Chapter I: Introduction to Sampled Data Systems
Automation and Control Engineering
October 2025
1 Introduction
The fundamental difference between a traditional analog control loop and a digital control
system lies in the presence of sampling and digital processing elements. Figure 1 compares
both architectures.
Figure 1: Comparison between analog and digital control systems
1.1 System Components
• v(t): Disturbance signal
• w(t), w(k): Analog and sampled reference signals
• y(t), y(k): Analog and sampled output signals
• A/D: Analog-to-Digital Converter
• D/A: Digital-to-Analog Converter
1
1.2 Advantages of Digital Control Systems
• Flexibility: Easy implementation of complex algorithms
• Resource Sharing: A single processor can control several loops
• Data Logging: Capability to record and analyze long-term data
• Adaptability: Parameters can be adjusted in real time
• User Interface: Integration with graphical tools and monitoring platforms
1.3 Main Challenge
• Information Loss: Sampling may cause aliasing and performance degradation
2 Sampling and Quantization
2.1 Sampling Process
The A/D converter transforms the continuous signal y(t) into discrete values y(kTs ) taken
at sampling instants defined by the sampling period Ts :
[y(0), y(Ts ), y(2Ts ), y(3Ts ), . . . , y(kTs ), . . .]
For constant sampling intervals, this is simplified to:
[y(0), y(1), y(2), y(3), . . . , y(k), . . .]
2.2 Quantization
The A/D converter delivers digital values, typically 8 or 16 bits, representing quantized
samples of the analog signal. The resulting signal is both discrete in time and quantized
in amplitude.
3 Determination of the Sampling Period
The sampling period Ts is a key design parameter directly influencing system stability
and performance.
3.1 Nyquist–Shannon Sampling Theorem
A continuous-time signal x(t) with maximum frequency fmax can be perfectly
reconstructed from its samples if it is sampled at a frequency:
fs > 2fmax
2
3.2 Practical Considerations
• Small Ts : High precision but heavy computational load
• Large Ts : Possible signal distortion and control delay
• Industrial rule: fs = 5–10× system bandwidth
Figure 2: Sampling period selection based on Shannon’s theorem
4 Sampling Period for Continuous Systems
When discretizing a continuous system, the choice of the sampling period (or frequency)
directly affects the closed-loop performance. A widely accepted empirical rule states:
For continuous systems, the sampling frequency fe is generally chosen between
6 and 24 times the closed-loop bandwidth frequency fBP,CL .
Depending on the transfer function structure, the following guidelines can be applied:
4.1 Sampling Period of a First-Order System
For a first-order system with transfer function
k
G(s) = ,
1 + τs
the sampling period Te should be selected based on the system time constant τ . To ensure
accurate digital representation of the continuous dynamics, the following practical rule is
applied:
τ
τ ≥ Te ≥ .
4
This range ensures that the sampling frequency is sufficiently high compared to the
dominant time constant of the system.
3
4.2 Sampling Period of a Second-Order System
For a second-order system, the sampling period Te should be chosen according to the
system’s natural frequency ω0 and damping ratio ζ. Depending on the damping ratio,
the following practical ranges apply:
0.25 ≤ ω0 Te ≤ 1; ζ = 0.7
0.4 ≤ ω0 Te ≤ 1.75; ζ=1
A general range that provides a good trade-off between performance and numerical
stability is:
0.25 ≤ ω0 Te ≤ 1.5; 0.7 ≤ ζ ≤ 1
This choice guarantees a sufficiently fast sampling rate to capture the dynamic be-
havior without excessive computational load.
4.3 System of Order n
For:
K
G(s) = ,
(1 + τ1 s)(1 + τ2 s) · · · (1 + τn s)
a practical range is:
τmin τmin
< Te < ,
10 5
where τmin = min(τi ).
5 Reconstruction of the Analog Signal
The digital controller outputs a discrete command u(k), which must be converted back
into a continuous signal u(t) to drive the analog plant.
5.1 Zero-Order Hold (ZOH)
The ZOH maintains the last sample value constant during the entire sampling period:
ua (t) = u[k], for kTs ≤ t < (k + 1)Ts
5.2 Higher-Order Holds
Higher-order holds (e.g., first-order holds using linear extrapolation) generate smoother
signals but increase computational complexity and delay.
4
Figure 3: Signal reconstruction using zero-order hold
Figure 4: First-order hold reconstruction
6 Mathematical Representation of Sampling
The sampling operation can be modeled as:
∞
X
xs (t) = x(t) δ(t − kTs )
k=−∞
where δ(t) denotes the Dirac delta function. This formalism allows frequency-domain
analysis using Fourier transforms and the concept of spectral replication.
7 References
• Franklin, G. F., Powell, J. D., & Workman, M. L. (1998). Digital Control of Dy-
namic Systems. Addison-Wesley.
5
• Åström, K. J., & Wittenmark, B. (1997). Computer-Controlled Systems: Theory
and Design. Prentice Hall.
• Oppenheim, A. V., & Schafer, R. W. (2009). Discrete-Time Signal Processing.
Pearson.
• Ogata, K. (2010). Modern Control Engineering. Prentice Hall.