Professional Documents
Culture Documents
3 - Digital Control Systems
3 - Digital Control Systems
288
Literature
Nise, Norman S.
Control Systems Engineering
2011, Wiley
ISBN 978-0-470646120
Gopal, M.
Control Systems
2011, McGraw Hill
ISBN 978-0-07-066879-9
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 289
Digital Computer Control System
Applications
• Total number of computer control
systems installed in industry has
grown over the past three decades
• Transistor density
• Powerful
• Relatively inexpensive
• Mobile capability
• Improved measurement sensitivity
(less signal noise)
• Flexibility
• Reconfigure control algorithm in
software
Source: INTEL
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 290
Digital Computer Control System
Applications
Difference to
conventional controller
Digital computer receives
and operates on signals in
digital (numerical) form
measurement data
converted from analog to
digital / output data
converted from digital to
analog
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 291
Digital Computer Control System
Applications
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 292
Digital Computer Control System
Applications
Difference to conventional controller
target
actual (cont)
actual (dig)
control (cont)
control (dig)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 293
Learning Outcomes
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 295
Signal Conversion
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 296
Signal Conversion
• Analog signal (a)
• Analog signal sampled at
periodic intervals and held
over the sampling interval
(b)
• Device so called zero-order
sample-and-hold (z.o.h.)
• After sampling and holding,
the A/D-converter converts
to a digital number (c)
• Quantization error
• Stability and transient
response are now
dependent on sampling rate
taking conversion into
account during modeling
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 297
Modeling
Sampler
• All number entering or leaving the
computer at the same fixed period
T, called sampling period
• Sequence of sample values
(discrete) r(kT) in contrast to r(t)
• Input r(t) / output r*(t), where nT is
the current sampling time
• 𝑟 ∗ 𝑡 = 𝑟(𝑛𝑇)𝛿(𝑡 − 𝑛𝑇)
• Sketch 𝑟 ∗ 𝑡 in the figure on the
right
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 298
Modeling
Zero-Order Hold
• Z.o.h. takes the value r(kT) and holds it
constant for 𝑘𝑇 ≤ 𝑡 < 𝑘 + 1 𝑇
• Sampler and z.o.h. accurately follow the
input signal if T is small compared to the
transient changes in the signal. Output will
approach input as T approaches zero
1 1 −𝑠𝑇 1−𝑒 −𝑠𝑇
• TF of z.o.h. 𝐺ℎ 𝑠 = − 𝑒 =
𝑠 𝑠 𝑠
• Quantization error and the error due to a
computers´s finite word size are small
relative to the amplitude, precision
limitations can be neglected
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 299
Modeling
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 300
Modeling
Ramp input to a PT1-system, T=0.1s
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 301
Modeling
Ramp input to a PT1-system, T=1s
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 302
Modeling
CL-control of PT1-system, T=0.05s
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 303
Modeling
CL-control of PT1-system, T=0.2s
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 304
Modeling
CL-control of PT1-system, T=0.7s
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 305
TF of a Sampled System
u(t) y(t)
u(t) y(t)
LTI
t t
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 306
TF of a Sampled System
Differential quotient approximation
𝑑𝑦 𝑦 𝑘𝑇 − 𝑦 𝑘 − 1 𝑇
≈
𝑑𝑡 𝑡=𝑘𝑇 𝑇
𝑦 𝑘𝑇 − 𝑦 𝑘 − 1 𝑇 𝑦 𝑘−1 𝑇 −𝑦 𝑘−2 𝑇
𝑑²𝑦 −
≈ 𝑇 𝑇
𝑑𝑡² 𝑇
𝑡=𝑘𝑇
𝑦 𝑘𝑇 − 2𝑦 𝑘 − 1 𝑇 + 𝑦 𝑘 − 2 𝑇
=
𝑇²
Difference equation
𝑦 𝑘 + 𝑎1 𝑦 𝑘 − 1 + 𝑎2 𝑦 𝑘 − 2 + ⋯ + 𝑎𝑚 𝑦 𝑘 − 𝑚
= 𝑏0 𝑢 𝑘 + ⋯ + 𝑏𝑚 𝑢(𝑘 − 𝑚)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 307
TF of a Sampled System
𝑦(𝑧) 𝑏0 + 𝑏1 𝑧 −1 + ⋯ + 𝑏𝑚 𝑧 −𝑚 𝐵 𝑧
𝐺 𝑧 = = −1 −𝑚
=
𝑢(𝑧) 1 + 𝑎1 𝑧 + ⋯ + 𝑎𝑚 𝑧 𝐴 𝑧
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 308
The z-Transform
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 309
The z-Transform
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 310
The z-Transform
𝑍 𝑓(𝑡) = 𝐹 𝑧 = 𝑓(𝑘𝑇)𝑧 −𝑘
𝑘=0
• 𝐹(𝑧) can be transformed to 𝑓(𝑘𝑇), or vice versa
• Paralleling the development of the Laplace
transform, we can form a table relating 𝑓(𝑘𝑇) to
𝐹(𝑧)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 312
The z-Transform
𝑍 𝑓(𝑡) =
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 313
The z-Transform
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 315
The z-Transform
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 316
The z-Transform
Example
Let´s consider the system in the block diagram for
𝑇 = 1. Determine the TF in the z-domain
𝑌(𝑧)
𝐺 𝑧 = .
𝑅(𝑧)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 317
The z-Transform
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 318
The z-Transform
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 319
Review Questions
Next steps
• Using the inverse z-transform to get 𝑓 𝑘𝑇 out of 𝐹(𝑧)
• Determine the stability of a system using the transfer
function in the z-domain 𝐹(𝑧)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 322
Controls
Prof. Dr.-Ing. Dirk Nissing
323
The Inverse z-Transform
• Determination of the sampled time function from
its z-transform inverse z-transform
• 𝑍 −1 𝐹(𝑧) = 𝑓(𝑘𝑇)
• For functions not in the table, we must perform
an inverse z-transform calculation similar to the
inverse Laplace transform by
- partial-fraction expansion or
- via power series method (division of
polynomials)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 324
Closed-Loop Feedback Sampled-
Data Systems
Block Diagram Reduction
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 331
Closed-Loop Feedback Sampled-
Data Systems
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 332
Closed-Loop Feedback Sampled-
Data Systems
Problem: Find the z-transform of the system shown
in the figure below
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 335
Closed-Loop Feedback Sampled-
Data Systems
b) Phantom sampler
S4 at the output
(output of a sampled-
data system can only
be found at the
sampling instants
anyway and the
signal is not an input
to any other block
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 336
Closed-Loop Feedback Sampled-
Data Systems
b) Add phantom
sampler S2 and S3
(sampled sum is
equivalent to the sum
of the sampled inputs
when all samplers
are synchronized)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 337
Closed-Loop Feedback Sampled-
Data Systems
c) Move sampler S1
and 𝐺(𝑠) to the right
past the kickoff point
motivation to yield
a sampler at the
input of 𝐺 𝑠 𝐻(𝑠)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 338
Closed-Loop Feedback Sampled-
Data Systems
d) 𝐺 𝑠 𝐻(𝑠) with
samplers S1 and
S3 becomes 𝐺𝐻(𝑧)
and 𝐺(𝑠) with
samplers S1 and
S4 becomes 𝐺(𝑧);
converting 𝑅∗ (𝑠) to
𝑅(𝑧) and 𝐶 ∗ (𝑠) to
𝐶(𝑧)
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 339
Closed-Loop
Feedback
Sampled-Data
Systems
e) Apply rules for block
diagrams
simplification
(feedback formula)
with only a change in
variables from 𝑠 to 𝑧
f) finally, multiplication
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 340
Stability
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 341
Stability
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 342
Stability
• A:
• B:
• C:
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 343
Stability
• Each region of the s-
plane can be
mapped to a
corresponding region
on the z-plane
• Positive values in the
right half of the s-
plane (region C)
points outside the
unit circle on the z-
plane
• Points on the Im-axis
(region B) points
on the z-plane with
magnitude = 1, the
unit circle
• Negative values in
the left half of the s-
plane (region A)
points inside the unit
circle on the z-plane
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 344
Stability
Example
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 345
Stability
Example
Determine if the systems are stable
𝑧
1. 𝐺 𝑧 =
𝑧²−𝑧+0.6322
𝑧−1
2. 𝐺 𝑧 =
𝑧²+2.31𝑧+3.012
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 347
Stability
Exercise
Let´s consider the open-loop system having the
discrete transfer function
𝐾(0.3678𝑧 + 0.2644)
𝐺 𝑧 =
𝑧² − 1.3678𝑧 + 0.3678
where K is the controller gain. Calculate if the
closed-loop system is stable/unstable, when
a) K=1
b) K=10
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 348
Stability
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 349
Virtually Continuous Control
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 351
Virtually Continuous Control
Time constant of the control Sampling period T
system (experimentally)
Delay time Tu ≤ 0.1 Tu
Compensating time Tg ≤ 0.05-0.1 Tg
Limiting factors:
-
-
-
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 352
Review Questions
Controls
07.01.2022 Prof. Dr.-Ing. Dirk Nissing 355