Digital Control Module 5 Lecture 4

Module 5: Design of Sampled Data Control Systems

Lecture Note 4

1 Frequency Domain Analysis

When a sinusoidal input is given to a stable LTI system it produces a sinusoidal output of
same frequency but with different magnitude and phase.

The variation of output magnitude and phase with input frequency is known as frequency
response of the system.

Frequency domain analysis provides a good design in presence of uncertainty in plant model.
Experimental results can be used to construct frequency response even if the plant model is
unknown.

Analysis of digital control systems in frequency domain depends on the extension of the
existing techniques in continuous time case.

Two most popular graphical representations in frequency domain are Nyquist plot and
Bode diagram.

1.1 Nyquist plot

The Nyquist plot of a transfer function, usually the loop transfer function GH(z), is a map-
ping of Nyquist contour in z-plane onto GH(z) plane which is in polar coordinates. Thus it
is sometimes known as polar plot.

Absolute and relative stabilities can be determined from the Nyquist plot using Nyquist
stability criterion.

Given the loop transfer function GH(z) of a digital control system, the polar plot of GH(z)
is obtained by setting z = ejωT and varying ω from 0 to ∞.

Nyquist stability criterion: The closed loop transfer function of single input single output

I. Kar 1
Digital Control Module 5 Lecture 4

digital control system is described by

G(z)
M (z) =
1 + GH(z)

Characteristic equation 1 + GH(z) = 0

The stability of the system depends on the roots of the characteristic equation or poles
of the system. All the roots of the characteristic equation must lie inside the unit circle for
the system to be stable.

Before discussing Nyquist stability criterion for the digital system, following steps are neces-
sary.
1. Defining the Nyquist path in the z-plane that encloses the exterior of the unit circle.
Here the region to the left of a closed path is considered to be enclosed by that path
when the direction of the path is taken anticlockwise.
2. Mapping the Nyquist path in z-plane onto the GH(z) plane which results in Nyquist
plot of GH(z).
3. Stability of the closed loop system is investigated by studying the behavior of Nyquist
plot with respect to the critical point (−1, j0) in the GH(z) plane.

Two Nyquist paths are defined. The Nyquist path z1 as shown in Figure 1 does not enclose
poles on the unit circle whereas the Nyquist path z2 as shown in Figure 2 encloses poles on
the unit circle.

These figures are the mapping of the Nyquist contours in s-plane where the entire right half
of the s-plane, without or with the imaginary axis poles, is enclosed by the contours.

Let us now define the following parameters.

Z−1 = number of zeros of 1 + GH(z) outside the unit circle in the z-plane.

P−1 = number of poles of 1 + GH(z) outside the unit circle in the z-plane.

P0 = number of poles of GH(z) (same as number of poles of 1 + GH(z)) that are on the unit
circle.

N1 = number of times the (−1, j0) point is encircled by the Nyquist plot of GH(z) corre-
sponding to z1.

N2 = number of times (−1, j0) point is encircled by Nyquist plot of GH(z) correspond-
ing to z2.

I. Kar 2
Digital Control Module 5 Lecture 4

Figure 1: Nyquist path that does not enclose poles on the unit circle

Figure 2: Nyquist path that encloses poles on the unit circle

According to the principle of argument in complex variable theory

N1 = Z−1 − P−1

I. Kar 3
Digital Control Module 5 Lecture 4

N2 = Z−1 − P−1 − P0

Now let us denote the angle traversed by the phasor drawn from (−1, j0) point to the
ωs
Nyquist plot of GH(z) as ω varies from to 0, on the unit circle of z1 excluding the small
2
indentations, by φ.

It can be shown that

φ = (Z−1 − P−1 − 0.5P0 )1800 (1)

For the closed loop digital control system to be stable, Z−1 should be equal to zero. Thus
the Nyquist criterion for stability of the closed loop digital control systems is

φ = −(P−1 + 0.5P0 )1800 (2)

Hence, we can conclude that for the closed loop digital control system to be stable,
the angle, traversed by the phasor drawn to the GH(z) plot from (-1, j0) point
ωs
as ω varies from to 0, must satisfy equation (2).
2
Example 1: Consider a digital control system for which the loop transfer function is given
as
0.095Kz
GH(z) =
(z − 1)(z − 0.9)

where K is a gain parameter. The sampling time T = 0.1 sec.

Since GH(z) has one pole on the unit circle and does not have any pole outside the unit
circle,
P−1 = 0 and P0 = 1

Nyquist path has a small indentation at z = 1 on the unit circle.

Nyquist plot of GH(z), as shown in Figure 3, intersects the negative real axis at −0.025K
ωs
when ω = = 31.4 rad/sec.
2
φ can be computed as

φ = −(0 + 0.5 × 1)1800 = −900

It can be seen from Figure 3 that for φ to be −900 , (−1, j0) point should be located at the
left of −0.025K point. Thus for stability

I. Kar 4
Digital Control Module 5 Lecture 4

GH(z) plane j Im GH(z)

ωs
ω=
2
−1 −0.025K −1 +1
Re GH(z)
0
φ = −90o φ = 90o
φ = 90o

ω→0

Figure 3: Nyquist plot for Example 1

−1 < −0.025K

⇒ K < 40

If K > 40, (−1, j0) will be at the right of −0.025K point, hence making φ = 90o .

If φ = 90o , we get from (1)

φ
Z−1 = + 0 + 0.5 = 1
180o

Thus for K > 40, one of the closed loop poles will be outside the unit circle.

If K is negative we can still use the same Nyquist plot but refer (+1, j0) point as the
critical point. φ in this case still equals +90o and the system is unstable. Hence the stable
range of K is

0 ≤ K < 40

More details can be found in Digital Control Systems by B. C. Kuo.

I. Kar 5