Feedback Control Systems

(KON 313E)
Week 6

Prof.Dr. Volkan Sezer

Control and Automation Engineering Department

Stability Analysis
Stability is the most important system property.

If a system is unstable, its transient state behaviour and steady state error cannot be mentioned.

The response of a system consists of a forced and natural solution.

𝐶(𝑡) = 𝐶𝑓𝑜𝑟𝑐𝑒𝑑(𝑡) +𝐶𝑛𝑎𝑡𝑢𝑟𝑎𝑙(𝑡)

The following definitions of stability can be made
Alternatively, it is possible to make the following
according to the natural response of a system.
definitions of stability by looking at the total
When 𝑡 → If nat. res. approaches zero , system is STABLE
response.
• If it gives a bounded output for every bounded
When 𝑡 → If nat. res. approaches infinity , system is UNSTABLE input, it is stable in the sense of bounded input-
bounded output (BIBO).
When 𝑡 → If nat. res. Remains constant, system is MARGINALLY STABLE • If any limited bounded input gives unbounded
output, the system is unstable in the BIBO sense.
Stability Relationship with the Location of the Poles
Stability Relationship with the Location of the Poles
• The stability of a system can be determined by the location of the poles.
Im
x
x Finding the roots of the denominator of
If the poles are in the left half plane, the x the transfer function of a system is not
Re
system is Stable. x easy for high order systems!
x
Therefore, Routh Hurwitz criteria are
Im applied for stability analysis!
x x
Marginally Stable systems have poles with only imaginary
x
axis poles of multiplicity 1 and poles in the left half-plane Re
x
x
Im Im Muliplicity greater than 1

x x
In Unstable Systems, at least one pole is in the x
x x
right half-plane and/or poles of multiplicity x Re Re
x
greater than 1 on the imaginary axis x
Stability Relationship with the Location of the Poles

Is the closed loop transfer function stable?

Routh-Hurwitz Criteria

1) The characteristic polynomial (denominator of the 2) All coefficients of the polynomial must be positive. (If all are
transfer function) is written as follows. negative, each side is converted to positive by multiplying by -1.) If
all are not positive, the system is unstable. If all are positive,
𝑎𝑛 𝑠 𝑛 + 𝑎(𝑛−1) 𝑠 𝑛−1 + 𝑎(𝑛−2) 𝑠 𝑛−2 +… 𝑎1 𝑠 + 𝑎0 = 0 proceed to step 3.

3) If any power of ‘s’ is missing in the polynomial, the 4) If all powers are present and all coefficients are
system is unstable or marginally stable. positive, the Routh table is created.
Routh Tablosu
𝑎𝑛 𝑠 𝑛 + 𝑎(𝑛−1) 𝑠 𝑛−1 + 𝑎(𝑛−2) 𝑠 𝑛−2 +… 𝑎1 𝑠 + 𝑎0

𝑎𝑛−1 𝑎𝑛−6 − 𝑎𝑛 𝑎𝑛−7… 𝑏1 𝑎𝑛−7 − 𝑎𝑛−1 𝑏4

𝑏3 = 𝑐3 = …
𝑎𝑛−1 𝑏1

The number of sign changes in the 1st column gives the number of poles in the right half-plane!
 7
TF= 3 2 Is this transfer function stable?
𝑠 +6𝑠 +12𝑠+8
 7
TF= 3 2 Is this transfer function stable?
2𝑠 +4𝑠 +4𝑠+12
Is the closed loop transfer function stable?
Special Case-1

If the first element of any row is '0' and there are element(s) other than
0 in the row, the operations are continued by replacing the first
element of the row with a very small positive value '∈'.
 10
TF= 5 4 3 2 Is this transfer function stable?
𝑠 +2𝑠 +3𝑠 +6𝑠 +5𝑠+3

Changed sign 2 times, unstable!

2 poles in the right half-plane,

3 pole in the left half plane
Special Case-2
If all elements of a row are '0', the auxiliary polynomial is taken from the upper neighbour
row. The coefficients of the derivative of the auxiliary polynomial are used.

The roots coming from the auxiliary polynomial must be symmetric with
respect to the origin.

• Considering this property, the location of the poles from auxiliary

polynomial can be understood. (The roots of the auxiliary polynomial are
some of the roots of the transfer function)
• If there is no sign change on the column under the auxiliary polynomial in
the Routh table, all of the roots are on the jw axis. If a sign change is
present, number of right half plane poles is equal to the number of sign
change, rest is on jw axis.
 𝑠+1
TF= 5 4 Is this transfer function stable?
𝑠 +2𝑠 +24𝑠 3 +48𝑠 2 +25𝑠+50
All 3 elements are '0'. Therefore, let's take the auxiliary
polynomial from the upper row.

Derivative
 𝑠+1
TF= 5 4 Is this transfer function stable?
𝑠 +2𝑠 +24𝑠 3 +48𝑠 2 +25𝑠+50
Auxiliary Polynomial Previous Part Total
There is no sign change. There is no sign 1 pole is in the left half
So there is no pole in the change, plane.
right half plane. So there is 1 pole 4 poles on jw axis.
If not in the right, not in the in the left half
left. (Symmetry feature) plane.
So there are 4 poles on the
jw axis.

Do auxiliary polynomial’s poles have multiplicity greater than 1 ?

To be multiplicity greater than 1, it must in the following format:

(𝑠 2 + 𝑎)2 = 𝑠 4 + 2𝑎 𝑠 2 + 𝑎2

2𝑠 4 + 48𝑠 2 + 50 = 0 does not fit in the given format.

It is clear that the auxiliary polynomial’s poles are not multiplicity

greater than 1 . Therefore the current system is 'marginally stable'.
Stable System Design with the Help of Routh-Hurwitz
R(s) E(s) U(s) 𝑠+1 C(s)
K Design the K to stabilise the closed loop system.
+ 𝑠(𝑠 − 1)(𝑠 + 6)
_

𝐾(𝑠 + 1)
𝑠 𝑠 − 1 (𝑠 + 6) 𝐾(𝑠 + 1)
𝐶𝐿𝑇𝐹 = = 3
𝐾(𝑠 + 1) 𝑠 + 5𝑠 2 + 𝑠(𝐾 − 6) + 𝐾
1+
𝑠 𝑠 − 1 (𝑠 + 6)

𝑠 3 + 5𝑠 2 + 𝑠 𝐾 − 6 + 𝐾

Result: K>7.5 makes closed loop system stable

K=7.5 condition should be analysed separately.
Stable System Design with the Help of Routh-Hurwitz
R(s) E(s) U(s) 𝑠+1 C(s)
K Design the K to stabilise the closed loop system.
+ 𝑠(𝑠 − 1)(𝑠 + 6)
_

5 ∗ 1.5 − 1 ∗ 7.5 The whole row becomes '0'. Let's use the auxiliary polynomial
𝑏1 = =0
5
5𝑠 2 + 7.5 = 0
K=7.5 Derivative
10𝑠 = 0

Auxiliary Polynomial Previous Part Total

No sign change. 1 pole in the left half plane 1 pole in the left half plane
No pole on the right. 2 poles on the jw axis
If not on the right, not on
the left (symmetry rule)
2 poles on the jw axis

Result: Closed loop system is stable for K>7.5 . For K=7.5 it is marginally stable!
Stable System Design with the Help of Routh-Hurwitz
Design the K to stabilise the closed loop system.

𝐾
𝑠 𝑠 + 7 (𝑠 + 11) 𝐾
𝐶𝐿𝑇𝐹 = =
𝐾 𝑠 3 + 18𝑠 2 + 77𝑠 + 𝐾
1+
𝑠 𝑠 + 7 (𝑠 + 11)
𝒔𝟑 + 𝟏𝟖𝒔𝟐 + 𝟕𝟕𝒔 + 𝑲

Result: 0<K<1386 makes closed loop system stable

K=1386 condition should be analysed separately.
Stable System Design with the Help of Routh-Hurwitz
Design the K to stabilise the closed loop system.

18 ∗ 77 − 1386 ∗ 1 The whole row becomes '0'. Let's use the

𝑏1 = = 0 auxiliary polynomial
18

18𝑠 2 + 1386 = 0
Derivative
36𝑠 = 0

Auxiliary Polynomial Previous Part Total

No sign change. No sign change 1 pole in the left half
No pole on the right. 1 pole in the left plane
If not on the right, not on half plane 2 poles on the jw axis
the left (symmetry rule)
2 poles on the jw axis

Result: Closed loop system is stable for 0<K<1386 . For K=1386 it is marginally stable!