Vous êtes sur la page 1sur 10

Lecture abstract

EE C128 / ME C134 – Feedback Control Systems


Lecture – Chapter 13 – Digital Control Systems
Topics covered in this presentation
Alexandre Bayen I Important considerations in CT-to-DT conversion that yield errors
I z-transform & its inverse
Department of Electrical Engineering & Computer Science
University of California Berkeley I 2 CT-to-DT conversion methods
I DT region of stability
I DT RL
I DT TR characteristics
I Designing DT compensators in CT

September 10, 2013

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 1 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 2 / 55

13 Digital control systems 13.1 Intro

Chapter outline

1 13 Digital control systems


1 13 Digital control systems 13.1 Introduction
13.1 Introduction 13.2 Modeling the digital computer
13.2 Modeling the digital computer 13.3 The z-transform
13.3 The z-transform 13.4 Transfer functions
13.4 Transfer functions 13.5 Block diagram reduction
13.5 Block diagram reduction 13.6 Stability
13.6 Stability 13.7 Steady-state error
13.7 Steady-state error 13.8 Transient response on the z-plane
13.8 Transient response on the z-plane 13.9 Gain design on the z-plane
13.9 Gain design on the z-plane 13.10 Cascade compensation via the s-plane
13.10 Cascade compensation via the s-plane 13.11 Implementing the digital compensator
13.11 Implementing the digital compensator

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 3 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 4 / 55

13 Digital control systems 13.1 Intro 13 Digital control systems 13.1 Intro

Intro, [1, p. 724] Definitions, [1, p. 724]


Concept
I Only frequency-domain analysis
& design
I Not state-space techniques
I I Analog: CT, dynamic variables retain a particular value for only an
TFs built with analog
components ! digital infinitesimally short amount of time
computer that performs I Digital: DT, dynamic variables evolve in between computer
calculations that emulate the measurements of outputs & control inputs, but the computer remain
physical compensator unchanged throughout each non-zero period of sampling time

Figure: a. analog; b. digital

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 5 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 6 / 55
13 Digital control systems 13.1 Intro 13 Digital control systems 13.2 Modeling the digital computer

Digital control implementation, [1, p. 725]


Concept 1 13 Digital control systems
I Digital computer
I
13.1 Introduction
Control of multiple loops at the same time
I Signals are sampled at specified intervals & held 13.2 Modeling the digital computer
I system performance / !sample 13.3 The z-transform
I A/D converter: Measured outputs sampler 13.4 Transfer functions
I 2-step process 13.5 Block diagram reduction
1. Analog signal ! sampled signal
2. Sampled signal ! sequence of binary numbers
13.6 Stability
I Not instantaneous, i.e, there is a delay 13.7 Steady-state error
I !sample > !Nyquist = 2!BW 13.8 Transient response on the z-plane
I D/A converter: Control inputs zero-order hold 13.9 Gain design on the z-plane
I Instantaneous 13.10 Cascade compensation via the s-plane
13.11 Implementing the digital compensator

Figure: Digital computer with A/D & D/A converters


Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 7 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 8 / 55

13 Digital control systems 13.2 Modeling the digital computer 13 Digital control systems 13.2 Modeling the digital computer

Modeling the sampler, [1, p. 728] Modeling the sampler, [1, p. 729]

Concept Concept
I ...simplification & Laplace transform...
I z-transform: Laplace transform
I Sampled waveform portion not dependent upon the sampling
replacement for sampled signals
I
waveform characteristics
Sampled waveform
1
X
fT⇤W (t) = f (t)s(t) f ⇤ (t) = f (kT ) (t kT )
1 k= 1
X
= f (t) [u(t kT )
k= 1
... u(t kT Tw )]

I Integer, k 2 [ 1, 1]
I Period of pulse train, T
I Period of pulse width, TW Figure: Views of uniform-rate sampling Figure: Modeling of sampling with a uniform rectangular pulse train

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 9 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 10 / 55

13 Digital control systems 13.2 Modeling the digital computer 13 Digital control systems 13.3 The z-transform

Modeling the ZOH, [1, p. 729]

1 13 Digital control systems


Concept
13.1 Introduction
I Zero-order hold (ZOH): Hold
13.2 Modeling the digital computer
the last sampled value of f (t)
until the next sample 13.3 The z-transform
I
13.4 Transfer functions
Staircase approximation to
f (t) 13.5 Block diagram reduction
I Sequence of step functions 13.6 Stability
whose amplitude is f (t) at 13.7 Steady-state error
the sampling instant, or 13.8 Transient response on the z-plane
f (kT ) 13.9 Gain design on the z-plane
I TF of the step that starts at
Figure: Ideal sampling & ZOH 13.10 Cascade compensation via the s-plane
t = 0 & ends at t = T
13.11 Implementing the digital compensator
Ts
1 e
Gh (s) =
s

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 11 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 12 / 55
13 Digital control systems 13.3 The z-transform 13 Digital control systems 13.3 The z-transform

Intro, [1, p. 730] z- & s-transforms, [1, p. 732]

Concept
f (t) F (s) F (z) f (kT )
I Stability & TR of a sampled-data system depend upon sampling rate
(t) 1 1
I Laplace transform of the sampled time waveform 1 z
u(t) s z 1
u(kT )
1
X 1 Tz
⇤ kT s tu(t) s2 (z 1)2
kT
F (s) = f (kT )e 
dn
tn u(t) sn+1
n!
lim ( 1)n z (kT )n
k=0 a!0 dan z e aT
at u(t) 1 z akT
e e
I Letting s+a z e aT 
Ts dn
z=e tn e at u(t) ( 1)n da n
z (kT )n e akT
z e aT
! z sin(!T )
I z-transform sin(!t)u(t) s2 +! 2 z2 2z cos(!T )+1
sin(!kT )
1
X z(z cos(!T ))
s
k cos(!t)u(t) cos(!kT )
F (z) = f (kT )z s2 +! 2 z 2 2z cos(!T )+1
at ! ze aT sin(!T ) akT
k=0 e sin(!t)u(t) (s+a)2 +! 2 z2 2ze aT cos(!T )+e 2aT
e sin(!kT )
z 2 ze aT cos(!T )
f (kT ) ⌧ F (z)
at s+a akT
e cos(!t)u(t) (s+a)2 +! 2 z2 2ze aT cos(!T )+e 2aT
e cos(!kT )

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 13 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 14 / 55

13 Digital control systems 13.3 The z-transform 13 Digital control systems 13.3 The z-transform

z-transform theorems, [1, p. 733] Definitions, [1, p. 733]

Name theorem

Linearity Z[c1 f1 (t) + c2 f2 (t)] = c1 F1 (z) + c2 F2 (z)

Frequency shift Z[e aT f (t)] = F (eaT z) I Inverse z-transform: Sampled time function from its z-transform
n F (z) I Only yields the values of the time function at the sampling instants
Time shift Z[f (t nT )] = z
I Results in closed-form time functions that are only valid at sampling
Complex scale Z[c k f (t)] = F (cz) where c 2 C
instants
Complex di↵erentiation Z[tf (t)] = T z dz
dF (z) I 2 approaches
" 1 #
X I Partial-fraction expansion PFE
Real convolution Z f1 (kT )f2 (nT kT ) = F1 (z)F2 (z) I Power series
k= 1

Initial value theorem f (0) = lim F (z)


z!1

Final value theorem f (1) = lim (z 1)F (z)


z!1

Note: t may be substituted for kT in the table

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 15 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 16 / 55

13 Digital control systems 13.3 The z-transform 13 Digital control systems 13.3 The z-transform

Approach – PFE, [1, p. 733] Example, [1, p. 733]


Procedure
1. Sampled exponential time functions are related to their z-transforms
z
e akT
⌧ aT
z e Example (Inverse z-transform via PFE)
I Problem: Find the sampled time function
2. Predict that a PFE should be of the following form
Az Bz 0.5z
F (z) = + + ... F (z) =
z z1 z z2 (z 0.5)(z 0.7)
I Solution: On the board
3. PFE of F (s) did not contain terms with s in numerator of partial
fractions
F (z)
4. Form z to eliminate z terms in numerator
F (z)
5. Perform a PFE of z
6. Multiply the result by z to replace the zs in the numerator
Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 17 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 18 / 55
13 Digital control systems 13.3 The z-transform 13 Digital control systems 13.3 The z-transform

Approach – power series, [1, p. 734] Example, [1, p. 734]

Procedure Example (Inverse z-transform via power series)


I Values of the sampled time function found directly from F (z) I Problem: Find the sampled time function
I Does not yield closed-form expressions for f (kT )
0.5z
F (z) =
1. Indicated division yields a power series for F (z) (z 0.5)(z 0.7)
2. Transform power series for F (z) into F ⇤ (s) and f ⇤ (t) I Solution: On the board

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 19 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 20 / 55

13 Digital control systems 13.4 TFs 13 Digital control systems 13.4 TFs

TFs of sampled-data systems, [1, p. 736]

1 13 Digital control systems


13.1 Introduction
13.2 Modeling the digital computer
13.3 The z-transform
13.4 Transfer functions
13.5 Block diagram reduction Concept
13.6 Stability I The output is conceptually
13.7 Steady-state error sampled in synchronization
13.8 Transient response on the z-plane with the input by a phantom
13.9 Gain design on the z-plane sampler
13.10 Cascade compensation via the s-plane
13.11 Implementing the digital compensator

Figure: Sampled-data systems: a. CT;


b. sampled input; c. input & output

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 21 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 22 / 55

13 Digital control systems 13.4 TFs 13 Digital control systems 13.4 TFs

Pulse TF, [1, p. 736] Example, [1, p. 737]


Concept
I Sampled input is a sum of impulses
1
X
r⇤ (t) = r(nT ) (t nT )
Example (Converting G(s) in cascade with ZOH to G(z))
n=0
I I Problem: Given a ZOH in cascade with the OL TF, G(s), find the
Output
1
X sampled-data TF, G(z), if the sampling time, T = 0.5 seconds
c(t) = r(nT )g(t nT )
1 e Ts s+2
n=0
I
ZOH = ; G(s) =
Sampled output s s+1
1
X
k
C(z) = c(kT )z I Solution: On the board
k=0
I ...substitution...
I t = kT and m = k n
1
X 1
X
m n
C(z) = G(z)R(z) = g(mT )z r(nT )z
m=0 n=0
Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 23 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 24 / 55
13 Digital control systems 13.5 Block diagram reduction 13 Digital control systems 13.5 Block diagram reduction

Methodology, [1, p. 739]

1 13 Digital control systems Procedure


I Find the CL sampled-data TF of an arrangement of subsystems
13.1 Introduction
I Be careful!
13.2 Modeling the digital computer
I E.g. z[G1 (s)G2 (s)] 6= G1 (z)G2 (z)
13.3 The z-transform
13.4 Transfer functions 1. Multiply s-domain functions before taking z-transform
13.5 Block diagram reduction 2. Place a phantom sampler at the output of any subsystem that has a
13.6 Stability sampled input
13.7 Steady-state error I Justification is that the output of a sampled-data system can only be
13.8 Transient response on the z-plane found at the sampling instants, and the signal is not an input to any
13.9 Gain design on the z-plane other block
13.10 Cascade compensation via the s-plane 3. Add phantom samplers at the input to summing junctions whose
outputs are sampled
13.11 Implementing the digital compensator I Justification is that the sampled sum is equivalent to the sum of the
sampled inputs, and that all samples are synchronized
4. Use block diagram manipulations to yield isolated TFs with input and
output samplers
Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 25 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 26 / 55

13 Digital control systems 13.5 Block diagram reduction 13 Digital control systems 13.5 Block diagram reduction

Sampled-data systems, [1, p. 739] Example, [1, p. 740]

Example (Pulse TF of a FB system)


I Problem: Find the z-transform

Figure: Sampled-data system

I Solution: On the board


Figure: Sampled-data systems and their z-transforms

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 27 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 28 / 55

13 Digital control systems 13.6 Stability 13 Digital control systems 13.6 Stability

Digital system stability via the z-plane, [1, p. 742]

1 13 Digital control systems Concept


13.1 Introduction I Sampling rate changes TR & stability
13.2 Modeling the digital computer I Relate s-plane stability to z-plane stability
13.3 The z-transform
I Substitution of z = eT s & s = ↵ + j!
13.4 Transfer functions
13.5 Block diagram reduction z = eT (↵+j!)
13.6 Stability
13.7 Steady-state error = e↵T ej!T
13.8 Transient response on the z-plane = e↵T (cos(!T ) + j sin(!T ))
13.9 Gain design on the z-plane = e↵T \!T Figure: Mapping regions
13.10 Cascade compensation via the s-plane of s-plane onto z-plane
13.11 Implementing the digital compensator I s-plane regions ! z-plane regions
I RHP ! region outside unit circle
I j!-axis ! unit circle
I LHP ! region inside unit circle

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 29 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 30 / 55
13 Digital control systems 13.6 Stability 13 Digital control systems 13.6 Stability

Digital system stability via the z-plane, [1, p. 743] Example, [1, p. 745]

Concept Example (Range of T for stability)


I Digital control system is I Problem: Determine the ranges of the sampling interval, T , that will
I Stable: All CL poles are inside the unit circle make the system stable and unstable
I Unstable: Any pole is outside the unit circle and/or there are poles of
multiplicity greater than 1 on the unit circle
I Marginally stable: Poles of multiplicity 1 are on the unit circle and all
other poles are inside the unit circle
I Tabular methods for determining stability, e.g., Routh-Hurwitz
stability criterion, exist for sampled-data system
I Raible’s tabular method Figure: Digital system
I Jury’s stability test
I Bilinear transformations ! Routh-Hurwitz stability criterion I Solution: On the board

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 31 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 32 / 55

13 Digital control systems 13.6 Stability 13 Digital control systems 13.6 Stability

Bilinear transformations, [1, p. 746] Digital system stability via the s-plane, [1, p. 747]
Concept
I Exact transformations:
Concept
I Stability bilinear transformations: Used to obtain stability information
Ts ln(z) about the digital system by working in s-plane. The resulting TR of
z=e and s=
T CT system, G(s), is not same as that of DT system, G(z).
I Bilinear transformations: Mappings from the complex plane where
one point, s, is mapped into another point, z, of the form z+1 s+1
s= and z=
z 1 s 1
as + b dz + b
z= and s=
cs + d cz a I s-plane regions ! z-plane regions
I j!-axis ! points on unit-circle
I I RHP ! points outside unit circle
Allow application of s-plane analysis & design to digital systems
I I LHP ! points inside unit circle
Yield linear arguments when transforming in both directions through
direct substitution and without the complicated z-transform I Transforms the denominator of the pulsed TF, D(z), to the
I Di↵erent values of a, b, c, & d have been derived for particular denominator of a CT TF, D(s), allowing the use of Routh-Hurwitz
applications & yield various degrees of accuracy when comparing stability criterion
properties of continuous & sampled functions
Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 33 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 34 / 55

13 Digital control systems 13.6 Stability 13 Digital control systems 13.7 Steady-state error

Example, [1, p. 748]

1 13 Digital control systems


13.1 Introduction
13.2 Modeling the digital computer
Example (Stability via Routh-Hurwitz stability criterion) 13.3 The z-transform
I Problem: Given the CL CE, D(z), i.e. the denominator of the CL TF, 13.4 Transfer functions
T (z), use Routh-Hurwitz stability criterion to find the number of 13.5 Block diagram reduction
z-plane poles T (z) inside, outside, and on the unit circle. Is the 13.6 Stability
system stable? 13.7 Steady-state error
D(z) = z 3 z 2 0.2z + 0.1 13.8 Transient response on the z-plane
13.9 Gain design on the z-plane
I Solution: On the board 13.10 Cascade compensation via the s-plane
13.11 Implementing the digital compensator

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 35 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 36 / 55
13 Digital control systems 13.7 Steady-state error 13 Digital control systems 13.7 Steady-state error

Intro, [1, p. 749] Intro, [1, p. 750]

Concept
I For digital systems, the placement of the sampler changes the OL TF.
Assume the typical placement of the sampler after the error and in
the position of the cascade controller.
I Sampled error
R(z)
E ⇤ (s) = E(z) =
1 + G(z)
I Final value theorem for discrete signals
Figure: a. Digital FB control system
e⇤ (1) = lim z 1
z E(z) for evaluation of steady-state errors; b. Figure: c. pushing G(s) and its
z!1
phantom samplers added samplers to the right past the picko↵
z 1 R(z) point; d. z-transform equivalent system
= lim z
z!1 1 + G(z)

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 37 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 38 / 55

13 Digital control systems 13.7 Steady-state error 13 Digital control systems 13.7 Steady-state error

Common inputs, [1, p. 750] Example, [1, p. 752]

Table: Sampled steady-state error


Example (Finding steady-state error)
Input R(s) R(z) Static error constant e⇤ (1) I Problem: Find the steady-state error for step, ramp, and parabolic
Unit step 1 z
Kp = lim G(z) 1 inputs
s z 1 1+Kp
z!1 10
Unit ramp 1 Tz
Kv = 1
lim (z 1)G(z) 1 G(s) =
s2 (z 1)2 T z!1 Kv s(s + 1)
2 T 2 z(z+1) 1 2 1
Unit parabolic Ka = lim (z 1) G(z) I
s3 2(z 1)3 T 2 z!1 Ka Solution: On the board

Note: Multiple pole placement at z = 1 reduces the steady-state error to


zero

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 39 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 40 / 55

13 Digital control systems 13.8 TR on the z-plane 13 Digital control systems 13.8 TR on the z-plane

Visual interpretation of the z-plane, [1, p. 753]

1 13 Digital control systems s-plane ! z-plane


13.1 Introduction I Constant Ts

13.2 Modeling the digital computer I Constant real part, = T4s


I Vert. lines ! conc. circles
13.3 The z-transform
I s = + j! ! z = rej!T
13.4 Transfer functions
I Constant Tp
13.5 Block diagram reduction
13.6 Stability I Constant im. part, ! = T⇡p
I Hori. lines ! radial lines
13.7 Steady-state error
I s = + j! ! z = e T ej✓
13.8 Transient response on the z-plane
I Constant %OS
13.9 Gain design on the z-plane
I Constant ⇣, ! =
13.10 Cascade compensation via the s-plane
13.11 Implementing the digital compensator tan(sin 1 (⇣)) = p⇣
1 ⇣2
I Radial lines ! spiral lines
I s = + j! ! ◆

⇣ Figure: Constant ⇣, normalized Ts , &
!T p normalized Tp plots on the z-plane
z=e 1 ⇣2
\!T
Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 41 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 42 / 55
13 Digital control systems 13.9 Gain design on the z-plane 13 Digital control systems 13.9 Gain design on the z-plane

Digital system RL, [1, p. 755]

1 13 Digital control systems Concept


13.1 Introduction I Plot RL & determine gain for
13.2 Modeling the digital computer stability & TR requirement
13.3 The z-transform I Same CT RL rules
13.4 Transfer functions I Stability divided by unit circle
13.5 Block diagram reduction rather than imaginary axis
13.6 Stability
I Superimpose TR curves on
13.7 Steady-state error
13.8 Transient response on the z-plane z-plane
13.9 Gain design on the z-plane I Same drawback with CT RL,
Figure: Generic digital FB control
13.10 Cascade compensation via the s-plane limited to simple gain
system
13.11 Implementing the digital compensator adjustment to accomplish
design objective
I Limitations solved with
compensation
Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 43 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 44 / 55

13 Digital control systems 13.9 Gain design on the z-plane 13 Digital control systems 13.9 Gain design on the z-plane

Example, [1, p. 755] Example, [1, p. 756]

Example (Stability design via RL) Example (Stability design via RL)
I Problem: Sketch the RL and determine the range of gain, K, for I Problem: Find the value of gain, K, to yield
stability from the RL plot I ⇣ = 0.7

Figure: Digital FB control Figure: Digital FB control

I Solution: On the board I Solution: On the board

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 45 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 46 / 55

13 Digital control systems 13.10 Cascade compensation via the s-plane 13 Digital control systems 13.10 Cascade compensation via the s-plane

Intro, [1, p. 758]

1 13 Digital control systems Concept


13.1 Introduction I No design directly in the z-domain
13.2 Modeling the digital computer
1. Design on the s-plane
13.3 The z-transform 2. s-plane design ! digital implementation, i.e., bilinear transformation
13.4 Transfer functions 3. Apply cascade compensator
13.5 Block diagram reduction I Tustin transformation: Bilinear transformation that can be performed
13.6 Stability
with hand calculations & yields a DT TF whose output response at
13.7 Steady-state error
sampling instants is approximately same as equivalent CT TF
13.8 Transient response on the z-plane
13.9 Gain design on the z-plane s+ 2
1+ T
2(z 1) T 2s
13.10 Cascade compensation via the s-plane s= and z= 2 = T
T (z + 1) s T 1 2s
13.11 Implementing the digital compensator
I # T ! DT compensator ⇡ CT compensator
I " T ! discrepancy higher frequencies

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 47 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 48 / 55
13 Digital control systems 13.10 Cascade compensation via the s-plane 13 Digital control systems 13.10 Cascade compensation via the s-plane

DT compensator design notes, [1, p. 759] Example, [1, p. 760]

I Selecting the sampling interval, T


I Astrom & Wittenmark, 1984 Example (Digital cascade compensator design)
I Problem: Design a digital lead compensator in the s-domain and
0.15 0.5
T ⇡ to transform the compensator to the z-domain
! M ! M
1
I Where ! M is the 0 dB frequency (rad/s) of the magnitude frequency Gp (s) =
response curve for the cascaded CT compensator & plant
s(s + 6)(s + 10)
I Rule of thumb
10 20 I %OS = 20%
T ⇡ to
!BW !BW I Ts = 1.1 seconds
I Solution: On the board
I Where !BW is the frequency at which the magnitude frequency
response is 3 dB below the magnitude at 0 frequency

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 49 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 50 / 55

13 Digital control systems 13.11 Implementing the digital compensator 13 Digital control systems 13.11 Implementing the digital compensator

Algorithm to emulate the compensator, [1, p. 762]

1 13 Digital control systems Concept


13.1 Introduction I 2nd -order example
I Compensator
13.2 Modeling the digital computer
13.3 The z-transform X(z) a3 z 3 + a2 z 2 + a1 z + a0
13.4 Transfer functions Gc (z) = =
E(z) b2 z 2 + b1 z + b0
13.5 Block diagram reduction
13.6 Stability I ...cross multiply, solve for X(s), inverse z-transform...
13.7 Steady-state error a3 ⇤ a2 a1 a0 ⇤
13.8 Transient response on the z-plane x⇤ (t) = e (t + T ) + e⇤ (t) + e⇤ (t T) + e (t 2T )
b2 b2 b2 b2
13.9 Gain design on the z-plane b1 ⇤ b0 ⇤
13.10 Cascade compensation via the s-plane x (t T ) x (t 2T )
b2 b2
13.11 Implementing the digital compensator
I To be physically realizable
I Present sample of compensator output, x⇤ (t), cannot be a function of
future sample of error, e⇤ (t + T ) ! a3 = 0
I Compensator TF numerator order  denominator order
Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 51 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 52 / 55

13 Digital control systems 13.11 Implementing the digital compensator 13 Digital control systems 13.11 Implementing the digital compensator

Intro, [1, p. 763] Example, [1, p. 764]


Concept
I The output is a weighted linear combination of several successive
values of the input & output
Example (Digital cascade compensator implementation)
I Problem: Develop a block diagram for the digital compensator

X(z)
Gc (z) =
E(z)
z + 0.5
= 2
z 0.5z + 0.7
I Solution: On the board

Figure: Flowchart for a 2nd -order DT compensator

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 53 / 55 Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 54 / 55
13 Digital control systems 13.11 Implementing the digital compensator

Bibliography

Norman S. Nise. Control Systems Engineering, 2011.

Bayen (EECS, UCB) Feedback Control Systems September 10, 2013 55 / 55