemented by taking difference between the signal and a its
filtered version as shown in Figure 10.2.
The transfer function of a PID controller with a filtered derivative is


sTd
1
+
.
(10.14)
C(s) = kp 1 +
sTi 1 + sTd /N
The high-frequency gain of the controller is K(1 + N ). Instead of filtering
just the derivative it is also possible to use an ideal controller and filter the
measured signal. The transfer function of such a controller with the filter is
then


1
1
C(s) = kp 1 +
+ sTd
.
(10.15)
sTi
(1 + sTf )2
where a second order filter is used.

Setpoint Weighting
The control system in equation (10.1) is called a system with error feedback
because the controller acts on the error, which is the difference between the
reference and the output. In the simulation of PID controllers in Figure 10.1
there is a large initial peak of the control signal, which is caused by the
derivative of the reference signal. The peak can be avoided by modifying
the controller equation (10.1) to
Z
 dr dy 




r( ) y( ) d + kd
u = kp r y + ki
(10.16)
dt
dt
0
In this controller, proportional and derivative actions act on fractions and
of the reference. Integral action has to act on the error to make sure that
the error goes to zero in steady state. The closed loop systems obtained for
different values of and respond to load disturbances and measurement
noise in the same way. The response to reference signals is different because
it depends on the values of and , which are called reference weights or
setpoint weights.
Figure 10.9 illustrates the effects of setpoint weighting on the step response. The figure shows clearly the effect of changing . The overshoot for
reference changes is smallest for = 0, which is the case where the reference is only introduced in the integral term, and increases with increasing

317

10.5. IMPLEMENTATION
1.5
0

|Gyr (j)|

10

1
0.5
0

10

10

10

10

1.5

|Gur (j)|

10

1
0.5
0

0.5

10

10

10

10

Figure 10.9: Time and frequency responses for system with PI controller and setpoint weighting. The curves on the left show responses in process output y and
control signal and the curves on the right show the gain curves for the transfer
functions Gyr (s) and Gur (s). The process transfer function is P (s) = 1/s, the controller gains are k = 1.5 and ki = 1, and the setpoint weights are = 0 (dashed)
0.2, 0.5 and 1 (dash dotted).

. Parameter it typically in the range of 0 to 1 and is normally zero to

avoid large transients in the control signal when the reference is changed.
The controller given by equation (10.16) is a special case of controller
with two degrees which will be discussed in Section 11.1.

Implementation based Operational Amplifiers

PID controllers have been implemented in many different technologies. Figure 10.10 shows how they can be implemented by feedback around operational amplifiers.
To show that the circuit in Figure 10.10b is a PID controller we will
use the approximate relation between the input voltage e and the output
voltage u of an operational amplifier in Section 3.3:
u=

Z1
e,
Z0

where Z0 is the impedance between the negative input of the amplifier and
the input voltage e, and Z1 is the impedance between the zero input of the

318

CHAPTER 10. PID CONTROL

C0
R0

R1

R0

C1

R1

C1

e
u

(a)

(b)

Figure 10.10: Schematic diagram of an electronic PI (left) and PID controllers

(right) based on feedback around an operational amplifier.

amplifier and the output voltage u. The impedances are given by

Z0 =

R0
1 + R0 C0 p

Z1 = R1 +

1
,
C1 p

and we find the following relation between the input voltage e and the output
voltage u:
R1 (1 + R0 C0 p)(1 + R1 C1 p)
Z1
e.
u= e=
Z0
R0
R1 C1 p
This is the input-output relation for a PID controller on the form (10.2)
with parameters
kp =

R1
R0

Ti = R1 C1

Td = R0 C0 .

The corresponding results for a PI controller is obtained by setting C0 = 0.

Computer Implementation
In this section we briefly describe how a PID controller may be implemented
using a computer. The computer typically operates periodically, with signals from the sensors sampled and converted to digital form by the A/D
converter, the control signal computed and then converted to analog form
for the actuators. The sequence of operation is as follows:
1. Wait for clock interrupt
2. Read input from sensor
3. Compute control signal
4. Send output to the actuator
5. Update controller variables

319

10.5. IMPLEMENTATION
6. Repeat

Notice that an output is sent to the actuators as soon as it is available. The

time delay is minimized by making the calculations in Step 3 as short as
possible and performing all updates after the output is commanded.
As an illustration we consider the PID controller in Figure 10.8, which
has a filtered derivative, setpoint weighting and protection against integral
windup. The controller is a continuous time dynamical system. To implement it using a computer, the continuous time system has to be approximated by a discrete time system.
The signal v is the sum of the proportional, integral and derivative terms
v(t) = P (t) + I(t) + D(t)

(10.17)

and the controller output is u(t) = sat(v(t)) where sat is the saturation
function that models the actuator. The proportional term is
P = kp (ysp y).
This term is implemented simply by replacing the continuous variables with
their sampled versions. Hence
P (tk ) = kp (yr (tk ) y(tk )) ,

(10.18)

where {tk } denotes the sampling instants, i.e., the times when the computer
reads its input. The integral term is
I(t) = ki

Zt
0

e(s)ds +

1
sat(v) v)
Tt

and approximating the integral by a sum gives

I(tk+1 ) = I(tk ) + ki h e(tk ) +

h
sat(v) v) .
Tt

The derivative term D is given by the differential equation

Tf

dD
+ D = kd y.
dt

Approximating this equation with a backward difference we find

Tf

y(tk ) y(tk1 )
D(tk ) D(tk1 )
+ D(tk ) = kd
,
h
h

(10.19)

320

CHAPTER 10. PID CONTROL

which can be rewritten as

D(tk ) =

Tf
kd
D(tk1 )
(y(tk ) y(tk1 )) .
Tf + h
Tf + h

(10.20)

The advantage of using a backward difference is that the parameter Tf /(Tf +

h) is nonnegative and less than one for all h > 0, which guarantees that the
difference equation is stable.
Reorganizing equations (10.17)(10.20), the PID controller can be described by the following pseudo code:
% Precompute controller coefficients
bi=ki*h
ad=Tf/(Tf+h)
bd=kd/(Tf+h)
br=h/Tt
% Control algorithm - main loop
while (running) {
r=adin(ch1)
% read setpoint from ch1
y=adin(ch2)
% read process variable from ch2
P=kp*(b*r-y)
% compute proportional part
D=ad*D-bd*(y-yold)
% update derivative part
v=P+I+D
% compute temporary output
u=sat(v,ulow,uhigh)
% simulate actuator saturation
daout(ch1)
% set analog output ch1
I=I+bi*(r-y)+br*(u-v)
% update integral
yold=y
% update old process output
sleep(h)
% wait until next update interval
}

Precomputation of the coefficients bi, ad, bd and br saves computer time

in the main loop. These calculations have to be done only when controller
parameters are changed. The main loop is executed once every sampling
period. The program has three states: yold, I, and D. One state variable
can be eliminated at the cost of a less readable code. Notice that the code
includes computing the derivative of the process output, proportional action
on a portion of the error (b 6= 1), and modeling of the actuator saturation
in the integral computation to give protection against windup.

10.6

Further Reading

The history of PID control is a very rich one and stretches back to the
beginning of the foundation of control theory. A very readable treatment is

321

10.7. EXERCISES

given by Mindel [Min02]. A comprehensive presentation of PID control is

given in [
AH95].

10.7

Exercises

1. Consider a second order process with transfer function

P (s) =

b
.
s2 + a1 s + a2

Find the gains for a PID controller that gives the closed loop system
a characteristic polynomial of the form
s2 + 20 s + 02 .
2. (Vehicle steering) Design a proportion-integral controller for the vehicle steering system that gives closed loop characteristic equation
s3 + 20 s2 + 20 s + 03 .

322

CHAPTER 10. PID CONTROL