CHAPTER 1

Introduction

1.1 Interfacing Computers to the Physical World

Useful computers need some sort of input and output. Computation that we can’t see doesn’t
provide much attraction. Early computers used primitive I/O devices: lights, paper tape, crude
displays. The development of new I/O devices has paralleled the development of CPUs.
I/O is particularly important to embedded computing systems. Embedded computers are
responsible for a wide range of devices, ranging from simple appliances to complex vehicles
and industrial equipment. This range of I/O requirements calls for a comprehensive
playbook of interfacing techniques.
Embedded system interfacing is the conceptual interface between electrical and computer
engineering—we require the skills of both fields to design good, practical interfaces. Computer
engineers don’t always have a lot of experience with traditional electrical engineering. As a
result, we will cover this territory thoroughly. Readers with expertise in circuits should feel free
to skip some sections and get straight to the use of circuits as interfaces to computers.
Embedded system interfacing is a good example of mixed-signal design—the design of
circuits that combine analog and digital elements. Mixing analog and digital provides powerful
capabilities but must be done carefully. Among other concerns, we must take care with the
circuit characteristics of digital logic such as drive and load, something that is less of a
problem in a purely digital design. Interfacing also requires hardware/software codesign,
mixing the capabilities of software running on the CPU with mixed-signal circuits.
First, Section 1.2 surveys the goals of interface design and the techniques we use to achieve
those goals. Section 1.3 surveys microprocessors. Section 1.4 signals introduces electrical
signals. Sections 1.5 and 1.6 review the laws of electrical engineering: first for resistors, then
generalizing to capacitors and inductors. Section 1.7 describes basic techniques for circuit
analysis. Section 1.8 looks at nonlinear and active devices. Section 1.9 steps back to consider
methodologies and tools for the design of interfaces. Section 1.10 walks through an outline
of the remainder of the book. These sections will outline some basic concepts and terms in
electrical engineering for later reference; we will flesh out these concepts as needed in
later chapters.

Embedded System Interfacing. https://doi.org/10.1016/B978-0-12-817402-9.00001-7

# 2019 Elsevier Inc. All rights reserved.
1
2 Chapter 1

1.2 Goals and Techniques

Embedded computer systems are used in all sorts of applications; one interesting way to think
about the categories of embedded computers and their interfaces is the numbers of copies of the
system that will be built. Experimenters and hobbyists build one system or perhaps a few.
Industrial applications such as factories may build one-off devices but they also make use
of specialized equipment that is manufactured at modest levels: hundreds to tens of thousands.
Consumer products are manufactured in much larger volumes, from tens of thousands to
tens of millions of units. Interface design skills are useful in all of these categories.
Many integrated circuits are systems-on-chip (SoCs) that include processing, I/O devices, and
some amount of onboard memory. The design of these devices and their connections to the
computing system is a critical aspect of the design of the SoC. While many SoCs do not include
analog circuits, the digital devices must be designed with the characteristics of the analog
devices to which they will be connected. Advanced packaging techniques allow the complete
system to be composed of multiple chips built with different manufacturing technologies.
However, not all design is focused on integrated circuits. Many high-volume devices are built
largely out of standard parts assembled on printed circuit boards—what engineers typically
call board-level design. The printed circuit board also is a mainstay of industrial electronic
design. A board design allows the design of a custom circuit with control over the components
and manufacturing technology, all with substantially less cost and time commitment than
is required to design an integrated circuit.
However, many designs require only a handful of the traditional primitives of electrical
engineering: transistors, resistors, capacitors, inductors. Most board-level design puts together
integrated circuits, each of which performs a specialized function. The op amp is a classic
example of an integrated circuit that encapsulates a sophisticated circuit in an easy-to-use form.
While designing circuits using transistors is fun, it is often not realistic. Not only do integrated
circuits save us time, but also they often provide better characteristics than circuits made
from discrete components. But it is still important and useful to understand the basic principles
of circuit design: we need to know how to evaluate the appropriateness of an integrated
circuit for a particular application; and we need to be able to verify that we have designed
the proper circuit connections to them. Providing insufficient current to the input of a logic gate,
for example, will cause it to malfunction.
In order to design board-level systems properly, we need to be able to write the specifications
for the design. We also need to understand the specifications of the components we use to
build the board. These specifications are fundamental characteristics of circuits:

• Gain;
• Frequency response;
 Introduction 3

• Nonlinear characteristics such as rise time, ringing;

• Noise.
Design is the process of finding an embodiment of those specifications using available
components. Circuit theory gives us important concepts in design:
• Drive and load;
• Filtering;
• Amplification gain and bandwidth.
We design the entire interface, which we often break down into smaller designs for the pieces of
the interface. We use top-down design techniques to refine the specifications into realizations;
bottom-up design methods allow us to estimate the characteristics of candidate designs.
We will see in Chapter 8 that an embedded system interface requires us to answer two
questions:
• Where is the software/hardware boundary? What goes in software on the CPU and what
goes into the interface?
• Where is the digital/analog boundary? What parts of the interface are performed with
digital hardware and which with analog circuits?

1.3 Varieties of Microprocessors

A microprocessor is a CPU on an integrated circuit; in the modern era, virtually all CPUs are
microprocessors. A computer system is more than a CPU—it requires memory, I/O devices,
and interconnect between them. The term platform is often used to describe the complete
computing hardware (and perhaps lower levels of the software stack as well). We often
categorize platforms based on their size and complexity.
A microcontroller is a complete computer system on a chip: CPU, memory, I/O devices, and
bus. We typically use this term for smaller systems: simpler CPUs, modest amounts of memory,
and basic I/O. Many microcontrollers provide 4-bit or 8-bit CPUs; some of them provide
less than a kilobyte of memory. The Cypress PSoC 5LP [16] is a microcontroller, although
one with a 32-bit CPU. It provides an ARM Cortex-M3 CPU, three types of memory (flash,
RAM, EPROM), and digital and analog peripherals.
A digital signal processor (DSP) is a microprocessor optimized for signal processing
applications. The original use of the term referred to the combination of a hardware multiplier
and Harvard-style separate program and data memories. Today, DSP optimizations include
addressing modes useful for array calculations.
The term system-on-chip (SoC) is typically applied to more complex chips. Smartphone
processors are a classic example of an SoC but complex platforms are built for a range of
4 Chapter 1

applications, including multimedia and automotive. The NXP S32V234 [46] is an SoC—a
vision processor for automotive applications. It includes four ARM Cortex-A53 CPUs with
SIMD instructions, two ARM Cortex-M4 CPUs, a vision accelerator, GPU, image sensor
processor, image sensor interfaces, and support for safety and security.

1.4 Signals
A signal is a physical state over an extended period. We represent a signal mathematically as a
value defined by a function over time.
We talk about signals with respect to their time values:
• DC (direct current) values do not change over time. Practically speaking, they change
only slowly.
• AC (alternating current) values change over time. The term comes from sinusoidal
signals but we apply it more generally to time-varying values.
We make this distinction because we use different techniques to analyze DC and AC
signals and circuits. DC analysis makes use of simpler techniques.
AC signals can have arbitrary waveforms or shapes. For purposes of analysis, we deal
with two major forms of signals. The sinusoidal signal is determined by its amplitude A,
frequency ω, and phase φ:
vðtÞ ¼ Asinωt + φ: (1.1)
The exponential signal is determined by its amplitude A and time constant τ:

vðtÞ ¼ Aet=τ : (1.2)

Fig. 1.1 shows examples of sinusoidal and exponential signals.
Noise is like weeds—any sort of undesired signal. Noise may come from random sources
that we cannot predict or sources that we understand. When undesired signals come from
predictable sources, we may use other terms, such as interference or crosstalk, to describe it.
We also talk about signals relative to the domain in which we view them:
• Time-domain signals are functions of time.
• Frequency-domain signals are functions of frequency.
These two representations are equivalent—we can transform a time-domain signal into its
frequency-domain equivalent and vice versa. We can use the Fourier transform and its
computational form the fast Fourier transform (FFT) to move between the time and
frequency domains. Fig. 1.2 shows both representations of a signal which is formed by the
product of two sinusoids, one fast and one slow. The frequency domain form shows the