Introduction to Digital Control Systems ME475 Lecture Notes by B.

Yao

DIGITAL CONTROL

PRACTICAL EXPERIMENTAL SETUP CONTINUOUS CONTROLLER vs DIGITAL CONTROLLER:

One important application area of advanced control systems is the precision Continuous Control System (e.g., analog circuit):
motion control of various devices. As an example, the linear motor driven high-
speed/high-accuracy X-Y table controlled by a digital signal processor (DSP)
based digital control system is shown below. The entire system has been set-up
in the School of Mechanical Engineering at Purdue University. The two axes (X Disturbance
and Y directions respectively) are driven by two different types of leading edge
y(t)
precision linear motors. Each axis of the table can move at a speed more than r(t) e(t)
C(s)
u(t)
P(s)
2m/s with a position measurement resolution less than 1 micron and acceleration Reference Value Input Output
of several gs. The entire system is controlled by an advanced dSPACE product Controller Plant or Process
based digital servo control system. A Pentium PC functions as the user interface.
For more information regarding the application, you can visit y(t)
https://engineering.purdue.edu/~byao under the research item. 1
Sensor
Continuous Controller

Digital Control System Using a Microprocessor

DS3001
Encoder Interface

Matlab/Simulink
Real-TimeWorkshop DS1003
TMS320C40 DSP

Real-Time Interface
DS2201
Multi-I/O Board
Compilers A/D, D/A, I/O

Advantages of Digital Control System

Flexible and Universal:
Can implement any feedback gains and higher order controller. Overall
system set-up is quite fixed.
Powerful:
Can implement advanced control strategies such as nonlinear controllers
Experimental Setup and complicated switching logic.

Chapter 2-1
Introduction to Digital Control Systems ME475 Lecture Notes by B. Yao

INTERFACE BETWEEN CONTINUOUS PHYSICAL SYSTEM AND

DIGITAL COMPUTER

Sampler D/A Converter

x(t) x(kT) x(kT) ud(kT) u(kT) stored in bits u(kT) u(kT) expressed in analogy format
x(t)
 
0 T 2T 3T 4T 0 T 2T 3T 4T 0 0
t t T 2T 3T 4T t T 2T 3T 4T t
x(t) T x(kT)
ud(kT) u(kT)
D/A

The sampler is used to obtain the values of a continuous signal at each D/A converter converts a digital number ud (kT ) , the digital control input,
sampling instant tk  kT , k = 0, 1, 2, , where T is the sampling period.
to an analog signal u(kT) , which has the same value, i.e., u(kT )  ud (kT ) .
The sampled value is denoted as x ( kT ), k = 0, 1, 2, .

A/D Converter Hold Circuit

x(kT) x(kT) expressed in analog format xd(kT) x(kT) stored in bits

u(kT) u(kT) u(t) u(t)

 
0 T 2T 3T 4T 0 T 2T 3T 4T
t t 0 T 2T 3T 4T 0 T 2T 3T 4T
x(kT) xd(kT) t t
A/D u(kT) u(t)
hold

The A/D converter converts x ( kT ) , the sampled analog signal value at

each sampling instant, into xd (kT ) , a digital number that a The Hold Circuit generates a continuous control input signal that
microprocessor can accept. xd (kT ) is stored in bits and is normally corresponds to the digital control input sequence u(kT ) at each sampling
chosen to be the digital number closest the analog value x ( kT ) . For instant as:
example, for a 12 bits A/D convert that accepts an analog value between u (t )  u (kT ) when kT  t  (k  1)T
 10 V, x ( kT ) can be any value between  10 V but a 12 bit A/D can only
store 212 = 4096 digital numbers. So such an A/D convert only has a i.e., holds the value u(kT ) until next sampling instant. Such a hold circuit is
resolution of
20 V  0.0048 V . The more bits the A/D convert has,
4096 normally referred to as zero-order hold (ZOH).
the smaller the resolution, and the closer xd (kT ) to x ( kT ) .

For simplicity, in this course, we think A/D’s resolution is very small and
assume that xd ( kT )  x ( kT ) .
Chapter 2-2
Introduction to Digital Control Systems ME475 Lecture Notes by B. Yao

DIGITAL CONTROLLER DESIGN via EMULATION

I/O Model:
The easiest way to design a digital controller is by emulation or digital U (s ) 2(s  1)
implementation of continuous controller, which proceeds as follows:  C (s )   u  3u  2(e  e)
E (s ) s3
Step 1:
Design a continuous compensation or controller as done in previous  u (kT )  3u (kT )  2  e (kT )  e (kT ) 
lectures.
Use Euler’s Method to approximate u ( kT ) and e ( kT ) :
Step 2:
Digitize the continuous compensation or transfer function via certain
approximation techniques such as the Euler’s method, Tustin’s Method,
u (k  1)  u (k )
 3u (k )  2
 e(k  1)  e(k )  e(k ) 
Matched Pole-zero (MPZ) method, etc. T  T 
Step 3:
Using discrete analysis, simulation, or experimentation to verify the
 u (k  1)  (1  3T ) u (k )  2e(k  1)  2(-1  T )e(k )
design.
Or u ( k )  (1  3T ) u ( k  1)  2e( k )  2( 1  T )e( k  1)
The above difference equation can be used to calculate the control input at next
DIGITIZATION via Euler’s METHOD sampling instant u(k+1) once given u(k), e(k), and e(k+1).

Euler’s method is to approximate the derivative by a forward rectangular rule:

Ex.D2: Digitization of State-Space Model
Approximate x ( kT ) by:
Differential Equation in Continuous Domain:

x ((k  1)T )  x (kT )  xc  Axc  Be  xc (kT )  Axc (kT )  Be(kT )
x (kT )    
T  u  Cxc  u ( kT )  Cxc (kT )

Short hand notation: Use the Euler’s Method to approximate xc ( kT ) :

 xc (k  1)  xc (k )
x ( k )  x ( kT )   Axc (k )  Be(k )
 T
 u (k )  Cxc (k )
x (k  1)  x (k ) 
x (k ) 
T  xc (k  1)  (1  TA) xc (k )  TB e(k )

 u (k )  C xc (k )
Ex.D1: Digitization of Transfer Function
2(s  1) The above difference equation can be used to calculate the controller output u(k)
e 2(s  1) u or C (s) = to be commanded and the controller state xc (k  1) at the next sampling
s3
s3 instance once given the feedback signal e(k) and the initial condition of the
controller state xc (0) .

Chapter 2-3