Fpw chapter 4 - digital ctrl of dynamic systems

cHAPTER 4
Discrete Equivalents to Continuous
Transfer Functions: The Digital Filter
4.1 INTRODUCTION
One of the exciting fields of application of digital systems! is in signal pro-
cessing and digital filtering. A filter is a device designed to pass desirable
elements and hold back or reject undesirable ones; in signal processing it
is common to represent signals as a sum of sinusoids and to define the
"desirable elements" as those signals whose frequency components are in a
specified band. Thus a radio receiver filter passes the band of frequencies
transmitted by the station we want to hear and rejects all others. We would
call such a filter a bandpass filter. In electrocardiography it often happens
that power-line frequency signals are strong and unwanted, so we design a
filter to pass signals between 1 and 500 Hz but to eliminate those at 60
Hz. The magnitude of the transfer function for this purpose may look like
Fig. 4.1 on a log-frequency scale, where the amplitude response between
59.5 and 60.5 Hz might reach 10-3. Here we have a band-reject filter with a
60-dB rejection ratio in a I-Hz band centered at 60 Hz.
In long-distance telephony some filters playa conceptually different role.
There the issue is that transmission media- wires or microwaves-introduce
distortion in the amplitude and phase of a sinusoid which must be removed.
Filters to accomplish this correction are called equalizers. And in control we
must control systems whose dynamics require modification in order that the
complete system have satisfactory dynamic response. We call the devices
that make these changes compensators.
1Including microprocessors and special-purpose, very large-scale integration (VLSI)
digital chips for signal processing, called DSP chips.
133

134 CHAPTER 4 DISCRETE EQUIVALENTS
20 log I HI t
0-
-3-
60
w
500 211"
Figure 4.1 Magnitude of a low-frequency bandpass filter with a narrow rejection
band.
Whatever the name- filter, equalizer, or compensator- many fields have
found use for devices having specified characteristics of amplitude and phase
transmission, and the trend is to perform these functions by digital means.
The design of electronic filters is a well-established subject that includes not
only very sophisticated techniques but also well-tested computer programs
[Van Valkenburg(1982)]. Much of the effort in digital filter design has been
directed toward the design of digital filters that have the same characteristics
(as nearly as possible) as those of a satisfactory continuous design. For digital
control systems we have much the same motivation: Because continuous-
control designs are well established, we should like to know how to take
advantage of a good continuous design and cause a digital ,computer to
produce a discrete equivalent to the continuous compensator. This method
of design is called emulation. Although much of our presentation in this
book is oriented toward direct digital design and away from emulation of
continuous designs with digital equivalents, it is important to understand the
techniques of discrete equivalents for purposes of comparison and because it
is widely used by practicing engineers.
Thus we are led to the specific problem of this chapter: Given a transfer
function, H(s), what discrete transfer function will have approximately the
same characteristics? We will present here three approaches to this task:
Method 1: numerical integration.
Method 2: pole-zero mapping.
Method 3: hold equivalence.

4.2 DESIGN OF DISCRETE EQUIVALENTS 135
4.2 DESIGN OF DISCRETE EQUIVALENTS
BY NUMERICAL INTEGRATION
The topic of numerical integration of differential equations is quite complex,
and only the most elementary techniques are presented here. For example,
we only consider formulas of low complexity and fixed step-size. The funda-
mental concept is to represent the given filter transfer function H (s) as a
differential equation and to derive a difference equation whose solution is an
approximation La that of the differential equation. For example, the system
U(s) _ H s _ a
E(s)- ()-s+a
is equivalent to the differential equation
u+au = ae.
(4.1)
(4.2)
Now, if we write (4.2) in integral form, we have a development much like
that of (2.5) in Chapter 2, except that the integral is more complex here:
u(t) = fat[-au(r) + ae(r)Jdr,
In
· kT-T lkT
u(kT) = [-au + aeJdr + [-au + ae]dr
o kT-T
= u(kT _ T) + {area of -au + ae }. (4.3)
over kT - T ~ r < kT
We can now develop many rules based on our selection of the approximation
of the incremental area term. The first approximation leads to the forward
rectangular rule2 wherein we approximate the area by the rectangle looking
forward from kT - T and take the amplitude of the rectangle to be the value
of the integrand at kT - T. The width of the rectangle is T. The result is
an equation in the first approximation, Ul:
ul(kT) = ul(kT - T) +T[-aul(kT - T) +ae(kT - TJ]
= (1 - aT)ul(kT - T) + aTe(kT - T).
2Also known as Euler's rule.
(4.4a)

The transfer function corresponding to the forward rectangular rule in this
case is
aTz- l
Hp(z) = 1 _ (1 - aT)z- l
a
= --...,........,.--
(z-1)/T+a
(forward rectangular rule). (4.4b)
A second rule follows from taking the amplitude of the approximating
rectangle to be the value looking backward from kT toward kT-T, namely,
-au(kT) + ae(kT). The equation for U2, the second approximation,3 is
u2(kT) = u2(kT - T) +T[-au2(kT) + ae(kT)]
= u2(kT - T) + aT e(kT).
1 + aT 1 + aT
(4.5a)
Again we take the z-transform and compute the transfer function of the
backward rule:
aT 1
HB (z) = 1 + aT 1 _ z-l / (1 + aT)
aTz
z(1 + aT)_- 1
a
(backward rectangular rule). (4.5b)
(z - 1)/Tz + a
O Uf final version of integration rules is the trapezoid rule found by taking
the area approximated in (4.3) to be that of the trapezoid formed by the
average of the previously selected rectangles. The approximating difference
equation is
u3(kT) = u3(kT - T)
T
+ "2[-au3(kT - T) + ae(kT - T) - aU3(kT) + ae(kT)]
1- (aT/2) aT/2
= 1 + (aT/2) u3(kT - T) + 1 + (aT/2) [eJ(kT - T) + e3(k~~.6a)
3It is worth noting that in order to solve for (4.5a) we had to eiiminate u(kT) from
the right-hand side where it entered from the integrand. Had (4.2) been nonlinear,
the result would have been an implicit equation requiring an iterative :solution.
This topic is the subject of predictor-corrector rules, which are beyond our scope
of interest. A discussion is found in most books on numerical analysis. See Golub
and Van Loan(1983).

The corresponding transfer function from the trapezoid rule is
H z _ aT(z + 1)
T( ) - (2 + aT)z + aT - 2
a
(trapezoid rule).
(2/T)[(z - 1)/(z + 1)J +a
Suppose we tabulate our results obtained thus far.
H(s) Method Transfer function
a
Forward rule
a
HF =
s+a (z - l)/T +'CL
a
Backward rule
a
HD =
s+a (z - l)/Tz + a.
a
Trapezoid rule If. _ a
s+a T - (2/T){(z - l)/(z + In + a
(4.6b)
(4.7)
From direct comparison of H(s) with the three approximations in this
tabulation we can see that the effect of each of our methods is to present
a discrete transfer function that can be obtained froIIf the given Laplace
transfer function H(s) by substitution of an approximation for the frequency
variable as shown below:
Method
Forward rule
Backward rule
Trapezoid rule
Approximation
z-l
S t- ----;y-
z-l
st---
Ti
2z-1
s t- - - -
Tz+l
(4.8)
The trapezoid-rule substitution is also known, especially in digital and
sampled-data control circles, as Tustin's method [Tustin (1947)J after the
British engineer whose work on nonlinear circuits stimulated a great deal of
interest in this approach. The transformation is also called the bilinear trans-
formation from consideration of its mathematical form. The design method
can be summarized by stating the rule: Given a continuous transfer function
(filter), H(s), a discrete equivalent can be found by the substitution
HT(z) = H( s) Is=(2/T)[(z-1)/(z+l)]. (4.9)

Each of the approximations given in·(4. ) can be viewed as a map from
the s-plane to the z-plane. A further understanding of the maps can be ob-
tained by considering them graphically. For example, because the (s = jw)-
axis is the boundary between poles of stabl system and pol s of unstable
systems, it would be interesting to know how the jw-axis is mapped by the
three rules and where th left (stable) half of the s-plane appears in the
z-plane. For this purpose we must solve the relations in (4.8) for z in terms
of s. We find
i) z = 1 +Ts, forward rectangular rule,
ii)
1
backward rectangular rule,z= I
I-Ts
iii)
1 +Ts/2
bilinear rule. (4.10)Z= ,
I-Ts/2
If we let s = jw in these equations, we obtain the boundaries of Lhe regions
in the z-plane which originate from the stable portion of the s-plane. The
shaded areas sketched in the z-plane ill Fig. 4.2 are these stable r gion
for each case. To show that rule (li) results in a ircle, ~ i add d Lo aiid
subtracted from the right-hand si e to yield
1 {I I}z = 2+ 1- Ts - 2
1 11 +Ts
2 21- Ts
(4.11)= - -_.
Now it is easy to see that with s = jw, the magnitude of z- ~ is constant,
and the curve is thus a circle as drawn in Fig. 4.2(b). Because the unit circle
is the stability boundary in the z-plane, it is apparent from Fig. 4.2 that the
forward rectangular rule could cause a stable continuous filter to be mapped
into an unstable digital filter.
It is especially interesting to notice that the bilinear rule maps the stable
region of the s-plane exactly into the stable region of the z-plane although
the entire jw-axis of the s-plane is stuffed into the 27f-length of the unit circle!
Obviously a great deal of distortion takes place in the mapping in spite of the
congruence of the stability regions. As our final rule deriving from numerical

--=,-y--_ Unit circle
Ca) (b) (e)
Figure 4.2 Maps of the left-half s-plane to the z-plane by the integration rules of
(4.8). Stable s-plane poles map into the shaded regions in the z-plane. The unit cir-
cle is shown for reference. (a) Forward rectangular rule. (b) Backward rectangular
rule. (c) Trapezoid or bilinear rule.
integration ideas, we discuss a formula that extends Tustin's rule one step
in an attempt to correct for the inevitable distortion of real frequencies
mapped by the rule. We begin with our elementary transfer function (4.1)
and consider the bilinear rule approximation
a
HT(Z) = (2/T)[(z - l)/(z +1)] + a'
The original H(s) had a pole at s = -a, and for real frequencies, s = jw,
the magnitude of H(jw) is given by
2
IH(jwW = 2 a 2
w +a
1
(4.12)
Thus our reference filter has a half-power point, IHI2 = ~, at w = a. It will
be interesting to know where HT(Z) has a half-power point.
As we saw in Chapter 2, signals with poles on the imaginary axis in
the s-plane (sinusoids) map into signals on the unit circle of the z-plane.
A sinusoid of frequep-cy WI corresponds to zi = ejw1T , and the response
of HT(Z) to a sinusoid of frequency WI is HT(ZI). We consider now (4.6b)
for HT(Zl) and manipulate it into a more convenient form for our present

purposes:
(
2 wIT )
=a/ Tjtan-2-+a .
The magnitude squared of HT will be ! when
or
2 wIT
-tan-- = a
T 2
(4.13)
(4.14)
Equation (4.14) is a measure of the frequency distortion or warping caused
by Tustin's rule. Whereas we wanted to have a half-power point at W = a,
we realized a half-power point at WI = (2/T) tan-1(aT/2). WI will be ap-
proximately correct only if aT/2 « 1 so that tan-1
(aT/2) ~ aT/2, that is,
if ws (= 27r/T) » a and the sample rate is much faster than the half-power
frequency. We can turn our intentions around and suppose that we really
want the half-power point to be at WI. Equation (4.14) can be made into
an equation of prewarping: If we select a according to (4.14), then, using
Tustiri"s bilinear rule for the design, the half-power point will be at WI. A
statement of a, complete set of rules for filter design via bilinear transforma-
tion with prewarping is:
a) Write the desired filter characteristic with transform variable s and crit-
ical frequency WI in the form H(s/wI).4
4The critical frequency need not be the band edge. We can use the band center of
a bandpass filter or the crossover frequency of a Bode plot compensator. However,
we must have WI < 11"IT if a stable filter is to remain stable after warping.

b) Replace WI by a such that
2 wIT
a = T tan-
2
-,
and in place of H(S/WI), consider the prewarped function H(s/a). For
more complicated shapes, such as bandpass filters, the specification fre-
quencies, such as band edges and center frequency, should be prewarped
before the continnous design is done; and then the bilinear transforma-
tion will bring all these points to their correct frequencies in the digital
filter.
c) Substitute
2z-1
s=-- -
Tz+1
in H(s/a) to obtain the prewarped equivalent Hp(z).
As a frequency substitution the result can be expressed as
(4.15)
It is clear from (4.15) that when W= WI, Hp(ZI) = H(j1) and the discrete
filter has exactly the same transmission at WI as the continuous filter has at
this frequency. This is the consequence of prewa~ping. We also note that as
the sampling period gets small, Hp(z) approaches H(jW/WI).
Example 4.1: The frequency responses of discrete filters designed
by the four rules studied in this section are plotted in Fig. 4.3. The
filters are discrete equivalents of a third-order lowpass Butterworth5
filter designed to have unity pass bandwidth (wp = ).). A simple fre-
quency scaling would of course translate these curves to any desired
passband frequency. The continuous filter transfer function is
5A description of the properties of Butterworth filters is given in most books on
filter design and briefly in Franklin, Powell and Emami-Naeini(1986).

0.6
JHI
0.4
0.2
0
0 0.5 1.5
0
-50 .
-LOO
L.H
-150
-200
-250
0 0.5 1.5
(a) Normalized frequency Wlwp
bilinear =0
warped = +
backward =*
forward = x
2
2
2.5
2.5
Figure 4.3(a) Response of third-order lowpass filter and digital equivalents for
ws/wp = 207r.
Computation of the discrete equivalents is numerically t dious, and
the state-space algorithms described below were used to generate
the functions and the response curves. In Fig 4.3(a) we see that at a
high sample rate (T = 0.1), where the ratio ofsampling frequency to
passband frequency is wslwp ~ 63, all the rules do reasonably well
but the rectangular rilles are already showing orne deviation.
From Fig. 4.3(b) we see that at ws/wp = 271" the re tangu.lar rules
are useless (the forward rule is unstable).
Finally, in Fig. 4.3(c) at ws/wp = 71" with a sampung period of
T =2 sec we see that only with prewarping do we have a design that
comes even close to the continuous response. Notice in this plot that
at the Nyquist radian frequency, w = 7I"/T, the magnitude response
of the discrete filter starts to repeat according to the periodic nature
of a discrete-transfer-function frequency response. Notice also that
the magnitude and phase of the prewarped designs match those of
the continuous filter exactly at the band edge, w = 1 for all these
cases. This is no surprise, because such matching was the whole idea
of prewarping.

~ : :I'''''''''"""
0.8 -
IHI
0.6
0.4
0.2
0
0 0.5
0
-50
-100
L.H
-150 ...
-200
-250
0 0.5 1.5 2 2.5
(b) Normalized frequency ro/rop
Figure 4.3(b) Response of third-order lowpass filter and digital equivalents for
ws/wp = 2'11'.
IHI
L.H
0.8
0.6
0.4
0.2
. ~
.". ...,,-,.. ...."
..
o
.. ·..··0.........-..:f: ''- -_._..._..... - _...
: + oo +
.....~.••.~ . • - +
•• 0
"I ~ : . 'i".~. or.'......'........_._.jj • .
°o~________-L__________L-__o_O~O~+UL~~~.~~~________~
0.5 1.5 2 2.5
0
-50
-100 ._--......
-150
-200 t..
·250
0 0.5
(c)
........• .c x'~..~ ,.
o ,'- ,,+ ... ...:.... ..:. ~. f, .
o. •
•n' .-
. .0 '0,0 '" ..
0
00
.-
0 0 " ..,
1.5
Normalized freqlency ro/rop
2 2.5
Figure 4.3(c) Response of third-order lowpass filter and digital equivalents for
ws/wp = '11'. Bilinear = 0, warp = +, backward = *, and forward = x.

These formulas for discrete equivalents are particularly simple and conve-
nient when expressed in state-variable form and used with a computer-aided
design package. For example, suppose we have a vector-matrix description
of a continuous design in the form of the equations
x= Ax+Be,
u = Cx+De.
The Laplace transform of this equation is
sX=AX+BE,
U= CX+DE.
(4.16)
(4.17)
We can now make the substitution for s in (4.17) of any of the forms in
z corresponding to an integration rule. Suppose we start with the forward
rectangular rule, for which we replace s with (z -l)/T from (4.8):
z-l
--rX=AX+BE,
U=CX+DE. (4.18)
In the time domain, the operator z corresponds to forward shift; that is,
zx(k) = x(k + 1). Thus the corresponding discrete equations in the time
domain are
x(k +1) - x(k) = TAx(k) +TBe(k),
x(k + 1) = (I + TA)x(k) + TBe(k),
u = Cx+De. (4.19)
Equation (4.19) is a state-space formula for the forward rule equivalent.
For the backward rule, we substitute s ~ (z - 1)/ zT with the result
and in the time domain,
z-l
--X=AX+BE,
Tz
x(k + 1) - x(k) = TAx(k + 1) + TBe(k + 1). (4.20)

In this equation, we have terms in k+1 on both the right- and left-hand sides.
In order to get an equation with such terms only on the left, we transpose
all k + 1 terms to the left and define them as a new state vector:
x(k + 1) - TAx(k + 1) - TBe(k + 1) = x(k)
I:;,. . )
= w(k + 1 . (4.21)
From this equation, solving for x in terms of wand e, we obtain
(I - AT)x = w +TBe
x = (I - AT)-lw + (I - AT)-lBTe. (4.22)
With this expression for x, we can rewrite (4.21) in standard"form as
w(k + 1) = (I - AT)-lw(k) + (I - AT)-lBTe(k), (4.23a)
and the output equation is now
u(k) = C(I - AT)-lw +{D + C(I - AT)-lBT}e. (4.23b)
Equation (4.23) is a state-space description of the backward rule equivalent
to (4.16).
Finally, for the trapezoid,or bilinear rule, the z:transform equivalent is
obtained from
2(z-1)X=AX BE
T(z + 1) +
AT . BT
(z -l)X = T(z+ l)X + T(z+ l)E
U=CX+DEj (4.24)
and the time domain equation for the state is
AT . . BT .
x(k + 1) - x(k) = T(x(k + 1) + x(k)) + T(e(k + 1) + e(k)). (4.25)

Once more, we collect all the k + 1 terms onto the left and define these as
w(k + 1) as follows:6 '
AT BT AT BTx(k + 1) - -2-x(k + 1) - Te(k + 1) = x(k) + 2 x (k) + Te(k)
~ VTw(k + 1). (4.26)
Writing the definition of w at time k, we can Rolve for x as before:
( AT) BT1- -2- x = VTw+ T e
( AT)-l (AT)-l BTx= 1- -2
- VTw+ 1- 2 Te. (4.27)
Substituting (4.27) into (4.26), we get
( AT) ( AT)-l{ BT} BTVTw(k + 1) = 1+ 2 1- 2 VTw(k) + T e + Te(k)
( AT) ( AT)-lw(k + 1) = 1+ - 2- 1 - 2 w(k)
{( AT) ( AT)-l }BVT+ 1+2 1- 2 +1 -2-e(k)
( AT) (AT)-l (AT)-l= 1+ - I - 2 w(k) + 1- 2 BVTe(k).
2 (4.28)
In following this algebra, it is useful to know that in deriving the last part
of (4.28), we expressed the identity I as (I - A{)(I - A!)-l and factored
out (I - A{)-l on the right.
To obtain the output equation for the bilinear equivalent, we substitute
(4.27) into the second part of (4.24):
( AT)-l { (AT)-l BT}u(k) = VTC 1 - -2- w(k) + D + C I - 2 T e(k).
6The scale factor of VT is introduced so that the gain of the discrete equivalent will
be balanced between input and output, a rather technical condition. See Al Saggaf
and Franklin (1986) for many more details.

4.3 ZERO-POLE MAPPING EQUIVALENTS 147
These results can be tabulated for convenient reference. Suppose we have
a continuous system described by
x(t) = Ax(t) + Be(t),
u(t) = Cx(t) + De(t).
Then a discrete equivalent at sampling period T will be described by
w(k + 1) = «pw(k) +re(k),
u(k) = HW(k) +Je(k),
where «P, r, H , and J are given as follows:7
Forward Backward Bilinear
C) (I + AT) (I - AT)-l (I + At)(I ~ Af)-l
r BT (I - AT)-lBT (I - A,{)-lB..;T
H C C(I - AT)-l ..;TC(I _ ~T)-l
J D D + C(I - AT)-lBT D + C(I - At)-lBTj2
4.3 ZERO-POLE MAPPING EQUIVALENTS
A very simple but effective method of obtaining a discrete equivalent to a
continuous transfer function is to be found by extrapolation of the relation
derived in Chapter 2 between the s- and z-planes. If we take the z-transform
of samples of a continuous signal e(t), then the poles of the discrete transform
E(z) are related to the poles of E(s) according to z = esT. We must go
through the z-transform process to locate the zeros of E(z), however. The
idea of the zero-pole mapping technique is that the map z = esT could
reasonably be applied to the zeros also. The technique consists of a set of
heuristic rules for locating the zeros and poles and setting the gain of a
z-transform that will describe a discrete, equivalent transfer function that
approximates the given H(s). The rules are as follows:
1. All poles of H (s) are mapped according to z = esT. If H (s) has a pole
at s = -a, then Hzp(z) has a pole at z = e-aT . If H(s) has a pole at
-a + jb then Hzp(z) has a pole at reiO, where r = e-aT and () = bT.
7These formulas are easily implemented in CAD packages; for example, see EQUIVNT
in Table E.1 in Appendix E.

2. All finite zeros are also mapped by z = esT. If H(s) has a zero at s = -a,
then Hzp(z) has a zero at z = e-
aT
, and ~o on.
3. The zeros of H(s) at s = 00 are mapped in Hzp(z) to the point z = -1.
a) One zero of H(s) at s = 00 is mapped into z = 00. That is, Hzp(z)
is left with the number of zeros one less than the number of poles in
the finite plane. The series expansion of H (z) in powers of z-l will
have no constant term, and thus the corresponding pulse response,
h(k), has a one-unit delay. This choice means that the computer has
one sample period to do the computation that corresponds to the
transfer function because there is no direct transmission term.
4. The gain of the digital filter is selected to match the gain of H(s) at the
band center or a similar critical point. In most control applications, the
critical frequency is s = 0, and hence we typically select the gain so that
The rationale behind rule 3 is that the map of real frequencies from
jw = 0 to increasing w is onto the unit circle at z = eiO = 1 until z =
ei1r = -1. Thus the point z = -1 represents, in a real way, the highest
frequency possible in the discrete transfer function, so it is appropriate that
if H(s) is zero at the highest (continuous) frequency, IHzp(z)Ishould be z.ero
at z = -1, the highest frequency that can be processed by the digital filter.
Example 4.2: Application of these rules to H(s) = aj(s+a) gives
H ( ) _ (z + 1)(1 - e-
aT
)
zp Z - 2(z _ e-aT) ,
(4.29)
or, using rule 3(a), we get
1- e-aT
Hzp(z) = T '
Z - e-a
(4.30)
A state-space algorithm to generate the zero-pole equivalent is als~ read-
ily constructed with the utilities of a control design package.8
The frequency
8See EQUIVNT in Table E.l.

4.4 HOLD EQUIVALENTS 149
,
response of the zero-pole equivalent was found for the third-order case of Ex-
ample 4.1 and is shown in Fig. 4.8 along with other equivalents for purposes
of comparison.
4.4 HOLD EQUIVALENTS
For this technique, we construct the 'ituation sketched in Fig. 4.4. The pur-
pose of the samplers in Fig. 4.4(b) is Lo require that HhO(z) has a discrete
signal at its input and produces a discrete signal at its output, and thus
Hho(z) can be realized as a discret transfer function. The philosophy of the
design is the following. We are asked to design a discrete system that, with
an input consisting of samples of e(t) , has an output that approximates the
output of the continuous filter H( ) whose input is the continuous e(t). We
generate the discrete equivalent by first approximating e(t) from the samples
e(k) and then putting this e{t) through the given H(s). First consider the
possibilities for approximation. These are techniques for taking a sequence of
samples and extrapolating or holding them to produce a continuous signa1.9
Suppose we have the e(t) as sketched in Fig. 4.5. This figure also shows a
sketch of a piecewise constant approximation to e(t) obtained by the oper-
ation of bolding e(t) constant at e(k) over the interval from kT to (k +1)T.
This operation is the ze1'O-o1'der hold (or ZOH) we've discuss d before. If we
use a first-order polynomial for extrapolation, we have a first-order hold (or
FOH), and so on for second-, and nth-order holds.
4.4.1 Zero-Order Hold Equivalent
If the approximating hold is the zero-order hold, then we have for our ap-
proximation exactly the same situation that in Chapter 2 was analyzed as
a sampled-data system.lO
Therefore, the zero-order-hold equivalent to H(s)
is given by (2.39), or
(4.31)
9Some books on digital-signal processing suggest using no hold at. all, using the
equivalent H(z) =Z{H(s)} . This choice is called the z-transform equivalent.
lORecall that we noticed in Chapter 3 that the signal e is, on the average, delayed
from e by T /2 sec. The size of this delay is one measw'e of the quality of the
appmximation and can be used as a guide to the selection of T.

e(/)
e(t)
·IL_H_(_S)_-,
(a)
u(t)
o
L -______________~y~--------------~
HhO(z)
(b)
u(k)
Figure 4.4 System construction for hold equivalents. a) A continuous transfer
function. b) Block diagram of an equivalent system.
An example will fix ideas.
Example 4.3: Suppose we again take the first-order filter
Then
H(s)
s
and
/
a
H(s)=-.
s+a
a 1 1
----- = - - ----
s(s + a) s s+a
(4.32)
Figure 4.5 A signal, its samples, and its approximation by a zero-order hold.

1.0
Figure 4.6 Impulse response of the extrapolation filter for the triangle hold.
and, by definition of the operation given in (4.32),
1
=---,-
1- z-l
(1 - e-aTz-l) - (1 - z-l)
(1-r1 )(1-e-aTz-1) .
(4.33)
Finally, substituting (4.33) in (4.31), we get the zero-order-hold
equivalent of H (s), namely,
(4.34)
We note that for the trivial example given, the zero-order-hold equivalent
of (4.34) is identical to the matched zero-pole equivalent given by (4.30).
HowevE'r, this is not generally true as is evident in the comparison with
other equivalents for the third-order example (4.1) in Fig. 4.8.
4.4.2 Triangle Hold Equivalent
An interesting hold equivalent can be constructed by imagining that we
have a noncausal hold impulse response, as sketched in Fig. 4.6. The result
is called the triangle-hold equivalent. The effect of this hold filter is to ex-
trapolate the samples so as to connect sample to sample in a straight line.
Although the continuous system is noncausal the discrete equivalent is not.

The Laplace transform of the extrapolation filter that follows the impulse
sampling is
eTs _ 2 + e-Ts
Ts2
Therefore the discrete equivalent that corresponds to (4.31) is
(z - 1)2 {H(S)}Htl'i(Z) = Z -')- .
Tz s-
(4.35)
Example 4.4: As a first example, consider H(s) = 1/s2. In this
case, from the tables, we have
T 2 (z2+4z+1)z
-
6 (z-1)4
(4.36)
and
H .(z) ~ (z - 1)2 T2 (z2 +4z + l)z
tn T z 6 (z - 1)'1
T2 z2 + 4z + 1
6'" (z -1)2 .
(4.37)
An alternative, convenient way to compute the triangle-hold equivalent
is again to consider the state-space formulation. The block diagram is shown
in Fig. 4.7.
The continuous equations are
X= Fx+ Gv,
v = w/T,
w= u(t + T)8(t + T) - 2u(t)8(t) + u(t - T)8(t - T),
~ e
Ts
_2+e-
Ts
~
Figure 4.7 Block diagram of the triangle-hold equivalent.
(4.38)

and, in matrix form,
(4.39)
where urepresents the input impulse functions. We define the large matrix
in (4.39) as FT, and the one-step solution to this equation is
because uconsists only of impulses at the sampling instants. If we define
(4.40)
then the equation in x becomes
With care, the last two equations of (4.38) can be integrated to show that
v(k) =u(k) and that w(k) = u(k + 1) - u(k). Now if we define a new state
{(k) = x(k) - l'2u(k), then the state equation for the triangle equivalent is
{(k + 1) = clJ({(k) +l'2u(k)) + (l'l - l'2)u(k)
= clJ{(k) + (l'l + clJl'2 - l'2)u(k).
The output equation is
y(k) = Hx(k) +Ju(k)
= H({(k) +l'2u(k)) + JU(k)
= H{(k) + (J + Hl'2)U(k).
(4.41)
(4.42)

0.8
-±06
IHI
0.4
02
a
0 0.5 1.5
a
.~
'---'
~
-50
-100
LH
-ISO
-200
- 250
o
0.8
0.6
IHI
0.4
-
0.5
990 N.
.~!
'i''i'~...........
T-'t~ 0 °++ °O n
1.5
Normalized frequen cy w/wp
(al
"'lp!¥~
2~~
++xlto "'"
+ x~",
"'-+ ><
2o~+
'" IS :'----
++ 1I l§ x x ~
2
2
xx15 11 "
I
...+ Vo ~ .
+++ o~oooo +++
0.2
a
a
-50
- IDa
LH
-ISO
-200
o 0.5 1.5
~+~oo
.~+,
~oo ~
+ 00
+
0.5
...,..
+ 00
+ 0 ~~xxxxx
+ 0
0+
0"
1.5
Normalized frequency w/wp
(b)
2
2
2.5
--
" ' v
2.5
nO
o ~x +
OX +
22~
~
+
+
±
2.5
"xx
2.5
Figure 4.8 Comparison of digital equivalents for sampling period (a) T = 1 and
ws/wp = 27r and (b) T = 2 and ws/wn = 7rj where ZOH = 0, zero-pole = +, and
triangle = x.

4.5 SUMMARY· 155
Thus the triangle equivalent of [F, G, H, J] with sample period T is
given by
A=~,
B = r 1 + ~r2 - r 2,
C=H,
D = J +Hr2,
where ~, r1 , and r 2 are defined by (4.40).11
(4.43)
In Fig. 4.8 the responses of the zero-pole, the zero-order hold, and the
triangle-hold equivalents are compared, again for the third-order lowpass
filter. We notice in particular that the triangle hold has excellent phase
responses, even with sampling period of T = 2, which corresponds to a
sampling frequency to passband frequency ratio of only ws/wp =7r-.
4.5 SUMMARY
In this chapter we have presented several techniques for the construction of
discrete equivalents to continuol1S transfer functions so that known design
methods for continuous systems-controls and filters- can be used as a basis
for the design of discrete systems. The methods presented were:
1. Numerical integration
a) Forward rectangular rule
b) Backward rectangular rule
c) Trapezoid or Tustin's rule
d) Bilinear transformation with prewarping
2. Zero-pole mapping
3. Hold equivalence
All methods, except the forward rectangular rule, guarantee a stable,
discrete system from a stable, continuous prototype. The bilinear trans-
formation with prewarping affords exact control over the transmission at a
selected critical frequency, which must be less than 1/2T. Zero-pole mapping
is the simplest method to apply computationally if the desired filter is given
USee-TItIANG in Table E.1.

in terms of its zeros and poles; but with a reasonable computer-aided-design
tool, the designer can select that method that best meets the requirements
of the design.
PROBLEMS AND EXERCISES
4.1 Sketch the zone in the z-plane where poles corresponding to the left half of the
s-plane will be mapped by the zero-pole mapping technique and the zero-order-hold
technique.
4.2 Show that (4.15) is true.
4.3 a) The following transfer function is a lead network designed to add about
60° phase lead at WI = 3 rad:
H(s) _ s + 1
O.ls + 1
For each of the following design methods compute and plot in the z-plane
the p_ole and zero locations and compute the amount of phase lead given
by the equivalent network at ZI = ejw1T
if T = 0.25 sec and the design is
via:
i) Forward rectangular rule
ii) Backward rectangular rule
iii) Bilinear rule
iv) Bilinear with prewarping (Use WI as the warping frequency.)
v) Zero-pole mapping
vi) Zero-order-hold equivalent
vii) Triangular-hold equivalent
b) Plot over the frequency range WI = 0.1 -t Wh = 100 the amplitude and
phase Bode plots for each of the above equivalents.
4.4 a) The following transfer function is a lag network designed to introduce
gain attenuation of a factor of 10 (20 dB) at WI = 3:
H(s)=lOs+1.
laOs + 1
For each of the following design methods, compute and plot on the z-plane
the zero-pole patterns of the resulting discrete equivalents and give the
gain attenuation at ZI = ejw1 T. Let T = 0.25 sec.
i) Forward rectangular rule
ii) Backward rectangular rule
iii) Bilinear rule

PROBLEMS AND EXERCISES 157
iv) Bilinear with prewarping (Use WI = 3 radians as the warping fre-
quency.)
v) Zero-pole mapping
vi) Zero-order-hold equivalent
vii) Triangle-hold equivalent
b) For each case computed, plot the Bode amplitude and phase curves over
the range WI = 0.01 ---+ Wh = 10 rad.

Fpw chapter 4 - digital ctrl of dynamic systems

Fpw chapter 4 - digital ctrl of dynamic systems

More Related Content

What's hot (20)

Similar to Fpw chapter 4 - digital ctrl of dynamic systems (20)

More from Aleksandar Micic (14)

Recently uploaded (20)

Fpw chapter 4 - digital ctrl of dynamic systems