4/5/2026

DEADBEAT RESPONSE AT SAMPLING

INSTANTS
• Control systems are often design with objective that
output response should reach the desired reference
value as quickly as possible and without any overshoot.
• Such type of response is generally referred to as a
Module-3 deadbeat response.

1 Figure (a) Figure (b) 2

• Figure (a) shows the deadbeat response of a • For a discrete-time system to present a deadbeat
continuous-time system to a unit step input. response at the sampling instants , the closed loop
• Figure (b) shows the deadbeat response of a transfer function should be of the form:
discrete-time system to a unit step input.
Y (z) 1
ØThe output sequence are shown to follow the 
R(z) z N
reference unit step input after two sampling
instants. • Where N is a positive integer.
ØIf response of Figure (b) is a sampled representation • Thus, for a unit step function, for example, the
of a continuous-time system, then deadbeat output, Y(z) is a expanded as:
characteristics at the sampling instants do not
Y (z)  0.3z 1  z 2  z 3  z 4 
guarantee that continuous-time response is also
deadbeat. • which corresponds to a deadbeat response for N = 2
3 4

1
 4/5/2026

DESIGN OF DISCRETE-DATA SYSTEMS WITH DEADBEAT ØOne must distinguish between the design of
RESPONSE deadbeat response for a digital control systems and
ØA deadbeat response is one that follows the for a sampled-data system.
reference input without error in minimum finite
time. ØAs the signal between the samples in the digital
ØDiscrete-data systems are capable of having control system are zero.
deadbeat responses due to the switching
operations on signals. ØWhereas the output of a sampled data system are
ØWithout swi tching or sampling operation, a continuous time function.
deadbeat response cannot be achieved in linear
continuous-data control systems.
ØTherefore, a deadbeat response at the sampling
ØIt is necessary that the characteristic equation of a instants of sampled-data system must guarantee
discrete-data system with deadbeat response be of that no ripples occurs between the sampling
form:  ( z )  z N where N is a positive integer.
5
instants. 6

DESIGN OF DIGITAL-DATA SYSTEMS

WITH DEADBEAT RESPONSE ØDesign approach can divided into two cases:
Ø The deadbeat response design is characterized by
the following design criteria: • Case-1 : When the process (plant) transfer function
(1) The system must have zero steady state error at has all poles and zeros within the unit circle.
the sampling instants for specified input signal.
(2) Time for the output to reach steady-state value • Case-2: When the process (plant) transfer function
should be finite and minimum. with poles or zeros on or outside the unit circle.
(3) The designed digital controller D(z) must be
physically realizable (D(z) must not have more zeros .
than poles)
7 8

2
 4/5/2026

Case-1: First consider the case where all the poles and
zeros are within the unit circle ØSolving for D(z), one gets:
ØConsider the block diagram of closed loop system of 1  M ( z) 
D(z) 
an-all digital system. Gp (z)  (1 M (z)) 

ØThe error signal : E(z)  R(z)  Y (z)

 R(z)1 M (z)

ØClosed loop transfer function becomes: ØThe z-transform of input, R(z) in general, be
Y (z) D(z)Gp (z) described by function:
M (z)   A(z)
R(z) 1 D(z)Gp (z) R(z) 
1 z1 
N
9 10

• where N is a posi tive integer a nd A (z) is a • As the polynomial A(z) does not contain any zeros at
polynomial in z1 with no zeros at z = 1. z = 1,
• For example: • necessary condition for steady state error to be zero
is that (1- M(z)) contain the factor 1  z 1  .
N
a) For a unit step input : A(z) = 1 and N = 1
b) For a unit ramp input ; A(z)  Tz1 and N = 2 • Thus, 1 - M(z) should have the form:
 
N
c) For a parabolic input; A(z)  T 2 z(z 1) and N = 3, 1  M ( z )  1  z 1 F ( z)

Ø For obtaining zero steady state error, • where F(z) is polynomial of z .

lim e( kT )  lim 1  z 1  E ( z ) ØSolving for M(z) in last equation:
k  z 1
Q( z )
 
 lim 1  z 1 R ( z ) 1  M ( z )  
M ( z )  1  1  z 1 
N
F ( z) 
z 1
zP
A( z )

 lim 1  z 1  1  M ( z )   0 Øwhere Q(z) is polynomial in z and P  N
1  z 
z 1 N
1
11 12

3
 4/5/2026

ØThis verifies that characteristic equation of the Ø“When the characteristic equation of digital system
system with zero steady-state in finite time is in control system is in form of z P  0 (all the poles of
form of: z P  0 closed loop system are at z = 0), the error signal of
• The Z-transform of error-function now becomes: system will go to zero in a finite number of sampling
A(z) periods”.
E(z)  1 M (z)
1 z1 
N
Physical Realizabilty Consideration
A(z) ØFor a given plant Gp(s) , the physical realizability
N  
1 N
 1 z F ( z ) condition imposes constraints on the selection of
1 z1  M(z).
 A(z)F (z) ØLet power series expansion of Gp(z) and M(z) are
ØAs A(z) and F(z) are polynomial of finite terms, so given as: G (z)  g zn  g zn1  g zn2 
E(z) will have a finite number of terms in its power- p n n1 n2

series expansions in the inverse powers of z. 13

M (z)  mk z  mk 1z  mk 2 zk 2 
k k 1
14

• where n and k are excess poles over zeros of Gp(z) • If the Gp(z) does not have poles or zeros on or
and M(z) respectively. outside the unit circle , closed loop transfer M(z)
1  M ( z)  can be chosen according to the type of input as:
• Then using D(z) 
Gp (z)  (1 M (z)) 
z 1
• We get the power series expansion of D(z) as: • Step Input R(z)  ; M (z)  n
z 1 z
D(z)  dk n z(k n)  dk n1z(k n1)  dk n2 z(k n2) 
Tz 2z 1
• For D(z) to be physically realizable , one should • Ramp Input R(z)  ; M (z)  n1
 z 1
2
z
have: k  n
Tz(z 1) 3z2  3z 1
• Parabolic Input R( z )  ; M ( z ) 
 z 1 zn2
3

ØThe excess of poles over zeros of M(z) must be at

least equal to the excess of poles over zeros of Gp(z). where n is the excess of the poles over zeros of Gp(z)
15 16

4
 4/5/2026

Case-2: Plant Transfer function with unstable poles or • This i mposes addi ti onal constrai ned on the
zeros selection of the closed loop transfer function M(z).
ØEnd result of case -1 procedure leads to the • Consider the plant transfer to be written as:
cancellation of the poles and zeros of the plant K

Gp(z) by the zeros and poles of the digital controller (1 z zi

1
)
D(z). Gp (z)  i 1
L
G1 (z)
ØIf Gp(z) has a pole or zero on or outside the unit (1 p z
j 1
j
1
)
circle, imperfect cancellation ( which is very likely to
occur in practice) would lead to an unstable closed • where zi , i  1, 2, , K and p j , j  1, 2, , L are the
loop system. zeros and poles, respectively, of Gp(s) that are
ØFor all practical purposes, no attempt should be outside or on the unit circle.
made to cancel the poles and zeros of Gp(z) that • G1(s) is a rational function in z 1 with poles and
are on or outside the unit circle. 17
zeros only inside unit circle. 18

• Using the equation:

 1  z z 
K
1
1  M ( z)  ØSo the M(z) must contain the factors:
D(z)  i
Gp (z)  (1 M (z))  i 1

 1  p z 
L
1
• One gets the controller transfer function as: Ø(1-M(z)) must contain the factors: j
L j 1

 (1  p z
j 1
j
1
)
 M ( z)  ØThus in general M(z) and (1-M(z)) have the following
D( z )  K  (1  M ( z ))  terms:
 (1  z z 1 ) G ( z ) 
i 1

K 
i 1

ØD(z) cannot contain the polesp j and zeros zi as

 
M ( z)   1 zi z 1  Mk z k  Mk 1z k 1  
 i1 
its zeros and poles respectively.

M(z)) and M(z)

L
19


P
 
ØThey must be cancelled by the zeros and poles of (1- 1 M ( z)   1 pi z 1  1 z 1 1 a1z 1  a2 z 2 
 i1
  20


5
 4/5/2026

Ø In the equation of M(z), k  n where n is the number of • Q) To design a deadbeat controller so that output
excess poles over zeros of Gp(z). y(kT) will follow a unit step input in minimum time.
Ø The power of the term1  z , P, should be equal to
1
Plot the response y(kT) to show the effect.
either the order of poles of R(z), N, or to the order of
the poles of Gp(z) at z = 1, whichever is greater. 0 .0125 ( z  0 .195 )( z  2 .821 )
G p ( z) 
Ø The truncation of last two equation can be done by z ( z  1)( z  1 .368 )( z  0 .8187 )
following conditions:
(1) The orders of poles of M(z) and (1-M(z)) must be
equal. • Step 1: plant has a poles , z =1 on the unit circle
(2) The total number of unknowns in and one pole z=1.368, One zero, z = -2.821 outside
the unit circle.
M k , M k 1 , and a1 , a2 
• Step 2:Multiply the num and den of Gp(z) by z 4
must be equal to the order of M(z) so that these
coefficient can be solved independently. 0 .0125 z 2 (1  0 .195 z 1 )(1  2 .821 z 1 )
G p ( z) 
21
(1  z 1 )(1  1 .368 z 1 )(1  0 .8187 z 1 )
22

Step 6: Finding value of P

Step 3: Zeros outside the unit circle
• No of Poles of Gp(z) at z = 1 is 1
M(z) must contains factor of (1  2 . 821 z  1 )
• Input is step function : N =1
• P=max(No of Poles of Gp(z) at z = 1 , N) = 1
Step 4 : Pole over zero excess of M(z)
Step 7: general form of M(z) and 1- M(z)
Gp(z) has two more poles than zeroes, n = 2
K 
so pole over zero excess of M(z) , k , should be:   
M ( z)   1 zi z 1  Mk z k  Mk 1z k 1 
k  n (select k to minimum value)  i1 
Therefore k = 2
Step 5: poles outside the unit circle
L
 
  P


1 M ( z)   1 pi z 1  1 z 1 1 a1z 1  a2 z 2 
 i1 

1- M(z) must contains factor of (1  1 . 368 z )
1

23 24

6
 4/5/2026

• Putting all the information in general form , one get:

M ( z)  (1  2.821z 1 )M 2 z 2  M3 z 3  M 4 z 4   • Step 9: solving the unkno
M ( z)  M2 z2  (M3  2.821M2 ) z3  2.821M3 z4
1  M ( z)  (1  1.368z )(1  z )1  a1z  a2 z  
1 1 1 2

1  M ( z)  (1  2.368z1  1.368z2 ) 
Step 8: Truncation condition
 (a1z1  2.368a1z2  1.368a1z3 )
Order of M(z) = Order of (1-M(z))
Total no .of unknown parameter = Order of M(z)  (a2 z2  2.368a2 z3  1.368a2 z4 )
• simplify
• Min order of M(z) possible= 4
1  M ( z)  1  (a1  2.368) z 1 
• Min order of 1- M(z) possible =4
M ( z)  (1  2.821z )M2 z  M3 z
1 2 3
 (1.368 2.368a1  a2 ) z 2 
1  M ( z)  (1 1.368z 1 )(1  z 1 )1  a1z 1  a2 z 2 
25
 (1.368a1  2.368a2 ) z 3  1.368a2 z 4 26

• Putting expression for M(z) in M(z) and equating the

coefficients of like-powers • Solve M2 and M3
( a1  2 .368 )  0
(1 .368  2 .368 a1  a 2 )  M 2 • Form M(z)

(1 .368 a1  2 .368 a 2 )  ( M 3  2 .821 M 2 ) M ( z)  (1  2.821z 1 )M2 z 2  M3 z 3 

1 .368 a 2  2 .821 M 3 • Obtain Y(z) = M(z) R(z)
 0 0 1 0 M2   2.368 
 1 1 M3  1.368
• Take inverse z-transfrom in power series and
0  2.368
      plot y(kT)
2.821 1 1.368  2.368 a1   0  1  M ( z) 
• Obtain Controller as D(z) 
     Gp (z)  (1 M (z)) 
 0  2.821 0 1.368  a2   0  27 28

7
 4/5/2026

Realization of Digital Controller • Taking cross multiplication

ØThere are three basic method of programming: a0 E2 ( z)  a1z 1E2 ( z)  a2 z 2 E2 ( z)    an z  n E2 ( z)
1. Direct Programming
 b0 E1 ( z)  b1z 1E1 ( z)  b2 z 2 E1 ( z )    bm z  m E1 ( z)
2. Cascade Programming
3. Parallel Programming • and then taking inverse z-transform.
1. Direct Programming a0e2* (t )  a1e2* (t  T )  a2e2* (t  2T )    ane2* (t  nT )
Ø Transfer function of digital controller can be given
as:  b0e1* (t )  b1e1* (t  T )  b2e1* (t  2T )    bne1* (t  nT )
1 2 m
E2 ( z) b0  b1z  b2 z    bm z
D( z)   • It can be written as:
E1 ( z) a0  a1z 1  a2 z 2    an z n n m
Ø where a0  0 if b0  0 and m and n are positive a0e2* (t )   ak e2* (t  kT )   bj e1* (t  jT )
integer with  n  m . 29
k 1 j 0
30

• Solving for the output signal of controller: The block diagram representation of the direct digital
programming:
m n
1 1
e2* (t )  
a0 j 0
bj e1* (t  jT )   ak e2* (t  kT )
a0 k 1
ØPresent values of e2* (t ) depends on the present and
*
past values of input, e1 (t ) as well as past information
of output e2* (t ).
ØTo implement thi s: Two basic mathemati cal
operation are required:
(1) Data Storages
(2) Arithmetic Manipulations
ØIt require a total of (n + m) data storage units.
31 32

8
 4/5/2026

• Data Storages are simply time delay units which • Assuming n = m, Block diagram of direct digital
delay the input data for one sampling period. programming by direct decomposition can given as:
• A s a n a l te r n a t i ve . O n e ca n u se t h e d i r e c t -
decomposition method for implementing the direct
digital programming.
• In that case, we have:
E2 ( z ) 
1
a0
 
b0  b1 z 1  b2 z  2    bm z  m X ( z )

X (z) 
1
a0
E1 ( z ) 
1
ao
 
a1 z 1  a 2 z  2    a n z  n X ( z )

• where X(z) is a dummy variable.

• Number of data-storage units can be is equal to the
greater of n and m. 33 34

2) Cascaded Digital Programming • where p is the greater of n and m.

ØThe transfer function D(z) can be written as a • In general , the transfer function D k ( z ) is assume
product of a number of a simple transfer functions. the following forms depending on the poles and
ØEach simple transfer function can be realizable by a zeros of D(z) and relative magnitude of m and n.
simple digital program. • Real Pole and Zero:
ØIn this case, digital programming of D(z) can be 1  c k z 1
Dk ( z )  K k
represented by series of cascaded digital programs 1  d k z 1
of the simple transfer functions. • Two Complex Conjugate Poles:
ØThe Controller transfer function can be written in 1
the factored form as: Dk ( z )  K k
1  d k z  f k z 2
1

p
D ( z )   Dk ( z ) • One Real Zero and Two Complex Conjugate Poles:
k 1
35 36

9
 4/5/2026

1  c k z 1 • Figure be low shows the block diagram

Dk ( z )  K k representation of the cascade digital programming
1  d k z 1  f k z  2
• Complex Conjugate Poles and Zeros: of D(z):

1  g k z  1  hk z  2
Dk ( z )  K k
1  d k z 1  f k z  2
• Real Zero (m > n):
D k ( z )  K k 1  c k z 1  • Each of simple transfer function can be realized by
use of direct digital programming method.
• Complex Conjugate Zeros ( m > n):
3) Parallel Digital Programming
D k ( z )  K k 1  g k z 1  hk z 2 
ØThe transfer function D(z) is first expanded by
• One Real Pole and Two Complex Conjugate zeros partial-fraction expansion into a sum of simple first-
(m > n) 1  g k z  1  hk z  2
Dk ( z )  K k or second-order transfer functions.
1  d k z 1 37 38

• In the general, The transfer function D(z) can be • Pole at z = 0; j = 1, 2, 3, … , N;

written as: p K
Dk ( z )  kj
D ( z )   Dk ( z ) z 
k 1
• where p is greater of m and n. • Each of these transfer functions can be realized by
• Depending on the nature of D(z), D k ( z ) can be of direct digital programming method.
the following forms: • The block diagram of parallel programming is as
1) Real pole of multiplicity j; j=1, 2, 3, …, N. shown below:
Kk
Dk ( z ) 
1  d k z 1  j
2) Complex poles of multiplicity j; j=1, 2, 3, …, N
K k 1  c k z 1 
Dk ( z ) 
1  d k z 1  f k z  2 j 39 40

10