1

Digital Control Systems

Lecture Summary #6

I. P ROPORTIONAL CONTROLLER FOR SECOND ORDER

Step Response
SYSTEMS
1.4

The transfer function of a proportion controller is a pure 1.2 K=15

gain. We first begin with a continuous time example and we

1
generalize our approach to discrete time systems in the z-
domain. The transfer function of proportional control is

Amplitude
0.8
C(s) = K (1)
0.6
An example of using proportional controller to control the K=2
closed loop system is shown below. 0.4
K=10

0.2

Example: 0
0 1 2 3 4 5 6 7 8 9 10
Time (sec)

A position control system has the following transfer func-

tion Fig. 1. Step response for the closed loop system (3) for different values of
K the gain
G(s) = (2)
s(s + 4)
Design a proportional controller for the system to obtain
proportional controller is constrained. The code allowing to
1) a specified damping ratio, ζ = 0.77 obtain and plot the closed loop response is as follows:
2) a specified undamped natural frequency, ωn = 3.25
d0 = tf ([1], [1 4 0]) (8)
The first step is to obtain the closed loop transfer function,
CL2 = f eedback(2 ∗ d0, 1) (9)
which is given by:
CL10 = f eedback(10 ∗ d0, 1) (10)
K
Gcl (s) = 2 (3) CL15 = f eedback(15 ∗ d0, 1) (11)
s + 4s + K
step(CL2), hold on (12)
Now, we match this transfer function with the standard form
characteristic polynomial: step(CL10), hold on (13)
step(CL15, ) hold on (14)
s2 + 4s + K = s2 + 2ζωn s + ωn2 (4)
Matlab command feedback does the closed loop transfer
1) For damping ratio of ζ = 0.77, we have
function.
2ζωn = 4, and ωn2 = K (5)
and the solution for the gain is given by: C HARACTERISTIC EQUATION FOR SECOND ORDER
SYSTEMS
4
K= = 6.75 (6) Desired characteristics of a control system include stability,
ζ2
minimizing steady state error and desired transient. Last time
Note that in this case, we cannot choose ωn , it is we saw that the standard form for continuous time second
constrained by its relationship with K. order systems is given by
2) Fur an undamped natural frequency of ωn = 3.25, we
have ωn2
K = ωn2 = 10.56 (7) G(s) = (15)
s2 + 2ζωn s + ωn2
The corresponding value for the damping ratio is ζ = The behavior of the closed loop system depends on the
0.6154. location of the poles. Our goal is to develop similar equations
by mapping the s-domain to the z-domain. The continuous
The step response for different values of the gain is shown
domain poles are
in figure 1. Clearly, achieving desired values for both the p
damping ratio and the undamped natural frequency using a s1,2 = −ζωn ± jωn 1 − ζ 2 (16)
Digital Control, spring 2018 Lecture 6

We already know that

Step Response

z = esT (17) 1.4

where T is the sampling time. We can write for the discrete 1.2
time poles:
1
z1,2 = es1,2 T (18)
= e−ζωn T e±jωd T (19)

Amplitude
0.8
−ζωn T
= e ∠±ωd T (20)
0.6

which can be written as

0.4
z1,2 = |z|∠ ± θ (21)
0.2
with
−ζωn T
|z| = e (22) 0
0 0.5 1 1.5 2 2.5 3 3.5
Time (sec)
and p
θ = ωd T = ωn T 1 − ζ 2 (23)
Fig. 2. Closed loop step response with settling time of 2s.
Now, our goal is to find the discrete time characteristic polyno-
mial. Knowing that the poles are z1 and z2 , the characteristic
polynomial is given by
2) Settling time of 2s.
Q(z) = (z − z1 )(z − z2 ) (24) 4
= 2 (32)
= (z − e−ζωn T ejωd T )(z − e−ζωn T e−jωd T ) (25) ωn ζ
= z 2 − (2 cos(ωd T )) e−ζωn T z + e−2ζωn T (26) ωn ζ = 2 rad/s (33)
(34)
P ROPORTIONAL CONTROLLER IN THE Z - DOMAIN Using the second equation, we get
Now we consider designing a proportional controller to
achieve achieve the following goals: (1) desired transient, and K = e−2(2)0.1 − 0.3 = 0.37 (35)
(2) desired steady state error.
Figure 2 shows the closed loop step response for this
value of the gain, which confirms the settling time.
A. Example 1
Consider the following open loop transfer function in the
z-domain R EDUCING SYSTEM ORDER
1
G(z) = (27)
(z − 1)(z − 0.3)
Sometimes higher order systems can be approximated
With T = 0.1s. Design a proportional controller to achieve by first or second order systems. Consider for example
the following (independently) z
1) A damped natural frequency of 5rad/s G(z) = (36)
(z − 0.1)(z − 0.2)(z − 0.9)
2) A settling time of 2s.
This is a third order system with poles {0.1, 0.2, 0.9}.
First, we find the closed loop characteristic equation. The The poles have different effects on the system’s re-
closed loop transfer function is given by sponse. By taking the inverse z-transform, it is possible
K to write
Gcl (z) = (28)
z 2 − 1.3z + 0.3 + K g(k) = a0.1k + b0.2k + c0.9k (37)
In order to solve, we need to match the coefficients in the where a, b, c are constants. when k = 3, it is possible to
characteristic polynomial with the standard form characteristic write
polynomial. This will give us:
g(k) = a · 0.001 + b · 0.008 + c · 0.729 (38)
1.3 = 2e−ζωn T cos(ωd T ) (29)
K + 0.3 = e−2ζωn T (30) After three iterations only, it can be seen that the effect
of the first and second pole is reduced considerably
1) For ωd = 5rad/s, we have compared to the effect of the third pole. The third pole
 2 is called the dominant pole. Therefore, it is possible to
1.3
e−2ζωn T = = K + 0.3 (31) reduce the order of the system by keeping the dominant
2 cos(ωd T ) pole only, which is the slowest pole in the system. An
Solving for the gain yields K = 0.248 example is shown below.

2
Digital Control, spring 2018 Lecture 6

Example
Consider the third order transfer function given by
z
G(z) = (39)
(z − 0.1)(z − 0.2)(z − 0.9)
• By keeping the dominant pole only, we get the first Step Response
order system given by 14
z
G1 (z) = (40)
z − 0.9 12

• By keeping two dominant poles, we get the second

10
order system given by
z

Amplitude
8
G2 (z) = (41)
(z − 0.2)(z − 0.9)
6 Third order system
The step response of the three systems is shown in figure Second order approximation
3. 4
First order approximation

The poles have an effect on the system’s transient

response but also on its final value as shown in figure 3. 2
The reduced order systems do not have the same final
value as the original system. In order to fix the problem 0
0 1 2 3 4 5 6
we perform gain adjustment so that the reduced order Time (seconds)

systems have the same final value as the original system.

Let K and Ki be the DC gains (z = 1) of the original Fig. 3. Third order system and approximation with lower order systems
system and the reduced order system, respectively, and without gain adjustment
Gapr is the reduced order transfer function. In order to
have the same final value, the approximation must be
scaled as follows
K
Gapr ← Gapr (42)
Ki
This simple gain adjustment allows for Gapr and the
original system to have the same final value.

Example
For the previous example, the DC gains are Step Response

  14
System DC Gain
 G(z) K = 13.89 12

 G1 (z)
 (43)
K1 = 10 
Third order system
G2 (z) K2 = 12.5 10
Second order approximation
First order approximation
Therefore,
Amplitude

z
G1 (z) = 1.389 (44) 6
z − 0.9
z 4
G2 (z) = 1.1112 (45)
(z − 0.2)(z − 0.9)
2
The systems’ response after gain adjustment is shown
in figure 4. 0
0 1 2 3 4 5 6
Time (seconds)

Fig. 4. Third order system and approximation with lower order systems with
gain adjustment