Outline

Asymptotic stability.
Stability of Digital Input-output stability.
Control Systems Internal stability.
Stability conditions.
M. Sami Fadali
Routh-Hurwitz criterion.
Professor of Electrical Engineering
Jury test.
UNR

1 2

Bounded-Input-Bounded-Output
Asymptotic Stability
(BIBO) Stability
Response due to any initial conditions The response due to any bounded input
decays to zero asymptotically in the steady
state
Lim

Marginal Stability: response due to any remains bounded.
initial conditions remains bounded but does
not decay to zero.

3 4
 Stable Z-Domain Pole Time Sequence
Locations Z
Sampled exponential and its z-transform
with real or complex Bounded sequence for poles in the closed unit
Z disc (i.e. on or inside the unit circle).
Sequence decays exponentially for poles in the
open unit disc (i.e. inside the unit circle).
Unbounded sequence for repeated poles on the
_| unit circle.

For real time sequences poles and partial fraction
coefficients are either real or complex conjugate
pairs.
5 6

Theorem: Asymptotic Stability Proof

Asymptotic Stability: response to any ICs Response of the system due to ICs 0 , 1 ,,
decays to zero asymptotically.
Thm. In the absence of pole-zero cancellation,
a LTI digital system is (i) asymptotically 1. Assume no pole-zero cancellation.
stable if its transfer function poles are in the 2. denominator= z-transfer function denominator.
open unit disc and (ii) marginally stable if the 3. Poles of = poles of transfer function.
poles are in the closed unit disc with no
4. due to ICs: bounded for system poles in the
repeated poles on the unit circle.
closed unit disc unit (no repeated poles on the unit
LTI Model: circle) & decays exponentially for system poles in
1 1
the open unit disc (inside the unit circle).
1 1
0,1,2, 5. No cancellation so all poles are considered.
7 8
 Example Solution ( )
Determine the asymptotic stability of the
following systems:
Pole outside the unit circle unstable

All poles inside the unit circle

Use Theorem 1 (a) and (b) without pole-zero asymptotically stable.
cancellation.
Ignore zeros, (do not affect response due to ICs)
9 10

Solution BIBO Stability

All poles inside the unit circle Is system BIBO stable if its impulse response is
asymptotically stable. bounded? NO
Counterexample: bounded & strictly positive impulse
response
One pole on the unit circle marginally Bounded input:
stable: not asymptotically stable.
Unbounded output:

11 12
 Theorem 4.2:BIBO Stability Proof of Necessity (Only if)

A discrete-time linear system is BIBO

stable if and only if its impulse response Assume the system is BIBO stable but the
sequence is absolutely summable i.e. impulse response is not absolutely summable.

Input

Contradiction: unbounded output with a bounded

input.
13 14

Proof of Sufficiency (If) Theorem 4.3: BIBO Stability

Assume an absolutely summable impulse
response and show that the system is BIBO A discrete-time linear system is BIBO
stable. stable if and only if the poles of its transfer
Use the input bound in the convolution function lie inside the unit circle.
summation

15 16
 Proof of Sufficiency
Proof of Necessity Assume exponentially decaying impulse
Z response (i.e. poles inside the unit circle).
is the coefficient of largest magnitude
is the largest pole magnitude.
Impulse response is bounded if the poles of the
transfer function are in the closed unit disc and The impulse response is bounded by
decays exponentially if the poles are in the open
unit disc.
Systems with a bounded impulse response that
does not decay exponentially are not BIBO
stable.

17 18

Example 4.2 Solution

Investigate the BIBO stability of systems BIBO stable since the impulse response
with the impulse response satisfies

where is a finite constant. Let = upper bound for any impulse

response of finite duration.
Any FIR system is BIBO stable.
19 20
 Example 4.3 Solution
Investigate BIBO stability for Example 4.1

After pole-zero cancellation

Use Theorem 4.3 with pole-zero cancellation. a) BIBO stable, all poles inside unit circle.
b) BIBO stable, all poles inside unit circle.
c) BIBO stable, all poles inside unit circle.
21 22

Solution Z-plane Stable Pole Locations

Im[z]

Unit Circle
Not BIBO stable, a pole on unit circle.
STABLE
Re[z]

23 24
MATLAB Stability Determination MATLAB: ddamp
Pole Magnitude Damping Frequency Time Constant
Obtain roots of polynomial: (rad/TimeUnit) (TimeUnit)
-5.00e-02 + 1.41e-01 7.11e-01 2.75e+00 5.11e-01
roots(den) % denominator coeffts. den 1.32e-01i
-5.00e-02 - 1.41e-01 7.11e-01 2.75e+00 5.11e-01
zpk(g) %g = transfer function 1.32e-01i

>> ddamp([1,.1,.02])
Stable for roots inside the unit circle. Gives the pole locations, and
Closed-loop transfer function
>> H = feedback(gforward, gfeedback, 1)
Default: negative feedback
25 26

Internal Stability Theorem 4-4

If all the transfer functions that relate the system
inputs ( and ) to the possible system outputs
( and ) are BIBO stable, then the system is
Assume no unstable pole-zero cancellation.
internally stable. The system of Figure 4-3 is internally stable if
and only if all the closed-loop poles are in the
1 1 open unit disc.
D(z
R(z) E(z) Y(z)
1 1 + U(z)
C(z) GZAS(z)

R(z) E(z)
D(z
Y(z)
+ U(z)
C(z) GZAS(z)

27 28
 Transfer Functions Proof (Necessity: only if)
Substitute

Coprime polynomials (no common factors)

Internally stable (all transfer functions are

asymptotically stable) + coprime
D(z
characteristic polynomial has
R(z) + E(z) U(z) Y(z)
C(z) GZAS(z)
no zeros on or outside the unit circle.

29 30

Proof (Sufficiency: if) Theorem 4-5

The system is internally stable if and only if:
Substitute
1. The characteristic polynomial

has no zeros on or outside the unit circle.

2. The loop gain has no pole-zero
Characteristic polynomial
cancellation on or outside the unit circle .
No zeros on or outside the unit circle R(z) E(z)
D(z
Y(z)
+ U(z)
C(z) GZAS(z)
the systems with the 4 TFs are asymptotically
stable closed-loop system is internally stable.
31 32
 Proof: Necessity Proof: Sufficiency
: no zeros on or outside the
unit circle for internal stability (Theorem 4.4)
(ii) cancels unstable i. No zeros (c.l. poles) on or outside the unit circle
ii. No unstable pole-zero cancellation for the
closed-loop system if there is no unstable pole-
zero cancellation for

Sufficient for internal stability (Theorem 4.4)

is both an open-loop and a closed-loop pole
33 34

Example 4-4 Solution

Transfer function of isothermal chemical Discretized process transfer function
reactor
Z

o Determine with
o Verify that the resulting feedback system is Transfer function from reference input to output
not internally stable with the feedback
controller
All its poles are inside the unit circle (some
cancel)
35 36
 Solution: Control Variable Solution: Theorem 4-5

Pole at 1.334 outside the unit circle. Violates condition (ii) of Theorem 4-5:
Unstable pole at 1.334 cancels in the loop
The control variable is unbounded even
gain.
when the reference input is bounded.
37 38

Routh-Hurwitz Criterion Geometric Interpretation

Im(s) z3 Im(z)
1. Transform the inside of the unit circle to X s3
z3 1
z2 1

the LHP (bilinear transformation). z3 1 z2

z2 1
s2 X
2. Use the Routh-Hurwitz criterion for the Re(s) Re(z)
1
investigation of discrete-time system s1 X
1
z1 1
z1 1
stability.
z1

s z 1 z 1
90 inside unit circle

s 90 outside unit circle
90 on unit circle
39 40
 Advantages/Disadvantages Example 4.4
Easy stability test for low-order polynomials.
Find stability conditions for
Difficult for high order z-polynomials.
For high order polynomials, use symbolic a) The first order polynomial
manipulation.

b) The second order polynomial

41 42

Solution: 1st order Solution: 2nd order

Solve for the root. ,

Stability conditions Stability determination by solving for roots is

difficult.
Monic polynomial
constant term = product of poles
For pole magnitudes < 1
Necessary stability condition
Sufficient for complex conjugate poles

43 44
 Bilinear Transformation Comments
Recall:
Sufficient for complex conjugate roots and
only necessary for real roots.
Routh-Hurwitz criterion: poles of 2nd order w-
polynomial remain in the LHP iff its coefficients are
Real roots: substituting the three conditions
all positive. in the z-domain characteristic polynomial
gives roots between 1 and +1.

Adding the first and third conditions gives the

condition obtained earlier
45 46

Stable Parameter Range for 2nd Jury Test

Order z-polynomial The roots of the polynomial

Conditions with a0 are inside the unit circle if and only if

: (1)
1 a0= 1
stable (2)
a0= a11

1 1
(3) 0
a1
a0= a11
1
(4)
(5)

(n+1)
47 48
 Table Entries Jury Table
Row z0 z1 z2 zn k zn 1 zn
1 a0 a1 a2 an k an 1 an
2 an an 1 an 2 ak a1 a0
3 b0 b1 b2 bn k bn 1
4 bn 1 bn 2 bn 3 bk b0

5 c0 c1 c2 cn 2
6 cn 2 cn 3 cn 4 c0

. . . .
. . . .
. . . .
2 n 5 s0 s1 s2 s3
2 n 4 s3 s2 s1 s0

2 n 3 r0 r1 r2

49 50

Remarks Remarks (cont.)

1. First row of Jury table: list the coefficients of the
polynomial in order of increasing power of . 7- Condition (3) with 1, requires the constant term of
2. Number of rows of table 2 3 is always odd and the the polynomial to be less than unity in magnitude. The
coefficients of each even row are the same as the odd constant term is simply the product of the roots and
row directly above it with the order of the coefficients must be smaller than unity for all the roots to be inside
reversed. the unit circle.
3. There are 1 conditions for 1 coefficients of . 8 For first and second order systems, the Jury stability
conditions reduce to the conditions derived earlier.
4. Conditions 3 through 1 are calculated using the
coefficient of the first column of the Jury table, together 9- For higher order systems, applying the Jury test by
with the last coefficient of the last row. hand is laborious and it is preferable to test the stability
of a polynomial using a CAD package.
5. The middle coefficient of the last row is never used and
need not be calculated. 10- If the coefficients of the polynomial are functions of
system parameters, the Jury test can be used to obtain
6. Conditions (1) and (2) are calculated from directly. If the stable ranges of the system parameters.
one of the first two conditions is violated, has roots
on or outside the unit circle (no need to construct the Jury
table or test the remaining conditions). 51 52
 Solution
Example
(1) 1 1 2.6 0.56 2.05 0.0775 0.35 1.4175 0
Test the stability of the polynomial
(2) 1 1 1 1 2.6 0.56 2.05 0.0775 0.35
0.3825 0
(3) 0.35 1
Row (4) 0.8775 0.8325
1 0.35 0.0775 2.05 0.56 2.6 1 (5) 0.0770 0.5151
2 1 2.6 0.56 2.05 0.0775 0.35 (6) 0.2593 0.3472
3 0.8775 2.5729 0.1575 1.854 0.8325 Conditions (2), (5) & (6) violated .
4 0.8325 1.854 0.1575 2.5729 0.8775
Condition (2) is sufficient to conclude instability.
5 0.0770 0.7143 0.2693 0.5151
No. of conditions violated No. of roots outside the unit circle.
6 0.5151 0.2693 0.7143 0.0770
Polynomial has roots on or outside the unit circle.
7 0.2593 0.0837 0.3472 0.7 0.5 0.5 0.8 2.5
Root at 2.5 outside the unit circle.
53 54

Example 4.6 Solution

Transfer function of analog subsystem, ADC and DAC
Find the stable range of the gain for
Z
the unity feedback digital control system
with analog plant
Z

Partial fraction expansion

with DAC and ADC if the sampling

period is 0.02 s.
Transfer function ( )

55 56
 Closed-loop System Example
Unity feedback, closed-loop characteristic
equation Find the stable range of the gain for the digital
position control system with the analog plant
transfer function
Stability conditions

and with DAC and ADC if .

Hence, the stable range of is

57 58

Solution C.L. Characteristic Equation

Z

Partial fraction expansion

Transfer function ( )

59 60
 Stability Testing Nyquist Criterion
Given the number of unstable open-loop poles.
1) 2 Count the number of unstable closed-loop
3 poles.
2) 2 Follow a closed contour and substitute each
3 point in (vector)
3) 0 Plot the resulting closed curve.
Count the number of counterclockwise turns
of the vector (clockwise is )

The stable range is
61 62

Mapping Inside/Outside Contour

1

1

closed-loop poles, open-loop poles
Count the number of counterclockwise turns of the vector Net rotation: zero for outside, 1 turn for inside
(clockwise is ) 63 64
 Thm. 4-7 Nyquist Criterion Proof Outline
Traverse contour clockwise and substitute in
=no. of counterclockwise encirclements of the to produce a plot.
point (1,0) for a loop gain when traversing
the stability contour (i.e. clockwise)
has open-loop poles inside the contour.
Then the system has closed-loop poles outside
the unit circle with given by ( ), or Net angle change: /zero, /pole inside the
contour and zero for those outside it.
Corollary: An open-loop stable system is closed- One encirclement =
loop stable iff the Nyquist plot does not encircle
the point (1,0) (i.e. ) = number of c.l. (resp. o.l.) poles inside
65 the contour. 66

Contour for Stability Shift by (

Nyquist Diagram

Determination 0.6
1+ L

L
0.4

0.2

Imaginary Axis
On the unit circle, is the frequency response for 0

angles 0, , or 0, / 0, /2 . -0.2

complex conjugate of (mirror image) -0.4

Points on the large circle map to zero or to a single -0.6

point. -1 -0.5 0 0.5 1 1.5 2

For infinitesimal, angle contributions of values on the No. encirclements of origin for
Real Axis

straight line portions close to the real axis cancel.

= No. of encirclements of ( of
67 68
 Example Nyquist Diagram
Modified Contour
0.8

0.6
Modified contour Small semicircle: 1
0.4
Avoid pole at unity Pole at 1 mapped to large
0.2

Maps to large semicircle. clockwise semicircle

Imaginary Axis

-0.2
Denominator angle: net effect =
1 counterclockwise semicircles
-0.4

-0.6

-0.8
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Real Axis

4.95 4.901
10
1.97 0.9704
, no open-loop poles outside the unit
circle: system is closed-loop stable. 69 70

>>gz=10/(z-1)/(z-0.1) Phase Margin and Gain Margin

Gain Margin: gain perturbation that makes
the system marginally stable.
Phase Margin: negative phase perturbation
that makes the system marginally stable.

R(s) + C(z)
G(z) G(z)

71 72
 Nyquist Plot: PM & GM MATLAB Plots and Margins
>> nyquist(gd) % Nyquist plot.
>> bode(gd) % Bode plot.
>> [gm,pm]=margin(gd) % Find PM & GM
>> margin(gd) % GM & PM on Bode plot
Nyquist plot: click on plot and select
Characteristics
All stability margins

73 74

Example 4.10 Solution

Digital control system for the furnace model

Using MATLAB Nyquist Diagram

Discrete-time first order actuator

>> gd=c2d(g,0.01) 25

20

>> gtd=gd*ga; 15

10

1. Determine the closed-loop stability. >> nyquist(gtd) 5

Imaginary Axis
>> hold on
0

2. How does an amplifier of gain affect -5

closed-loop stability? >> nyquist(5*gtd) -10

-15

Change scale -20

-25

75 -5 0 5 10 15 20 25
Real Axis
30 35 40 45 50
76
 GM & PM
Nyquist Diagram
2

stable 1.5

0.5
System: untitled1
Real: -1.41
Imag: 0.000741
Imaginary Axis

>>[gm,pm]=margin(gtd)
Frequency (rad/sec): -5.33
0

gm = -0.5

3.4817 -1
System: gdt
Real: -0.795
Imag: -0.624
Frequency (rad/sec): 2.59

pm =
37.5426
-1.5

-2
-5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
Real Axis
77