Tuning Controllers &

Types of Industrial Controllers

Dr. Kevin Craig

Professor of Mechanical Engineering
Rensselaer Polytechnic Institute
Tuning Controllers & Industrial Controllers

K. Craig

Tuning a Control System

Tuning is the process of setting controller gains to optimize
system performance.
It is a difficult task, as control systems have inherent
limitations in response and stability.
High gains increase responsiveness but also move the system
closer to instability.
All feedback systems can become unstable if improperly
designed. In all real-world components there is some kind of
lagging behavior between the input and output, characterized
by the and n in a transfer function. Instantaneous response
is impossible in the real world!
Tuning Controllers & Industrial Controllers

K. Craig

 Instability in a feedback control system results from an

improper balance between the strength of the corrective
action and the system dynamic lags. When signals traverse a
control loop, inherent phase lags in the control system can
add enough delay to cause instability. Reducing delays
within the control loop is a sure way to make room for higher
gains; therefore, fast sampling and high-speed sensors are
highly desirable.
Instability that results from the accumulation of phase lag
around the control loop usually occurs at just one frequency;
this is why unstable control systems oscillate. All systems
have noise and noise contains virtually all frequencies. If a
system is unstable at any frequency, nature will find that
frequency, usually in a few milliseconds.
Tuning Controllers & Industrial Controllers

K. Craig

 Each component in a control loop generates a certain

amount of phase lag that varies with frequency: the
controller, the amplifier, the plant, the feedback element.
For most control systems, there will exist at least one
frequency where the phase lag accumulates to 180, but this
alone does not cause instability. To cause instability, the
loop gain must also be equal to unity. Similar to phase lag,
each block in the loop contributes gain; the total gain
through the loop is the accumulation of the block gains,
i.e., the sum of the gains when measured in dB.
Self-sustained oscillations occur when two conditions are
met: the loop gain is unity (0 dB) and the phase lag around
the loop is 180.

Tuning Controllers & Industrial Controllers

K. Craig

Closed-Loop System Block Diagram

R(s)

+-

E(s)

+
Gc(s) +

B(s)

Open-Loop Transfer Function

B(s)
= G c (s)G(s)H(s)
E(s)
Feed-Forward Transfer Function

C(s)
= G c (s)G(s)
E(s)

D(s)
G(s)

C(s)

H(s)
Closed-Loop Transfer Functions

G c (s)G(s)
C(s)
=
R(s) 1 + G c (s)G(s)H(s)
C(s)
G(s)
=
D(s) 1 + G c (s)G(s)H(s)

G(s)
C(s) =
[G c (s)R(s) + D(s)]
1 + G c (s)G(s)H(s)

Tuning Controllers & Industrial Controllers

K. Craig

Example Problem
Plant

1
1
500
G (s) = =
=
Js 0.002s
s

K t = 1 (Nm/A)

2
Amplifier G amp ( s ) = 2
s + 2s + 2

= 5027 (rad/s) = 800 (Hz)

= 0.707
Tuning Controllers & Industrial Controllers

K. Craig

Sensor

2
H (s) = 2
s + 2s + 2

= 1885 (rad/s) = 300 (Hz)

= 0.707

PI Controller

s + 100
G c ( s ) = 0.58
s

K
KP s + I
KP
K I K Ps + K I

KP +
=
=
s
s
s
Tuning Controllers & Industrial Controllers

K. Craig

gain crossover

Open-Loop
Bode Plots
phase crossover

GM = 14.3 dB @ 1340 rad/s

PM = 53.7 @ 305 rad/s
Tuning Controllers & Industrial Controllers

K. Craig

Closed-Loop
Bode Plots

Tuning Controllers & Industrial Controllers

K. Craig

Closed-Loop Step Response

Overshoot: 23.1%

Rise Time: 0.0031 sec

2% Settling Time: 0.0266 sec

Tuning Controllers & Industrial Controllers

K. Craig

10

 Two open-loop performance criteria in common use to specify

relative stability are gain margin and phase margin.
The open-loop frequency response is defined as (B/E)(i).
One could open the loop by removing the summing
junction at R, B, E and just input a sine wave at E and
measure the response at B. This is valid since (B/E)(i) =
GcGH(i). Open-loop experimental testing has the
advantage that open-loop systems are rarely absolutely
unstable, thus there is little danger of starting up an untried
apparatus and having destructive oscillations occur before
it can be safely shut down.
The utility of open-loop frequency-response rests on the
Nyquist stability criterion.

Tuning Controllers & Industrial Controllers

K. Craig

11

 Gain margin (GM) and phase margin (PM) are in the nature of
safety factors such that (B/E)(i) stays far enough away from
1 -180 on the stable side.
Gain margin is the multiplying factor by which the steady
state gain of (B/E)(i) could be increased (nothing else in
(B/E)(i) being changed) so as to put the system on the
edge of instability, i.e., (B/E)(i)) passes exactly through
the -1 point. This is called marginal stability.
Phase margin is the number of degrees of additional phase
lag (nothing else being changed) required to create
marginal stability.
Both a good gain margin and a good phase margin are needed;
neither is sufficient by itself.
Tuning Controllers & Industrial Controllers

K. Craig

12

 It is important to realize that, because of model uncertainties, it

is not merely sufficient for a system to be stable, but rather it
must have adequate stability margins.
Stable systems with low stability margins work only on
paper; when implemented in real time, they are frequently
unstable.
The way uncertainty has been quantified in classical control is
to assume that either gain changes or phase changes occur.
Typically, systems are destabilized when either gain
exceeds certain limits or if there is too much phase lag (i.e.,
negative phase associated with unmodeled poles or time
delays).
As we have seen these tolerances of gain or phase uncertainty
are the gain margin and phase margin.
Tuning Controllers & Industrial Controllers

K. Craig

13

 Although the measurements of PM and GM are objective,

determining the desired values for these measures requires
judgment.
Applications vary in the amount of margin they require. Systems
responding to demanding commands, e.g., step command, require
higher stability margins than those that must respond only to gentler
commands. Some applications can tolerate more overshoot than others.
Also, some control methods require more PM or GM than others for
equivalent response.

Experience teaches that GM should be between 10 and 25 dB,

depending on the application and controller type; PM should
be between 35 and 80. All things being equal, more PM is
better.
Basic Rule: Eliminate unnecessary phase lag
Unnecessary phase lags limit the ultimate performance of the control
system.
Tuning Controllers & Industrial Controllers

K. Craig

14

 The challenge in tuning is that multiple gains must be varied

and each affects many of the performance measures.
It would be desirable if we could decouple the multiple tuning gains so
that they may be adjusted individually. This can be done by
considering the effects of each gain as being dominant over a certain
frequency range.

Lets consider the PI controller first applied to a simplified

system.
1
G (s) =
Js

GC (s ) = K P +

KI
s

H (s) = 1

Closed-Loop Transfer Function

G c (s)G(s)
K Ps + K I
C(s)
=
= 2
R(s) 1 + G c (s)G(s)H(s) Js + K P s + K I
Tuning Controllers & Industrial Controllers

K. Craig

15

 While the numerator does have an impact on response, the

denominator determines the overall stability and response of
the system.
In the high-frequency range, s is large and the Js2 term dominates the
denominator.
In the middle-frequency range, the KPs term will dominate the
denominator. Here the proportional gain dominates; the frequency is
still too high for the integral gain to have much impact.
In the low-frequency range, as the frequency approaches zero, the KI
term will dominate.

To apply a frequency-range-based approach to tuning, tune the

highest frequency terms first.
Assuming the plant gain cannot be changed, tune KP first. When
optimizing KP, you are balancing responsiveness (larger KP) against
stability (smaller KP).

Tuning Controllers & Industrial Controllers

K. Craig

16

 Integral gain is important as it removes steady-state error. When tuning

KI, its value will have little bearing on the stability of the highfrequency range; changing KI will not require returning to change KP.

What about changes in plant gain? The plant gain here is 1/J,
which could change as J changes.
The plant gain is usually the most difficult to know and control.
The effect of changes in the plant gain is equivalent to that of changes
in the loop gain as long as the system remains out of saturation.
Variations of 20% in the plant gain usually have little impact on
control system performance. Larger variations may cause the system to
be either unresponsive or marginally stable at the extremes. In such
cases, it is difficult to find a single set of servo gains that will
accommodate the range of variation.

Tuning Controllers & Industrial Controllers

K. Craig

17

 If changes in the plant can be anticipated with reasonable

accuracy, a technique known as Gain Scheduling may be used.
This requires knowledge of the gain in question and a controller with
gains that can be modified during operation.

If this is not the case, it might be possible to sense the changes

in the plant using the controller.
At the time the controller is commissioned, it can generate signals that
excite the plant and then use the feedback to estimate the parameter.
This technique is called Auto-Tuning. This, however, is usually not an
option during normal operation, as strong excitation signals during
normal operation would disturb the system. To sense gain for
scheduling in this condition, a technique known as Adaptive Control is
sometimes used.

Tuning Controllers & Industrial Controllers

K. Craig

18

 Multiple loops are common in control systems, e.g., a current

loop may reside in a velocity loop which may reside in a
position loop.
When tuning multi-loop control systems, we follow a similar
procedure. The inner loops operate in the next higher frequency range
to the outer loop. After an inner loop is tuned, it acts like a low-pass
filter within the outer loop. Once an inner loop is tuned, there is little
need to return to it, as each loop operates over a different frequency
range.
When tuning, expect the bandwidth of the outer loop to be between
20% and 40% of that of the inner loop. Tune each loop to be as
responsive as possible because it becomes the barrier for the next outer
loop.

Tuning Controllers & Industrial Controllers

K. Craig

19

 What about Saturation?

Saturation is the most prevalent nonlinear behavior in industrial controllers
and often causes a system to overshoot excessively, when it is actually
quite stable.
While you should avoid nonlinear behavior in system design, all systems
have limits as to how much energy they can supply to the plant. During
saturation, the control system is applying all the power that is available.
All systems have power converter limitations, and most will enter
saturation immediately when those limits are exceeded.
Saturation causes a serious problem for integral controllers called windup.

Avoid saturation when tuning.

Saturation is a nonlinear effect; it does not indicate either stability or
responsiveness because the system is locked full on. When generating
step response or Bode plots, ensure that the system does not enter
saturation by monitoring the control signal; it should never reach its
maximum value.
Tuning Controllers & Industrial Controllers

K. Craig

20

Integral Wind-Up and Its Correction

Tuning Controllers & Industrial Controllers

K. Craig

21

 Let's consider the situation of integral windup and its correction.

Integral control may be degraded significantly by saturation
effects.
For example, as seen in the figure, a large sustained error causes the
integral controller to ramp its output pressure up to the 20-psig supply
pressure. The diaphragm valve, sized to be wide open at 15 psig (the upper
end of the 3 to 15 psig control range) saturates at 15 psig.
The integral signal beyond t = 7.5 seconds is really useless since it asks for
a motion that the valve cannot produce. When the error reverses at t = 10
seconds, the valve cannot respond to this change until the integral signal
(which has "wound up" to 20 psig) is "unwound" back to the 15-psig level
at t = 12.5.

This delayed response is called integral wind-up.

Tuning Controllers & Industrial Controllers

K. Craig

22

 Note that this delay is in addition to the normal lagging behavior of

integral control and can cause excessive overshooting and stability
problems.

Integral windup is of course not a problem in every application

of integral control. If difficulty is anticipated, the controller
can be modified in different ways to give various degrees of
improvement. Basically, one wants to disable the integrator
whenever its output signal causes saturation in the final control
element.
In this example, the integrator is disabled when its output pressure
reaches 15 psig, preventing any windup.
When the error reverses at t = 10 seconds the integrator and valve
immediately respond to the negative error since there is no windup that
needs to first be unwound.

Tuning Controllers & Industrial Controllers

K. Craig

23

Types of Industrial Controllers

The PID controller is the most common controller in general
use. It can be simplified by setting one or two of the three
gains to zero.
Lets explore variations of the P, I, and D gains.
In choosing a controller for an application, the designer must
weigh complexity vs. performance. More complex controllers
require more processing capability, e.g., faster processors for
digital control or more components for analog control. They
are also more difficult to tune.
How much performance is worth paying for?

Tuning Controllers & Industrial Controllers

K. Craig

24

 The basic issues in control systems vary little between digital

and analog controllers. For all control systems, gain and phase
margins must be maintained, and phase loss around the loop
should be minimized.
First the focus will be on velocity control of a singleintegrating loop, i.e., control of torque to produce velocity.
Then we will focus on a double-integrating position loop.
To facilitate side-by-side comparison of different controller
types, we evaluate response to a square wave input, as this is
the signal of choice for exposing marginal stability; testing
with gentler signals may allow marginal stability to pass
undetected.

Tuning Controllers & Industrial Controllers

K. Craig

25

 For each of the controllers discussed, we will use frequencyrange-based tuning. The P and D gains combine to determine
behavior in the high-frequency range and they should be set
first; simultaneously tuning is required. The I gain and a
command filter determine behavior in the low-frequency
range.
The high-frequency range is limited by the plant, amplifier,
and sensor. The low-frequency range is limited primarily by
the high-frequency range.
Time delays, i.e., computation delay, sample-and-hold delay,
sensor delay, can all be added as needed to the control system.
We focus here on command response, although disturbance
response is just as important.

Tuning Controllers & Industrial Controllers

K. Craig

26

 Command response is usually preferred for determining

stability as commands are easier to generate. When tuning, the
command should be as large as possible to maximize the
signal-to-noise ratio, as this supports accurate measurements;
however, the power converter must remain out of saturation
during these tests.

Proportional Control
Virtually all controllers have a large proportional gain. While
derivative gain can provide incremental improvements at high
frequencies, and integral gain improves performance at lower
frequencies, the proportional gain is the primary actor across
the entire frequency range of operation.

Tuning Controllers & Industrial Controllers

K. Craig

27

 Here the manipulating variable M is directly proportional to

the actuating signal E.
We assume that the dynamics associated with the real
controller are negligible relative to other system dynamics.
The corrective effort is made proportional to system "error";
large errors engender a stronger response than do small ones.
We can vary in a continuous fashion the energy and/or
material sent to the controlled process.
m(t) = K P e(t)
M(s)
= K P = Proportional Gain
E(s)

Tuning Controllers & Industrial Controllers

K. Craig

28

 Relative to on-off control, the advantage is a lack of limitcycling behavior. The disadvantages are general complexity,
higher cost, and lower reliability of hardware.
Proportional control exhibits nonzero steady-state errors for
even the least-demanding commands and disturbances.
Why is this so? Suppose for an initial equilibrium operating
point xc = xv and steady-state error is zero. Now ask xc to go
to a new value xvs. It takes a different value for the
manipulated input M to reach equilibrium at the new xc. When
the manipulated input M is proportional to the actuating signal
E, a new M can only be achieved if E is different from zero
which requires xc xv; thus, there must be a steady-state error.

Tuning Controllers & Industrial Controllers

K. Craig

29

 How do you tune a proportional controller?

Set KP low.
Apply a square wave command at about 10% of the desired
loop bandwidth. Use large amplitude, but avoid saturation.
Raise KP for little of no overshoot.
The loss of stability is a consequence of phase loss in the
loop, and the proportional gain will rise to press that limit.
Be aware, however, that other factors, primarily noise, often
ultimately limit the proportional gain below what the
stability criterion demands.
Independent of its source, noise will be amplified by the
high-frequency gains in the controller, such as the
proportional gain.
Setting the proportional gain requires balancing the need for
performance and the elimination of noise.
Tuning Controllers & Industrial Controllers

K. Craig

30

Proportional Control

R(s)

+-

1
1
500
G (s) = =
=
Js 0.002s
s
2
G amp ( s ) = 2
s + 2s + 2

D(s)
E(s)
B(s)

+
Gc(s) +

G(s)

C(s)

H(s)

= 5027 (rad/s) = 800 (Hz)

= 0.707

K t = 1 (Nm/A)
H (s) = 1

G C ( s ) = K P = 1.2

Tuning Controllers & Industrial Controllers

K. Craig

31

Closed-Loop Step-Response Plot

2 % Settling Time = 0.00561 sec

Rise Time = 0.00298 sec

Tuning Controllers & Industrial Controllers

K. Craig

32

Open-Loop Bode Plots: PM = 80.3 and GM = 21.5 dB

Tuning Controllers & Industrial Controllers

K. Craig

33

Closed-Loop Bode Plots: Closed-Loop Bandwidth = 726 rad/s

Tuning Controllers & Industrial Controllers

K. Craig

34

 Integral Control
When a proportional controller can use large loop gain and
preserve good relative stability, system performance, including
those on steady-state error, may often be met.
However, if difficult process dynamics such as significant
dead times prevent use of large gains, steady-state error
performance may be unacceptable.
When human process operators notice the existence of steadystate errors due to changes in desired value and/or disturbance
they can correct for these by changing the desired value ("set
point") or the controller output bias until the error disappears.
This is called manual reset.

Tuning Controllers & Industrial Controllers

K. Craig

35

 Integral control is a means of removing steady-state errors

without the need for manual reset. It is sometimes called
automatic reset.
dm(t)
M(s) K i
= K i e(t)
=
dt
E(s)
s
t

m(t) = K i e()d
0

If the value of e(t) is doubled, then the value of m(t) varies

twice as fast.
For e(t) = 0, m(t) remains stationary.
We have seen why proportional control suffers from steadystate errors. We need a control that can provide any needed
steady output (within its design range, of course) when its
input (system error) is zero.

Tuning Controllers & Industrial Controllers

K. Craig

36

Comparison: Proportional vs. Integral Control

Integral control has the undesirable side effects of

reducing response speed and degrading stability
Tuning Controllers & Industrial Controllers

K. Craig

37

 Although integral control is very useful for removing or

reducing steady-state errors, it has the undesirable side effect
of reducing response speed and degrading stability.
Why? Reduction in speed is most readily seen in the time
domain, where a step input (a sudden change) to an integrator
causes a ramp output, a much more gradual change.
Stability degradation is most apparent in the frequency domain
(Nyquist Criterion) where the integrator reduces the phase
margin by giving an additional 90 degrees of phase lag at
every frequency, rotating the (B/E)(i) curve toward the
unstable region near the -1 point.

Tuning Controllers & Industrial Controllers

K. Craig

38

 Occasionally an integrating effect will naturally appear in a

system element (actuator, process, etc.) other than the
controller.
These gratuitous integrators can be effective in reducing
steady-state errors. Although controllers with a single
integrator are most common, double (and occasionally triple)
integrators are useful for the more difficult steady-state error
problems, although they require careful stability augmentation.
Conventionally, the number of integrators between E and C in
the forward path has been called the system type number.

Tuning Controllers & Industrial Controllers

K. Craig

39

In addition to the number of

integrators, their location (relative
to disturbance injection points)
determines their effectiveness in
removing steady-state errors.
In Figure (a) the integrator gives
zero steady-state error for a step
command but not for a step
disturbance.
By relocating the integrator as in
Figure (b), either or both step
inputs Vs and Us can be
"canceled" by M without requiring
E to be nonzero.

Integrators must be located upstream

from disturbance-injection points if they
are to be effective in removing steadystate errors due to disturbances.
Location is not significant for steadystate errors caused by commands.

Tuning Controllers & Industrial Controllers

K. Craig

40

 Integral control can be used by itself or in combination with

other control modes.
Proportional + Integral (PI) Control is the most common
mode.
Kp t
m(t) = K p e(t) +
e()d

TI 0

M(s)
1
= K p 1 +

E(s)
T
s
I

TI = integral time
1/TI = reset rate = number of times per minute that the
proportional part of the control action is duplicated = repeats
per minute

Tuning Controllers & Industrial Controllers

K. Craig

41

 Integral gain provides DC and low-frequency stiffness. When

a DC error occurs, the integral gain will move to correct it.
The higher the gain, the faster the correction. Fast correction
implies a stiffer system.
Dont confuse DC stiffness with dynamic stiffness. A system
can be quite stiff at DC and not stiff at all at high frequencies!
Higher integral gains will provide higher DC stiffness but will
not substantially improve stiffness at or above the loop
bandwidth.
PI controllers are more complicated to implement than P
controllers. Saturation becomes more complicated, as wind-up
must be avoided. In analog controllers, clamping diodes must
be added, and in digital controllers, saturation algorithms must
be coded.
Tuning Controllers & Industrial Controllers

K. Craig

42

 Integral gain can cause instability. In the open loop, the

integral, with its 90 phase lag, reduces PM. In the time
domain, the common result of adding integral gain is
overshoot and ringing. As a result, larger integral gains
usually reduce bandwidth.
PI controllers have two frequency ranges: high and low. The
high range is served by KP and the low by KI.
To tune a PI controller, set the P gain as it was in the P
controller. After the higher frequency range is complete, KI
can be tuned. Raise KI for 15% overshoot to a step input; a
modest amount of overshoot to a step input is tolerable in most
applications.

Tuning Controllers & Industrial Controllers

K. Craig

43

Proportional-Integral Control

R(s)

+-

1
1
500
G (s) = =
=
Js 0.002s
s
2
G amp ( s ) = 2
s + 2s + 2
K t = 1 (Nm/A)
H (s) = 1

E(s)
B(s)

D(s)
+
Gc(s) +

G(s)

C(s)

H(s)

= 5027 (rad/s) = 800 (Hz)

= 0.707

K
s + KI
s + 110
=
G C ( s ) = K P 1 + I = K P
1.2

s
s

Tuning Controllers & Industrial Controllers

K. Craig

44

Closed-Loop Step-Response Plot

Overshoot = 12.2%

Rise Time = 0.0021 sec

2% Settling Time = 0.0212 sec

Tuning Controllers & Industrial Controllers

K. Craig

45

Open-Loop Bode Plots: PM = 69.9 and GM = 21.2 dB

Tuning Controllers & Industrial Controllers

K. Craig

46

Closed-Loop Bode Plots

Tuning Controllers & Industrial Controllers

Closed-Loop Bandwidth = 851 rad/s

Peak Response = 1.08 dB @ 223 rad/s

K. Craig

47

Analog PI Controller
and Lag Controller

eout
R FCs + 1
=
ein
R IN Cs
RF

R2

ein

C2

R1
e0
ein

C1

-V
Lead or Lag
Controller

Tuning Controllers & Industrial Controllers

+V
-V

eout

PI Controller

+V

RIN

e0

eout

eout
R 2 R1C1s + 1
=
ein
R 1 R 2 C 2s + 1
K. Craig

48

 PI+ Control
This is an enhancement to PI control. Usually, because of the
overshoot, the integral gain in PI control is limited to relatively
small values.
PI+ control uses a low-pass filter on the command signal to
remove overshoot. In this way, the integral gain can be raised
to higher values. PI+ control is useful in applications where
the rejection of DC disturbances is paramount. The primary
shortcoming of PI+ control is that the command filter also
reduces the controllers command response.

Tuning Controllers & Industrial Controllers

K. Craig

49

PI+ Controller
D(s)
V(s)

R(s)
A(s)

M(s)

E(s)

+
B(s) _

Gc(s)

G(s)

C(s)

H(s)

Command Filter

A ( s ) = K F + (1 K F )

KI
s + KI

Tuning Controllers & Industrial Controllers

KF = 1: all filtering removed

KF = 0: filtering most severe
K. Craig

50

 When KF = 0, the command is filtered by a single-pole, lowpass filter, with the pole at KI. This allows the highest
integral gain but will also most severely limit the controller
command response.
Typically, KF = 0 will allow an increase of almost three times
in the integral gain but will reduce the bandwidth by when
compared with KF = 1 (PI control).
Finding the optimal value of KF depends on the application,
but a value of 0.65 has been found to work in many
applications. This value typically allows the integral gain to
more than double while reducing the bandwidth by only 15% 20%.
Why select the frequency of the command low-pass filter to be
KI? Why not set this higher or lower?
Tuning Controllers & Industrial Controllers

K. Craig

51

 The reason is that this frequency is excellent at canceling the

peaking caused by the integral gain.
One way to look at PI+ control is that it uses the command
filter to attenuate the peaking caused by PI control. The
peaking caused by KI can be cancelled by the attenuation of a
low-pass filter with a break of KI.
How do you tune a PI+ Controller?
Tuning a PI+ controller is similar to tuning a PI controller
except that you must choose the amount of filtering (KF)
before tuning the integral gain.
KF < 0.4
KF = 0.65
KF > 0.9
Tuning Controllers & Industrial Controllers

high DC stiffness
general
fastest response
K. Craig

52

Closed-Loop Step-Response Plot

KI = 300
KP = 1.2
KF =0.65

Overshoot = 11%

2% Settling Time = 0.0111 sec

Rise Time = 0.00258 sec

Tuning Controllers & Industrial Controllers

K. Craig

53

Open-Loop Bode Plots: PM = 54.8 and GM = 20.7 dB

Tuning Controllers & Industrial Controllers

K. Craig

54

Closed-Loop Bode Plots

Tuning Controllers & Industrial Controllers

Closed-Loop Bandwidth = 731 rad/s

Peak Response = 0.841 dB @ 284 rad/s

K. Craig

55

 Derivative Control
On-off, proportional, and integral control actions can be used
as the sole effect in a practical controller.
But the various derivative control modes are always used in
combination with some more basic control law. This is
because the derivative mode produces no corrective effect for
any constant error, no matter how large, and therefore would
allow uncontrolled steady-state errors.
One of the most important contributions of derivative control
is in system stability augmentation. If absolute or relative
stability is the problem, a suitable derivative control mode is
often the answer.
The stabilization or "damping" aspect can easily be understood
qualitatively from the following discussion.
Tuning Controllers & Industrial Controllers

K. Craig

56

 Invention of integral control may have been stimulated by the

human process operators desire to automate their task of
manual reset. Derivative control hardware may first have been
devised as a mimicking of human response to changing error
signals. Suppose a human process operator is given a display
of system error E and has the task of changing manipulated
variable M (say with a control dial) so as to keep E close to
zero.

Tuning Controllers & Industrial Controllers

K. Craig

57

 If you were the operator, would you produce the same value of
M at t1 as at t2? A proportional controller would do exactly
that.
A stronger corrective effect seems appropriate at t1 and a lesser
one at t2 since at t1 the error E is E1,2 and increasing, whereas
at t2 it is also E1,2 but decreasing.
The human eye and brain senses not only the ordinate of the
curve but also its trend or slope. Slope is clearly dE/dt, so to
mechanize this desirable human response we need a controller
sensitive to error derivative.
Such a control can, however, not be used alone since it does
not oppose steady errors of any size, as at t3, thus a
combination of proportional + derivative control, for example,
makes sense.
Tuning Controllers & Industrial Controllers

K. Craig

58

 The relation of the general concept of derivative control to the

specific effect of viscous damping in mechanical systems can
be appreciated from the figure below.
Here an applied torque T tries to control position of an
inertia J. The damper torque on J behaves exactly like a
derivative control mode in that it always opposes velocity
d/dt with a strength proportional to d/dt making motion less
oscillatory.

Tuning Controllers & Industrial Controllers

K. Craig

59

 Derivatives of E, C, and almost any available signal in the

system are candidates for a useful derivative control mode.
First derivatives are most common and easiest to implement.
The noise-accentuating characteristics of derivative operations
may often require use of approximate (low-pass filtered)
derivative signals.
Derivative signals can sometimes be realized better with
sensors directly responsive to the desired value, rather than
trying to differentiate an available signal.
In addition to stability augmentation, derivative modes may
also offer improvements in speed of response and steady-state
errors.

Tuning Controllers & Industrial Controllers

K. Craig

60

 The derivative gain advances the phase of the loop by virtue of

the 90 phase lead of a derivative. Using derivative gain will
usually allow the system responsiveness to increase, allowing the
bandwidth to nearly double in some cases.
Derivative gain has high gain at high frequencies. So while some
derivative gain does help the phase margin, too much hurts the
gain margin by adding gain at the phase crossover frequency,
typically a high frequency. This makes the derivative gain
difficult to tune. The designer sees overshoot improve because
of increased PM, but a high-frequency oscillation, which comes
from reduced GM, becomes apparent.

Tuning Controllers & Industrial Controllers

K. Craig

61

 Derivatives are also very sensitive to noise. The derivative gain

needs to followed by a low-pass filter to reduce noise content.
However, the lower break frequency of the filter, the less benefit
can be gained from the derivative gain.
de(t)
Proportional + Derivative Control
m(t) = K p e(t) + K p Td
dt
M(s)
= K p (1 + Td s )
E(s)
Td = derivative time = time interval by which the rate action
advances the effect of the proportional control action
Derivative control has an anticipatory character, however, it can
never anticipate any action that has not yet taken place.
Derivative control amplifies noise signals and may cause a
saturation effect in the actuator.
Tuning Controllers & Industrial Controllers

K. Craig

62

 Tuning a PD Controller
Zero KD and set KP low.
Apply a square wave command at about 10% of the desired
loop bandwidth. Use large amplitude, but avoid saturation.
Raise KP for some overshoot, but no ringing.
Raise KD to eliminate most overshoot.
If it is too noisy, reduce noise at the source or lower KD or
lower KP.

Tuning Controllers & Industrial Controllers

K. Craig

63

Proportional-Derivative Control

R(s)

+-

1
1
500
G (s) = =
=
Js 0.002s
s
2
G amp ( s ) = 2
s + 2s + 2
K t = 1 (Nm/A)
H (s) = 1

E(s)
B(s)

D(s)

+
Gc(s) +

G(s)

C(s)

H(s)

= 5027 (rad/s) = 800 (Hz)

= 0.707

K Ds
0.0002s

G C ( s ) = K P 1 +

= 1.7 1 +
5
s + 1
8.0 (10 ) s + 1

Tuning Controllers & Industrial Controllers

K. Craig

64

Closed-Loop Step-Response Plot

2% Settling Time = 0.0044 sec

Rise Time = 0.00232 sec

Tuning Controllers & Industrial Controllers

K. Craig

65

Open-Loop Bode Plots: PM = 85.8 and GM = 21.6 dB

Tuning Controllers & Industrial Controllers

K. Craig

66

Closed-Loop Bode Plots

Tuning Controllers & Industrial Controllers

Closed-Loop Bandwidth = 940 rad/s

K. Craig

67

 Proportional + Integral + Derivative (PID) Control

Kp

de(t)
m(t) = K p e(t) +
e()d + K p Td

Ti 0
dt

M(s)
1
= K p 1 +
+ Td s
E(s)
Ti s

About 90% of the industrial controllers is use today

utilize PID or modified PID control schemes.
Analog PID controllers are mostly hydraulic,
pneumatic, electric, and electronic types or their
combinations.

K Ds 2 + s + K I
KI

G c ( s ) = K P 1 + K Ds + = K P

s
s

Tuning Controllers & Industrial Controllers

K. Craig

68

 Digital PID control through the use of microprocessors

is becoming more common.
Most PID controllers are adjusted on site and many
different tuning rules have been proposed.
PID controllers with automatic tuning, automatic
generation of gain schedules, and continuous adaptation
are now possible.
Modified forms of PID control (e.g., I-PD and twodegrees-of-freedom PID control) are currently in use in
industry.
Many practical methods for bumpless switching (from
manual operation to automatic operation) and gain
scheduling are commercially available.
Usefulness of PID control lies in its general
applicability to most control systems.
Tuning Controllers & Industrial Controllers

K. Craig

69

 How to Tune a PID Controller

A PID controller is a two-zone controller. The P and D gains
jointly form the higher frequency zone. The I gain forms the
low-frequency zone.
The benefit of the D gain is that it allows the P gain to be set
higher than it could be otherwise.
The first step is to tune the controller as if it were a P
controller, but to allow more overshoot than normal,
understanding that the D gain will cure the problem.
Typically, the P gain can be raised 25% to 50% over the value
from the P and PI controllers.
The next step is to add a little D gain to cure the overshoot
induced by the higher than normal P gain.
The P & D gains together form the high-frequency zone.
Tuning Controllers & Industrial Controllers

K. Craig

70

 Next, the I gain is tuned, much as it was in the PI controller.

The expectation is that the P and I gains will be about 30%
higher than they were in the PI controller.
The phase margin compared to that of a PI controller will be
about the same, however, the gain margin will be less because
the high-frequency zone of the PID controller is so much
higher than that of the PI controller as evidenced by the
higher bandwidth of the PID controller. Reduced gain margin
is a concern because the gains of plants often change during
normal operation and is of particular concern in systems
where the gain can increase, e.g., saturation of an inductor in
a current controller, declining inertia in a motion system, or
declining thermal mass in a temperature controller. These all
raise the gain of the plant and reduce GM.

Tuning Controllers & Industrial Controllers

K. Craig

71

 Given the same plant and amplifier, a PID controller will

provide faster response than a PI controller but will often
be harder to control and more sensitive to changes in the
plant.
The problems with noise in the PI controller are
exacerbated by the use of a differential gain, as the gain of
a true derivative increases without bound as the frequency
increases. In most working systems, a low-pass filter is
placed in series with the derivative to limit gain at the
highest frequencies. If the noise content of the feedback or
command signals is high, the best cure is to reduce the
noise at its source. Beyond that, lowering the frequency of
the derivatives low-pass filter will help, but it will also
limit the effectiveness of the D gain.

Tuning Controllers & Industrial Controllers

K. Craig

72

 Noise can also be reduced by reducing the D gain directly,

but this is usually a poorer alternative than lowering the
low-pass filter frequency.
If the signal is too noisy, the D gain may need to be
abandoned altogether.
Ziegler-Nichols Tuning Method
A popular method for tuning P, PI, and PID controllers is
the Ziegler-Nichols method.
This method starts by zeroing the I and D gains and then
raising the P gain until the system is unstable.
The value of Kp at the point of instability is called KMAX;
the frequency of oscillation is called f0 (rad/s).

Tuning Controllers & Industrial Controllers

K. Craig

73

 The method then backs off the P gain a predetermined

amount and sets the I and D gains as a function of f0.
KP

KI

KD

0.5 KMAX

PI

0.45 KMAX

1.2 f0

PID

0.6 KMAX

2.0 f0

0.125 / f0

It is assumed that KP is in series with KI and KD.

Tuning Controllers & Industrial Controllers

K. Craig

74

 The Ziegler-Nichols method is too aggressive (small

stability margins) for many industrial control systems.
Also, in general, the gains from the Ziegler-Nichols method
will be higher than from then other methods discussed here.

Tuning Controllers & Industrial Controllers

K. Craig

75

Proportional-Integral-Derivative Control
D(s)

R(s)

+-

E(s)
B(s)

+
Gc(s) +

G(s)

H(s)

1
1
500
G (s) = =
=
Js 0.002s
s
= 5027 (rad/s) = 800 (Hz)
2
G amp ( s ) = 2
s + 2s + 2 = 0.707
K t = 1 (Nm/A)
H (s) = 1

C(s)

KP =1.7
KI = 160
KD = 0.0002

1
= 2000 Hz

= 12566 rad/s

K Ds K I

G C ( s ) = K P 1 +
+
s + 1 s
( + K D ) s 2 + ( K I + 1) s + K I
= KP

+
s
s
1
(
)

Tuning Controllers & Industrial Controllers

K. Craig

76

Closed-Loop Step-Response Plot

2% Settling Time = 0.00443 sec

Rise Time = 0.00232 sec

Tuning Controllers & Industrial Controllers

K. Craig

77

Open-Loop Bode Plots: PM = 75.2 and GM = 21.5 dB

Tuning Controllers & Industrial Controllers

K. Craig

78

Closed-Loop Bode Plots

Tuning Controllers & Industrial Controllers

Closed-Loop Bandwidth = 942 rad/s

K. Craig

79

 PID+ Control
A PID+ controller is a PID controller modified with the
command filter.
As with the PI+ controller, the goal for the PID+ controller
is to allow higher integral gains for improved DC stiffness.
Again, the input filter cancels peaking caused by high
integral gains.
As with PI+ control, the command response suffers as the
stiffness improves.
Tuning a PID+ controller is the same as tuning a PID
controller except that the value of KF must be selected
before tuning the I gain (similar to the PI+ controller).

Tuning Controllers & Industrial Controllers

K. Craig

80

PID+ Controller
V(s)

D(s)
R(s)

A(s)

M(s)

E(s)

+
B(s) _

Gc(s)

C(s)

G(s)

H(s)

Command Filter

KI
A ( s ) = K F + (1 K F )
s + KI

KF = 1: all filtering removed

KF = 0: filtering most severe
KF < 0.4 high DC stiffness
KF = 0.65 general
KF > 0.9 fastest response

K Ds K I

G C ( s ) = K P 1 +
+
s + 1 s
( + K D ) s 2 + ( K I + 1) s + K I
= KP

+
s
s
1
(
)

Tuning Controllers & Industrial Controllers

KI = 400
KI = 1.7
KD = 0.0002
KF =0.65

1
= 2000 Hz

= 12566 rad/s

K. Craig

81

Closed-Loop Step-Response Plot

2% Settling Time = 0.00447 sec

Rise Time = 0.00232 sec

Tuning Controllers & Industrial Controllers

K. Craig

82

Open-Loop Bode Plots: PM = 60.5 and GM = 21.4 dB

Tuning Controllers & Industrial Controllers

K. Craig

83

Closed-Loop Bode Plots

Tuning Controllers & Industrial Controllers

Closed-Loop Bandwidth = 954 rad/s

K. Craig

84

 Choosing A Controller
The simple P controller provides performance suitable for many
applications.
The introduction of the I term provides DC stiffness but reduces
PM.
The command filter in PI+ and PID+ control allows even higher
DC stiffness but reduces bandwidth.
The D term provides higher responsiveness but erodes gain
margin and adds phase shift, which is a disadvantage if this
loop is to be enclosed in an outer loop.
In selecting a controller, first determine whether the application
needs the D gain; if not, avoid it as it adds complexity,
increases noise susceptibility, and steals GM.

Tuning Controllers & Industrial Controllers

K. Craig

85

 Next, make sure the application can support D gains;

systems that are noisy configured as PI controllers may not
work well with a D gain.
After that, examine the application for the needed DC
stiffness. If none is required, avoid the I gain. If some is
needed, use the standard form PI; if maximum DC stiffness
is required, add the input filter by using PI+ control.

Tuning Controllers & Industrial Controllers

K. Craig

86

Approximate Control Modes

Proportional + Integral (PI)
Phase-Lag Compensation
Proportional + Derivative (PD)
Phase-Lead Compensation
Proportional + Integral + Derivative (PID)
Lead / Lag Compensation

Tuning Controllers & Industrial Controllers

K. Craig

87

 We have introduced the basic control modes: on-off,

proportional, integral, and derivative. Each of these has its
own advantages and drawbacks, and thus it is not
surprising that many practical applications are best served
by some combination of basic modes.
We have also considered the most basic or idealized
versions of the modes so that their essential features could
be brought out most clearly without confusing side issues.
Practical versions of some controllers are not able to
realize completely the ideal behavior and also may require
a modified design technique. Sometimes a non-ideal
controller can meet specifications with simpler hardware or
software. For these reasons, approximate forms of control
modes should be considered.
Tuning Controllers & Industrial Controllers

K. Craig

88

Phase-Lag Compensation
PI control provides the steady-state-error benefits of
pure integral control with faster response and improved
stability.
Phase-Lag Compensation is the approximate version of
PI Control realized in many practical controllers. It
cannot attain the zero steady-state errors possible with
perfect integral control but this is not a fatal defect
because realistic error specifications always must allow
some steady-state error.

Tuning Controllers & Industrial Controllers

K. Craig

89

 Phase-Lead Compensation
Since derivative control is never used alone and we
have already briefly discussed PD control, let's
concentrate on the approximate version, phase-lead
compensation.
If a basic system has had its gain set for desired relative
stability and we then find that its response speed is too
slow, phase-lead compensation may be helpful. Also, if
a basic system is structurally unstable (gain setting does
not provide stability), phase-lead compensation may
stabilize the system. Usually, phase-lead compensation
also provides a modest gain increase, so steady-state
errors are reduced whether this was a problem or not.

Tuning Controllers & Industrial Controllers

K. Craig

90

 Proportional + Integral + Derivative (PID) Control

This combination of basic control modes can improve
all aspects (stability, speed, steady-state errors) of
system performance and is the most complex method
available as an off-the-shelf general-purpose controller.
If we look at analog pneumatic and electronic
controllers, their microprocessor-based digital versions,
or the individual control loops implemented in a large
general-purpose digital process computer, over and
over again we see successful applications of P, PI, PD,
and PID controls. The basis of the strength of the PID
modes is their simplicity; they "make sense."

Tuning Controllers & Industrial Controllers

K. Craig

91

 Lag/Lead Compensation
The approximate version of PID control implemented
in many practical controllers is called lag/lead
compensation. Mathematically it is exactly a cascading
of the phase-lag and phase-lead controllers already
discussed.
The effects on system performance are also a
superposition of the two separate effects, thus a lag/lead
controller can improve all aspects of performance (as
can a PID): stability, speed, and steady-state errors.
Selection of the parameters is performed by essentially
designing the two compensators separately.

Tuning Controllers & Industrial Controllers

K. Craig

92