Frequency Response Methods : Stability Frequency Response Methods : Stability

EE
• Why study stability in terms of frequency response?

EE
• Nyquist stability criterion.
Examples on using Nyquist stability criterion.

E1 E1
• Contour mapping in the complex plane.

• Gain margin.
• Cauchy’s theorems. → amount of additional gain to make the system

51 51
unstable.
– Cauchy’s Residue Theorem.
– Cauchy’s Principle of the Argument.

• Phase margin.
• Why did the mathematician name his dog Cauchy? → amount of additional phase shift to make the system
unstable.
Frequency Response c
2003 M.C. Ramos Frequency Response c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute

Stability in the Frequency Domain Stability in the Frequency Domain

EE
• Why study stability again? Enough already?
– roots of the characteristic equation.
– Routh-Hurwitz stability test.
EE
• Try an alternative method based on frequency response.
Can we extract stability information without using
Bode plots?
– root locus.

E1 E1
Save time and we do not have to worry about the
accuracy of our identified transfer function.

51 51
• We have studied Bode plots.
• This is where the Nyquist stability criterion comes in.
We can infer stability by identifying the transfer Use the Nyquist diagram to determine stability.
function given the frequency response.

frequency response ⇒ transfer function ⇒ stability • But before going into Nyquist stuff, let us look at some
mental appetizers. Contour mapping.
Frequency Response c
2003 M.C. Ramos Frequency Response c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute
 Contour Mapping in the Complex Plane Contour Mapping in the Complex Plane

EE
• Mapping of a point.

w = f (z) where
j
y

x
f (z) =
1
z
v

0.5
w-plane

EE
• Mapping of a region : rectangle 7→ disk.

y v
w-plane

E1 E1
z = x + jy and 1 −0.5j f (z) =
1
w z
w = u + jv. z-plane x u
c

1
2c

51 51
z-plane
• Mapping of a contour : line 7→ circle.
y v
w-plane
1
f (z) =
x
z
u
• Complex analysis :
c
Field of mathematics concerned with the study of
1
z-plane 2c complex variables and complex functions.
Complex Analysis c
2003 M.C. Ramos Complex Analysis c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute

Contour Mapping in the Complex Plane Cauchy’s Theorems

EE
• Mapping of closed contours.
A closed contour is a contour which has the same initial
and final points.
EE
• Cauchy, Augustin Louis.
– Born: 21 Aug 1789 in Paris, France.
– Died: 23 May 1857 in Sceaux (near Paris), France.

E1
A closed contour on the z-plane maps to a closed
contour on the w-plane.

E1
• Numerous terms in mathematics bear Cauchy’s name.

51 51
– Cauchy integral theorem.
• Conformal mapping. – Cauchy-Kovalevskaya existence theorem.
Mathematical technique used to map (or convert) a – Cauchy-Riemann equations and Cauchy sequences.
mathematical problem and its solution to another.

• Cauchy produced 789 mathematics papers.

Complex Analysis c
2003 M.C. Ramos Cauchy’s Theorems c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute
 Cauchy’s Theorems Cauchy’s Theorems

EE
• Cauchy’s Residue Theorem.
I
f (z)dz = j2π
n
X
Res(zi)
EE
• What can we do with Cauchy’s Residue Theorem?

E1 E1
Γ i=1
• Determine if there are poles in a certain region.
where the zi’s are poles of f (z) enclosed by Γ. How about the right-half plane?
The residue Res(zi) of pole zi of multiplicity mi is

51 51
1 dmi−1
Res(zi) = lim [(z − zi)mi f (z)] • Fundamental idea.
z→zi (mi − 1)! dz mi−1
– choose a contour enclosing the right-half plane.
– perform the contour integration.
• There is a relationship between the value of the contour – does this reveal poles within the contour?
integral and the poles that reside within the contour. – deduce stability.
Cauchy’s Theorems c
2003 M.C. Ramos Cauchy’s Theorems c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute

Cauchy’s Theorems Cauchy’s Theorems

EE
• Can we derive something easier to use?
⇒ Principle of the Argument.

EE
• Using the properties of the logarithm,

f (s) =
d
[log g(s)] =
α
X mi
−
β
X ni

E1 E1
ds s − zi s − pi
i=1 i=1
• Let us move to the s-plane. Let use f (s) such that
g ′(s) d

51 51
f (s) = = [log g(s)] • From Cauchy’s Residue Theorem,
g(s) ds
α β
1
I
where X X
Qα mi
f (s)ds = mi − ni = Z − P
i=1(s − zi) j2π Γ
g(s) = i=1 i=1
Qβ ni
i=1(s − pi) where Z and P are the total number of zeros and poles
enclosed by the contour Γ, respectively.
Cauchy’s Theorems c
2003 M.C. Ramos Cauchy’s Theorems c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute
 Cauchy’s Theorems Cauchy’s Theorems

EE
• Let us now evaluate the contour integral.
I
f (s)ds =
I
d
Γ ds
[log g(s)]ds =
I
d[log g(s)]
EE
• The integral reduces to
1
I

j2π Γ
d[log g(s)] =
1
2π
[∠g(s2) − ∠g(s1)]

E1 E1
Γ Γ

• Using log g(s) = log |g(s)| + j · ∠g(s), we get • Since the contour is closed, the net change angle will be

51 51
1
I
1
s2
 a multiple of 2π. Thus, we can write
d[log g(s)] = {log |g(s)| + j · ∠g(s)} 1
I
1
I
j2π Γ j2π s1 f (s)ds = d[log g(s)] = N
j2π Γ j2π Γ
where N is number of encirclements of the origin by
g(s) as s traverses the contour Γ once.
• Contour Γ is closed ⇒ log |g(s1)| = log |g(s2)|.
Cauchy’s Theorems c
2003 M.C. Ramos Cauchy’s Theorems c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute

Cauchy’s Theorems Cauchy’s Theorems

EE
• A more useful form of the Principle of the Argument is

Given a function g(s) and a closed contour Γ such that

EE
• Examples.
2
s-plane 2
s-plane

imaginary axis
imaginary axis
1.5 1.5

E1 E1
g(s) is analytic on and within Γ (except at the finite 1

0.5
1

0.5

poles of g(s) within Γ) and which does not vanish on Γ,

0 0

−0.5
→ {s − 2.5} → −0.5

then
−1 −1

−1.5 −1.5

−2 −2
0 0.5 1 1.5 2 2.5 3 3.5 4 −2.5 −2 −1.5 −1 −0.5 0 0.5 1 1.5

N = Z − P real axis real axis

51 51
s-plane s-plane
where N is number of encirclements of the origin by 2 1

imaginary axis
imaginary axis
0.8
1.5

0.6

g(s) as s traverses the contour Γ in the positive 1

0.5

1
 0.4

0.2

direction, and Z and P are the total number of zeros 0

→ → 0

−0.2

s − 1.5
−0.5

−0.4

and poles enclosed by the contour Γ, respectively.

−1
−0.6

−1.5
−0.8

−2 −1
0 0.5 1 1.5 2 2.5 3 3.5 4 −2 −1.5 −1 −0.5 0 0.5 1

real axis real axis

Cauchy’s Theorems c
2003 M.C. Ramos Cauchy’s Theorems c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute
 Nyquist Stability Criterion Nyquist Stability Criterion

EE
• Adapt Cauchy’s Principle of the Argument

EE
• We usually work with a closed-loop system with an
open-loop transfer function G(s).
⇒ investigate the characteristic equation.

E1
• Stability : are there poles in the right half plane?
⇒ we choose a contour ΓRHP enclosing the RHP.

E1 R
+

-
E
G Y Y (s)
R(s)
=
G(s)
1 + G(s)

51 51
s = +j∞ s = +j∞ s = +j∞
imag imag imag
infinity infinity infinity

ω=0 ω=0 ω=0 • Closed-loop stability depends on the location of the

real real real
closed-loop poles (= zeros of {1 + G(s)}).
s-plane s-plane s-plane
ΓRHP ΓRHP ΓRHP We want to know if there are closed-loop poles (or zeros
s = −j∞ s = −j∞ s = −j∞
of {1 + G(s)}) in the RHP?
Nyquist Stability Criterion c
2003 M.C. Ramos Nyquist Stability Criterion c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute

Nyquist Stability Criterion Nyquist Stability Criterion

EE
• Then, if we set g(s) = 1 + G(s) and use the contour
ΓRHP enclosing the RHP, we can find out something
about the number of zeros and poles in the RHP by
EE
• For the closed-loop system to be stable, there should be
no zeros of g(s) = 1 + G(s) in the RHP (Z = 0).
Thus,

E1 E1
applying the Principle of the Argument (N = Z − P ).
N = Z − P ⇒ N = −P

• Also, note that the poles of g(s) = 1 + G(s) are

exactly the poles of G(s).

encirclements about −1 + j0.

Nyquist Stability Criterion
51
• Instead of looking at g(s) = 1 + G(s) and the
encirclements about the origin, look at G(s) and the

c
2003 M.C. Ramos
• Nyquist Stability Criterion.

the number of poles of G(s) in the RHP.

Nyquist Stability Criterion
51
A feedback control system is stable if and only if, for the
clockwise contour enclosing the RHP, the number of
counterclockwise encirclements of −1 + j0 is equal to

c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute
 Nyquist Stability Criterion Summary

EE
• Example. Determine the stability of a unity gain
feedback closed-loop system with an open-loop TF
Nyquist plot

EE
• Stability in realm of frequency response.

E1
1 1

E1
• Cauchy’s residue theorem.

imaginary axis
G(s) =
s − 0.5
0

G(s) has one pole in

51 51
the RHP ⇒ P = 1.
−1 • Nyquist stability criterion.
−2 −1 0
real axis

number of encirclements : one CCW ⇒ N = − 1.

of −1 + j0 • Why did the mathematician name his dog Cauchy?
Because he left a residue at every pole.
N = Z − P ⇒ Z = 0 ⇒ closed-loop system
is stable.
Nyquist Stability Criterion c
2003 M.C. Ramos Nyquist Stability Criterion c
2003 M.C. Ramos
EEE 151 UP EEE Institute EEE 151 UP EEE Institute