CHAPTER 6 
APPLICATION OF CONVENTIONAL TECHNIQUES TO 
THE DESIGN OF SAMPLED-DATA CONTROL SYSTEMS 
After the techniques of analysis as described in the previous chapters 
have been applied to a sampled-data system, one’s attention naturally 
turns to the problems of design or synthesis. With reference to Fig. 6.1, 
the design problem for a single-loop system may be described as that of 
specifying the operation of the digital controller, D, and the continuous 
network, N, such that the over-all system performance meets the design- 
er’s requirements. The approach used to solve this problem depends to 
a large degree upon the basic performance specification and upon the sys- 
tem constraints in terms of plant characteristics and controller limitations. 
It is neither possible nor practical to enter into a detailed discussion of all 
these factors here because of the enormous variation among individual 
cases and the lack of a unique solution of even one of these cases. It is 
possible, however, to indicate several alternative approaches which have 
promise in a fairly broad class of problems and thereby to aid the indi- 
vidual designer select that method which most nearly fits his particular 
needs. Since this is a book on sampled-data control systems, the primary 
emphasis here is placed on those particular characteristics of the problem 
which are introduced by the sampling process. In order to meet this aim 
with the least possible confusion, attention will be centered on the single- 
loop error-sampled system of Fig. 6.1, and no extension of these tech- 
niques to more complicated cases will be attempted. In general, so long 
as only linear control systems are considered, the single-loop case dis- 
plays almost all of the essential peculiarities of the sampled-data design 
problem. 
Two specific aspects of the design problem will be postponed until later 
chapters. These are the ripple, or intersample behavior, of the output 
and multirate systems where 7; and 7, in Fig. 6.1 are not equal. Each of 
these problems requires special techniques and is treated separately from 
the basic single-rate design problem. Of course every design requires 
satisfactory performance between sampling instants, but, as shown in the 
next chapter, this problem can be handled by proper design at sampling 
instants. 
117 
