DSP Integrated Circuits

Lars Wanhammar

• Publisher: Academic Press
• Year: 1999
• Print ISBN: 9780127345307
• eBook ISBN: 9780080504773
• Topic: Technology & Engineering
• Subtopic: Electrical Engineering & Telecommunications
• Index: Technology & Engineering

DSP Integrated Circuits establishes the essential interface between theory of digital signal processing algorithms and their implementation in full-custom CMOS technology. With an emphasis on techniques for co-design of DSP algorithms and hardware in order to achieve high performance in terms of throughput, low power consumption, and design effort, this book provides the professional engineer, researcher, and student with a firm foundation in the theoretical as well as the practical aspects of designing high performance DSP integrated circuits. Centered around three design case studies, DSP Integrated Circuits thoroughly details a high-performance FFT processor, a 2-D Discrete Cosine Transform for HDTV, and a wave digital filter for interpolation of the sampling frequency. The case studies cover the essential parts of the design process in a top-down manner, from specification of algorithm design and optimization, scheduling of operations, synthesis of optimal architectures, realization of processing elements, to the floor-planning of the integrated circuit. - Details the theory and design of digital filters - particularly wave digital filters, multi-rate digital filters, fast Fourier transforms (FFT's), and discrete cosine transforms (DCT's) - Follows three complete "real-world" case studies throughout the book - Provides complete coverage of finite word length effects in DSP algorithms - In-depth survey of the computational properties of DSP algorithms and their mapping to optimal architectures - Outlines DSP architectures and parallel, bit-serial, and distributed arithmetic - Presents the design process in a top-down manner and incorporates numerous problems and solutions

1

DSP Integrated Circuits

1.1 INTRODUCTION

Rapid advancements in electronics, particularly in manufacturing techniques for integrated circuits, have already had, and will undoubtedly continue to have, a major impact on both industry and society as a whole. In this book we will discuss various approaches to designing integrated circuits for digital signal processing (DSP) applications. Modern DSP systems are often well suited to VLSI implementation. Indeed, they are often technically feasible or economically viable only if implemented using VLSI technologies. The large investment necessary to design a new integrated circuit can only be justified when the number of circuits to be manufactured is large, or when the necessary performance requirements are so high that they cannot be met with any other technology. In practice, we often find that both arguments are valid, particularly in communication and consumer applications. Advances in integrated circuit technology also open new areas for DSP techniques, such as intelligent sensors, robot vision, and automation, while simultaneously providing a basis for continuing advancements in traditional signal processing areas, such as speech, music, radar, sonar, audio, video, and communications.

Integrated circuit technology has had a profound effect on the cost, performance, and reliability of electronic circuits. Manufacturing cost is almost independent of the complexity of the system. The cost per integrated circuit (unit cost) for large-volume applications using large chips is dominated by the cost of the chip, while for small and medium size chips the package cost tends to dominate. The whole system cost for small-volume applications is often dominated by the development cost. Unfortunately, the development cost is often difficult to estimate accurately. Increase in system complexity and integration of the manufacturing and design processes tend to increase development costs and cause long design times. However, these adverse effects can be mitigated by extensive use of computer-aided design tools and the use of efficient design methodologies. Today, computer-aided design (CAD) and computer-aided manufacturing (CAM) are used extensively in almost all aspects of electronic engineering. To explore VLSI technology optimally it is necessary that the design team cover all aspects of the design, specification, DSP algorithm, system and circuit architecture, logic, and integrated circuit design. Hence, changes in classical design methodologies and in the organization of design teams may be necessary. We will therefore discuss the most common design methodologies used for the design of DSP systems. We will also present a novel methodology and apply it to some common DSP subsystems.

The problem of designing special-purpose DSP systems is an interesting research topic, but, more important, it has significant industrial and commercial relevance. Many DSP systems (for example, mobile phones) are produced in very large numbers and require high-performance circuits with respect to throughput and power consumption. Therefore, the design of DSP integrated circuits is a challenging topic for both system and VLSI designers. DSP integrated circuits are also of economic importance to the chip manufacturers.

1.2 DIGITAL SIGNAL PROCESSING

Signal processing is fundamental to information processing and includes various methods for extracting information obtained either from nature itself or from man-made machines. Generally, the aim of signal processing is to reduce the information content in a signal to facilitate a decision about what information the signal carries. In other instances the aim is to retain the information and to transform the signal into a form that is more suitable for transmission or storage. The DSP systems of interest here are the so-called hard real-time systems, where computations must be completed within a given time limit (the sample period). An unacceptable error occurs if the time limit is exceeded.

Modem signal processing is mainly concerned with digital techniques, but also with analog and sampled-data (discrete-time) techniques, which are needed in the interfaces between digital systems and the outside analog world [9,11]. Sampled-data systems are generally implemented using switched capacitor (SC) [10] or switched current (SI) circuits. Most A/D and D/A converters are today based on SC circuit techniques. An important advantage of SC circuits is that they can easily be integrated with digital CMOS circuits on the same chip. Recently, analog circuits such as anti-aliasing filters have also become possible to implement on the same chip. A fully integrated system-on-a-chip is therefore feasible by using a suitable combination of circuit techniques. This will affect both performance and cost of DSP systems.

Generally, complex signal processing systems are synthesized using subsystems that perform the basic DSP operations. Typical operations are frequency selective and adaptive filtering, time-frequency transformation, and sample rate change. In Chapters 3 and 4, we will review some of the most common signal processing functions used in such subsystems. The aim is to provide a background for three typical DSP subsystems that will be used as case studies throughout the book.

1.3 STANDARD DIGITAL SIGNAL PROCESSORS

In principle, any DSP algorithm can be implemented by programming a standard, general-purpose digital signal processor [1]. The design process involves mainly coding the DSP algorithm either using a high-level language (for example, the C language) or directly in assembly language. Some high-level designtools allow the user to describe the algorithm as a block diagram via a graphicuser interface. The tool automatically combines optimized source codes for the blocks, which are stored in a library, with code that calls the blocks according tothe block diagram. Finally, the source code is compiled into object code that can be executed by the processor. This approach allows rapid prototyping, and the achieved performance in terms of execution speed and code size is reasonably good since the codes for the blocks are op...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright page
• Preface
• 1: DSP Integrated Circuits
• 2: VLSI Circuit Technologies
• 3: Digital Signal Processing
• 4: Digital Filters
• 5: Finite Word Length Effects
• 6: DSP Algorithms
• 7: DSP System Design
• 8: DSP Architectures
• 9: Synthesis of DSP Architectures
• 10: Digital Systems
• 11: Processing Elements
• 12: Integrated Circuit Design
• Index