The Stability Of Linear Feedback Systems

In the S-plane, if a pole is located at the left side of the plane then the
system is stable.
𝑆 = −𝛼 ± 𝑗𝜔𝑑 If −α is negative then the system is stable.

𝑆+2
i.e. 𝐺(𝑠) = 𝑆 3 +2𝑆 2 +50𝑆+1

If one pole is at the unstable region, then the system is unstable.

𝑆 3 + 2𝑆 2 + 50𝑆 + 1 = 0 is called the characteristic equation.

Note: For system stability the following conditions must be satisfied (but not
sufficient):
1) All coefficients of the characteristic equation must have the same sign.
2) None of the polynomial coefficients be zero.

Not sufficient means we immediately know that the system is unstable if

any of these conditions is not satisfied.

Consider:
𝑁(𝑠) 𝑁(𝑠)
𝑇(𝑠) = 𝑄(𝑠) = 𝑎 𝑛 +𝑎 𝑛−1 +𝑎 𝑛−2 +.......+𝑎 𝑆+𝑎
𝑛𝑆 𝑛−1 𝑆 𝑛−2 𝑆 1 0

48
Notes:
1) The # of roots of Q(s) which are positive (unstable) real roots = the # of
sign change in the first column.

2) A stable system should have no sign change.

Case (1): No zeros in the first column.

𝑆+0.532
Ex: 𝑇(𝑠) = 𝑆 3 +𝑆 2 +2𝑆+24

S3 1 2
49
 S2 1 24
S b1 b2
1 C1

Two sign change in the first column ---> there are two roots in the positive
s-plane (unstable)

Case (2) : some zeros in the first column with at least one non-zero
element in the same row containing the zero.

𝑆+0.356
EX: 𝑇(𝑠) = 𝑆 5 +2𝑆 4 +2𝑆 3 +4𝑆 2 +11𝑆+10

50
Since b1 is zero, we assume it a small number (𝜖 ) which has the sign of the
element located directly above it.
The first column has no sign change with a zero, then the system is
marginally stable.

Case (3): All the elements in a row are zero.

EX: 𝑄(𝑠) = 𝑆 3 + 2𝑆 2 + 4𝑆 + 𝐾

A- For a stable system, we require that:

1) 𝐾>0
8−𝐾
2) >0
2
⇒ 0<𝐾<8

51
B- When K=8 we obtain a row with all zeros, so we define the auxiliary
polynomial U(s) which is the equation of the row preceding the row of
zeros.
𝑈(𝑠) = 2𝑆 2 + 8𝑆 0
𝑑𝑈(𝑠)
= 4𝑆 + 0 , This equation replaces the all zero row.
𝑑𝑠

This occurs when factors are symmetrically located about the origin.

Notice that the auxiliary polynomial has no roots on the right half plane
(which means stable!!). However, for the auxiliary, it is necessary to factor
its equation to determine its poles.
2𝑆 2 + 8 = 0
𝑆 = ±𝑗2

52
The general form

𝑌(𝑠) 𝐾𝐺(𝑠)
=
𝑅(𝑠) 1 + 𝐾𝐺(𝑠)
𝛥(𝑠) = 1 + 𝐾𝐺(𝑠) = 0
𝐾𝐺(𝑠) = −1 ± 𝑗0
|𝐾𝐺(𝑠)| = 1
𝑎𝑛𝑔𝑙𝑒(𝐾𝐺(𝑠)) = ±180 + 𝑛 360𝑜 , 𝑛 = 0, ±1, ±2, . . . . . ..

For the closed loop,

1
𝛥(𝑠) = 1 + 𝐾𝐺(𝑠) = 1 + 𝐾 =0
𝑆(𝑆 + 2)
𝛥(𝑠) = 𝑆 2 + 2𝑆 + 𝐾 = 0
𝑆1,2 = −1 ± 𝑗√𝐾 − 1

Notes:
If 𝐾 = 0 ⇒ 𝑆1,2 = 0, −2 open loop poles
If 𝐾 = 5 ⇒ 𝑆1,2 = −1 ± 𝑗2
If 𝐾 = 1 ⇒ 𝑆1,2 = −1
If 𝐾 = 0.5 ⇒ 𝑆1,2 = −1.7, −0.3

53
If we are looking for the K value (knowing the root value and S is given),
𝐾 = |𝑆||𝑆 + 2|
@𝑆 = −1 + 𝑗2 ⇒ 𝐾 = | − 1 + 𝑗2|| − 1 + 𝑗2 + 2| = 5

Procedure for the root locus:

1) Write the characteristic equation as Δ(s) = 1 + KP(s)

54
EX:

For this example K=10,

1
𝛥(𝑠) = 1 + 10 =0
𝑆(𝑠 + 𝑎)
𝑆(𝑆 + 𝑎) + 10 = 0
𝑆 2 + 𝑎𝑆 + 10 = 0
𝑎𝑆 = −(10 + 𝑆 2 )
𝑎𝑆
= −1
𝑆 2 + 10
⇒1+𝑎
𝑆 = 1 + 𝐾𝑃(𝑠)
2
𝑆 + 10
2) Factor P(s) for poles and zeros
∏𝑛𝑧
𝑖=1 (𝑆+𝑍𝑖 )
𝛥(𝑠) = 1 + 𝐾 𝑛
𝑝
∏𝑗=1 (𝑆+𝑃𝑗 )
3) Locate on the S-plane the poles and zeros.

Important Note:
Locus of the roots (poles) of (1 + KP(s) = 0) begins at the poles of P(s) and
ends at the zero of P(s) as K increases (zero at ∞)

55
4) 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.

EX: A single loop feedback control system processes the characteristic

equation
2(𝑆 + 2)
1+𝐾 =0
𝑆(𝑆 + 4)
Here we have,
𝑍1 = −2
𝑃1 = 0
𝑃2 = −4

5) Determine the number of separate loci( SL=# of poles).

6) Root locus must be symmetrical around the real ( X-axis).
7) The loci proceeds to the zero at ∞ along asymptotes centered at (σA )
with angle (φA ) if 𝑁 = 𝑛𝑝 − 𝑛𝑧 , then an N sections of locies proceed to
the zero at ∞ along an asymptote as K → ∞

𝑛𝑝 𝑛𝑧
∑𝑖=1 (𝑝𝑖 )−∑𝑗=1 (𝑍𝑗 )
Where 𝜎𝐴 = 𝑛𝑝 −𝑛𝑧
𝑛𝑝 : # of poles
𝑛𝑧 : # of zeros

2𝑞 + 1
𝜑𝐴 = 180𝑜 , 𝑞 = 0,1,2, . . . . . , (𝑛𝑝 − 𝑛𝑧 − 1)
𝑛𝑝 − 𝑛𝑧

56
 1
EX: 𝛥(𝑠) = 1 + 𝐾 𝑆(𝑆+2) = 0

𝑛𝑧 = 0
𝑛𝑝 = 2 , 𝑃 = 0, −2

(0 + −2) − 0
𝜎𝐴 = = −1
2−0
2(0) + 1
𝜑0 = 180 = 90𝑜
2
2(1) + 1
𝜑1 = 180 = 270𝑜
2
* Determine the Breakaway point(B.P) where the root locus leaves the X-
axis.

For example, consider a unity feedback closed loop system with an open
𝐾
loop transfer function 𝐺(𝑠) = (𝑆+2)(𝑆+4)
1
𝛥(𝑠) = 1 + 𝐾 =0
(𝑆 + 2)(𝑆 + 4)
𝑃(𝑠) = 𝐾 = −(𝑆 + 2)(𝑆 + 4)
𝑑𝑃(𝑠)
𝐵. 𝑃 = 𝑟𝑜𝑜𝑡𝑠( = 0) = 𝑟𝑜𝑜𝑡𝑠(2𝑆 + 6 = 0) ⇒ 𝑆 = −3
𝑑𝑆
* Determine the angle of departure from the complex poles:

For a point in the S-plane to be on the root locus:

∑ (all the angles from finite zeros)
− ∑ (all the angles from finite poles)
57
= ±180 + 𝑟360
𝐾
EX: 𝛥(𝑠) = 1 + 𝑆 4 +12𝑆 3 +64𝑆 2 +128𝑆 = 0
𝑛𝑧 = 0
𝑛𝑝 = 4
𝑃1 = 0
𝑃2 = −4
𝑃3,4 = −4 ± 𝑗4
0 − 4 − 4 + 𝑗4 − 𝑗4 − 4
𝜎𝐴 = = −3
4−0
𝑞 = 0,1,2,3
1
𝜑0 = 180 = 45𝑜
4
𝜑1 = 135𝑜
𝜑2 = 225𝑜
𝜑3 = 315𝑜
𝐵. 𝑃 ⇒ 𝑃(𝑠) = −(𝑆 4 + 12𝑆 3 + 64𝑆 2 + 128𝑆)
𝑑𝑃(𝑠)
= 0 = −(4𝑆 3 + 36𝑆 2 + 128𝑆 + 128)
𝑑𝑠
𝑆1,2 = −3.71 ± 𝑗2.5
𝑆3 = −1.576 we take this value because it’s located on the X-axis as
the definition.

Angle of departure for complex poles:

0 − (𝜃 + 90 + 90 + 135) = 180
𝜃 = −135𝑜

58
59
60
61
62
Parameter Design by the Root Locus Method (Two Parameters)
EX:

Find K1 , K 2 such that the following are satisfied:

1) 𝑒𝑆𝑆 ≤ 35% of a ramp input slope.
1
2) 𝜁≥
√2
3) 𝑡𝑠 ≤ 3 𝑠𝑒𝑐.
Sol:
𝐺(𝑠) 𝐾 1
Let 𝐺1 (𝑠) = 1+𝐺(𝑠)𝐻 = 2
1 (𝑠) 𝑆 +2𝑆+𝐾 1 𝐾2 𝑆
𝑒𝑆𝑆 = 𝑙𝑖𝑚𝑆𝐸(𝑠)
𝑆→0
1 𝐴
𝑒𝑆𝑆 = 𝑙𝑖𝑚𝑆[(1 − 𝑇(𝑠))𝑅(𝑠)] = 𝑙𝑖𝑚[𝑆 ]
𝑆→0 𝑆→0 1 + 𝐺1 (𝑠) 𝑆 2
𝐴
= 𝑙𝑖𝑚[ ]
𝑆→0 𝑆 + 𝑆𝐺1 (𝑠)
𝐴
𝑒𝑆𝑆 = 𝑙𝑖𝑚
𝑆→0 𝐾1
𝑆+𝑆 2
𝑆 + 2𝑆 + 𝐾1 𝐾2 𝑆
𝑒𝑆𝑆 2 + 𝐾1 𝐾2
= ≤ 0.35
𝐴 𝐾1

63
For 𝑡𝑠≤ 3 𝑠𝑒𝑐,
4
𝑇𝑆 = 4𝜏 = ≤3
𝜁𝜔𝑛
‫؞‬ 4
𝜁𝜔𝑛 ≥ 3 = 1.33
Keep in mind: 𝑆 1,2 = −𝛼 ± 𝑗𝜔𝑑 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 √1 − 𝜁 2

Given that 𝜁 = 0.707

4
𝜁𝜔𝑛 ≥ 3
⇒ 𝜔𝑛 = 1.88
𝑆1,2 = −1.33 ± 𝑗1.88√1 − ( 1 )
2
⇒
√2

Now,
𝐺1 (𝑠)
𝑇(𝑠) =
1 + 𝐺1 (𝑠)
𝐾1
1+ 2 =0
𝑆 + 2𝑆 + 𝐾1 𝐾2 𝑆
𝛥(𝑠) = 𝑆 2 + 2𝑆 + 𝐾1 𝐾2 𝑆 + 𝐾1 = 0
𝛥(𝑠) = 𝑆 2 + 2𝑆 + 𝛽𝑆 + 𝛼 = 0
Since we have two parameters that we have to find, we can do:
𝛽 = 0,
1)
⇒ 𝛥(𝑠) = 𝑆 2 + 2𝑆 + 𝛼 = 0
64
 1
1 + 𝛼 𝑆(𝑆+2) = 0
2) Choose any value of 𝛼 on the
root locus (such that = 20 )

𝑆1,2 = −1 ± 𝑗4.36

3) Back to 𝛽
𝑆
1+𝛽 2
𝑆 + 2𝑆 + 20

𝛽 = 𝐾1 𝐾2
4.3 = 20𝐾2
𝐾2 = 0.215

65
PID Controller
P: Proportional (Kp)
1
I : Integrator (K I )
S
D: Differentiator (K D S)

P controller:

𝑉𝑜𝑢𝑡 = −𝑅𝑓 = 𝐾 = 𝐺 (𝑠)

𝑝 𝑐
𝑉𝑖𝑛 𝑅𝑖𝑛
Standard closed second order system:
𝑌 𝜔𝑛2
=
𝑅 𝑆 2 + 2𝜁𝜔𝑛 𝑆 + 𝜔𝑛2
Standard open second order system:
𝑌 𝜔𝑛2
= = 𝐺(𝑠)
𝑅 𝑆(𝑆 + 2𝜁𝜔𝑛 )

66
PD controller:
𝐺𝑐 (𝑠) = 𝐾𝑝 + 𝐾𝐷 𝑆

𝑉𝑜 𝑅2
𝐺𝑐 (𝑠) = = −[ + 𝑅2 𝐶𝑆]
𝑉𝑖𝑛 𝑅1
−𝑅2
⇒ 𝐾𝑝 = , 𝐾𝐷 = −𝑅2 𝐶
𝑅1
The open second order system transfer function will be:
𝑌 𝜔𝑛2 (𝐾𝑝 + 𝐾𝐷 𝑆)
=
𝑅 𝑆(𝑆 + 2𝜁𝜔𝑛 )
⇒Here a new zero has been added to the system.

PI controller:
𝐾𝐼
𝐺𝑐 (𝑠) = 𝐾𝑝 +
𝑆

𝑉𝑜 𝑅2 1
𝐺𝑐 (𝑠) = = −[ + ]
𝑉𝑖𝑛 𝑅1 𝑅1 𝐶𝑆

67
 −𝑅2 −1
⇒ 𝐾𝑝 = , 𝐾𝐼 =
𝑅1 𝑅1 𝐶
The open second order system transfer function will be:
𝑌 𝑆𝐾𝑝 + 𝐾𝐼 𝜔𝑛2
=
𝑅 𝑆 𝑆(𝑆 + 2𝜁𝜔𝑛 )
⇒Here one zero and one pole have been added.
PID controller:
𝐾𝐼 𝐾𝐷 𝑆 2 + 𝐾𝑝 𝑆 + 𝐾𝐼
𝐺𝑐 (𝑠) = 𝐾𝑝 + + 𝐾𝐷 𝑆 =
𝑆 𝑆
⇒Here two zeros and one pole have been added.

68