5.8 DSP I:

Processors Architecture

Dr. Tarek A. TutunjiMechatronics Engineering Department

Philadelphia University, Jordan

Overview

Microprocessors and Microcontrollers were discussed in the previous sequences

In this sequence, Digital Signal Processors (DSP) will be introduced

Introduction

Digital Signal Processing (DSP) is the arithmetic processing of discrete-time signals

A signal is a physical quantity that varies with time, frequency, or space

Instead of using op-amps, resistors, and other analog electronics to process an analog signal, a microprocessor (or DSP chip) can be used to perform mathematical operations on digital signals to achieve the same (or better) effect.

A/D is needed for analog signals

The increasing importance of DSP in the 1980’s led major electronics manufacturers, such as Texas Instruments and Motorola, to develop Digital Signal Processor chips

Introduction

DSP chips are specialized microprocessors (i.e. programmable devices with its own instruction code) with architectures designed to reduce the number of instructions and operations necessary for efficient signal processing.

The programmable flexibility of DSP chips enable developers to implement complex algorithms in software

DSP chips are used to perform computationally efficient and fast algorithms, such as Digital Filtering, Spectral Analysis, Parameter Estimation, and Data Compression

Implementing DSP, especially in real-time, is generally treated separately from the theory

DSP Architecture Features

Feature Benefit

Single-Cycle Instructions Executes advanced control systems in real-time

Pipelined Architecture Controls high-bandwidth systems

Harvard Architecture Accesses Data and Instructions Simultaneously and

therefore increases speed

Hardware Multiplier Minimizes computational delays

Hardware Shifter Have large dynamic range

Hardware Stack Support fast interrupt processing

32-bit register Minimizes truncation errors

DSP Architecture Advantages

Common features for DSP:• Use a lot of mathematics (multiplying and adding signals)

Majority of signal processing functions multiply two series of numbers and sum the results: result = x1 * c1 + x2 * c2 + x3 * c3 …xn * cn

• Deal with signals that come from the real world • Require a response in a certain time

Signals are measured from the real world and usually need a reaction in real time

Application Features of DSP• High-speed processing applications such as real-time control• Accurate and Complex Control Systems• Frequency-based applications• Iterative Algorithms• Matrix operations

DSP Control Advantage

DSP chips are built for speed, enabling developers to close the control loop quickly.• FIFOs serving communication peripherals reduce the burden of

interrupt servicing by the CPU• Ultra-fast A/D with dual sample and hold enable controllers to

capture data efficiently with low peripheral overhead.

DSP chips have a special architecture to accelerate control loop processing common to most control applications, enabling controllers to do more each loop iteration while consuming fewer cycles.• A multi-bus pipeline architecture enables the controller to read

and write multiple data values in a single cycle, compared to traditional MCUs which can only work with a single data value at a time

Floating vs. Fixed point DSPs

Fixed-Point representation is a generalization of the decimal representation of a number as a string of digits with a decimal point

The problem of fixed point processors is quantization error, caused by the limited fixed point precision

Floating point representation consists of a mantissa M (0.5<M<1) multiplied by an exponential factor 2E.

IEEE 754 standard representation of numbers (-1)s 2E-127 M

s E M

0 1 8 9 31

Digital Signal Processors

Modern processors achieve high performance through enabling technologies such as parallel processing, deep pipelines, specialized internal compute engines, and integrated peripherals. Performance of these processors is often measured in millions or billions of operations per second.

DSP Processor characteristics fall into three categories:• specialized high speed arithmetic • data transfer to and from the real world • multiple access memory architectures

The basic DSP operations are:• additions and multiplications • delays • array handling

Digital Signal Processors Additions and multiplications operations require the

following:• fetch two operands • perform the addition or multiplication• store the result or hold it for a repetition

delay operation require the following:• hold a value for later use

Array operation require the following:• fetch values from consecutive memory locations • copy data from memory to memory

To suit these fundamental operations DSP processors often have:• parallel multiply and add • multiple memory accesses (to fetch two operands and store

the result) • lots of registers to hold data temporarily • efficient address generation for array handling • special features such as delays or circular addressing

Texas Instruments:

TMS320 DSP Family Highest Performance: TMS320C6000™ DSP platform

• TMS320C6000 DSP platform offers the industry’s highest performance fixed- and floating-point DSPs ideal for video, imaging, broadband infrastructure and performance audio applications.

Best Power Efficiency: TMS320C5000™ DSP Platform• With standby power consumption as low as 0.12 mW and

performance up to 900 MIPS, C5000™ DSPs are ideal for for personal and portable products like digital music players, GPS receivers, portable medical equipment, MIPS-intensive voice and data processing, and extremely cost effective single and multi-channel applications

Control Optimized: TMS320C2000™ DSP Platform• TMS320C2000™ DSP Platform provides the digital control

industry with the control peripheral integration and ease of use of a microcontroller with the processing power and efficiency of TI’s leading DSP technology

TMS320C2801 Features Buses:

• Eight buses: 1 Program, 3 Data (2 RD & 1 WR), 4 Address• Control Bus

High-Performance 32-Bit CPU• 100 MHz (10-ns cycle time)• 16 x 16 and 32 x 32 MAC Operations • 16 x 16 Dual MAC • Harvard Bus Architecture • Atomic Operations (read/write/modify)

optimize raw power, control, and programming

• Fast Interrupt Response and Processing • Code-Efficient & Optimized (in C/C++ and Assembly)• 32-bit ALU (Arithmetic Logic Unit)• Two 32-bit accumulators• Three 32-Bit CPU Timers

Pipeline Operation• Fetch, Decode, Execute can be done in parallel rather than

seria• Six level deep instruction pipeline

TMS320C2801 Features On-Chip Memory: C2801: 16K X 16 ROM, 6K X 16 SARAM Standard Math Tables Clock and System Control Dynamic PLL On-Chip Oscillator Peripheral Interrupt Expansion supports 43 Peripheral Interrupts

• RMW makes it so that an interrupt cannot occur in the middle of a read/modify/write action without having to mask interrupts

Enhanced Control Peripherals • Up to 16 PWM Outputs • Up to Six 32-bit/Six 16-bit Timers

Serial Port Interface 12-Bit ADC, 16 Channels Up to 35 Individually Programmable, Multiplexed General-Purpose

Input/Output (GPIO) Pins With Input Filtering Development Support Includes

• ANSI C/C++ Compiler/Assembler/Linker • Supports TMS320C24x™/240x Instructions • Digital Motor Control and Digital Power Software Libraries

DSP Applications

Industrial Drives Appliances Optical Networking Cooling Systems Power Management Automotive Consumer Goods Fuel Pumps Tunable Lasers UPS Intelligent Sensors Video and Mobile (Data Compression)

DSP in Control Systems Electronic power steering for automotive applications

• eliminates the traditional hydraulic power steering system’s pump, hoses, hydraulic fluid, drive belt, and pulley while improving overall fuel economy by an average of 5%

Variable speed motors• controlled digitally for applications such as refrigeration

compressor drives improve energy efficiency, reduce current draw, eliminate wear items such as brushes, and increase overall system durability

Multiple motor control for HVAC applications • increases furnace efficiency, improves comfort level, more

easily maintains a constant airflow, and eliminates the need for an exhaust pressure switch

Digital motor drive control • for white goods such as washing machines eliminates

speeded/current sensors and mechanical gearing, enabling the use of smaller DC Link capacitors through ripple compensation, and reduces EMC filter size through power factor correction (PFC) to result in overall higher spin speeds, larger baskets in the same size enclosure, and less noise and vibration

Summary

Digital Signal Processors were introduced. This included:

• DSP Definition

• Architecture Features

• Motorola & Texas Instruments DSP

• Applications