ELEC-C8201: Control and Automation

5. The Root Locus Method

&
PID Controllers

Themistoklis Charalambous
Office number: 2561, TUAS Building
Email: themistoklis.charalambous@aalto.fi
Office hours: Wednesday 16.00-18.00

1
 In the previous lecture…

You:

• Understood the concept of stability of dynamic systems

• Got introduced to the Routh-Hurwitz method as a tool for assessing system

stability.

2
 Learning outcomes

By the end of this lecture, you should be able to:

• Understand the concept of the root locus and its role in control system design

• Know how to obtain a root locus plot by sketching or using MATLAB

• Be familiar with the PID controller as a key element of many feedback systems

3
 The closed-loop characteristic equation
5.2 Closed-Loop Control (Feedback Control)
Input Output
For Example:
disturbance disturbance

d̄i (s) d̄o (s)

Demanded
Error
Output
Signal Control Controlled
(Reference)
Signal Output
+ +
+ Controller + “Plant” +
Σ K(s) Σ G(s) Σ
r̄ (s) − ē(s) ū(s) ȳ(s)

• The dynamic performance of a closed-loop

Figure 5.1
control system is described by the
closed-loop transfer function
1 ¯ G(s) ¯ K(s)G(s)
ȳ(s) = 5.2.1 Derivation
do (s)
of
+
the closed-loop
di (s) + functions:
transfer
r̄(s)
1 + K(s)G(s) 1 + K(s)G(s) 1 + K(s)G(s)
" #
• Note: All the closed-loop transfer functions have the same denominator.
ȳ(s) = d̄o (s) + G(s) d̄i (s) + K(s)ē(s)

ē(s) = r̄ (s) − ȳ(s)

" $ %#
=⇒ ȳ(s) = d̄o (s) + G(s)
4 d̄i (s) + K(s) r̄ (s) − ȳ(s)
• The closed loop poles are the roots of the closed-loop characteristic equation
1 + K(s)G(s) = 0

• The performance of a feedback system

‣ Closed-loop stability of the system

‣ Characteristics of the closed-loop system’s transient response

(e.g., speed of response, presence of any resonances etc)
can be described in terms of the location of the roots of the characteristic
equation in the s-plane.

• The poles lie in the s-plane, and given s is a complex is a complex variable, the
characteristic equation may be rewritten in polar form as

|K(s)G(s)|\ K(s)G(s) = 1 + j0

and therefore it is necessary that

|K(s)G(s)| = 1 and \ K(s)G(s) = 1800 + k360o

5
 The root locus plot

• Root locus plot: a graph showing how the roots of the characteristic equation
move around the s-plane as a single parameter varies

‣ a powerful tool for designing and analyzing feedback control systems

‣ frequently necessary to adjust one or more system parameters in order to obtain

suitable root locations → determine how the roots of the characteristic equation
of a given system migrate about the s-plane as the parameters are varied

‣ provides the engineer with a measure of the sensitivity of the roots of the system
to a variation in the parameter being considered

• The root locus plot may be used in conjunction with the Routh-Hurwitz
criterion

6
 The root locus concept - Proportional control
5.5.2 Proportional Control

• In this case, K(s) = kp K(s) = kp

r̄ + ē ū ȳ
Σ kp G(s)
−

Typical result of increasing the gain kp , (for control systems where

• Typical result
G(s) isofitself stable): the gain kp, (for systems where G(s) is stable):
increasing

✓ IncreasedIncreased
accuracy of control
accuracy of control. “good”

- IncreasedIncreased
control control
action action.
- ReducedReduced
damping

- PossiblePossible
(→ more oscillations)
damping.

loss of closed-loop
loss of closed-loop stability
stability gain kkpp .
for large
for large
} “bad”

Example:
1
G(s) =
(s + 1)2
(A critically damped 2nd order system)
7
Example 1:
• Consider the following critically damped 2nd-order system
1
G(s) =
(s + 1)2

Then
1
kp G(s) kp (s+1) 2
ȳ(s) = r̄(s) = 1 r̄(s)
1 + kp G(s) 1 + kp (s+1)2
kp
= 2 r̄(s)
s + 2s + 1 + kp
“root locus plot”
The characteristic equation representing this system is

(s) = s2 + 2s + 1 + kp = s2 + 2⇣!n s + !n2 = 0 xx

1

The closed-loop poles are s = − 1 ± j kp

Movement of closed-loop
poles as kp increases

8
• The damping factor and natural frequency can be computed by
( ( p
⇣!n = 1 !n = 1 + kp
)
!n2 = 1 + kp ⇣=p1
1+kp
2.6. Behavior of continuous 2nd order systems with unit step input

Consider the following block diagram with a standard 2nd order system,
1
Reminder: R(s) = ! 2 Y (s)
R (s) = 1/s s
<latexit sha1_base64="ef1k4MP7EHW0j16VeX8IiMS9uY0=">AAAB+XicdVDLSgNBEOyNrxhfqx69DAYhXsJujHERhIAXj1HMA5IQZiezyZDZBzOzgbDsn3jxoIhX/8Sbf+MkWUFFCxqKqm66u9yIM6ks68PIrayurW/kNwtb2zu7e+b+QUuGsSC0SUIeio6LJeUsoE3FFKedSFDsu5y23cn13G9PqZAsDO7VLKJ9H48C5jGClZYGpnlXkqdXPU9gkthpItOBWbTK1gLIKp9VqlXH0aRWc5zaObIzqwgZGgPzvTcMSezTQBGOpezaVqT6CRaKEU7TQi+WNMJkgke0q2mAfSr7yeLyFJ1oZYi8UOgKFFqo3ycS7Es5813d6WM1lr+9ufiX142V5/QTFkSxogFZLvJijlSI5jGgIROUKD7TBBPB9K2IjLFOQemwCjqEr0/R/6RVKdua31aL9cssjjwcwTGUwIYLqMMNNKAJBKbwAE/wbCTGo/FivC5bc0Y2cwg/YLx9Aj6/k2Y=</latexit>

H(s) ω
= 2 nY (s) <latexit sha1_base64="DFI1wxtp8KaTZ2f4hJuymIY53fE=">AAAB63icdVBNS8NAEJ3Ur1q/qh69LBahXkpSaw2eCl48VrAf0oay2W7apbtJ2N0IpfQvePGgiFf/kDf/jZs2goo+GHi8N8PMPD/mTGnb/rByK6tr6xv5zcLW9s7uXnH/oK2iRBLaIhGPZNfHinIW0pZmmtNuLCkWPqcdf3KV+p17KhWLwls9jakn8ChkASNYp9JdWZ0OiiW7Yi+A7MpZtVZzXUPqddetnyMns0qQoTkovveHEUkEDTXhWKmeY8fam2GpGeF0XugnisaYTPCI9gwNsaDKmy1unaMTowxREElToUYL9fvEDAulpsI3nQLrsfrtpeJfXi/RgevNWBgnmoZkuShIONIRSh9HQyYp0XxqCCaSmVsRGWOJiTbxFEwIX5+i/0m7WnEMv6mVGpdZHHk4gmMogwMX0IBraEILCIzhAZ7g2RLWo/VivS5bc1Y2cwg/YL19Aq99jfs=</latexit>

H (s) = 2 n
s 2 +2 2⇣! s + ! 2
r=1 s + 2ζω n s + ω n n n
<latexit sha1_base64="Hxm3c0WEuRO4gV2mw7eJvvUMSmw=">AAACI3icbVDLSsNAFJ34rPVVdelmsAiKIEkQlIJQcNNlBfuAJobJ9KYOTiZhZiLU0H9x46+4caEUNy78F6cPoVoPDBzOOZc794QpZ0rb9qe1sLi0vLJaWCuub2xubZd2dpsqySSFBk14ItshUcCZgIZmmkM7lUDikEMrvL8a+a0HkIol4kb3U/Bj0hMsYpRoIwWlSu1IHeNL7EWS0NxLYuiRQNy6g1zduieu9wia/KhYneCZRFAq26f2GHieOFNSRlPUg9LQ6yY0i0FoyolSHcdOtZ8TqRnlMCh6mYKU0HvSg46hgsSg/Hx84wAfGqWLo0SaJzQeq7MTOYmV6sehScZE36m/3kj8z+tkOrrwcybSTIOgk0VRxrFO8Kgw3GUSqOZ9QwiVzPwV0zti2tKm1qIpwfl78jxpuqeO4ddn5WplWkcB7aMDdIQcdI6qqIbqqIEoekIv6A29W8/WqzW0PibRBWs6s4d+wfr6Biimo1w=</latexit>

• The
The behavior
behavior of theof the issystem
system as follows:is as follows:

The
The behavior
behavior of the
of the system is system is fully
characterized
fully characterized by ζby:
,
which is called the damping
• ζ the dumping factor
ratio, and ωn , which is called
•ωnatural
the n the natural
frequency. frequency

27
Prepared by Ben M. Chen

9
Reminder (continued):
34 CHAPTER 1. NATURAL RESPON
p
b ± b2 4ac
s1,2 =
2a p “root locus plot”
2⇣!n ± (2⇣!n )2 4!n2 {}
=
p2 X
= ⇣!n ± !n ⇣ 2 1
p
= ⇣!n ±j !n 1 ⇣ 2
| {z } | {z }
!d

{}

10
Figure 1.23: Pole locations in the s-plane for second-order mechanical sys
 System specifications
Reminder (continued):
• Typical
Typical specifications
specifications forfor
thethe step
step response:
response:

• Steady-state accuracy ess

? Rise time (10% ! 90%): tr ⇡ 1.8/!0 p
<latexit sha1_base64="7BHxj6NGKH9NbuHwMSgVyhPH/+U=">AAAB7XicdVBNSwMxEM3Wr1q/qh69BIvgqWxrreut4MVjBfsB7VKy6Wwbm02WJCuUpf/BiwdFvPp/vPlvTNsVVPTBwOO9GWbmBTFn2rjuh5NbWV1b38hvFra2d3b3ivsHbS0TRaFFJZeqGxANnAloGWY4dGMFJAo4dILJ1dzv3IPSTIpbM43Bj8hIsJBRYqzUhkGq9WxQLLlldwHsls+qtZrnWVKve179HFcyq4QyNAfF9/5Q0iQCYSgnWvcqbmz8lCjDKIdZoZ9oiAmdkBH0LBUkAu2ni2tn+MQqQxxKZUsYvFC/T6Qk0noaBbYzImasf3tz8S+vl5jQ81Mm4sSAoMtFYcKxkXj+Oh4yBdTwqSWEKmZvxXRMFKHGBlSwIXx9iv8n7Wq5YvlNrdS4zOLIoyN0jE5RBV2gBrpGTdRCFN2hB/SEnh3pPDovzuuyNedkM4foB5y3T0s4j5M=</latexit>

• Rise time (10% — 90%) tr = 1.8/(!n⇡⇣/

) 1 ⇣2
? Peak overshoot: Mp ⇡ e p <latexit sha1_base64="BcdC+PGiYmhQOZxcj/TsUc0znHU=">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</latexit>

✓ ◆
Mp in %
? • Peaktime
Settling overshoot
(to 1%): Mtps⇡=e 4.6/(⇣!0 ) or ⇣
⇡⇣/ 1 ⇣ 2 0.6 1
<latexit sha1_base64="z83g2HgzaqykkfCZnFSA3Hce9qo=">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</latexit>

<latexit sha1_base64="zSqrZDvQDhS4rQgFwEd6saQLb5Y=">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</latexit>
100
? Steady statetime
• Settling error(1%)
to unit step: t e=ss4.6/(⇣! )
s n <latexit sha1_base64="J22rseOowdEBSuquWMpBq0Zkdr0=">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</latexit>

? Phase margin: P M ⇡ 100⇣

4 - 16
11
• Steady-state errors using the Final Value Theorem (FVT):
kp 1
lim y(t) = lim sȳ(s) = lim s 2
t!1 s!0 s!0 s + 2s + 1 + kp s

kp kp
= lim 2 =
s!0 s + 2s + 1 + kp 1 + kp

Therefore, the steady-state error is

1
lim e(t) = 1 lim y(t) =
t!1 t!1 1 + kp

• Hence, in this example, increasing kp gives

✓ smaller steady-state errors
- but a larger and more oscillatory transient response

12
The root
5.5.3 locus concept
Proportional - Proportional
+ Derivative + Derivative control
(PD) Control

• In this case, K(s) = kp + kds K(s) = kp + kd s

r̄ + + ū ȳ
Σ kp Σ G(s)
− +

kd s

Typical result of increasing the gain kd , (when G(s) is itself stable):

• Typical result of increasing the gain kd, (for systems where G(s) is stable):
Increased Damping.
✓ Increased damping
- Greater sensitivity
Greater to noise
sensitivity to noise.

(It is usually better to measure the rate of change of the error directly
if possible – i.e. use velocity feedback)

1
Example: G(s) = 2
, 13 K(s) = kp + kd s
(s + 1)
Example 2:
• Consider the following critically damped 2nd-order system
1
G(s) =
(s + 1)2

Then
1
(kp + kd s)G(s) (kp + kd s) (s+1) 2
ȳ(s) = r̄(s) = 1 r̄(s)
1 + (kp + kd s)G(s) 1 + (kp + kd s) (s+1)2
(kp + kd s)
= 2 r̄(s)
s + (2 + kd )s + 1 + kp

The characteristic equation representing this system is

(s) = s2 + (2 + kd )s + 1 + kp = s2 + 2⇣!n s + !n2 = 0

14
• Solving for the damping factor
( 8 p
2⇣!n = 2 + kd < !n = 1 + kp
)
!n2 = 1 + kp :⇣ = p
2+kd
2 1+kp

kp = 3, kd = 0
x
Movement of closed-loop
poles for increasing kd !n

2 1

“root locus plot”

15
 The root locus procedure
• We will develop an orderly procedure of 6 steps that facilitates the rapid
sketching of the root locus

• Step 1: Locate the poles and zeros on the s-plane with selected symbols
(by convention, we use 'x' to denote poles and 'o' to denote zeros)

For example, consider the case 1 + kpG(s) = 0. We can write it in the form
Qm
i=1 (s zi )
1 + kp Qn =0
j=1 (s pj )

By rewriting the above equation as

n
Y m
Y
(s pj ) + k p (s zi ) = 0
j=1 i=1

it is easy to observe that for kp

= 0 the poles of the characteristic equation are
the same as the poles of system G(s)

16
Alternatively, the above equation, for kp ≠ 0, can be written as
n m
1 Y Y
(s pj ) + (s zi ) = 0
kp j=1 i=1

Therefore, as kp → ∞ the roots of the characteristic equation are the zeros of

system G(s)

Therefore,

‣ the locus of the roots of the characteristic equation 1 + kpG(s) = 0 begins at the
poles of G(s) and ends at the zeros of G(s) as kp increases from zero to infinity

‣ For most systems G(s) that we will encounter, several of the zeros of G(s) lie at
infinity in the s-plane.This is because most of our systems have more poles than
zeros. With n poles and m zeros and n > m, we have n − m branches of the root
locus approaching the n − m zeros at infinity

17
• Step 2: Locate the segments of the real axis that are root loci. The root locus
on the real axis always lies in a section of the real axis to the left of an odd
number of poles and zeros.

This can be observed by the phase requirement: ∠(kpG(s)) = 180o + k360o .

Therefore,
m
X n
X
\(s zi ) \(s pi ) = 180o + k360o
i=1 j=1

Only to the left of an odd number and zeros the equation above is satisfied on
the real axis!

18
Example 3:
• A single-loop feedback control system has the characteristic equation
2(s + 2)
1 + kp G(s) = 1 + kp =0
s(s + 4)
Step 1: The zeros and the poles of G(s) are shown in the figure below
zero

x o x
4 2 0

poles

Step 2: Due to the angle criterion, the locus begins at the pole and ends at the
zeros, and therefore the locus of roots appears as below

x o x
4 2 0

19
Observations:
‣ Since the loci begin at the poles and end at the zeros, the number of separate loci
is equal to the number of poles since the number of poles is greater than or equal
to the number of zeros

‣ The root loci must be symmetrical with respect to the horizontal real axis because
the complex roots must appear as pairs of complex conjugate roots

‣ Since the system in this example has two real poles and one real zero, the second
locus segment ends at a zero at negative infinity

‣ To evaluate the gain kp at a specific root location on the locus, we use the
magnitude criterion. For example, to have a root at s1 = − 1, we have

2|s1 + 2|
kp =1
|s1 ||s1 + 4|
2| 1 + 2|
kp =1
| 1|| 1 + 4|
2 3
kp = 1 ) kp =
3 2

20
• Step 3: The loci proceed to the zeros at infinity along asymptotes centered at
σA and with angles ϕA. When the number of finite zeros of G(s), m, is less than
the number of poles, n, by the number N = n − m, then N sections of loci
proceed to the zeros at infinity along asymptotes as kp approaches infinity.

The asymptotes are centered at a point (asymptote centroid) on the real axis
given by
Pn Pm
j=1 pj i=1 zi
A =
n m

The angle of the asymptotes with respect to the real axis is

(2k + 1)180o
A = , k = 0, 1, 2, . . . , n m 1
n m

(proof can be found in reference book, pp. 452-453)

21
Example 4:
• A single-loop feedback control system has the characteristic equation
1
1 + kp G(s) = 1 + kp =0
s(s + 2)

Step 3: Since n − m = 2, we expect two loci to end at zeros at infinity. The

asymptotes of the loci are located at a center
Pn Pm
j=1 pj i=1 zi 0 + ( 2)
A = = = 1
n m 2

and at angles
(2k + 1)180o (2k + 1)180o
A = = = (2k + 1)90o , k = 0, 1
n m 2

x 1
x
2 0

22
Example 5:
• A single-loop feedback control system has the characteristic equation
s+1
1 + kp G(s) = 1 + kp 2
=0
s(s + 2)(s + 4)

Step 1: Locate the poles and zeros:

xx x o x
4 2 1 0

Step 2: Locate the segments of the real axis that are root loci:

xx x o x
4 2 1 0

we would expect an asymptote centroid

around there

23
Step 3: Since n − m = 3, we expect 3 loci to end at zeros at infinity. The
asymptotes of the loci are located at a center
Pn Pm
j=1 pj i=1 zi 0 + ( 2) + 2( 4) ( 1) 9
A = = = = 3
n m 3 3
and at angles
(2k + 1)180o (2k + 1)180o
A = = = (2k + 1)60o , k = 0, 1, 2
n m 3

xx x o x
4 3 2 1 0

24
• Step 4:Determine the break-in and breakaway points on the real axis (if any).
For each s = σ on a real-axis segment of the root locus,
1
1 + kp G( ) = 0 ) kp =
G( )
Real-axis break-in and breakaway points are the real values of σ for which
dkp ( )
=0
d
breakaway
point

x 1
x
2 0

Alternatively, it can be shown that a breakaway or break-in point satisfy

X 1 X 1
=
i
zi j
pj

25
Proof: Consider the characteristic equation
Y (s)
1 + kp G(s) = 0 ) 1 + kp = 0 ) X(s) + kp Y (s) = 0
X(s)

For a small change in kp, we have

kp Y (s)
X(s) + (kp + kp )Y (s) = 0 ) 1 + =0
X(s) + kp Y (s)

Since the denominator is the original characteristic equation, a multiplicity r of

roots exists at a break-in or a breakaway point. Hence,
Y (s) Ci Ci
= r
=
X(s) + kp Y (s) (s si ) ( s)r

Therefore,
Ci kp ( s)r 1
1 + kp r
=0) =
( s) s Ci
As we allow δkp → 0, we obtain
dkp
=0
ds
26
Example 6:
• A single-loop feedback control system has the characteristic equation
(s 3)(s 5)
1 + kp G(s) = 1 + kp =0
(s + 1)(s + 2)
Root Locus ELEC304-Alper Erdogan 1
So, on the real-axis segments we have
( + 1)( + Break-away
2) 2 and
+3 + 2 Break-in Points
kp ( ) = = 2
( 3)( 5) 8 + 15

Taking the derivative with respect to σ,

Break-away
2
dkp ( ) 11 26 61 point Break-in
= =0 point
d ( 2 8 + 15) 2
(
1 = 1.45
)
2 = 3.82

27
• Step 5: Determine where the locus crosses the imaginary axis (if it does so).

1st way: Using the Routh-Hurwitz criterion:

‣ When we have jω-axis crossings, the Routh-table has all zeros at a row

‣ Find the value of kp for which a row of zeros is achieved in the Routh-table.

2nd way: Alternatively, If s = jω is a closed-loop pole on the imaginary axis,

then
1 + kp G(j!) = 0

The real and imaginary parts of the equation above provide us with 2 equations
with two unknowns kp and ω (i.e., the critical gain beyond which the system
goes unstable, and the oscillation frequency at the critical gain)

28
 • Find the K value for which a row of zeros is achieved
in the Routh-table.
Example 7 (1st way):
Example: Consider
• A single-loop feedback control system with closed-loop transfer function
K(sK(s
+ 3) + 3)
TT(s) = 44 3 3 2 2
(s) =
ss ++
7s7s+ 14s
+ 14s
+ (8 + (8 ++K)s
+ K)s 3K + 3K
The Routh table
The is given table
Routh by

The row s 1 is zero for K = 9.65. For this K, the previous row polynomial is
The row s1 is (90
zero K)s
for 2K+ = 9.65.
21K = 0
For this K, the
previous row polynomial is
whose roots are s = ± j1.59 2
(90 ° K)s + 21K = 80.35s2 + 202.7 = 0
29
Example 7 (2nd way):
• A single-loop feedback control system has the characteristic equation
s+3
1 + kp G(s) = 1 + kp =0
s(s + 1)(s + 2)(s + 4)

So, on the imaginary-axis we have Root Locus sketching rules

KG

kp G(j!) = 1) ! 4 + j7! 3 + 14!•2 Rule

j(kp7:+Imaginary
8)! 3K axis
=crossings
0
jω KG(j

Separating real and imaginary parts, j3

( Asymptote
s-plane −ω 4
system response
! 4 + 14! 2 3K = 0 j2
contains undamped S
terms at this point
7! 3 (kp + 8)! = 0 j1

Asymptote
In the second equation, we can discard X X X X σ
In t
t
-4 -3 -2 -1 0 1 2
the trivial solution ω = 0. It then yields
-j1

2 kp + 8
! = -j2
µ
7 Asymptote −
8

And substitute in the first to find kp -j3

Of the t
Figure by MIT OpenCourseWare. discard th
Thus,
30
• Step 6: Determine the angle of departure of the locus from a pole and the
angle of arrival of the locus at a zero, using the phase angle criterion.

The angle of locus departure from a pole is the difference between the net
angle due to all other poles and zeros and the criterion angle of
± (2k + 1)180o, and similarly for the locus angle of arrival at a zero.

breakaway −1 + j
x
point

x 1
x x x
2 0 2 0

x−1 − j

Due to the phase criterion, the tangents to the loci at the breakaway point are
equally spaced over 360°. Therefore,
‣ in left figure, the two loci at the breakaway point are spaced 180° apart

‣ in right figure, the four loci are spaced 90° apart

31
 5.5.5 Proportional + Integral + Derivative (PID) Control
Proportional + Integral + Derivative control
ki
• In this case, K(s) = kp + ki /s + kps
K(s) = kp + + kd s
s

ki /s

+
r̄ + + ū ȳ
Σ kp Σ G(s)
− +

kd s

• The continuous-time PID controller (in time domain) is

Z t
Characteristic equation: de(t)
u(t) = Kp e(t) + Ki e(⌧ )d⌧ + Kd
0 dt
• Taking Laplace transforms: 1 + G(s)(kp + kd s + ki /s) = 0
✓ ◆
Ki of both derivative and
• can potentially combine the advantages
ū(s) = Kp + + Kd s ē(s)
integral action: s
but can be difficult to “tune”.
There are many empirical rules for32
tuning PID controllers
 K(s) = kp +
s
Proportional + Integral control

• In this case, K(s) = kp + ki /s

r̄ + + ū ȳ
Σ kp Σ G(s)
− +

ki /s

• result of having integral control ki:

Example:
Typical
1
✓ (if stabilizing) G(s)
always results(s + 1)
, K(s) = kperror,
= in zero2steady-state + ki /s
in the presence of
constant disturbances and demands

33
Example 8:
• Consider the following critically damped 2nd-order system
1
G(s) =
(s + 1)2

Then
1
(kp + ki /s)G(s) (kp + ki /s) (s+1) 2
ȳ(s) = r̄(s) = 1 r̄(s)
1 + (kp + ki /s)G(s) 1 + (kp + ki /s) (s+1)2
(kp s + ki )
= 2
r̄(s)
s(s + 1) + kp s + ki

Applying the FVT:

(kp s + ki ) 1
lim y(t) = lim sȳ(s) = lim s 2
=1
t!1 s!0 s!0 s(s + 1) + kp s + ki s

Therefore, there is no steady-state error!

34
 P-Controller

• The obvious method - proportional control

• This method fails if, for instance, the error corresponds to more than a single
task or the system changes; hence, for the same error, different gains are
needed.
No control
Increasing K
KCp (Kc=0)
(K p = 0)

0
Time

• That's where the integral and derivative terms play their part.

35
 I-Controller

• An integral term increases action in relation not only to the error, but also the
time for which it has persisted. So, if applied control action is not enough to
bring the error to zero, this control action will be increased as time passes.

• A pure "I" controller could bring the error to zero, however, it would be both
slow reacting at the start, brutal, and slow to end, prompting overshoot and
oscillations.

Increasing t
decreasing
I Ki increasing Kp
Increasing KC
y
y

0 0
time time

Time Time
(a) (b)
• Alternative formulation: change the error in small persistent steps - over time
the steps accumulate and add up dependent on past errors;
this is the discrete-time equivalent to integration.

36
 D-Controller

• Aims at flattening the error trajectory into a horizontal line, damping the control
applied, and so reduces overshoot

• Ideal derivative control cannot (and must not) be realized in a PID-controller.

Practical systems always contain high frequency disturbances (e.g., white
noise), which are amplified by derivative control.

• Because of that a lag term is usually added to the derivation.

Kd s
Kd s
1 + Kd s/N

• Other practical modification is to derivate only the output (not the reference,
not the error signal)

37
PID-Controller

• Top-left: P-controller effect

(ID-controllers kept constant)

• Bottom-left: I-controller effect

(PD-controllers kept constant)

• Bottom-right: D-controller effect

(PI-controllers kept constant)

38
 Tuning PID controllers

• The structure of the used discrete PID algorithm must always be told together
with the tuning parameters kp, ki, kd.

• Controller design is based on heuristic design methods for selecting the

controller parameters.

• The principal design goal is stability: The system is stable when the closed
loop poles are on the left-half of s-plane

• Secondary criteria are, for example, rise, overshoot, settling time, and steady
state error. These can be analyzed graphically from impulse, step and ramp
responses of the close loop system
Effects of increasing a parameter independently
Parameter Rise time Overshoot Settling time Steady-state error Stability

Kp Decrease Increase Small change Decrease Degrade

Ki Decrease Increase Increase Eliminate Degrade

Kd Minor change Decrease Decrease No effect in theory Improve if small

39
PID-Controllers

40
 Learning outcomes

By the end of this lecture, you should be able to:

• Understand the concept of the root locus and its role in control system design

• Know how to obtain a root locus plot by sketching or using MATLAB

• Be familiar with the PID controller as a key element of many feedback systems

41