Download - Easy Learn to Verilog HDL

VLSI DESIGN Reference Material

By Verilog Course Team

Where Technology and Creativity Meet

Contact Us

VERILOG COURSE TEAM

Email:[email protected] Blog: www.vlsiprojects.blogspot.com Web: www.verilogcourseteam.com Phone: +91 98942 20795 Revision: 1

For hardcopies drop a mail or contact us. Disclaimer:

Due care and diligence has been taken while editing of this material. Verilog Course Team does not warrant or assume any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed. No warranty of any kind, implied, expressed or statutory, including to fitness for a particular purpose and freedom from computer virus, is given with respect to the contents of this material or its hyperlinks to other Internet resources. The material acts as just a reference to move forward and understand the concept. Reference in this material to any specific commercial products, processes, or services, or the use of any trade, firm or corporation name is for the information, and does not constitute endorsement, recommendation, or favoring.

About Verilog Course Team

Verilog Course Team is a Electronic Design Services (EDS) for VLSI /

EMBEDDED and MATLAB, delivering a wide variety of end-to-end services ,

including design , development, & testing for customers around the world .With

proven expertise across multiple domains such as Consumer Electronics Market

,Infotainment, Office Automation, Mobility and Equipment Controls. Verilog Course

Team is managed by Engineers / Professionals possessing significant industrial

experience across various application domains and engineering horizontals . Our

engineers have expertise across a wide range of technologies, to the efforts of

engineering our clients. Leveraging standards based components and investments in

dedicated test lab infrastructure; we offer innovative, flexible and cost-effective

Services and solutions.

Our Mission

Our mission is to provide cost effective, technology independent, good quality

reusable Intellectual Property cores with quality and cost factor are our

important constraints so as to satisfy our customers ultimately. We develop and

continuously evaluate systems so as to pursue quality in all our deliverables. At our

team, we are completely dedicated to customer’s requirements. Our products are

designed and devoted to empower their competitive edge and help them succeed.

Visit www.verilogcourseteam.com for more details.

Preface

The India Semiconductor Association (ISA), an Indian semiconductor industry organization, has briefed growth, trends and forecasts for the Indian semiconductor market in collaboration with a U.S. consulting company Frost & Sullivan.

The report titled as "ISA-Frost & Sullivan 2007/2008 Indian Semiconductor Market Update."

According to the report, total semiconductor consumption in India (total value of semiconductors used for devices marketed in India) was $2.69 billion (USD) in 2006. The $2.69 billion represents 1.09% of the global semiconductor market. Of the total semiconductor consumption in India, consumption by local Indian set manufacturers accounted for $1.26 billion.

The overall Indian semiconductor consumption will grow at an average rate of 26.7% per year in 2006 through 2009. Based on the actual consumption in 2006, the overall Indian semiconductor consumption is forecast to be $5.49 billion in 2009. This represents 1.62% of the global semiconductor market in 2009.

Semiconductor consumption by local Indian set manufacturers is predicted to increase at 35.8% per year in 2006 through 2009 and amount to $3.18 billion in 2009.

This material is the result of the Verilog Course Team’s practical experience both in Design/Verification and Training. Many of the examples illustrated throughout the material are real designs models. With Verilog Course Team’s training experience has led to step by step presentation, which addresses common mistakes and hard-to-understand concepts in a way that eases learning. Verilog Course Team invites suggestion and feedbacks from both students and faculty community to improve the quality, content and presentation of the material.

VLSI DESIGN

Verilog Course Team www.verilogcourseteam.com Dream IT, We make U to Deliver

UNIT-I CMOS TECHNOLOGY

1. An overview of silicon semiconductor technology 1

1.1 The Fabrication of a Semiconductor Device 1

1.1.2 Wafer Fabrication 2

1.1.3 Assembly 6

1.2 Basic CMOS Technology 8

1.2.1 A Basic n-well CMOS Process 9

1.2.2 A Basic p-well CMOS Process 13

1.2.3 Twin-Tub (Twin-Well) CMOS Process 13

1.2.4 Silicon On Insulator (SOI) Process 14

1.3 INTERCONNECT 18

1.3.1 Metal Interconnect 18

1.3.2 Polysilicon/Refractory Metal Interconnect 19

1.3.3 Local Interconnect 20

1.4 CIRCUIT ELEMENTS 21

1.4.1 Resistors 21

1.4.2 Capacitors 21

1.4.3 Electrically Alterable ROMs 23

1.4.4 Bipolar Transistors 24

1.4.5 LatchUp 26

1.4.5.1 The Physical Origin of Latchup 26

1.4.5.2 Latchup Triggering 28

1.4.6 Latchup Prevention 29

1.5. LAYOUT DESIGN RULES 30

1.5.1 Layer Representations 31

1.5.2 CMOS n-well Rules 32

1.5.3 Scribe Line 34

VLSI DESIGN


1.5.4 SOI Rules 34

1.5.5 Layer Assignments 35

1.6 PHYSICAL DEISGN 35

1.6.1 Basic Concept 35

1.6.2 CAD Tools sets 37

1.6.3 Physical Design-The Inverter 38

1.6.4 Physical Design-The NOR 38

1.6.5 Physical Design-The NAND 39

1.7 DESIGN STRATEGIES 39

1.7.1 Structured Design Strategies 40

1.7.2 Hierarchy 40

UNIT 2 MOS TRANSISTOR THEORY

2 .1 NMOS ENHANCEMENT TRANSISTOR 41

2.2 PMOS ENHANCEMENT TRANSISTOR 45

2.3 THRESHOLD VOLTAGE 45

2 .3 .1 Threshold Voltage Equations 46

2.4 BODY EFFECT 48

2.5 MOS Device Design Equations 48

2.5.1 Basic DC Equations 48

2.5.2 Second Order Effects 50

2.5.2.1 Threshold Voltage-Body Effect 51

2.5.2.2 Subthreshold Region 51

2.5.2.3 Channel-length Modulation 52

2.5.2.4 Mobility Variation 52

2.6 MOS MODELS 53

2.7 SMALL SIGNAL AC CHARACTERISTICS 54

VLSI DESIGN


2.8THE COMPLEMENTARY CMOS INVERTER –

DC CHARACTERISTICS 55

2.8.1 βn/βp ratio 61

2.8.2 Noise Margin 62

2.9 THE TRANSMISSION GATE 64

2.10 THE TRISTATE INVERTER 68

UNIT 3 SPECIFIFCATION OF VERILOG HDL

3. HISTORY OF VERILOG 69

3.1 BASIC CONCEPTS 69

3.1.1 Hardware Description Language 69

3.1.2 VERILOG Introduction 69

3.1.3 VERILOG Features 70

3.1.4 Design Flow 70

3.1.5 Design Hierarchies 73

3.1.5.1 Bottom up Design 73

3.1.5.2 Top-Down Design 74

3.1.6 Lexical Conventions 74

3.1.6.1 Whitespace 75

3.1.6.2 Comments 75

3.1.6.3 Identifiers and Keywords 76

3.1.6.4 Escaped Identifiers 76

3.1.7 Numbers in Verilog 76

3.1.7.1 Integer Numbers 77

3.1.7.2 Real Numbers 77

3.1.7.3 Signed and Unsigned Numbers 77

3.1.8 Strings 78

3.1.9 Data types 79

VLSI DESIGN


3.1.9.1 Data Types Value set 79

3.1.9.2 Nets 79

3.1.9.3 Vectors 80

3.1.9.4 Integer, Real and Time Register Data Types 80

3.1.9.5 Arrays 81

3.1.9.6 Memories 82

3.1.9.7 Parameters 82

3.1.9.8 Strings 82

3.2 MODULES 83

3.2.1 Instances 84

3.3 PORTS 84

3.3.1 Port Declaration 85

3.3.2 Port Connection Rules 85

3.3.3 Ports Connection to External Signals 86

3.4 GATE DELAYS 87

3.4.1 Rise, Fall, and Turn-off Delays 87

3.4.2 Min/Typ/Max Values 88

3.5 MODELING CONCEPTS 89

3.6 SWITCH LEVEL MODELING 90

3.6.1 Switch level primitives 91

3.6.2 MOS switches 92

3.6.3 CMOS Switches 93

3.6.4 Bidirectional Switches 94

3.6.5Power and Ground 95

3.6.6 Resistive Switches 95

3.8 Delay Specification on Switches 96

3.8.1 MOS and CMOS switches 96

3.8.2 Bidirectional pass switches 97

VLSI DESIGN


3.9 GATE LEVEL MODELING 101

3.9.1 Gate Types 101

3.10 BEHAVIORAL AND RTL MODELING 108

3.10.1 Operators 108

3.10.1.1 Arithmetic Operators 108

3.10.1.2 Relational Operators 109

3.10.1.3 Bit-wise Operators 110

3.10.1.4 Logical Operators 112

3.10.1.5 Reduction Operators 113

3.10.1.6 Shift Operators 114

3.10.1.7 Concatenation Operator 115

3.10.1.8 Replication Operator 116

3.10.1.9 Conditional Operator 116

3.10.1.10 Equality Operators 117

3.10.2 Operator Precedence 119

3.10.3 Timing controls 119

3.10.3.1 Delay-based timing control 119

3.10.3.2 Event based timing control 122

3.10.3.3 Level-Sensitive Timing Control 124

3.10.4 Procedural Blocks 124

3.10.5 Procedural Assignment Statements 125

3.10.6 Procedural Assignment Groups 126

3.10.7 Sequential Statement Groups 128

3.10.8 Parallel Statement Groups 128

3.10.9 Blocking and Nonblocking assignment 129

3.10.10 assign and deassign 130

3.10.11 force and release 131

3.10.12 Conditional Statements 131

VLSI DESIGN


3.10.12.1 The Conditional Statement if-else 131

3.10.12.2 The Case Statement 132

3.10.12.3 The casez and casex statement 134

3.10.13 Looping Statements 136

3.10.13.1 The forever statement 136

3.10.13.2 The repeat statement 136

3.10.13.3 The while loop statement 137

3.10.13.4 The for loop statement 138

3.11 DATA FLOW MODELING AND RTL 139

3.11.1 Continuous Assignment Statements 139

3.11.2 Propagation Delay 141

3.12 STRUCTURAL GATE LEVEL DESCRIPTION 141

3.12.1 2 to 4 Decoder 141

3.12.2 Comparator 142

3.12.3 Priority Encoder 144

3.12.4 D-latch 144

3.12.5 D Flip Flop 145

3.12.6 Half adder 145

3.12.7 Full adder 146

3.12.8 Ripple Carry Adder 146

UNIT 4 CMOS CHIP DESIGN

4.1 INTRODUCTION TO CMOS 148

4.2 LOGIC DESIGN WITH CMOS 149

4.2.1 COMBITIONAL LOGIC 149

4.2.2 INVERTER 150

4.2.3 The NAND Gate 151

4.2.4 The NOR Gate 152

VLSI DESIGN


4.3 TRANSMISSION GATES 153

4.3.1Multiplexers 153

4.3.2 Lathes 153

4.4 CMOS CHIP DESIGN OPTIONS 154

4.4.1 ASIC 154

4.4.2 Uses of ASICs 155

4.4.3 Full Custom ASICs 155

4.4.5 Semi-Custom ASICs 156

4.4.6 Standard- Cell-Based ASIC 156

4.4.7 Gate Array Asic 157

4.4.8 Channeled Gate Array 158

4.4.9 Channelless Gate Array 158

4.4.10 Structured Gate Array 159

4.5 PROGRAMMABLE LOGIC 159

4.5.1 Programmable Logic Structures 160

4.5.2 Programmable of PALs 161

4.5.3 Fusible Links 161

4.5.4 UV-erasable EPROM 161

4.5.5 EEPROM 161

4.5.6 Programmable Interconnect 162

4.6 ASIC DESIGN FLOW 163

UNIT-5 CMOS TEST METHODS

5.1 THE NEED FOR TESTING 165

5.1.1 Functionality Tests 166

5.2 MANUFACTURING TEST PRINCIPLS 166

5.2.1 FAULT MODELS 167

5.2.1.1 Stuck-At-Faults 167

VLSI DESIGN


5.2.1.2 Short-Circuit and Open-Circuit Faults 168

5.2.2 Observability 170

5.2.3 Controllability 171

5.2.4 Fault Coverage 171

5.2.5 Automatic Test Pattern Generation (Atpg) 171

5.2.6 Fault Grading And Fault Simulation 177

5.2.7 Delay Fault Testing 178

5.2.8 Statistical Fault Analysis 179

5.2.9 Fault Sampling 180

5.3 DESIGN STRATEGIES FOR TEST 180

5.3.1 Design for Testability 180

5.3.2 Ad-Hoc Testing 181

5.3.3 Scan-Based Test Techniques 184

5.3.3.1 Level Sensitive Scan Design (LSSD) 185

5.3.3.2 Serial Scan 187

5.3.3.3 Partial Serial Scan 188

5.3.3.4 Parallel Scan 190

5.3.4 Self-Test Techniques 191

5.3.4.1 Signature Analysis and BILBO 191

5.3.4.2 Memory Self-Test 193

5.3.4.3 Iterative logic array testing 194

5.3.5 IDDQ testing 194

5.4 CHIP-LEVEL TEST TECHNIQUES 194

5.4.1 Regular Logic Array 194

5.4.2 Memories 195

5.4.3 Random Logic 196

5.5 SYSTEM-LEVEL TEST TECHNIQUES 196

5.5.1 Boundary Scan 196

VLSI DESIGN


5.5.1.1 Introduction 196

5.5.1.2 The Test Access Port (TAP) 197

5.5.1.3 The Test Architecture 197

5.5.1.4 The TAP controller 198

5.5.1.5 The Instruction Register (IR) 198

5.5.1.6 Test-Data Registers 199

5.5.1.7 Boundary Scan Registers 199

VLSI DESIGN CMOS TECHNOLOGY

Verilog Course Team www.verilogcourseteam.com Dream IT, We make U to Deliver 1

UNIT-I

An overview of silicon semiconductor technology

Silicon in its pure or intrinsic state is a semiconductor, having a bulk electrical resistance somewhere between that of a conductor and an insulator. The conductivity of silicon can be varied over several orders of magnitude by introducing impurity atoms onto silicon crystal lattice. These dopants may either supply free electrons or holes. Impurity elements that use electrons are referred to as acceptors, since they accept some of the electrons already in the silicon, leaving vacancies or holes. Similarly, donor elements provide electrons. Silicon that contains a majority of donors is known as n-type and that which contains a majority are brought together, the region where the silicon changes from n-type and p-type materials are brought together, the region where the silicon changes from n-type to p-type is called a junction. By arranging junctions in certain physical structures and combining these with other physical structures, various semiconductor devices may be constructed. Over the years, silicon semiconductor processing has evolved sophisticated techniques for building these junctions and other structures having special properties.

An integrated circuit is a small but sophisticated device implementing several electronic functions. It is made up of two major parts: a tiny and very fragile silicon chip (die) and a package which is intended to protect the internal silicon chip and to provide users with a practical way of handling the component. The various steps in manufacturing processes of transistor both in “front-end” and “back-end” is taken as example, because it uses the MOS technology. Actually, this technology is used for the majority of the ICs manufacturing companies.

1.1 The Fabrication of a Semiconductor Device

The manufacturing phase of an integrated circuit can be divided into two steps. The first, wafer fabrication, is the extremely sophisticated and intricate process of manufacturing the silicon chip. The second, assembly, is the highly precise and automated process of packaging the die. Those two phases are commonly known as “Front-End” and “Back- end”. They include two test steps:

• Wafer probing and Final test.

The flow chart is shown in figure 1.1.



Figure 1.1 Manufacturing Flow Chart of an Integrated Circuit

1.1.2 Wafer Fabrication (Front-End)

Identical integrated circuits, called die, are made on each wafer in a multi-step process. Each step adds a new layer to the wafer or modifies the existing one. These layers form the elements of the individual electronic circuits. The main steps for the fabrication of a die are summarized in the following table. Some of them are repeated several times at different stages of the process. The order given here doesn't reflect the real order of fabrication process.

PhotoMasking

This step shapes the different components. The principle is quite simple (see drawing on next page). Resin is put down on the wafer which is then exposed to light through a specific mask. The lighten part of the resin softens and is rinsed off with solvents (developing step).

Etching

This operation removes a thin film material. There are two different methods: wet (using a liquid or soluble compound) or dry (using a gaseous compound like oxygen or chlorine).

Diffusion

This step is used to introduce dopants inside the material or to grow a thin oxide layer onto the wafer. Wafers are inserted into a high temperature furnace (up to 1200 ° C) and doping gazes penetrate the silicon or react with it to grow a silicon oxide layer.



Ionic Implantation

It allows to introduce a dopant at a given depth into the material using a high energy electron beam.

Metal Deposition

It allows the realization of electrical connections between the different cells of the integrated circuit and the outside. Two different methods are used to deposit the metal: evaporation or sputtering.

Passivation

Wafers are sealed with a passivation layer to prevent the device from contamination or moisture attack. This layer is usually made of silicon nitride or a silicon oxide composite.

Back-lap

It’s the last step of wafer fabrication. Wafer thickness is reduced (for microcontroller chips, thickness is reduced from 650 to 380 microns), and sometimes a thin gold layer is deposited on the back of the wafer.

Initially, the silicon chip forms part of a very thin (usually 650 microns), round silicon slice: the raw wafer. Wafer diameters are typically 125, 150 or 200 mm (5, 6 or 8 inches). However raw pure silicon has a main electrical property: it is an isolating material. So some of the features of silicon have to be altered, by means of well controlled processes. This is obtained by "doping" the silicon.

Dopants (or doping atoms) are purposely inserted in the silicon lattice, hence changing the features of the material in predefined areas: they are divided into “N” and “P” categories representing the negative and positive carriers they hold. Many different dopants are used to achieve these desired features: Phosphorous, Arsenic (N type) and Boron (P type) are the most frequently used ones. Semiconductors manufacturers purchase wafers predoped with N or P impurities to an impurity level of.1 ppm (one doping atom per ten million atoms of silicon). There are two ways to dope the silicon. The first one is to insert the wafer into a furnace. Doping gases are then introduced which impregnate the silicon surface. This is one part of the manufacturing process called diffusion (the other part being the oxide growth). The second way to dope the silicon is called ionic implantation. In this case, doping atoms are introduced inside the silicon using an electron beam. Unlike diffusion, ionic implantation allows to put atoms at a given depth inside the silicon and basically allows a better control of all the main



parameters during the process. Ionic implantation process is simpler than diffusion process but more costly (ionic implanters are very expensive machines).

Figure 1.2 Diffusion and Ionic Implantation Processes

PhotoMasking (or masking) is an operation that is repeated many times during the process. This operation is described in figure 1.3. This step is called photomasking because the wafer is “masked” in some areas (using a specific pattern), in the same way one “masks out” or protects the windscreens of a car before painting the body. But even if the process is somewhat similar to the painting of a car body, in the case of a silicon chip the dimensions are measured in tenth of microns. The photoresist will replicate this pattern on the wafer. The exposed part of the photoresist is then rinsed off with a solvent (usually hydrofluoric or phosphoric acid).

Figure 1.3 Photo Masking Process Metal deposition is used to put down a metal layer on the wafer surface. There are two ways to do that. The process shown in the figure 1.4, is called sputtering. It consists first in creating a plasma with argon ions. These ions bump into the target surface (composed of a metal, usually aluminium) and rip metal atoms from the target. Then, atoms are projected in all the directions and most of them condense on the substrate surface.



Figure 1.4 Metal Deposition Process Etching process is used to etch into a specific layer the circuit pattern that has been defined during the photomasking process. Etching process usually occurs after deposition of the layer that has to be etched. For instance, the poly gates of a transistor are obtained by etching the poly layer. A second example is the aluminium connections obtained after etching of the aluminum layer.

Figure 1.5 Etching Process

Photomasking, ionic implantation, diffusion, metal deposition, and etching processes are repeated many times, using different materials and dopants at different temperatures in order to achieve all the operations needed to produce the requested characteristics of the silicon chip. The resolution limit (minimal line size inside the circuit) of current technology is 0.35 microns. Achieving such results requires very sophisticated processes as well as superior quality levels.

Backlap is the final step of wafer fabrication. The wafer thickness is reduced from 650 microns to a minimum of 180 microns (for smartcard products).

Wafer fabrication takes place in an extremely clean environment, where air cleanliness is one million times better than the air we normally breathe in a city, or some orders of magnitude better than the air in a heart transplant operating theatre. Photomasking, for example, takes place in rooms where there’s maximum



one particle whose diameter is superior to 0.5 micron (and doesn’t exceed 1 micron) inside one cubic foot of air.

All these processes are part of the manufacturing phase of the chip itself. Silicon chips are grouped on a silicon wafer (in the same way postage stamps are printed on a single sheet of paper) before being separated from each other at the beginning of the assembly phase.

Wafer Probing. This step takes place between wafer fabrication and assembly. It verifies the functionality of the device performing thousands of electrical tests, by means of special microprobes. Wafer probing is composed of two different tests:

1. Process parametric test: This test is performed on some test samples and checks the wafer fabrication process itself.

2. Full wafer probing test: This test verifies the functionality of the finished product and is performed on all the dies. The bad dies are automatically marked with a black dot so they can be separated from good die after the wafer is cut. A record of what went wrong with the non-working die is closely examined by failure analysis engineers to determine where the problem occurred so that may be corrected. The percentage of good die on an individual wafer is called its yield.

Figure 1.6 Description of the Wafer Probing Operation

1.1.3 Assembly (Back-End)

Figure 1.7



The first step of assembly is to separate the silicon chips: this step is called die cutting (figure 1.7). Then, the dies are placed on a lead frame: the “leads” are the chip legs (which will be soldered or placed in a socket on a printed circuit board. On a surface smaller than a baby's fingernail we now have thousands (or millions) of electronic components, all of them interconnected and capable of implementing a subset of a complex electronic function. At this stage the device is completely functional, but it would be impossible to use it without some sort of supporting system. Any scratch would alter its behavior (or impact its reliability), any shock would cause failure. Therefore, the die must be put into a ceramic or plastic package to be protected from the external world.

Figure 1.8 Description of The Assembly Process

Figure 1.9 Wire Bonding Wires thinner than a human hair (for microcontrollers the typical value is 33 microns) are required to connect chips to the external world and enable electronic signals to be fed through the chip. The process of connecting these thin wires from the chip’s bond pads to the package lead is called wire bonding.

The chip is then mounted in a ceramic or plastic package. The package not only protects the chip from external shocks, but also makes the whole device easier to handle. These packages come in a variety of shapes and sizes depending on the die itself and the application in which it will be used.



Figure 1.10 Wire Bonding Operation

Products are then marked with a “traceability code” which is used by the manufacturer and the user to identify the function of the device (and its date of fabrication). At the end of the assembly process, the integrated circuit is tested by automated test equipment. Only the integrated circuits that passed the tests will be packed and shipped to their final destination.

Figure 1.11 Different Kinds of Plastic Packages

1.2 Basic CMOS Technology

Complementary metal–oxide–semiconductor (CMOS) (pronounced "see-moss), is a major class of integrated circuits. CMOS technology is used in microprocessors, microcontrollers, static RAM, and other digital logic circuits. CMOS technology is also used for a wide variety of analog circuits such as image sensors, data converters, and highly integrated transceivers for many types of communication. Frank Wanlass got a patent on CMOS in 1967 (US Patent 3,356,858).



CMOS is also sometimes referred to as complementary-symmetry metal–oxide–semiconductor. The words "complementary-symmetry" refer to the fact that the typical digital design style with CMOS uses complementary and symmetrical pairs of p-type and n-type metal oxide semiconductor field effect transistors (MOSFETs) for logic functions.

Two important characteristics of CMOS devices are high noise immunity and low static power consumption. Significant power is only drawn when the transistors in the CMOS device are switching between on and off states. Consequently, CMOS devices do not produce as much waste heat as other forms of logic, for example transistor-transistor logic (TTL) or NMOS logic, which uses all n-channel devices without p-channel devices. CMOS also allows a high density of logic functions on a chip.

The four main CMOS technologies are; • n-well process. • p-well process. • twin-tub process. • Silicon on insulator.

1.2.1 A Basic n-well CMOS Process

The basic process steps for pattern transfer through lithography, and having gone through the fabrication procedure of a single n-type MOS transistor, the generalized fabrication sequence of n-well CMOS integrated circuits, as shown in figure. 1.12 In the following figures, some of the important process steps involved in the fabrication of a CMOS inverter will be shown by a top view of the lithographic masks and a cross-sectional view of the relevant areas. The n-well CMOS process starts with a moderately doped (with impurity concentration typically less than 1015 cm-3) p-type silicon substrate. Then, an initial oxide layer is grown on the entire surface. The first lithographic mask defines the n-well region. Donor atoms, usually phosphorus, are implanted through this window in the oxide.

Figure 1.12



Once the n-well is created, the active areas of the nMOS and pMOS transistors can be defined. Figures 1.13 through 1.18 illustrate the significant milestones that occur during the fabrication process of a CMOS inverter. Following the creation of the n-well region, a thick field oxide is grown in the areas surrounding the transistor active regions, and a thin gate oxide is grown on top of the active regions. The thickness and the quality of the gate oxide are two of the most critical fabrication parameters, since they strongly affect the operational characteristics of the MOS transistor, as well as its long-term reliability.

Polysilicon Gate Connections Figure 1.13

The polysilicon layer is deposited using chemical vapor deposition (CVD) and patterned by dry (plasma) etching. CVD Chemical Reactions

• SiH4(gas) + O2(gas) SiO2(solid) + 2H2 (gas) • SiH4(gas) + H2(gas) +SiH2(gas) 2H2(gas) + PolySilicon (solid) •

Figure 1.14



Isolation layer

Figure 1.15

The created polysilicon lines will function as the gate electrodes of the nMOS and the pMOS transistors and their interconnects. Also, the polysilicon gates act as self-aligned masks for the source and drain implantations that follow this step. Using a set of two masks, the n+ and p+ regions are implanted into the substrate and into the n- well, respectively. Also, the ohmic contacts to the substrate and to the n-well are implanted in this process step.

Figure 1.16

An insulating silicon dioxide layer is deposited over the entire wafer using CVD. Then, the contacts are defined and etched away to expose the silicon or polysilicon contact windows. These contact windows are necessary to complete the circuit interconnections using the metal layer, which is patterned in the next step.

Figure 1.17



Metal (aluminum) is deposited over the entire chip surface using metal evaporation, and the metal lines are patterned through etching.

Figure 1.18

Since the wafer surface is non-planar, the quality and the integrity of the metal lines created in this step are very critical and are ultimately essential for circuit reliability. The composite layout and the resulting cross-sectional view of the chip, showing one nMOS and one pMOS transistor (built-in n-well), the polysilicon and metal interconnections. The final step is to deposit the passivation layer (for protection) over the chip, except for wire-bonding pad areas. The patterning process by the use of a succession of masks and process steps is conceptually summarized in Figure. 1.19. It is seen that a series of masking steps must be sequentially performed for the desired patterns to be created on the wafer surface. An example of the end result of this sequence is shown as a cross-section on the right.

Figure 1.19



1.2.2 A Basic p-well CMOS Process

N-well processes have emerged in popularity in recent years. Prior to this p-well process was one of the most commonly available forms of CMOS. Typical p-well fabrication steps are similar to an n-well process, except that a p-well is implemented rather than an n-well. The first masking step defines the p-well regions. This is followed by a low-dose boron implant driven in by a high-temperature step for the formation of the p-well. The well depth is optimized to ensure against n-substrate to n+ diffusion breakdown, without compromising p-well to p+ separation. The next steps are to define the devices and other; to grow field oxide; contact cuts; and metallization. A p-well mask is used to define the p-channel transistors and Vss contacts. Alternatively, an n-plus mask to define the n-channel transistors, because the masks usually are the complement of each other. P-well process are preferred in circumstances where the characteristics of the n- and p-transistors are required to be more balanced than that achievable in an n-well process. Because the transistor that resides in the native substrate tends to have better characteristics, the p-well process has better p devices than an n-well process. Because p-devices inherently have lower gain than n-devices, the n-well process exacerbates this difference while a p-well process moderates the difference. 1.2.3 Twin-Tub (Twin-Well) CMOS Process

Twin-tub technology provides the basis for separate optimization of the nMOS and pMOS transistors, thus making it possible for threshold voltage, body effect and the channel transconductance of both types of transistors to be tuned independently. Generally, the starting material is a n+ or p+ substrate, with a lightly doped epitaxial layer on top. This epitaxial layer provides the actual substrate on which the n-well and the p-well are formed.

Figure 1.20 Twin-well CMOS process cross section Since two independent doping steps are performed for the creation of the well regions, the dopant concentrations can be carefully optimized to produce the desired device characteristics. The aim of epitaxy is to grow high-purity silicon



layers of controlled thickness with accurately determined dopant concentration distributed homogenously throughout the layer. The electrical properties of this layer are determined by the dopant and its concentration in the silicon. The process sequence, which is similar to the n-well process apart from the tub formation where both p-well and n-well are utilized, entails the following steps,

• Tub formation. • Thin-oxide construction. • Source and drain implantations. • Contact cut definition. • Metallization.

In the conventional n-well CMOS process, the doping density of the well region is typically about one order of magnitude higher than the substrate, which, among other effects, results in unbalanced drain parasitics. The twin-tub process (figure 1.20) also avoids this problem. 1.2.4 Silicon On Insulator (SOI) Process

Silicon on insulator technology (SOI) refers to the use of a layered silicon-insulator-silicon substrate in place of conventional silicon substrates in semiconductor manufacturing, especially microelectronics, to reduce parasitic device capacitance and thereby improve performance. SOI-based devices differ from conventional silicon-built devices in that the silicon junction is above an electrical insulator, typically silicon dioxide or (less commonly) sapphire. The choice of insulator depends largely on intended application, with sapphire being used for radiation-sensitive applications and silicon oxide preferred for improved performance and diminished short channel effects in microelectronics devices. The insulating layer and topmost silicon layer also vary widely with application. The first implementation of SOI was announced by IBM in August 1998. Rather than using silicon as the substrate, the technologies have sought to use an insulating substrate to improve process characteristics such as latchup and speed. Hence the emergence of Silicon On Insulator (SOI) technologies. SOI CMOS processes have several potential advantages over the traditional CMOS technologies. These include closer packing of p- and n- transistors, absence of latchup problems, and lower parasitics substrate capacitances. In the SOI process a thin layer of single-crystal silicon film is epitaxially grown on an insulator such as sapphire or magnesium aluminium spinal. Alternatively, the silicon may be grown on SiO2 that has been in turn grown on silicon. This option has proved more popular in recent years due to the compatibility of the starting material with conventional silicon CMOS fabrication. Various masking and doping techniques



(figure 1.21) are then used to form p-channel and n-channel devices. Unlike the more conventional CMOS approaches, the extra steps in well formation do not exist in the technology. The steps used in typical SOI CMOS process are as follows. A thin film (7-8 µm) of very lightly –doped n-type Si is grown over an insulator, Sapphire or SiO2 is commonly used insulator (figure 1.21 a). • An anisotropic etch is used away the Si except where a diffusion area (n or p) will be needed. The etch must be anisotropic since the thickness of the Si is much greater than the spacing desired between the Si “islands: (figure 1.21 b, c). • The p-islands are formed next by masking the n-islands with a photoresist. A p-type dopant, boron, for example is then implanted. It is masked by the photoresist, but forms p-islands at the unmasked islands. The p-islands will become the n-channel devices (figure 1.12 d). • The p-islands are then covered with a photoresist and an n-type dopant- phosphorus, for example is implanted to form the n-islands. The n-islands will become the p-channel devices (figure 1.12 e). • A thin gate oxide (around 100-250 A) is grown over all of the Si structures, this is normally done by thermal oxidation. • A polysilicon film is deposited over the oxide. Often the polysilicon is doped with phosphorus to reduce its resistivity (figure 1.12f). • The polysilicon is then patterned by photomasking and is etched. This defines the polysilicon layer in the structure (figure 1.12 g). • The next step is to form the n-doped source and drain of the n-channel devices in the p-islands. The n-islands are covered with a photoresist and an n-type dopant, normally phosphorus is implanted. The dopant and an n-type dopant, normally phosphorus is implanted. The dopant will be blocked at the n-islands by the photoresist, and it will be blocked from the gate region of the p-islands by the polysilicon. After this step the n-channel devices are complete (figure 1.12 h). • The p-channel devices are formed next by masking the p-islands and implanting a p-type dopant such as boron. The polysilicon over the gate of the n-island will block the dopant from the gate, thus forming the p-channel devices (figure 1.12 i). • A layer of phosphorus glass or some other insulator such as silicon dioxide is then deposited over the entire structure. • The glass is etched as contact –cut locations. The metallization layer is formed next by evaporating aluminum over the entire surface and etching it to leave only the desired metal wires. The aluminium will flow through the contact cuts to make contact with the diffusion or polysilicon regions (figure 1.12 j). • A final passivation layer of phosphorus glass is deposited and etched over bonding pad locations (not shown in figure).



Because the diffusion regions extend to the insulating substrate, only “sidewall” areas associated with source and drain diffusion contribute to the parasitic junction capacitance. Since sapphire and SiO2 are extremely good insulators, leakage currents between transistors and substrate and adjacent devices are almost eliminated.

In order to improve the yield, some processes use “preferential etch” in which he island edges are tapered. Thus aluminium or poly runners can enter and leave the islands with a minimum step height. This is contrasted to “fully anisotropic etch” in which the undercut is brought to zero, as shown in figure 1.13.An” isotropic etch” is also shown in the same diagram for the comparison.

Figure 1.12 SOI Process Flow



The advantages of SOI technology are as follows,

• Due to absence of wells, transistor structures denser than bulk silicon are feasible. Also direct n-to-p connections may be made. • Lower substrate capacitances provide the possibility for faster circuits. • No field-inversion problems exist( insulating substrate) • There is no latchup because of the isolation of the n-and p-transistors by the insulating substrate. • Because there is no conducting substrate, there are no body-effect problems. However the absence of a backside substrate contact could lead to odd device characteristic such as the “kink” effect in which the drain current increases abruptly at around 2 to 3 volts. Some of the disadvantages are, • Due to absence of substrate diodes, the inputs are somewhat more difficult to protect. Because device gains are lower, I/O structures have to be larger. • Single crystal sapphire, spinel substrate, and silicon SiO2 are considerably more expensive than silicon substrate and their processing techniques tend to be less developed than bulk silicon techniques.

Figure 1.13 Classification of Etching processes



1.3 INTERCONNECT

The most important additions for CMOS logic processes are additional signal- and power-routing layers. This eases the routing (especially automated netting) of logic signals between modules and improves the power and clock distribution to modules. Improved mutability is achieved through additional layers of metal or by improving the existing polysilicon interconnection layer.

1.3.1 Metal Interconnect

A second level of metal is almost mandatory for modern CMOS digital. A third layer is becoming common and is certainly required for leading-edge high-density, high-speed chips. Normally, aluminum is used for the metal layers. I f some form of planarization is employed the second-level metal pitch can be the same as the first. As the vertical topology becomes more varied, the width and spacing of metal conductors has to increase so that the conductors do not thin and hence break at vertical topology jumps (step coverage). Contacting the second-layer metal to the first-layer metal is achieved by a via, as shown in figure 1.14. If further contact to diffusion or polysilicon is required, a separation between the via and the contact cut is usually required. This requires a first-level metal tab to bridge between metal2 and the lower-l e v e l conductor. It is important to realize that in contemporary processes first level metal must be involved in any contact to underlying areas. A number of contact geometries are shown in figure 1.15.

Figure 1.14 Two-level metal process cross section

Processes usually require metal borders around the via on both levels of metal although some process require none. Processes may have no restrictions on the placement of via with respect to underlying layers



(figure 1.15a) or they may have to be placed inside (figure 1.15b) or outside (f igure1.15c) the underlying polysilicon or diffusion areas. Aggressive processes allow the stacking of vias on top of contacts, as shown in figure 1.15 (d). a b c d

Figure 1.15 Two-level metal /via contact geometrics

Consistent with the relatively large thickness of the intermediate isolation layer, the vias might be larger than contact cuts and second-layer metal may need to be thicker and require a larger via overlap although modern processes strive for uniform pitches on metal I and metal2. The process steps for a two-metal process are briefly as follows: • The oxide below the first-metal layer is deposited by atmospheric chemical vapor deposition (CVD). • The second oxide layer between the two metal layers is applied in a similar manner. • Depending on the process, removal of the oxide is accomplished using a plasma etcher designed to have a high rate of vertical ion bombardment. This allows fast and uniform etch rates. The structure of a via etched using such a method is shown in figure1.14.

1.3.2 Polysilicon/Refractory Metal Interconnect

The polysilicon layer used for the gates of transistors is commonly used as i t interconnect layer. However, the sheet resistance of doped polysilicon is between 20Ω and 40Ω/square. If used as a long distance conductor, a polysilicon wire can represent a significant delay.



One method to improve this that requires no extra mask levels is to reduce the polysilicon resistance by combining it with a refractory metal. Three such approaches are illustrated in figure 1.16.In figure 1.16(a) a silicide (e.g., silicon and tantalum) is used as the gate material. Sheet resistances of the order of 1 to 5Ω/square may be obtained. This is called the, silicide gate approach.

Figure 1.16 Refractory metal interconnect

Silicides are mechanically strong and may be dry ached in plasma reactors. Tantalum silicide is stable throughout standard processing and has the advantage that it may be retrofitted into existing process lines. Figure 1.16(b) uses a sandwich of sil icide upon polysilicon, which is commonly called the polycide approach. Finally, the silicide/polysilicon approach may he extended to include the formation of source and dra in r eg ions us ing the s i l i c ide . This is called the salicide process (Self Aligned SILICIDE) (f igure 1.16c). The effect of all of these processes is to reduce the "second layer" interconnect resistance, allowing the gate material to be used as a moderate long-distance interconnect. This is achieved by minimum perturbation of an existing process. An increasing trend in process is to use the salicide approach to reduce the resistance of both gate and source/drain conductors.

1.3.3 Local Interconnect

The silicide itself may be used as a "local interconnect" layer for connection within cel ls . TiN is used as an example. Local interconnect allows a direct connection between polysilicon and diffusion, thus alleviating the need for area intensive contacts and metal. Figure 1.17 shows a portion (p-devices only) of a six transistor SRAM cell that uses local interconnect. The local interconnect has been used to make the polysilicon-to-diffusion connections within the cell, thereby alleviating the need to use metal (and contacts). Metal2 (not shown) bit lines run over the cell vertically. Use of local in te rconnect in this RAM reduced the cell area by 25%.



Figure 1.17 Local interconnect as used in a RAM cell In general, local interconnect if available can be used to complete intracell routing, leaving the remaining metal layers for global wiring.

1.4 CIRCUIT ELEMENTS

1.4.1 Resistors

Polysilicon, if left undoped, is highly resistive. This property is used to build resistors that are used in static memory cells. The process step is achieved by preventing the resistor areas from being implanted during normal processing. Resistors in the tera-Ω (1012 Ω) region are used. A value of 3TΩ results in a standby current of 2µA for a 1 Mbit memory. For mixed signal CMOS (analog and digital), a resistive metal such as nichrome may be added to produce high-value, high-quality resistors. The resistor accuracy might be further improved by laser trimming the result resistors on each chip to some predetermined test specification. In this process a high-powered laser vaporizes areas of the metal resistor until it meets a measurement constraint. Sheet resistance values in the KΩ/square are normal. The resistors have excellent temperature stability and long-term reliability.

1.4.2 Capacitors

Good quality capacitors are required for switched-capacitor analog circuits while small high-value/area capacitors are required for dynamic memory cells. Both types of capacitors are usually added by using at least one extra layer of polysilicon, although the process techniques are very different.



Polysilicon capacitors for analog applications are the most straightforward. A second thin-oxide layer is required in order to have an oxide sandwich between the two polysilicon layers yielding a high-capacitance/unit area. Figure 1.18 shows a typical polysilicon capacitor. The presence of this, second oxide can also be used to fabricate transistors. These may differ, characteristics from the primary gate oxide devices. For memory capacitors recent processes have used three dimensions to increase the capacitance/area.

Figure 1.18 Polysilicon Capacitor

One popular structure is the trench capacitor, which has evolved considerably over the years to push memory densities to 64Mbits and beyond. A typical trench structure is shown in figure 1.19(a). The sides of the trench are doped n+ and coated with a thin 1Onm oxide. Sometimes oxynitride is used because its high dielectric constant increases capacitance.

a b

Figure 1.19 Dynamic memory capacitors The trench is filled with a polysilicon plug, which forms the bottom plate of the cell storage capacitor. This is held at VDD/2 via a metal connection at the



edge of the array. The sidewall n+ forms the other side of capacitor and one side of the pass transistor that is used to enable data onto the bit lines. The bottom of the trench has a p+ plug that forms a channel stop region to isolate adjacent capacitors. The trench is 4µm deep and has a capacitance of 90fF. Rather than building a trench, figure 1.19(b) shows a fintype- capacitor used in a 64-Mb DRAM. The storage capacitance is 20 to 30 fF. The fins have the additional advantage of reducing the bit capacitance by shielding the bit lines. The fabrication of 3D-process structures such as these is a constant reminder of the skill, perseverance, and ingenuity of the process engineer. 1.4.3 Electrically Alterable ROMs

Electrically alterable/erasable R O M ( E A R O M / E E P R O M ) i s added to CMOS processes to yield permanent but reprogrammable s torage to a process. This is usually added by adding a polysilicon layer. Figure 1.20 shows a typical memory structure, which consists of a stacked-gate s t ruc ture . The normal gate is left floating, while a control gate is placed above the floating gate. A very thin oxide called the tunnel oxide separates the floating gate from the source, drain, and substrate.

Figure 1.20 EEPROM technology

This is usually 10 nm thick. Another thin oxide separates the control gate from the floating gate. By controlling the control-gate, source, and drain voltages, the thin tunnel oxide between the floating gate and the drain of the device is used to allow electrons to "tunnel" to or from the floating gate to turn the cell or on, respectively, using Fowler-Nordheim tunneling. Alternatively, by setting the appropriate voltages on the terminals, "hot electrons" can be induced to charge the floating gate, thereby programming the transistor. In non-electrically alterable versions of the technology, the process can be reversed by illuminating the gate with UV light. In these the chips are usually housed in glass-lidded packages.



1.4.4 Bipolar Transistors The addition of the bipolar transistor to the device repertoire forms the basis for BiCMOS processes. Adding an npn-transistor can markedly aid in reducing the delay times of highly loaded signals, such as memory word lines microprocessor busses. Additionally, for analog applications bipolar transistors may be used to provide better performance analog functions than MOS alone. To get merged bipolar/CMOS functionality,

Figure 1.21 Typical mixed signal BiCMOS process cross section

Figure 1.22 BiCMOS process steps for the cross section shown in

figure 1.21



MOS transistors can add to a bipolar process or vice versa. In past days, MOS processes always had to have excellent gate oxides while bipolar processes had to have precisely controlled diffusions. A BiCMOS process has to have both. A mixed signal BiCMOS process cross section is shown in figure 1.21. This process features both npn- and pnp-transistors in addition to pMOS and nMOS transistors. The major processing steps are summarized in figure 1.22, showing the particular device to which they correspond. The base layers of the process are similar to the process shown in figure 1.12. The starting material is a lightly-doped p-type substrate into which antimony or arsenic are diffused to form an n+ buried layer. Boron is diffused to form a buried p+ layer. An n-type epitaxial layer 4.0 µm thick is then grown. N-wells and p-wells are then diffused so that they join in the middle of the epitaxial layer. This epitaxial layer isolates the pnp-transistor in the horizontal direction, while the buried n+ layer isolates it vertically. The npn-transistor is junction-isolated. The base for the pnp is then ion-implanted using phosphorous. A diffusion step follows this to get the right doping profile. The npn-collector is formed by depositing phosphorus before LOCOS. Field oxidation is carried out and the gate oxide is grown. Boron is then used to form the p-type base of the npn transistor. Following the threshold adjustment of the pMOS transistors, the polysilicon gates are defined. The emitters of the npn-transistors employ polysilicon rather than a diffusion. These are formed by opening windows and depositing polysilicon. The n+ and p+ source/drain implants are then completed. This step also dopes the npn-emitter and the extrinsic bases of the npn- and pnp-transistors (extrinsic because this is the part of the base that is not directly between collector and emitter). Following the deposition of PSG, the normal two-layer metallization steps are completed. Representative of a high-density digital BiCMOS process is that represented by the cross section shown in figure 1.23. The buried-layer-epitaxial layer-well structure is very similar to the previous structure. However because this is a 0.8µm process, LDD structures must be constructed for the p-transistors and the n-transistors. The npn is formed by a double-diffused sequence in which both base and emitter are formed by impurities that diffuse out of a covering layer of polysilicon. This process, intended for logic applications, has only an npn-transistor. The collector of the npn is connected to the n-well, which is in turn connected to the VDD supply. Thus all npn-collectors are commoned. A typical npn-transistor with a 0.8µm-square emitter has a current gain of 90 and an ft. of 15 GHz.



Figure 1.23 Digital BiCMOS process cross section

1.4.5 LatchUp

If every silver lining has a cloud, then the cloud that has plagued CMOS is a parasitic circuit effect called "latchup." The result of this effect is the shorting of the VDD and Vss lines, usually resulting in chip self-destruction or at least system failure with the requirement to power down. This effect was a critical factor in the lack of acceptance of early CMOS processes, but in cur-rent processes it is controlled by process innovations and well-understood circuit techniques.

1.4.5.1 The Physical Origin of Latchup

The source of the latchup effect may be explained by examining the process cross section of a CMOS inverter, shown in figure 1.24(a), on which is overlaid an equivalent circuit. The schematic depicts, in addition to the expected nMOS and pMOS transistors, a circuit composed of an npn-transistor, a pnp-transistor, and two resistors connected between the power and ground rails (figure 1.24b). Under the right conditions, this parasitic circuit has the VI characteristic shown in figure 1.24(c), which indicates that above some critical voltage (known as the trigger point) the circuit "snaps" and draws a large current while maintaining a low voltage across the terminals (known as the holding voltage). This is, in effect, a short circuit. As mentioned, the bipolar devices and resistors shown in figure 1.24 (b) are parasitic, that is an unwanted byproduct of producing pMOS and nMOS transistors. From the figure 1.24(a) reveals how these devices are constructed. The figure shows a cross-sectional view of a typical (n-well) CMOS process. The (vertical) pnp-transistor has its emitter formed by the p+ source/drain implant used in the pMOS transistors. Note that either the drain or source may act as the emitter although the source is the only terminal that can maintain the latchup condition. The base is formed by the n-well, while the collector is the p-



substrate. The emitter of the (lateral) npn-transistor is the n+ source/drain implant, while the base is the p-substrate and the collector is the n-well. In addition, substrate resistance Rsubstrate and well resistance Rwell are due to the resistivity of the semiconductors involved.

Figure 1.24 The origin model, and VI characteristics of CMOS Latchup Consider the circuit shown in figure 1.24(b). If a current is drawn from the npn-emitter, the emitter voltage becomes negative with respect to the base until the base emitter voltage is approximately 0.7 volts. At this point the npn-transistor turns on and a current flows in the well resistor due to common emitter current amplification. This raises the base emitter voltage of the pnp-transistor, which turns on when the pnp Vbe = -0 .7 volts. This in turn raises the npn base voltage causing a positive feedback condition, which has the characteristic shown in figure 1.24(c). At a certain npn-base-emitter voltage, called the trigger point, the emitter voltage suddenly "snaps back" and enters a stable state called the ON state. This state will persist as long as the voltage across the two transistors is greater than the holding voltage shown in the figure. As the emitter of the npn is the source/drains of the n-transistor, these terminals are now at roughly 4 volts. Thus there is about 1 volt across the CMOS inverter, which will



most likely cause it to cease operating correctly. The current drawn is usually destructive to metal lines supplying the latched up circuitry.

1.4.5.2 Latchup Triggering

For latchup to occur the parasitic npn-pnp circuit has to be triggered and the holding state has to be maintained. Latchup can be triggered by transient cur-rents or voltages that may occur internally to a chip during power-up or externally due to voltages or currents beyond normal operating ranges. Radiation pulses can also cause latchup. Two distinct methods of triggering are possible, lateral triggering and vertical triggering.

Lateral triggering occurs when a current flows in the emitter of the lateral npn-transistor. The static trigger point is set by

Intrigger ~ Vpnp-on (1.1) αnpn Rwell where Vpnp_on~ 0.7 volts the turn-on voltage of the vertical pnp-transistor anpn = common base gain of the lateral npn-transistor Rwell = well resistance.

Vertical triggering occurs when a sufficient current is injected into the emitter of the vertical-pnp transistor. Similar to the lateral case, this current is multiplied by the common-base-current gain, which causes a voltage drop across the emitter base junction of the npn transistor due to the resistance, Rsubstrate. When the holding or sustaining point is entered, it represents a stable operating point provided the current required to stay in the state can he maintained.

Current has to be injected into either the npn- or pnp-emitter to initiate latchup. During normal circuit in internal circuitry this may occur due to supply voltage transients, but this is unlikely. However, these conditions may occur at the I/O circuits employed on a CMOS chip, where the internal circuit voltages meet the external world and large currents can flow. Therefore extra precautions need to be taken with peripheral CMOS circuits.



a

b

Figure 1.25 Externally included latchup

Figure 1.25(a) illustrates an example where the source of an nMOS output transistor experiences undershoot with respect to Vss due to some external circuitry. When the output dips below Vss by more than 0.7V, the drain of the nMOS output driver is forward biased, which initiates latchup. The complementary case is shown in figure 1.25(b) where the pMOS output transistor experiences an overshoot more than 0.7V beyond VDD. Whether or not in these cases latchup occurs depends on the pulse widths and speed of the parasitic transistors. 1.4.6 Latchup Prevention

For latchup to occur an analysis of the circuit in figure 1.25(b) finds the following inequality has to be true βnpnβpnp> 1+ (βnpn+1) IRsubstrate+IRwellβpnp) (1.2)

IDD - IRsubstrate Where IRsubstrate == Vbe npn Rsubstrate IRwell = Vbe pnp Rwell

IDD =total supply current



This equation yields the keys to reducing latchup to the point where it should never occur under normal circuit conditions. Thus, reducing the resistor values and reducing the gain of the parasitic transistors are the basis for eliminating latchup. Latchup may be prevented in two basic ways:

• Latchup resistant CMOS processes. • Layout techniques.

A popular process option that reduces the gain of the parasitic transistors is the use of silicon starting-material with a thin epitaxial layer on top of a highly doped substrate. This decreases the value of the substrate resistor and also provides a sink for collector current of the vertical pnp-transistor. As the epi layer is thinned, the latchup performance improves until a point where the up-diffusion of the substrate and the down-diffusion of any diffusions in subsequent high-temperature procession steps thwart required device doping profiles. The so-called retrograde well structure is also used. This well has a highly doped area at the bottom of the well, whereas the top of the well is more lightly doped. This preserves good characteristics for the pMOS (or nMOS in p-well) transistors but reduces the well resistance deep in the well. A technique linked to these two approaches is to increase the holding voltage above the VDD supply. This guarantees that latchup will not occur.

It is hard to reduce the betas of the bipolar transistors to meet the condi-tion set above. Nominally, for a 1µ n-well process, the vertical pnp has a beta of 10-100, depending on the technology. The lateral npn-current-gain which is a function of n+ drain to n-well spacing , i s b e t w e e n 2 and 5.

1.5 LAYOUT DESIGN RULES

Layout rules, also referred to as design rules, can be considered as a pre-scription for preparing the photomasks used in the fabrication of integrated circuits. The rules provide a necessary communication link between circuit designer and process engineer during the manufacturing phase. The main objective associated with layout rules is to obtain a circuit with optimum yield (functional circuits versus nonfunctional circuits) in as small an area as possible without compromising reliability of the circuit.

In general, design rules represent the best possible compromise between performance and yield. The more conservative the rules are, the more likely it is that the circuit will function. However, the more aggressive the



rules are, the greater the probability of improvements in circuit performance. This improvement may be at the expense of yield. Design rules specify to the designer certain geometric constraints on the layout artwork so that the patterns on the processed wafer will preserve the topology and geometry of the designs. It is important to note that design rules do not represent some hard boundary between correct and incorrect fabrication. Rather, they represent a tolerance that ensures very high probability of correct fabrication and subsequent operation. For example, one may find that a layout that violates design rules may still function correctly, and vice versa. Nevertheless, any significant or frequent departure (design-rule waiver) from design rules will seriously prejudice the success of a design.

Two sets of design-rule constraints in a process relate to line widths and interlayer registration. If the line widths are made too small, it is possible for the line to become discontinuous, thus leading to an open circuit wire. On the other hand, if the wires are placed too close to one another, it is possible for them to merge together; that is, shorts can occur between two independent circuit nets. Furthermore, the spacing between two independent layers may be affected by the vertical topology of a process. The design rules primarily address two issues: (1) The geometrical reproduction of features that can be reproduced by the mask- making and litho-graphical process and (2) The interactions between different layers.

There are several approaches that can be taken in describing the design rules. These include 'micron' rules stated at some micron resolution, and lambda (λ) based rules. Micron designs rules are usually given as a list of minimum feature sizes and spacings for all masks required in a given process.

1.5.1 Layer Representations

The advances in the CMOS processes are generally complex and somewhat inhibit the visualization of all the mask levels that are used in the actual fabrication process. Nevertheless the design process can be abstracted to a manageable number of conceptual layout levels that represent the physical features observed in the final silicon wafer. At a sufficiently high conceptual level all CMOS processes use the following features:

• Two different substrates. • Doped regions of both p- and n-transistor-forming material. • Transistor gate electrodes.



• Interconnection paths. • Interlayer contacts.

The layers for typical CMOS processes are represented in various figures in terms of:

• A color scheme proposed by JPL based on the Mead-Conway colors. • Other color schemes designed to differentiate CMOS structures

(e.g., the colors as used on the from cover of this hook) • Varying stipple patterns. • Varying line styles.

Some of these representations are shown in below table.

1.5.2 CMOS n-well Rules

In this section a version of n-well rules based on the MOSIS CMOS Scalable Rules and compares those with the rules for a hypothetical commercial 1µ CMOS process shown in below table. The MOSIC rules are expressed in terms of λ. These rules allow some degree of scaling between processes as, in principal, we only need to reduce the value of λ and the designs will be valid in the next process down in size. Unfortunately, history has shown that processes rarely shrink uniformly. Thus industry usually uses the actual micron-design rules and codes designs in terms of these dimensions, or uses symbolic layout systems to target the design rules exactly. At this time, the amount of polygon pushing is usually constrained to a number of frequently used



standard cells or memories, where the effort expended is amortized over many designs. Alternatively, the designs are done symbolically, thus relieving the designer of having to deal directly with the actual design rules.

The rules are defined in terms of: • Feature sizes. • Separations and overlaps.



1.5.3 Scribe Line The scribe line is specifically designed structure that surrounds the completed chip and is the point at which the chip is cut with a diamond saw. The construction of the scribe line varies from manufacturer to manufactures

1.5.4 SOI Rules

SOI rules closely follow bulk CMOS rules except the n+ and p+ regions can abut. This allows some interesting and latch circuits. A spacing rule between the poly and island edges. This can be caused by thin or faculty oxide covering over the islands.



1.5.5 Layer Assignments The below table lists the MOSIS Scalable CMOS design-rule layer assignments for the Caltech Intermediate Form (CIF) and Calma stream format.

1.6. PHYSICAL DEISGN 1.6.1 Basic Concept

Figure 1.26 shows part of the design flow, the physical design steps, for an ASIC (omitting simulation, test, and other logical design steps that have already been covered). Some of the steps in Figure 1.26 might be performed in a different order from that shown. For example, depending on the size of the system, perform system partitioning before any design entry or synthesis. There may be some iteration between the different steps too. First to apply system partitioning to divide a microelectronics system into separate ASICs.

In floorplanning sizes estimate and set the initial relative locations of the various blocks in our ASIC (sometimes we also call this chip planning).



At the same time to allocate space for clock and power w i r i n g a n d decide on the location of the I/O and power pads. Placement defines the location of the logic cells within the flexible blocks and sets aside space for the interconnect to each logic cell. Placement for a gate-array or standard-cell design assigns each logic cell to a position in a row.

Figure 1.26 Part of ASIC Design Flow For an FPGA, placement chooses which o f the fixed logic resources on the chip are used for which logic cells. Floorplanning and placement are closely related and are sometimes combined in a single CAD tool.

Routing makes the connections between logic cells. Routing is a hard problem by itself is normally split into two distinct steps, called global and local routing. Global routing determines where the interconnections between the placed logic cells and blocks will be situated. Only the routes to he used by the interconnections within the wiring areas.

Global routing is sometimes called loose routing for this reason. Local routing joins the logic cells with interconnections. Information on which interconnections areas to use comes from the global router. Only at this stage o f layout d, finally decide on the width, mask layer, and exact location of the interconnections local routing is also known as detailed routing.



1.6.2 CAD Tools sets

In order to develop a CAD tool it is necessary to convert each of the physical do steps to a problem with well-defined goals and objectives. The goals for each physical design step are the things to achieve. The objectives for each step things to meet goals. Some examples of goals and objectives for each of the ASIC physical design steps are as explained below,

System partitioning

• Goal: Partition a system into a number of ASICs. • Objectives: Minimize the number of external connections between the ASICs. Keep each ASIC smaller than a maximum size.

Floorplanning

• Goal: Calculate the sizes of all the blocks and assign them locations. • Objective: Keep the highly connected blocks physically close to each other.

Placement

• Goal: Assign the interconnect areas and the location of all the logic cells within the flexible blocks. • Objectives: Minimize the ASIC area and the interconnect density.

Global routing

• Goal: Determine the location of all the interconnect. • Objective: Minimize the total interconnect area used.

Detailed routing

• Goal: Completely route all the interconnect on the chip. • Objective: Minimize the total interconnect length used. There i s no magic recipe involved in the choice o f the ASIC physical design steps. These steps have been chosen simply because, as tools and techniques have developed historically, these steps proved to be the easiest way to split up the larger problem if ASIC physical design.



1.6.3 Physical Design-The Inverter

1.6.4 Physical Design-The NOR



1.6.5 Physical Design-The NAND

1.7 DESIGN STRATEGIES

The economic viability of an IC is in large part affected by the productivity that can be brought to hear on the design. This in turn depends on the efficiency with which the design may be converted from concept to architecture, to logic and memory, to circuit and hence to a physical layout. A good VLSI design system should provide for consistent in all three description domains (behavioral, structural and physical) and at all relevant levels of abstraction (architecture, RTL, logic, circuit). The means by which this is accomplished may be measured in various terms that differ in importance based on the application. These design parameters may be summarized in terms of

• Performance-speed, power, function, flexibility. • Size of die (hence cost of die). • Time to design (hence cost of engineering and schedule). • Ease of test generation and testability (hence cost of engineering and

schedule). Design is a continuous trade-off to achieve adequate results for all of the above parameters. As such, the tools and methodologies used for a particular chip will be a function of these parameters. Certain end results have to be met (i.e., the chip must conform to performance specifications), but other constraints may be a function of economics (i.e., size of die affecting yield) or even subjectivity (i.e., what one designer finds easy, another might find incomprehensible).



Given that the process of designing a system on silicon is complicated, the role of good VLSI-design aids is to reduce this complexity, increase productivity, and assure the designer of a working product. A good method of simplifying the approach to a design is by the use of constraints and abstractions. By using constraints the tool designer has some hope of automating procedures and taking a lot of the "legwork" out of a design. By using abstractions, the designer can collapse details and arrive at a simpler concept with which to deal. 1.7.1 Structured Design Strategies

The successful implementation of almost any integrated circuit requires an attention to the details of the engineering design process. Over the years a number of structured design techniques have been developed to deal with both complex hardware and software projects. Not surprisingly the techniques have a great deal of commonality. Rigorous application of these techniques can drastically alter the amount of effort that has to be expended on a given project and also in all likehood, the chances of a successful conclusion. Whether under consideration is a small chip designed by a single designer or a large system designed by a team of designers, the basic principles of structured design will improve the prospects of success.

1.7.2 Hierarchy

The use of hierarchy, or "divide and conquer," involves dividing a module into submodules and then repeating this operation on the submodules until the complexity of the submodules is at an appropriately comprehensible level of detail. This parallels the software case where large programs are split into smaller and smaller sections until simple subroutines, with well defined functions and interfaces can be written. A design may be expressed in terms of three domains. A "parallel hierarchy" in each domain to document the design. For instance, an adder may have a subroutine that models the behavior, a gate-connection diagram that specifies the circuit structure, and a piece of layout that specifies the physical nature of the adder. Composing the adder into other structures can proceed in parallel for all three domains, with domain-to-domain comparisons ensuring that the representations are consistent. At a system level, the use of hierarchy allows one to specify single designer projects, at which level the schedule is proportional to the number of available personnel.

VLSI DESIGN MOS TRANSISTOR THEORY


UNIT-2

2.1 NMOS ENHANCEMENT TRANSISTOR

The structure for an n-channel enhancement-type transistor, shown in figure 2.1, consists of a moderately doped p-type silicon substrate into which two heavily doped n+ regions, the source and the drain, are diffused. Between these two regions there is a narrow region of p-type substrate called the channel, which is covered by a thin insulating layer of silicon dioxide (SiO2), called gate oxide. Over this oxide layer is a polycrystalline silicon (polysilicon) electrode, referred to as the gate. Polycrystalline silicon is silicon that is not composed of a single crystal. Since the oxide layer is an insulator, the DC current from the gate to channel is essentially zero. Because of the inherent symmetry of the structure, there is no physical distinction between the drain and source regions. Since SiO2 has relatively low loss and high dielectric is strength, the application of high gate fields is feasible. In operation, a positive voltage is applied between the source and the Drain (Vdy). With zero gate bias (Vs = 0), no current flows from source to drain because they are effectively insulated from each other by the two reversed biased pn junctions shown in figure 21 (indicated by the diode symbols). However, a voltage applied to the gate, which is positive with respect to 'he source and the substrate, produces an electric field E across the substrate, which attracts electrons toward the gate and repels holes. If the gate voltage is sufficiently large, the region under the gate changes from p-type to n-type (due to accumulation of attracted electrons) and provides a conduction path between the source and the drain.

Figure 2.1 Physical structure of an nMOS transistor

Under such a condition, the surface of the underlying p-type silicon is said to be inverted. The term n-channel is applied to the structure. This concept is further illustrated by figure 2.2(a), which shows the initial distribution of mobile positive holes in a p-type silicon substrate of MOS



structure for a voltage, Vgs, much less than a voltage, Vt, which is the threshold voltage. This is termed the accumulation mode. As Vgs is raised above Vt in potential, the holes are repelled causing a depletion region under the gate. Now the structure is in the depletion mode (figure 2.2b). Raising Vgs further above Vt. results in electrons being attracted to the region of the substrate under the gate. A conductive layer of electrons in the p substrate gives rise to the name inversion mode (figure 2.2c).

Figure 2.2 Accumulation, Depletion and Inversion modes in an MOS

structure The difference between a pn junction that exists in a bipolar transistor or diode (or between the source or drain and substrate) and the inversion layer substrate junction is that in the pn junction, the n-type conductivity is brought about by a metallurgical process; that is, the electrons are introduced into the semiconductor by the introduction of donor ions. In an inversion layer substrate junction, the n-type layer is induced by the electric field E applied to the gate. Thus, this junction, instead of being a metallurgical junction, is a field-induced junction.



Electrically, an MOS device therefore acts as a voltage-controlled switch that conducts initially when the gate-to-source voltage, Vgs, is equal to the threshold voltage, Vt. When a voltage V d s is applied between source and drain, with Vgs. = Vt , the horizontal and vertical components of the electrical held due to the source-drain voltage and gate-to-substrate voltage interact, causing conduction to occur along the channel. The horizontal component of the electric field associated with the drain-to-source voltage (i.e., Vds > 0) is responsible for sweeping the electrons in the channel from the source toward the drain. As the voltage from drain to source is increased, the resistive drop along the channel begins to change the shape of the channel characteristic. This behavior is shown in figure 2.3. At the source end of the channel, the full gate voltage is effective in inverting the channel. However, the drain end of the channel, only the difference between the gate and n voltages is effective. When the effective gate voltage (Vgs. – Vt) is greater than the drain voltage, the channel becomes deeper as V g s is increased. This is termed the

Figure 2.3 nMOS device behavior under the influence of different terminal

voltages



"linear," "resistive," "nonsaturated," or "unsaturated" region, where the channel current Ids is a function of both gate and drain voltages. If Vds > Vgs – Vt, then Vgd < Vt (Vgd is the gate to drain voltage), and the channel becomes pinched off- the channel no longer reaches the drain. This is illustrated in figure 2.3(c). However, in this case, conduction is brought about by a drift mechanism of electrons under the influence of the positive drain voltage. As the electrons leave the channel, they are injected into the drain depletion region and are subsequently accelerated toward the drain. The voltage across the pinched-off channel tends to remain fixed at (Vgs-Vt). This condition is the "saturated" state in which the channel current is controlled by the gate voltage and is almost independent of the drain voltage. For the fixed drain-to-source voltage and fixed gate voltage, the factors influence the level of drain current, Ids, flowing between source and drain are:

• the distance between source and drain • the channel width • the threshold voltage V, • the thickness of the gate-insulating oxide layer • the dielectric Constant of the insulator • the carrier (electron or hole) mobility µ

The normal conduction characteristics of an MOS transistor can be cat-egorized as follows:

• "Cut-off' region: where the current flow is essentially zero (accumulation region).

• "Nonsaturated" region: weak inversion region where the drain current is dependent on the gate and the drain voltage (with respect to the substrate).

• "Saturated" region: channel is strongly inverted and the drain current flow is ideally independent of the drain-source voltage (strong inversion region).

An abnormal conduction condition called avalanche breakdown or punch-through can occur if very high voltages are applied to the drain. Under these circumstances, the gate has no control over the drain current. 2.2 PMOS ENHANCEMENT TRANSISTOR

As discussed in previous topic toward nMOS; a reversal of n-type and p-type regions yields a p-channel MOS transistor. This is illustrated by figure 2.4. Application of a negative gate voltage (w.r.t. source) draws holes into the region below the gate, resulting in the channel changing



from n-type to p-type. Thus, similar to nMOS, a conduction path is created between the source and the drain. In this instance, how-ever, conduction results from the movement of holes (versus electrons) in the channel. A negative drain voltage sweeps holes from the source through the channel to the drain.

Figure 2.4 Physical structure of a pMOS transistor

2.3 THRESHOLD VOLTAGE

The threshold voltage, Vt, for an MOS transistor can be defined as the voltage applied between the gate and the source of an MOS device below which he drain-to-source current Ids , effectively drops to zero. The word “effectively” is used because the drain current never really is zero but drops to a very small value that may be deemed insignificant for the current application .In general, the threshold voltage is a function of a number of parameters including the following

• Gate conductor material. • Gate insulation material. • Gate insulator thickness-channel doping. • Impurities at the silicon-insulator interface. • Voltage between the source and the substrate, Vsb

In addition, the absolute value of the threshold voltage decreases with an increase in temperature. This variation is approximately - 4 mV/°C for high substrate doping levels, and - 2 mV/°C for low doping levels.

2 .3 .1 Threshold Voltage Equations

Threshold voltage, Vt, may be expressed as

(2.1)



where Vt_mos., is the ideal threshold voltage of an ideal MOS capacitor and Vfb is what is termed the flat-band voltage. Vt_mos is the threshold where there is no work function difference between the gate and substrate materials. The MOS threshold voltage, Vt-mos, is calculated by considering the MOS capacitor structure that forms the gate of the MOS transistor (see for example or3). The ideal threshold voltage may be expressed as

where is the oxide capacitance and

which is called the bulk charge term. The symbol øb is the bulk potential, a term that accounts for the doping of the substrate. It represents the difference between the Fermi energy level of the doped semiconductor and the Fermi energy level of the intrinsic semiconductor. The intrinsic level is midway between the valence-band edge and the conduction band edge of the semiconductor. In a p type semiconductor the Fermi level is closer to the valence hand, while in an n-type semiconductor it is closer to the conduction band. NA is the density of carriers in the doped semiconductor substrate, and N i is the carrier concentration in intrinsic silicon .Ni is equal to 1.45 x 1010 cm-3 at 3000K. The lowercase k is Boltzmann’s constant (1.380 x 10-23 J/°K).T is the temperature (°K) and q is the electronic charge (1.602 x10-19 Coulomb). The expression kT/q equals 0.02586 Volts at 3000K. The term Cox is the permittivity of silicon (1.06 x 10-12 Farads/cm). The term Co x is the gate-oxide capacitance, which is inversely proportional to the gate-oxide thickness (tox).The threshold voltage, Vt-mos is positive for n-transistors and negative for p-transistors. The flatband voltage,Vfb, is given by Vfb=øms-(Qfc/Cox)

The term Vfb is the flat-band voltage. The term Qfc represents the fixed charge due to surface states that arise due to imperfections in the silicon-oxide interface and doping. The term øms is the work function difference

(2.2)

(2.3)



between the gate material and the silicon substrate (øgate-øsi), which may the calculated for an n+ gate over a p substrate as follows

and T is the temperature (°K). For an n+ poly gate on an n-substrate

From these equations it may be seen that for a given gate and substrate material the threshold voltage may be varied by changing the doping concentration of the substrate (NA), the oxide capacitance (Cox), or the surface state charge (Qfc.). In addition, the temperature variation mentioned above may be seen.

It is often necessary to adjust the native (original) threshold voltage of an MOS device. Two common techniques used for the adjustment of the threshold voltage entail varying the doping concentration at the silicon-insulator interface through ion implantation or using different insulating material for gate. The former approach introduces a small doped region at the oxide/substrate interface that adjusts the flat-band voltage by varying the Qfc term in equation (2.3). In the latter approach for instance, a layer of silicon nitride (Si3N4) is combined with a layer of silicon dioxide resulting in an effective relative permittivity of about 6, which is substantially larger than the dielectric constant SiO2. Consequently, for the same thickness as an insulating layer consisting of only silicon dioxide, the dual dielectric process will be electrically equivalent to a thinner layer of SiO2 leading to a higher Cox value.

2.4 BODY EFFECT

In general, all devices comprising an MOS device are made on a common substrate. As a result, the substrate voltage of all devices is normally equal. (In some analog circuits this may not be true.) However, in ranging the devices to form gating functions it might be necessary to con-nect several devices in series as shown in figure. 2.5. This may result in an increase in source-to-substrate voltage as we proceed vertically along the series chain (Vsb1=0, Vsb2≠0).

(2.4a)

(2.4b)



Under normal conditions that is, when Vgs>Vt the depletion-layer width remains constant and charge carriers are pulled into the channel from the source. However, as the substrate bias Vsb (Vsource-Vsubstrate) is increased, the width of the channel substrate depletion layer also increases, resulting in an increase in the density of the trapped carriers in the depletion laver. For charge neutrality to hold, the channel charge must decrease. The resultant effect is that the substrate voltage, Vsb, adds to the channel-substrate junction potential. This increases the gate-channel voltage drop. The overall effect is an increase in the threshold voltage Vt (Vt2-Vt1)

Figure 2.5 The effect of substrate bias on series-connected n-transistors

2.5 MOS DEVICE DESIGN EQUATIONS

2.5.1 Basic DC Equations

The MOS transistors have three regions of operation:

• Cutoff or subthreshold region. • Nonsaturation or linear region. • Saturation region.

The ideal (first order, Shockley) equations describing the behavior of an nMOS device in the three regions are: The cutoff region: Ids=0 Vgs≤ Vt (2.5a) The nonsaturation, linear or triode region:

The saturation region:



Ids=(β/2)(Vgs-Vt)2 (2.5c)

where Ids is the drain-to-source current, Vgs is the gate-to-source voltage, Vt is the device threshold, and β is the MOS transistor gain factor. The last factor is depen-dent on both the process parameters and the device geometry, and is given by β=(µє/tox)(W/L) (2.6) where µ is the effective surface mobility of the carriers in the channel, є is the permittivity of the gate insulator, tox is the thickness of the gate insulator, W is the width of the channel, and L is the length of the channel. The gain factor β thus consists of a process dependent factor µє/tox, which contains all the process terms that account for such factors as doping density and gate-oxide thickness and a geometry dependent term (W/L), which depends on the actual layout dimensions of the device. The process dependent factor is sometimes written as µCox, where Cox= є / t o x is the gate oxide capacitance. The geometric terms in Eq. (2.6) are illustrated in figure 2.6 in relation to the physical MOS structure.

Figure 2.6 Geometric terms in the MOS device equation

Figure 2.7 VI characteristics for n- and p-transistors



The voltage-current characteristics of the n- and p-transistors in the non-saturated and saturated regions are represented in figure 2.7 (with the SPICE circuit for obtaining these characteristics for an n-transistor). Note that we use the absolute value of the voltages concerned to plot the characteristics of the p- and n-transistors on the same axes. The boundary between the linear and saturation regions corresponds to the condition | Vds| = | Vgs - Vt | and appears as a dashed line in figure 2.7. The drain voltage at which the device becomes saturated is called Vdsat, or the drain saturation voltage. In the above equations that is equal to Vgs-Vt

2.5.2 Second Order Effects

Equation 2.5 represents the simplest view of the MOS transistor DC voltage current equations. There have been many research papers published on more detailed and accurate models that have been created to fill a variety of requirements, such as accuracy, computational efficiency, and the conservation of charge. The circuit simulation program SPICE and its commercial and proprietary derivations generally use a parameter called LEVEL to specify which model equation, are LEVEL1 models build on those defined in Eq. (2.5) and include some important second order effects. LEVEL 2 models calculate the currents based on physics. LEVEL 3 is a semiempirical approach that relies on parameters selected on the basis of matching the equations to real circuits. The MOS device equations in terms of the LEVEL 1 parameters used in SPICE will be covered here.

First the term µє/tox(µCox) is defined as the process gain factor. In SPICE this is referred to as KP. Depending on the vintage of the process and the type of transistor, KP may vary from 10-100 pA/V2. In addition, it is not unusual to expect a variation of 10%-20% in KP within a given process as a result of variations in starting materials and variation in SiO2 growth. 2.5.2.1 Threshold Voltage-Body Effect

The threshold voltage Vt is not constant with respect to the voltage difference between the substrate and the source of the MOS transistor. This is known as the substrate-bias effect or body effect. The expression for the threshold voltage may be modified to incorporate Vsb, the difference between the source and the substrate.

(2.7)



where Vsb is the substrate bias, Vto is the threshold voltage for Vsb=0 , and γ is the constant that describes the substrate bias effect. The term Øb is defined in Eq 2.2. Typical values for γ lie in the range of 0.4 to 1.2. It may be expressed as

(2.8) in which q is the charge on an electron, єox, is the dielectric constant of the silicon dioxide, єsi; is the dielectric constant of the silicon substrate, and NA is the doping concentration density of the substrate. The term γ is the SPICE parameter called GAMMA. Vto is the parameter VTO, NA is he parameter NSUB, and øs=2øb is PHI, the surface potential at the onset of strong inversion. Thus the threshold shifts by approximately half a volt with the source at 2.5 volts for these process parameters. The type of CMOS process can have a large impact on this parameter for both n- and p-transistors. The increase in threshold voltage leads to lower device currents, which in turn leads to slower circuits. 2.5.2.2 Subthreshold Region

The cutoff region described by Eq. (2.5a) is also referred to as the subthreshold region where Ids increases exponentially with Vds and Vgs. Although the value of Ids is very small (Ids=0), the finite value of Ids may be used to advantage to construct very low power circuits or it may adversely affect circuits such as dynamic-charge storage nodes. As an approximation, Level 1 SPICE models set the subthreshold current to 0.

2.5.2.3 Channel-length Modulation

The simplified equations that describe the behavior of an MOS device assume that the carrier mobility is constant, and do not take into account the variations in channel length due to the changes in drain-to-source voltage, Vds . For long channel lengths, the influence of channel variation is of little consequence. However, as devices are scaled down, this variation should be taken into account. When an MOS device is in saturation, the effective channel length actually is decreased such that

Leff=L-Lshort (2.9)

where

The reduction in channel length increases the (W/L) ratio, thereby increasing β as the drain voltage increases. Thus rather than appearing as a constant current



source with infinite output impedance, the MOS device has a unite output impedance. An approximation that takes this behavior into account is represented by the following equation:

(2.10)

Where k is the process gain factor µє/tox and λ is an empirical channel-length modulation factor having a value in the range 0.02V-1 to 0.005V-1

2.5.2.4 Mobility Variation

The mobility µ, describes the ease with which carriers drift in the substrate material. It is defined by

µ=average carrier drift velocity (V)/ Electric Field (E) (2.11)

If the velocity, V, is given in cm/sec, and the electric field, E, in V/cm, the mobility has the dimensions cm2/V-sec. The mobility may vary in a number of ways. Primarily, mobility varies according to the type of charge carrier. Electrons (negative-charge carriers) in silicon have a much higher mobility than holes (positive-charge carriers), resulting in n-devices having higher current-producing capability than the corresponding p-devices. Mobility decreases with increasing doping-concentration and increasing temperature. The temperature variation becomes less pronounced as the doping density increases. In SPICE is specified by the parameter UO. 2.6 MOS MODELS

In the previous section the ideal equations that describe the behavior of MOS transistors. While these incorporate some non ideal effects (channel-length modulation, threshold-voltage variation), they may not accurately model a specific device in a particular process. That is especially true for devices that have very small dimensions (gate lengths, gate widths, oxide thicknesses) as the modeling process becomes increasingly 3D in nature.

Parameter nMOS pMOS Units Description

VTO 0.7 0.7 Volt Threshold voltage

KP 8x10-5 2.5x10-

5 A/V2 Transconductance

coefficient

GAMMA 0.4 0.5 V0.5 Bulk threshold parameter



PHI 0.37 0.36 volt Surface potential at strong inversion

LAMBDA 0.01 0.01 Volt-1 Channel length modulation parameter

LD 0.1 x 10-6

0.1x10-

6 Meter Lateral diffusion

TOX 2x10-8 2x10-8 Meter Oxide thickness

NSUB 2x1016 4x1016 1/cm3 Substrate doping density

Table 2.1 SPICE DC Parameters Researchers have developed and refined a wide range of MOS models in an effort to predict more accurately the performance of MOS devices before they are fabricated for varying design scenarios. For instance, one might predict DC currents very accurately from raw process parameters, thus helping predict the behavior of an as yet untested device. However, because of the complexity of the model, it might not be appropriate for a fast execution time model that might be needed for digital simulation purposes. In that case, a model based on parameters measured from an actual process might be appropriate. Depending on the particular circuit level simulator that may be available, a wide variety of MOS simulation models may be used. For instance in one commercial circuit simulators there are over 10 different MOS models. Many semiconductor vendors expend a great deal of effort to model the device they manufacture. Many times these efforts are aimed at internal circuit simulators and proprietary models. Most CMOS digital foundry operation have been standardized on the LEVEL 3 models in SPICE as the level of circuit modeling that is required for CMOS digital system design. Table 2.1 is a summary of the main SPICE DC parameters that are used in Levels 1, 2, and 3 with representative values for a 1µ n-well CMOS process.



2.7 SMALL SIGNAL AC CHARACTERISTICS

The MOS transistor can be represented by the simplified (Vsb = 0) small signal equivalent model shown in figure 2.8 when biased appropriately. Here the MOS transistor is modeled as a voltage-controlled current source (gm), an output conductance (gds), and the interelectrode capacitances. These values may be used, for instance, to calculate voltage amplification factors (gain) or bandwidth characteristics when considered along with other circuit elements.

Figure 2.8 Small signal model for an MOS transistor

The output conductance (gds) in the linear region can be obtained by dif-ferentiating Eq. (2.5b) with respect to Vds, which results in an output drain-source conductance of

gds=β((Vgs- Vt)-2 Vds) ≈β(Vgs-Vt) (2.12)

Note that consistent with Eq. (2.5b), Vds must be small compared to Vgs for the MOS device to be in a linear operating regime. On rearrangement, the channel resistance Rc is approximated by

Rc(linear) =1/ β(Vgs-Vt) (2.13)

which indicates that it is controlled by the gate-to-source voltage. The relation defined by Eq. (2.13) is valid for gate to source voltages that maintain constant mobility in the channel. In contrast, in saturation [i.e., Vds (Vgs - Vt)], the MOS device behaves like a current source, the current being almost independent of Vds. This may be verified from Eq. (2.5c) since

(2.14)



In practice, however, due to channel shortening (Eq. 2.9) and other effects, the drain-current characteristics have some slope. This slope defines the gds of the transistor. The output conductance can be decreased by lengthening the channel (i.e., L).

The transconductance gm expresses the relationship between output current Ids and the input voltage Vgs and is defined by

(2.15)

It is used to measure the gain of an MOS device. In the linear region gm given by gm(linear) =βVds (2.16) and in the saturation region by gm(sat)= β(Vgs-Vt) (2.17) Since transconductance must have a positive value, the absolute value is used for voltages applied to p-type devices.

2.8 THE COMPLEMENTARY CMOS INVERTER – DC CHARACTERISTICS

A complementary CMOS inverter is realized by the series connection of a p and an n-device, as shown in figure 2.11. In order to derive the DC-transfer characteristics for the inverter (output voltage, Vout, as a function of the inverter Vin), from the Table 2.1, which outlines various regions of operation for the n- and p-transistors. In this table, Vtn is the threshold voltage of the n channel device, and Vtp, is the threshold voltage of the p-channel device. The objective is to find the variation in output voltage (Vout) for changes in the input voltage (Vin).

Figure 2.9 A CMOS inverter (with substrate connections)



The graphical representation of the simple algebraic equations described by Eq. (2.5) for the two inverter transistors shown in figure 2.12(a). The absolute value of the p-transistor drain current Ids inverts this characteristic. This allows the VI characteristics for the p-device to he reflected about the x-axis (figure. 2.12b). This step is followed by taking the absolute value of the p-device, Vds, and superimposing the two characteristics yielding the resultant curves shown in Fig. 2.12(c).

Table 2.2 Relations Between Voltages for the Three Regions Of Operation Of

A CMOS Inverter The input/output transfer curve may now be determined by the points of common Vgs intersection in figure 2.10(c). Thus, solving for Vinn = Vi n p and Idsn=Idsp gives the desired transfer characteristics of a CMOS inverter as illustrated in figure 2.13. The switching point is typically designed to be 50 percent of the magnitude of the supply voltage: = VDD/2. During transition, both transistors in the CMOS inverter are momentarily "ON," resulting in a short pulse of current drawn from the power supply. This is shown by the dotted line in figure 2.11

The operation of the CMOS inverter can be divided into five regions (figure 2.13). The behavior of n- and p-devices in each of the regions may be found by using Table 2.2.

Region A. This region is defined by 0<Vin<Vtn in which the n-device is cut off (Idsn = 0), and the p-device is in the linear region. Since Idsp=- Idsp the drain-to-source current Idsp

for the p-device is also zero. But for Vdsp = Vout - VDD, with Vdsp = 0, the output voltage is Vout = VDD (2.18)



Figure 2.10 Graphical derivation of CMOS inverter characteristic

Region B. This region is characterized by Vtn< Vin < VDD/2 in which the p-device is in its nonsaturated region (Vds ≠ 0) while the n-device is in saturation. The equivalent circuit for the inverter in this region can be represented by a resistor for the p-transistor and a current source for the n-transistor as shown in figure 2.12(a).

The saturation current Idsn for the n-device is obtained by setting Vgs=Vin. This result in

Where βn=(µnε/tox)(Wn/Ln)

and Vtn=threshold voltage of n-device µn =mobility of electrons Wn=channel width of n-device Ln=channel length of n-device

The current for the p-device can be obtained by noting that Vgs=Vin-VDD and

Vds=Vout-VDD and therefore



Idsp=-βp (Vin-VDD-Vtp)(Vout-VDD)- (Vout-VDD)2/2 (2.19)

Where βp=(µpε/tox)(Wp/Lp)

and Vtp=threshold voltage of p-device µp =mobility of electrons Wp=channel width of p-device Lp=channel length of p-device

Substituting Idsp=-Idsn The output voltage Vout can be expressed as

(2.20)

Figure 2.11 CMOS inverter DC transfer characteristic and operating

regions

Figure 2.12 Equivalent circuits for operating regions of a CMOS

inverter



Region C. In this region both the n and p device are in saturation. This is represented by schematic in figure 2.12(b) which shows two current sources in series. The saturation currents for the two devices are given by

Idsp=-(βp/2)(Vin-VDD-Vtp)2 Idsn=-(βn/2)(Vin-Vtn)2

With Idsp= -Idsn This yields

(2.21) By setting βn=βp and Vtn=-Vtp We obtain

Vin=VDD/2 (2.22) Which implies that region C exists only for one value of Vin. The possible values of Vout in this region can be deduced as follows

n-channel: Vin-Vout<Vtn Vout>Vin-Vtn p-channel : Vin-Vout >Vtp Vout<Vin-Vtp Combining the two inequalities results in Vin-Vtn< Vout < Vin-Vtp (2.23)

This indicates that with Vin =VDD/2, Vout varies within the range shown. An MOS device in saturation behaves like an ideal current source with drain-to-source current being independent of Vds In reality, as Vds increases, Ids also increases slightly thus region C has a finite slope. The significant factor to be noted is that in region C we have two current sources in series, which is an “unstable” condition. Thus a small input voltage has a large effect at the output. This makes the output transition very steep, which contrasts with the equivalent nMOS inverter characteristic. The relation defined by equation is particularly useful since it provides the basis for defining the gate threshold Vinv, which corresponds to the state where Vout=Vin. This region also defines the “gain” of the CMOS inverter when used as a small signal amplifier.



Region D. This region is described by VDD/2 < Vin ≤ VDD+Vtp. The p-device is in saturation while the n-device is operating in its nonsaturated region. This condition is represented by equivalent circuit shown in figure. The two currents may be written as

Idsp= (-βp/2)(Vin-VDD-Vtp)2

and Idsn=βn (Vin-Vtn)Vout-(Vout

2/2) with Idsp=-Idsn

The output voltage becomes (2.24)

Region E. This region is defined by input condition Vin ≥VDD-Vtp , in which the p-device is cut off (Idsp=0), and the n-device is in the linear mode. Here , Vgsp=Vin-VDD, which is more positive than Vtp.

Region

Condition p-device n-device Output

A 0≤ Vin <Vtn Nonsaturated

Cutoff Vout=VD

D

B Vtn≤Vin<VDD/2

Nonsaturated

Saturated equation

C Vin=VDD/2 Saturated Saturated Vout≠f(Vin)

D VDD/2<Vin≤VDD-|Vtp|

Saturated Nonsaturated

Equation

E Vin>VDD-|Vtp| Cutoff Nonsaturated

Vout=VS

S

Table2.3 summary of CMOS inverter operation



The output in this region is Vout=0 (2.25)

From the transfer curve of figure 2.13 it may be seen that the transition between the two states is very steep. This characteristic is very desirable because the noise immunity is maximized. For convenience, the characteristics associated with the five regions are summarized in Table 2.3.

2.8.1 βn/βp ratio

In order to explore the variations of the transfer characteristics as a function of βn/βp, the transfer curve for several values of βn/βp are plotted in figure 2.15. The gate-threshold voltage Vinv where Vin=Vout is dependent on βn/βp . Thus for a given process, if we want to change βn/βp we need to change the channel dimensions, i.e., channel-length L and channel-width W. from figure it can be seen that as the ratio βn/βp is decreased the transition region shifts from left to right ; however, the output voltage transition remains sharp. For the CMOS inverter a ratio of

βn/βp=1 (2.26)

may be desirable since it allows a capacitive load to charge and discharge in equal times by providing equal current-source and sink capabilities. For inverter transfer curve is also plotted for Wn/Wp .This shows a relative shift to the left compared with β ratioed case because the p-device has inherently lower gain.

Figure 2.13 Influence of βn/βp on inverter DC transfer characteristic

Temperature also has an effect on the transfer characteristic of an inverter. As the temperature of an MOS device is increased, the effective carrier mobility µ decreases. This results in a decrease in β, which is related to temperature T by

β α T-1.5 (2.27)

Therefore



Ids α T-1.5 (2.28)

Since the voltage characteristics depend on the ratio βn/βp, and the mobility of both holes and electrons are similarly affected, this ratio is independent of temperature to a good approximation. Both Vtn and Vtp decrease slightly as temperature increases, and the extent of region A is reduced while the extent of region E increases. Thus the overall transfer characteristics of figure shift to the left as temperature increases. If the temperature rises by 500C, the threshold drop by 200mV each. This would cause a 0.2V shift in the input threshold of the inverter.

2.8.2 Noise Margin

Noise margin is a parameter closely related to the input-output voltage characteristics. This parameter allows us to determine the allowable noise voltage on the input of a gate so that the output will not be affected. The specification most commonly used to specify noise margin is in terms of two parameters – the LOW noise margin, NML and the HIGH noise margin NMH . NML is defined as the difference in magnitude between the maximum LOW output voltage of the driving gate and the maximum input LOW voltage recognized by the driven gate. Thus

NML =| VILmax - VOLmin| (2.29)

The value of NMH is the difference in magnitude between the minimum HIGH output voltage of the driving gate and the minimum input HIGH voltage recognized by receiving gate. Thus

NMH =| V OHmin - V IHmin| (2.30)

where

V IHmin =minimum HIGH input voltage VILmax =maximum LOW input voltage VOHmin =minimum HIGH output voltage VOLmax =maximum LOW output voltage

Figure 2.14 Noise margin definitions



Generally it is desirable to have VIH =VIL and for this to be a value that is midway in the “logic swing”, VOL to VOH . This implies that the transfer characteristics should switch abruptly, that is, there should be high gain in the transition region. For the purpose of calculating noise margins, the transfer characteristic of a typical inverter and the definition of voltage levels VIL, VOL, VIH , VOH are shown in figure 2.15. To determine VIL note that the inverter is in region B of operation, where the p-device is in its linear region while the n-device is in saturation. The VIL is found by using the unity gain point at the VOL end of the characteristic. For the inverter shown the NML is 2.3 volts while the NMH is 1.7 volts.

Note that if either NML or NMH for a gate are reduced, then the gate may be susceptible to switching noise that may be present on the inputs. Apart from considering a single gate, one must consider the net effect of noise sources and noise margins on cascaded gates in assessing the overall noise immunity of a particular system. This is the reason to keep track of noise margins.

Figure 2.15 CMOS inverter noise margins

2.9 THE TRANSMISSION GATE

The transistor connection for a complementary switch or transmission gate is reviewed in figure 2.18. It consists of an n-channel transistor and a p-channel transistor with separate gate connections and common source and drain con-nections. The control signal is applied to the gate of the n-device, and its complement is applied to the gate of the p-device. The operation of the trans-mission gate can be best explained by considering the characteristics of both the n-device and p-device as pass transistors individually. We will address this by treating the charging and discharging of a capacitor via a transmission gate.



Figure 2.16 Transistor connection for CMOS transmission gate

nMOS Pass Transistor:

Referring to figure2.17 (a), the load capacitor Cload is initially discharged (i.e., Vout = VSS). With S = 0 (Vss) (i.e.,Vgs= 0 volts), Ids = 0, then Vout = VSS irrespective of the state of the input Vin. When S = 1 (VDD), and Vin = 1, the pass transistor begins to conduct and charges the load capacitor toward VDD, i.e., initially Vgs, = VDD. Since initially Vin is at a higher potential than Vout the current flows through the device from left to right . As the output voltage approaches VDD-Vin, the n-device begins to turn off. Load capacitor, Cload, will remain charged when S is changed back to 0. Therefore the output voltage Vout, remains at VDD-Vtn(Vdd). V (tnVdd) is the n-transistor body affected threshold of logic one is degraded as it passes through the gate. With Vin=0,S=1, and Vout=VDD-VtnVdd, the pass transistor begins to conduct and discharge the load capacitor toward VSS, i.e., Vgs=VDD. Since initially Vin is at a lower potential than Vout , the current flows through the device from right to left. As the output voltage approaches VSS, the n device current diminishes. Because Vout falls to VSS, the transmission of a logic zero is not degraded.

pMOS Pass Transistor:

Once again a similar approach can be taken in analyzing the operation of a pMOS pass transistor as shown in figure 2.17(b). With -S =1 (S = 0), Vin = VDD, and Vout = VSS, the load capacitor Cload remains unchanged. When -S = 0 (S = 1), current begins to flow and charges the load capacitor toward VDD. However, when Vin= VSS and Vout= VDD,the load capacitor discharges through the p-device until Vout=Vtp(Vss), at which point the transistor ceases conducting. Thus transmission of a logic zero is somewhat degraded through the p-device. The resultant behavior of the n-device and p-device are shown in Table 2.4. By combining the two characteristics we can construct a transmission gate that can transmit both a logic one and a logic zero without degradation. As can be deduced from the discussion so far, the operation of the transmission gate requires both the true and the complement version of the control signal. DEVICE TRANSMISSION OF ‘1’ TRANSMISSION OF ‘0’

n poor good p good poor

Table2.4 Transmission Gate Characteristics



The overall behavior can be expressed as: n-device =off p-device =off S=0(-S=1); Vin=VSS, Vout=Z Vin=VDD, Vout=Z Where Z refers to a high impedance state and n-device =on p-device =on S=1(-S=0); Vin=VSS, Vout=VSS Vin=VDD, Vout=VDD

Vin

a

Figure2.17 nMOS and PMOS transistor operation in transmission gate

The transmission gate is a fundamental and ubiquitous component in MOS logic. It finds use as a multiplexing element, a logic structure, a latch element, and an analog switch. The transmission gate acts as a voltage controlled resistor connecting the input and the output. Figure 2.18(a) shows a typical circuit configuration for a transmission gate in which the output is connected to a capacitor and the input to an inverter. The control input is shown turning the transmission gate on. That is, the gate of the n-channel transmission gate switch is changing from 0 1 and the gate of the p-channel is changing from 1 0. First consider the case where the control input changes rapidly, the inverter input is low (VS S), the inverter output is high (VDD), and the capacitor on the transmission gate output is discharged (VSS). The currents that flow in this situation may be modeled by circuit shown in figure 2.18(b) in which the currents in the pass



transistors are monitored. In reality, the capacitor charge would be exponential, but a linear ramp serves to show what happens to the pass transistor currents. As Vout rises, the p-transistor current follows a constant Vgs of -5 volts. That is, it starts out in saturation and transitions to the nonsaturated case when |Vgsp-Vtp|<|Vdsp|. The n-transistor is always in the saturated region as Vdsn=Vgsn and Vgsn-Vtn < Vdsn. When Vout reaches a Vtn delow VDD, the n-transistor turns off. Thus there are three regions of operation:

Figure 2.18 Transmission gate output characteristic for control input

changing

Region A. n saturated, p saturated (Vout < |Vtp|) Region B. n saturated, p nonsaturated (|Vtp|<Vout<VDD-Vtn) Region C. n off, p saturated (VDD-Vin<Vout)

In region A, approximately the p-current as a constant current while the n-current varies quadradically with Vout. Hence the total current is roughly linear with Vin. In region B both currents yield a sum that varies almost linearly with Vout. Finally in region C the p-current varies linearly with Vout. Thus the transmission gate acts as a resistor, with contributions to its resistance from both n and p transistors. This can be seen in figure (Idn5+Idp5). Similar simulations may be carried out for Vin=VSS and Vout=VDD VSS.

Another operation mode that the transmission gate encounters in lightly loaded circuits is where the output closely follows the input, such as shown in figure 2.19(a). Figure 2.19(b) shows a model of this while figure 2.19 (c) shows the SPICE circuit used to model this condition including current monitoring voltage sources. Figure 2.19(d) shows the n and p pass transistor currents for Vout-Vin=-0.1 volts. It can be seen that again there are three regions of operation:

Region A. n nonsaturated, p off Region B. n nonsaturated, p nonsaturated Region C. n off, p nonsaturated



The total current decreases in magnitude as Vin increases until Vin=|Wtp(body-

affected)|. Here the p-transistor turns on and in this case slows the decrease of current. When Vin>VDD-Vtn(body-affected), the current starts to increase in magnitude as the p current continues to increase while the n transistor is off. In this simulation the p and n gains were matched. For the region |Vtp|<Vin<VDD-Vtn, the transmission gate will have a roughly constant resistance. The effect of having only one polarity transistor in the transmission gate is also seen. If only an n-transistor is used, the output will rise to an n threshold below VDD as current stops flowing at this point. Similarly, with a single p-transistor, the output would fall to a p threshold above VSS, as current stops flowing in the p-transistor at this point. Note also that as either the p or n current approaches zero, the speed of any circuit would be prejudiced. If the surrounding circuitry can deal with these imperfect high and low values, then single polarity transmission gates may be used. Figure 2.20, shows a plot of the transmission gate "on" resistance for the test circuit shown in figure 2.19(c).

Figure 2.19 Transmission gate output characteristic for switched

input changing

Figure 2.20 Resistance of a transmission gate for conditions in Figure 2.19



2.10 THE TRISTATE INVERTER

By cascade a transmission gate with an inverter the tristate inverter shown in figure 2.21(a) is constructed. when C=0 and –C=1 , the output of the inverter is in a tristate condition. When C=1 and and -C=0 the output Z is equal to the complement of A. The connection between the n and p transistors may be omitted and the operation remains substantially the same. Figure 2.21(c) shows the schematic icon that represents the tristate inverter. For the same n and p devices, this inverter is approximately half the speed of the inverter shown in figure 2.9.

Figure 2.21 Tristate inverter

VLSI DESIGN SPECIFICATION USING VERILOG HDL


UNIT 3

HISTORY OF VERILOG

Verilog was started in the year 1984 by Gateway Design Automation Inc as a proprietary hardware modeling language. It is rumored that the original language was designed by taking features from the most popular HDL language of the time, called HiLo, as well as from traditional computer languages such as C. At that time, Verilog was not standardized and the language modified itself in almost all the revisions that came out within 1984 to 1990.

Verilog simulator first used in 1985 and extended substantially through 1987. The implementation of Verilog simulator sold by Gateway. The first major extension of Verilog is Verilog-XL, which added a few features and implemented the infamous "XL algorithm" which is a very efficient method for doing gate-level simulation.

Later 1990, Cadence Design System, whose primary product at that time included thin film process simulator, decided to acquire Gateway Automation System, along with other Gateway products., Cadence now become the owner of the Verilog language, and continued to market Verilog as both a language and a simulator. At the same time, Synopsys was marketing the top-down design methodology, using Verilog. This was a powerful combination.

In 1990, Cadence organized the Open Verilog International (OVI), and in 1991 gave it the documentation for the Verilog Hardware Description Language. This was the event which "opened" the language.

3.1 BASIC CONCEPTS

3.1.1 Hardware Description Language

Two things distinguish an HDL from a linear language like “C”: Concurrency:

• The ability to do several things simultaneously i.e. different code-blocks can run concurrently.

Timing: • Ability to represent the passing of time and sequence events accordingly

3.1.2 VERILOG Introduction

• Verilog HDL is a Hardware Description Language (HDL). • A Hardware Description Language is a language used to describe a digital

system; one may describe a digital system at several levels. • An HDL might describe the layout of the wires, resistors and transistors

on an Integrated Circuit (IC) chip, i.e., the switch level.



• It might describe the logical gates and flip flops in a digital system, i.e., the gate level.

• An even higher level describes the registers and the transfers of vectors of information between registers. This is called the Register Transfer Level (RTL).

• Verilog supports all of these levels. • A powerful feature of the Verilog HDL is that you can use the same

language for describing, testing and debugging your system.

3.1.3 VERILOG Features • Strong Background:

Supported by OVI, and standardized in 1995 as IEEE std 1364

• Industrial support: Fast simulation and effective synthesis (85% were used in ASIC foundries by EE TIMES)

• Universal : Allows entire process in one design environment (including analysis and verification)

• Extensibility : Verilog PLI that allows for extension of Verilog capabilities 3.1.4 Design Flow

The typical design flow is shown in figure 3.1,

Design Specification

• Specifications are written first-Requirement/needs about the project • Describe the functionality overall architecture of the digital circuit to be

designed. • Specification: Word processor like Word, Kwriter, AbiWord and for drawing

waveform use tools like wave former or test bencher or Word. RTL Description

• Conversation of Specification in coding format using CAD Tools.

Coding Styles:

• Gate Level Modeling • Data Flow Modeling • Behavioral Modeling • RTL Coding Editor : Vim, Emacs, conTEXT, HDL TurboWriter



Figure 3.1 VLSI Design Flow I0 I1 I2 Out I3 S0 S1

Figure 3.2 Black Box View of 4:1 MUX



Functional Verification &Testing

• Comparing the coding with the specifications. • Testing the Process of coding with corresponding inputs and outputs. • If testing fails – once again check the RTL Description. • Simulation: Modelsim, VCS, Verilog-XL,Xilinx.

Figure 3.3 Simulation Output View of 4:1 MUX Using Modelsim Wave form

Viewer

Logic Synthesis

• Conversation of RTL description into Gate level -Net list form. • Description of the circuit in terms of gates and connections. • Synthesis: Design Compiler, FPGA Compiler, Synplify Pro, Leonardo

Spectrum, Altera and Xilinx.

Figure 3.4 Synthesis of 4:1 MUX Using Leonardo Spectrum Logical Verification and Testing

• Functional Checking of HDL coding by simulation and synthesis. If fails –check the RTL description.



Floor Planning Automatic Place and Route

• Creation of Layout with the corresponding gate level Net list. • Arrange the blocks of the net list on the chip • Place & Route: For FPGA use FPGA' vendors P&R tool. ASIC tools require expensive P&R tools like Apollo. Students can use LASI, Magic

Physical Layout

• Physical design is the process of transforming a circuit description into the physical layout, which describes the position of cells and routes for the interconnections between them.

Layout Verification

• Verifying the physical layout structure. • If any modification –once again check Floor Planning Automatic Place and

Route and RTL Description.

Implementation

• Final stage in the design process. • Implementation of coding and RTL in the form of IC.

Figure 3.5 Layout view of a system

3.1.5 Design Hierarchies

3.1.5.1Bottom up Design

The Traditional method of electronic design is bottom up. Each design is performed at the gate level using the standard gates .With increasing complexity of new designs this approach is nearly impossible to maintain. It gives way to new structural, hierarchical design methods. Without these new design practices it would be impossible to handle the new complexity



Figure 3.6 Bottom Up Design

3.1.5.2 Top-Down Design

A real top down design allows early testing, easy change of different technologies, a structured system design and offers many other advantages. But it is very difficult to follow a pure top-down design. Due to this fact most designs are mix of both the methods. Implementing some key elements of both design styles

Figure 3.7 Top-Down Design 3.1.6 Lexical Conventions

The basic lexical conventions used by Verilog HDL are similar to those in the C programming language. Verilog contains a stream of tokens. Tokens can be comments, delimiters, numbers, strings, identifiers, and keywords. Verilog HDL is a case-sensitive language. All keywords are in lower case.



3.1.6.1 Whitespace

White space can contain the characters for blanks, tabs, newlines, and form feeds. These characters are ignored except when they serve to separate other tokens. However, blanks and tabs are significant in strings.

White space characters are: • Blank spaces (\b) • Tabs(\t) • Carriage returns(\r) • New-line (\n) • Form-feeds (\a)

Example

3.1.6.2 Comments

Comments can be inserted in the code for readability and documentation. There are two forms to introduce comments.

• Single line comments begin with the token // and end with a carriage return

• Multi line comments begin with the token /* and end with the token */ Example



3.1.6.3 Identifiers and Keywords

Identifiers are names used to give an object, such as a register or a function or a module, a name so that it can be referenced from other places in a description. Keywords are reserved to define the language constructs.

• Identifiers must begin with an alphabetic character or the underscore character (a-z A-Z _ )

• Identifiers may contain alphabetic characters, numeric characters, the underscore, and the dollar sign (a-z A-Z 0-9 _ $ )

• Identifiers can be up to 1024 characters long. • Keywords are in lowercase.

Examples of legal identifiers

• data_input mu • clk_input my$clk • i386

Examples of keywords

• always • begin • end

3.1.6.4 Escaped Identifiers

Verilog HDL allows any character to be used in an identifier by escaping the identifier. Escaped identifiers provide a means of including any of the printable ASCII characters in an identifier (the decimal values 33 through 126, or 21 through 7E in hexadecimal).

• Escaped identifiers begin with the back slash ( \ ) • Entire identifier is escaped by the back slash. • Escaped identifier is terminated by white space (Characters such as

commas, parentheses, and semicolons become part of the escaped identifier unless preceded by a white space)

• Terminate escaped identifiers with white space, otherwise characters that should follow the identifier are considered as part of it.

3.1.7 Numbers in Verilog

Numbers in Verilog can be specified constant numbers in decimal, hexadecimal, octal, or binary format. Negative numbers are represented in 2's complement form. When used in a number, the question mark (?) character is the Verilog alternative for the z character. The underscore character (_) is legal anywhere in a number except as the first character, where it is ignored.



3.1.7.1 Integer Numbers

Verilog HDL allows integer numbers can be specified as

• Sized or unsized numbers (Unsized size is 32 bits) • In a radix of binary, octal, decimal, or hexadecimal • Radix and hex digits (a,b,c,d,e,f) are case insensitive • Spaces are allowed between the size, radix and value

Integer numbers are represented as <size>’<base format> <number>

<size> is wrriten only in decimal and specifies the number of bits in the number. Legal base formats are decimal (‘d or ‘D), hexadecimal(‘h or ‘H), binary (‘b or ‘B) and octal (‘o or ‘O). the number is specified as consecutive digits from 0,1,2,3,4,5,6,7,8,9,a,b,c,d,e,f. only a subset of these digits is legal for a particular base. Uppercase letters are legal for number specification.

4’b1111 –this is a 4-bit binary number 12’habc – this is a 12-bit hexadecimal number 16’d255 – this is a 16-bit decimal number 8’o44-this is 8 bit octal number

3.1.7.2 Real Numbers

• Verilog supports real constants and variables • Verilog converts real numbers to integers by rounding • Real Numbers can not contain 'Z' and 'X' • Real numbers may be specified in either decimal or scientific notation • < value >.< value > • < mantissa >E< exponent > • Real numbers are rounded off to the nearest integer when assigning to an

integer.

Example

Real Number

Decimal notation

1.2 1.2 0.6 0.6 3.5E6 3,500000.0 3.1.7.3 Signed and Unsigned Numbers

Verilog supports both types of numbers, but with certain restrictions. Like in C language Verilog don't have int and unint types to say if a number is signed



integer or unsigned integer. Any number that does not have negative sign prefix is a positive number. Or indirect way would be "Unsigned".

Negative numbers can be specified by putting a minus sign before the size for a constant number, thus they become signed numbers. Verilog internally represents negative numbers in 2's complement format. An optional signed specifier can be added for signed arithmetic.

Example

Number Description

32'hDEAD_BEEF Unsigned or signed positive number

-14'h1234 Signed negative number

Example

module signed_number;

reg [31:0] a; initial begin a = 14'h1234; $display ("Current Value of a = ‰h", a); a = -14'h1234; $display ("Current Value of a = ‰h", a); a = 32'hDEAD_BEEF; $display ("Current Value of a = ‰h", a); a = -32'hDEAD_BEEF; $display ("Current Value of a = ‰h", a); #10 $finish; end

endmodule

3.1.8 Strings

A string is a sequence of characters that are enclosed by double quotes. The restriction on a string is that it must be contained on a single line, that is, without a carriage return. It cannot be on multiple lines, Strings are treated as a sequence of one-byte ASCII values.

Examples

“hello Verilog world” “a/b” “aa+a”



3.1.9 Data types

Every signal has a data type associated with it:

• Explicitly declared with a declaration in your Verilog code. • Implicitly declared with no declaration when used to connect structural

building blocks in your code. Implicit declaration is always a net type "wire" and is one bit wide.

3.1.9.1 Data Types Value set

Verilog supports four values and eight strengths to model the functionality of real hardware. The four levels are listed in table Value level Condition in hardware circuits

0 logic zero, false condition 1 logic one, true condition x unknown value z High impedance, floating state

3.1.9.2 Nets

Nets represent connections between hardware elements. Just as in real circuits, nets have values continuously driven on them by the outputs of devices that they are connected to.

Types of Nets

Each net type has a functionality that is used to model different types of hardware (such as PMOS, NMOS, CMOS, etc)

Example

Net Data Type Functionality

wire, tri Interconnecting wire - no special resolution function

wor, trior Wired outputs OR together (models ECL)

wand, triand Wired outputs AND together (models open-collector)

tri0, tri1 Net pulls-down or pulls-up when not driven

supply0, supply1 Net has a constant logic 0 or logic 1 (supply strength)

Trireg Retains last value, when driven by z (tristate).



Register Data Types

• Registers store the last value assigned to them until another assignment statement changes their value.

• Registers represent data storage constructs. • You can create regs arrays called memories. • Register data types are used as variables in procedural blocks. • A register data type is required if a signal is assigned a value within a

procedural block • Procedural blocks begin with keyword initial and always.

Example

Data Types Functionality reg Unsigned variable integer Signed variable - 32 bits time Unsigned integer - 64 bits real Double precision floating point variable 3.1.9.3 Vectors

Nets or reg data types can be declared as vectors. If bit width is not specified, the default is scalar (1-bit).Vectors can be declared at [high#: low#] or [low#: high#], but the left number in the squared brackets is always the most significant bit of the vector.

Examples

wire a // scalar net variable default wire [7:0]bus; // 8-bit bus wire [31:0] busA, busB, busC; // 3 buses of 32-bit width reg clock // scalar register busA[7]; // bit #7 of vector busA bus [2:0] // three least significant bits of vector bus

3.1.9.4 Integer, Real and Time Register Data Types

Integer

An integer is a general purpose register data type used for manipulating quantities. Integers are declared by the keyword integer. Although it is possible to use reg as a general-purpose variable, it is more convenient to declare an integer variable for purposes such as counting. The default width for an integer is the host-machine word size, which is implementation specific but is at least 32



bits. Registers declared as data type reg store values as unsigned quantities, whereas integers store value as signed quantities.

Example

integer counter; // general purpose variable used as a counter

initial counter=-1; // A negative one is stored in the counter

Real

Real number constants and real register data types are declared with the keyword real. They can be specified in decimal notation (e.g., 5.12) or in scientific notation (e.g., 5e6, which is 5x10^6). Real numbers cannot have a range declaration, and their default value is 0. When a real value is assigned to an integer, the real number is rounded off to the nearest integer.

Example

real delta; // define a real variable called delta

initial begin delta=4e10; // delta is assigned in scientific notation delta=2.13 ; // delta is assigned a value 2.13 end integer i; // define an integer i Time

Verilog simulation is done with respect to simulation time. A special time register data type is used in Verilog to store simulation time. A time variable is declared with the keyword time. The width for time register data types is implementation specific but is at least 64 bits. The system function $time is invoked to get the current simulation time.

Example

time save_sim_time; // define a time variable save_sim_time

initial save_sim_time=$time; // save the current simulation time

3.1.9.5 Arrays

Arrays are allowed in Verilog for reg, integer, time and vector register data types. Arrays are not allowed for real variables. Arrays are accessed by <array_name>[<subscript>]. Multidimensional arrays not permitted in Verilog.



Example integer count[7:0]; // an array of 8 count variables

reg bool[31:0]; // array of 32 one-bit Boolean register variables

integer matrix[4:0][4:0]; // illegal declaration Multidimensional array 3.1.9.6 Memories

In digital simulation, one often needs to model register files, RAMs and ROMs. Memories are modeled in Verilog simply as an array of registers. Each element of the array is known as a word. Each word can be one or more bits. It is important to differentiate between n 1-bit registers and one n-bit register. A particular word in memory is obtained by using the address as a memory array subscript.

Example

reg mem1bit[0:1023]; // memory mem1bit with 1K 1-bit words reg [7:0]membyte[0:1023]; // memory membyte with 1K 8-bit words 3.1.9.7 Parameters

Verilog allows constants to be defined in a module by the keyword parameter. Parameters cannot be used as variables. Parameter values for each module instance can be overridden individually at compile time. This allows the module instances to be customized.

Example

parameter port_id = 5; //Defines a constant port_id parameter cache_line_width= 256; // Constant defines width of cache line 3.1.9.8 Strings

Strings can be stored in reg. The width of the register variables must be large enough to hold the string. Each character in the string takes up 8 bits (1 byte). If the width of the register is greater than the size of the string, Verilog fills bits to left of the string with zeros. If the register width is smaller than the string width, Verilog truncates the leftmost bits of the string. It is always safe to declare that is slightly wider than necessary.

Example

reg [8*81:1] string_value; // declare a variable that is 18 bytes wide

initial string_value=”hello Verilog course team”; // string can be stored in variable



3.2 MODULES

A module in Verilog consists of distinct parts as shown in figure 1.8. A module definition always begins with the keyword module. The module name, port list, port declarations, and optional parameters must come first in a module definition. Port list and port declarations are present only if the module has any ports to interact with the external environment. The five components within a module are;

• variable declarations, • dataflow statements • instantiation of lower modules • behavioral blocks • tasks or functions.

These components can be in any order and at any place in the module definition. The endmodule statement must always come last in a module definition. All components except module, module name, and endmodule are optional and can be mixed and matched as per design needs. Verilog allows multiple modules to be defined in a single file. The modules can be defined in any order in the file.

Figure 3.8 Components of Verilog Module



Example Module Structure :

module <module name>(<module_terminals_list>); ….. <module internals> …. Endmodule

3.2.1 Instances

A module provides a template from which you can create actual objects. When a module is invoked, Verilog creates a unique object from the template. Each object has its own name, variables, parameters and I/O interface. The process of creating objects from a module template is called instantiation, and the objects are called instances. In Example below, the top-level block creates four instances from the T-flip-flop (T_FF) template. Each T_FF instantiates a D_FF and an inverter gate. Each Instance must be given a unique name.

Example

// Define the top-level module called ripple carry // counter. It instants 4 T-filpflops. // four instances of the module T_FF are created. Each has a unique name. // each instance is passed a set of signals

module ripple_carry_counter(q,clk,reset); output [3:0]q; input clk,reset; T_FF tff0(q[0],clk,reset); T_FF tff1(q[1],q[0],reset); T_FF tff2(q[2],q[1],reset); T_FF tff3(q[3],q[2],reset); endmodule // define the module T_FF. it instantiates a D-filpflop. module T_FF(q,clk,reset); output q; input clk,reset; wire d; D_FF dff0(q,d,clk,reset); not n1(d,q); endmodule

3.3 PORTS

Ports provide the interface by which a module can communicate with its environment. For example, the input/output pins of an IC chip are its ports. The environment can interact with the module only through its ports. The internals



of the module are not visible to the environment. This provides a very powerful flexibility to the designer. The internals of the module can be changed without affecting the environment as long as the interface is not modified. Ports are also referred to as terminals. 3.3.1 Port Declaration

All ports in the list of ports must be declared in the module. Ports can be declared as follows Verilog Keyword Type of Port input Input port output Output port inout Bidirectional port Each port in the port list is defined as input, output, or inout, based on the direction of the port signal. 3.3.2 Port Connection Rules

One can visualize a port as consisting of two units, one unit that is internal to the module another that is external to the module. The internal and external units are connected. There are rules governing port connections when modules are instantiated within other modules. The Verilog simulator complains if any port connection rules are violated. These rules are summarized in figure 1.9.

Figure 3.9 Port connection Rules

Inputs: • Internally must be of net data type (e.g. wire) • Externally the inputs may be connected to a reg or net data type

Outputs: • Internally may be of net or reg data type • Externally must be connected to a net data type

Inouts: • Internally must be of net data type (tri recommended) • Externally must be connected to a net data type (tri recommended)



3.3.3 Ports Connection to External Signals

There are two methods of making connections between signals specified in the module instantiation and ports in a module definition. The two methods cannot be mixed.

• Port by order list • Port by name

Port by order list

Connecting port by order list is the most intuitive method for most beginners. The signals to be connected must appear in the module instantiation in the same order as the ports in the ports list in the module definition.

Syntax for instantiation with port order list: module_name instance_name (signal, signal...);

From the below example, notice that the external signals a, b, out appear in exactly the same order as the ports a, b, out in the module defined in adder below. Example

Port by name For larger designs where the module have ,say 5o ports , remembering the order of the ports in the module definition is impractical and error prone. Verilog provided the capability to connect external signals to ports by the port names, rather than by position.

Syntax for instantiation with port name:

module_name instance_name (.port_name(signal), .port_name (signal)… );

From the below example, note that the port connections in any order as long as the port name in the module definition correctly matches the external signal.



Example

Another advantage of connecting ports by name is that as long as the port name is not changed, the order of ports in the port list of a module can be rearranged without changing the port connections in the module instantiations.

3.4 GATE DELAYS

In real circuits, logic gates have delays associated with them. Gate delays allow the Verilog user to specify delays through the logic circuits. Pin-to-pin delays can also be specified in Verilog. 3.4.1 Rise, Fall, and Turn-off Delays

There are three types of delays from the inputs to the output of a primitive gate

Rise delay

The rise delay is associated with a gate output transition to a 1 from another value.

Fall delay

The fall delay is associated with a gate output transition to a 0 from another value.



Turn-off delay

The turn-off delay is associated with a gate output transition to the high impedance value (z) from another value.

If the value changes to X, the minimum of the three delays is considered.

Three types of delay specifications are allowed. If only one delay is specified, this value is used for all transitions. If two delays are specified, they refer to the rise and fall delay values. The turn-off delay is the minimum of the two delays. If all three delays are specified, they refer to rise, fall, and turn-off delay. If no delays are specified, the default value is zero.

Example for Delay Specification

3.4.2 Min/Typ/Max Values

Verilog provides an additional level of control for each type of delay mentioned. For each type of delay-rise, fall, and turn-off-three values, min, typ and, max can be specified. Any one value can be chosen at the start of the simulation. Min/ typ /max values are use to model devices whose delays vary within a minimum and maximum range because of the IC fabrication process variations.

Min value

The min value is the minimum delay value that the designer expects the gate to have.

Typ value

The typ value is the typical delay value that the designer expects the gate to have.

Max value The max value is the maximum delay value that the designer expects the gate to have.



Min, typ, or max values can he chosen at Verilog run time. Method of choosing a min/typ/max value may vary for different simulators or operat ing systems. (For Verilog-XLTM, the values are chosen by specifying options + m a x d e l a y , +typdelay, and +mindelays at run time. If no option is specified, the typical delay value is the default). This allows the designers the f l e x i b i l i t y of building three delay values for each transition into their design. The designer can experiment with delay values without modifying the design. Examples of Min, Max and Typical Delay Values

3.5 MODELING CONCEPTS

Verilog is both a behavioral and a structural language. Internals of each module to be defined at four levels of abstraction, depending on the needs of the design. The module behaves identically with the external environment irrespective of the level of abstraction at which the module is described. The internals of the module hidden from the environment. Thus, the level of abstraction to describe a module can be changed without any change in the environment. The levels are defined below

• Behavioral or algorithmic level

This is the highest level of abstraction provided by Verilog HDL. A module can be implemented in terms of the desired design algorithm without concern for the hardware implementation details. Designing at this level is very similar to C programming

• Dataflow level



At this level the module is designed by specifying the data flow. The designer is aware of how data flows between hardware registers and how the data is processed in the design.

• Gate level

The module is implemented in terms of logic gates and interconnections between these gates. Design at this level is similar to describing a design in terms of a gate-level logic diagram.

• Switch level

This is the lowest level of abstraction provided by Verilog. A module can be implemented in terms of switches, storage nodes, and the interconnections between them. Design at this level requires knowledge of switch-level implementation details.

Verilog allows the designer to mix and match all four levels of abstractions in a design. In the digital design community, the term register transfer level (RTL) is frequently used for a Verilog description that uses a combination of behavioral and dataflow constructs and is acceptable to logic synthesis tools.

If a design contains four modules, Verilog allows each of the modules to be written at a different level of abstraction. As the design matures, most modules are replaced with gate-level implementations. ,

Normally, the higher the level of abstraction, the more flexible and technology Independent the design. As one goes lower toward switch-level design, the design becomes technology dependent and inflexible. A small modification can cause a significant number of changes in the design. Comparing the analogy with C programming and assembly language programming. It is easier to program in higher-level language such as C. The program can be easily ported to any machine. However, if the design at the assembly level, the program is specific for that machine and cannot be easily ported to another machine.

3.6 SWITCH LEVEL MODELING

Usually, transistor level modeling is referred to model in hardware structures using transistor models with analog input and output signal values. On the other hand, gate level modeling refers to modeling hard-ware structures wing gate models with digital input and output signal values between these two modeling schemes is referred to as switch level modeling. At this level, a hardware component is described at the transistor level, but transistors only exhibit digital behavior and their input, and output signal values are only limited to digital values. At the switch level, transistors behave as on-off switches- Verilog uses a 4 value logic value system,



so Verilog switch input and output signals can take any of the four 0, 1, Z, and X logic values.

3.6.1 Switch level primitives

Table 1.1 shows Verilog switch and pull primitives. Switches are unidirectional or bidirectional and resistive or nonresistive. For each group those primitives that switch on with a positive gate like an NMOS transistor and those that switch on with a negative gate like a PMOS transistor. Switching on means that logic values flow from input transistor to its input. Switching off means that the output of a transistor is at Z level regardless of its input value.

A unidirectional transistor passes its input value to its output when it is switched on. A bidirectional transistor conducts both ways. A resistive structure reduces the strength of its input logic when passing it to its output. In addition to switch level primitives, pull-primitives that are used as pull-up and pull-down resistors for tri-state outputs.

Table 3.1

Figure 1.10 shows standard switches; pull primitives, and tri-state gates that behave like nmos and pmos. Instantiations of these primitives and their corresponding symbols are also shown. Cmos is a unidirectional transmission gate with a true and complemented control lines. Nmos and pmos are unidirectional pass gates representing NMOS and PMOS transistors respectively.

When such a resistive switch conducts, the strength of its output signal is one or two levels below that of its input signal. Delay values for transition to 1, transition to 0, and transition to Z can be specified in the #(to-1, to-0, to-z) format for unidirectional switches. Bidirectional tran switches shown in figure are functionally equivalent to unidirectional switches shown in the adjacent column of this figure. When conducting, the two inout ports are connected and logic values flow in both directions.



3.6.2 MOS switches

Two types of MOS switches can be defined with the keywords nmos and pmos. Keyword nmos is used to model NMOS transistors, Keyword pmos is used to model PMOS transistors. The symbols for nmos and pmos switches are shown in figure 3.11.

Figure 3.10

Figure 3.11 PMOS and NMOS Switches

In Verilog nmos and pmos switches are instantiated as shown in below,

nmos n1(out, data, control); // instantiate a nmos switch pmos p1(out, data, control); // instantiate a pmos switch

Since switches are Verilog primitives, like logic gates, the name of the instance is optional. Therefore, it is acceptable to instantiate a switch without assigning an instance name.



nmos (out, data , control); // instantiate nmos switch ; no instance name pmos (out, data, control); // instantiate pmos switch; no instance name Value of the out signal is determined from the values of data and control signals. Logic tables for out are shown in table. Some combinations of data and control signals cause the gates to output to either a 1 or 0 or to an z value without a preference for either value. The symbol L stands for 0 or Z; H stands for 1 or z.

Table 3.2 Logic Tables of NMOS and PMOS Thus, the nmos switch conducts when its control signal is 1. If control signal is 0, the output assumes a high impedance value. Similarly a pmos switch conducts if the control signal is 0.

3.6.3 CMOS Switches

CMOS switches are declared with the keyword cmos. A cmos device can be modeled with a nmos and a pmos device. The symbol for a cmos switch is shown in figure

Figure 3.12 CMOS switch

A CMOS switch is instantiated as shown in below, cmos cl(out, data, ncontrol, pcontrol);//instantiate cmos gate

or cmos (out, data, ncontrol, pcontrol); //no instance name given



The ncontrol and pcontrol are normally complements of each other. When the ncontrol signal is 1 and pcontrol signal is 0, the switch conducts. If ncontrol is 0 and pcontrol is 1, the output of the switch is high impedance value. The cmos gate is essentially a combination of two gates: one nmos and one pmos. Thus the cmos instantiation shown above is equivalent to the following.

nmos (out, data, ncontrol); //instantiate a nmos switch pmos (out, data, pcontrol); //instantiate a pmos switch Since a cmos switch is derived from nmos and pmos switches, it is possible derive the output value from Table 1.2, given values of data, ncontrol, and pcontrol signals. 3.6.4 Bidirectional Switches

NMOS, PMOS and CMOS gates conduct from drain to source. It is important to have devices that conduct in both directions. In such cases, signals on either side of the device can be the driver signal. Bidirectional switches are provided for this purpose. Three keywords are used to define bidirectional switches: tran, tranif0, and tranifl. Symbols for these switches are shown in figure 3.13 below.

Figure 3.13 Bidirectional switches

The tran switch acts as a buffer between the two signals inoutl and inout2. Either inoutl or inout2 can be the driver signal. The tranif0 switch connects the two signals inoutl and inout2 only if the control signal is logical 0. If the control signal is a logical 1, the nondriver signal gets a high impedance value z. The driver signal retains value from its driver. The tranifl switch conducts if the control signal is a logical 1. These switches are instantiated as shown in below

tran tl(inoutl, inout2); //instance name tl is optional tranifO (inoutl, inout2, control); //instance name is not specified



tranifl (inoutl, inout2, control); //instance name is not specified

Bidirectional switches are typically used to provide isolation between buses or signals. 3.6.5Power and Ground

The power (Vdd, logic 1) and Ground (Vss, logic 0) sources are needed when transistor-level circuits are designed. Power and ground sources are defined with keywords supplyl and supply0.

Sources of type supplyl are equivalent to Vdd in circuits and place a logical 1 on a net. Sources of the type supply0 are equivalent to ground or Vss and place a logical o on a net. Both supplyl and supply0 place logical 1 and o continuously on nets throughout the simulation. Sources supply1 and supply0 are shown below,

supply1 vdd; supply0 gnd; assign a=vdd; // connect a to vdd assign b=gnd; // connect b to gnd 3.6.6 Resistive Switches

MOS, CMOS, and bidirectional switches discussed before can be modeled corresponding resistive devices. Resistive switches have higher source-to-drain impedance than regular switches and reduce the strength of signals passing through them. Resistive switches are declared with keywords that have “x” prefixed to the corresponding keyword for the regular switch. Resistive have the same syntax as regular switches

rnmos rpmos rcmos rtran rtranif0 rtranifl

There are two main differences between regular switches and resistive switches: their source-to-drain impedances and the way they pass signal strengths. • Resistive devices have a high source-to-drain impedance. Regular to have a

low source-to-drain impedance. • Resistive switches reduce signal strengths when signals pass through them.

The changes are shown below. Regular switches retain strength levels of signals from input to output. The exception is that if the input is of supply, the output is of strength strong. Table 1.3 shows the strength reduction due to resistive switches



Input strength Output strength supply pull strong pull pull weak weak medium large medium medium small small small high high

Table 3.3 Strength reduction by resistive switches

3.8 Delay Specification on Switches

3.8.1 MOS and CMOS switches

Delays can be specified for signals that pass through these switch-level elements. optional and appear immediately after the keyword for the switch.

Switch element Delay specification

Examples

Pmos,nmos,rpmos,rnmos

Zero (no delay)

One (same delay on all transitions)

Two (rise,fall)

Three (rise,fall,turnoff)

pmos p1(out,data,control);

pmos#(1)

p1(out,data,control);

nmos #(1,2) p2(out,data,control);

nmos #(1,3,2) p2(out,data,control);

Cmos,rcmos Zero, one,two or three delays (same as above)

cmos #(5) c2(out,data,nctrl,pctrl);

cmos #(1,2) c1(out,data,nctrl,pctrl);

Table 3.4 Delay Specification on NMOS and CMOS switches



3.8.2 Bidirectional pass switches

Delay specification is interpreted slightly differently for bidirectional pass switches. These switches do not delay signals passing through them. Instead, they have turn-on and turn-off delays while switching. Zero, one, or two delays can b specified for bidirectional switches, as shown in Table 3.5.

Switch element

Delay specification

Examples

tran, rtran No delay specification allowed

tranif1,rtranif1

tranif0,rtranif0

Zero (no delay)

One (both turn-on and turn-off)

Two (turn-on, turn-off)

rtranif0 rt1(inout1,inout2,control);

tranif0#(3) T(inout1,inout2,control);

tranif1#(1,2) t1(inout1,inout2,control);

Table 3.5 Delay Specification for Bidirectional Switches

Example-CMOS Nor Gate

Verilog has a nor gate primitive, to design nor gate, using CMOS switches, the gate and the switch-level circuit diagram for the nor gate is shown in Figure 1.14

Figure 3.14 Gate and Switch diagram for Nor gate



Switch level Verilog for Nor gate

module my_nor(out, a, b); output out; input a,b; wire c; supply1 pwr; supply0 gnd; pmos (c,pwr,b); pmos (out,c,a); nmos (out, gnd, a); nmos (out,gnd, b); endmodule Example-CMOS NAND

Figure 3.14 Switch diagram for NAND gate

module my_nand (Out,A,B);

input A,B; ouput Out;

wire C;

supply1 Vdd; supply0 Vss;

pmos (Out,A,Vdd) pmos (Out,B,Vdd); nmos (Out,A,C); nmos(C,Vss,B);

endmodule



Example -2-to-1 Multiplexer A 2-to-1 multiplexer can be defined with CMOS switches. The circuit diagram for the multiplexer is shown in figure 3.15

Figure 3.15 2-to-1 Multiplexer, using switches

The 2-to-1 multiplexer passes the input I0 to output OUT if S=0 and passes I1 to OUT if S=1. The switch-level description for the 2-to-1 multiplexer is shown in below

module my_mux(out, s, i0, i1);

output out; input s,i0,i1;

wire sbar;

my_nor nt(sbar, s , s ); cmos (out, io , sbar, s); cmos (out, i1, s, sbar);

endmodule

Example-Simple CMOS Flip-Flop

Combinatorial elements are designed in the previous examples. Let us now define a memory element which can store a value. The diagram for a level-sensitive CMOS flip-flop is shown in figure 3.16. The switches C1 and C2 are CMOS switches. Switch C1 is open if clk=1, and switch C2 is open if clk=0. Complement of the clk is fed to the ncontrol input of C2. The CMOS inverters can be defined by using MOS switches as shown in figure 3.17



Figure 3.16 CMOS flip-flop

Figure 3.17CMOS inverter

The Verilog module description for the CMOS inverter from the switch-level circuit from figure 3.17 is given below ,

module my_not(out,in); output out; input in; supply1 pwr; supply0 gnd; pmos (out, pwr, in); nmos (out, gnd, in); endmodule

The CMOS flip-flop can be defined using the CMOS switches and my_hot inverters. The Verilog description for the CMOS flip-flop is given below. module cff(q, qbar, d, clk); output q,qbar; input d,clk; wire e; wire nclk; my_not nt(nclk, clk);



cmos (e, d, clk, nclk); cmos (e, q, nclk, clk); my_not nt1(qbar,e); my_not nt2(q, qbar); endmodule

3.9 GATE LEVEL MODELING

Verilog has built in primitives like gates, transmission gates, and switches. These are rarely used in design (RTL Coding), but are used in post synthesis world for modeling the ASIC/FPGA cells; these cells are then used for gate level simulation. Also the output netlist format from the synthesis tool, which is imported into the place and route tool, is also in Verilog gate level primitives. 3.9.1 Gate Types

A logic circuit can be designed by use of logic gates. Verilog supports logic gates as predefined primitives. Theses primitives are instantiated like modules except that they are predefined in Verilog and do not need a module definition. All circuit can be designed by using basic gates. There are two classes of basic gates: and/or gates and buf/not gates. And/Or Gates

and/or gates have one scalar output and multiple scalar inputs. The first terminal in the list of gate terminals is an output and the other terminals are inputs. The output of a gate is evaluated as soon as one of the inputs changes. The and/or gates available in Verilog are shown below.

and or xor nand nor xnor The corresponding logic symbols for these gates are shown in figure 1.18. We consider gates with two inputs.

These gates are instantiated to build logic circuits in Verilog. Examples of gate Instantiations are shown below. In the below example, for all instances, OUT is connected to the output out, and IN1 and IN2 are connected to the two inputs i1 and i2 of the gate primitives.

Figure 1.18 Gates symbol



The instance name does not need to be specified for primitives. More than two inputs can be specified in gate instantiation Gates with more two inputs are instantiated by simply adding more ports in the gate instantiation. Verilog automatically instantiates the appropriate gate.

Example Gate Instantiation of And/Or gates The truth tables for these gates are given below, assuming two inputs. Outputs of gates with more than two inputs are computed by applying the truth table iteratively.

Buf/Bufif1/Bufif0/Not/Notfif1/Notfif0 Gates

Buf/not gates have one scalar input and one or more scalar output. The lasts terminal in the port list is connected to the input. Other terminals are connected the outputs. Bufif1, Bufif0, Notif1, Notifo gates propagate only if their control signal is asserted. Such a situation is applicable when multiple drivers drive the signal. These drivers are designed to drive the signal on mutually exclusive control. They propagate z if their control signal is deasserted.

buf not bufif1 notfif1 bufif1 bufif0

__________________________________________

Figure 3.19 Gates But, Not, Bufif1, Bufif0, Notif1, Notifo



Table 1.6Truth Table

Example Gate Instantiation of Buf/Not gates

From the above example, notice that theses gates can have multiple outputs but exactly one inputs, which is the last terminal in the port list. The truth tables for these gates are shown below.



Examples

AND Gate from NAND Gate

Figure 3.20 Structural model of AND gate from two NANDS

module and_from_nand();

reg X, Y; wire F, W; // Two instantiations of the module NAND nand U1(W,X, Y); nand U2(F, W, W);



// Testbench Code initial begin $monitor ("X = %b Y = %b F = %b", X, Y, F); X = 0; Y = 0; #1 X = 1; #1 Y = 1; #1 X = 0; #1 $finish; end

endmodule

Simulation Output X = 0 Y = 0 F = 0 X = 1 Y = 0 F = 0 X = 1 Y = 1 F = 1 X = 0 Y = 1 F = 0 D-Flip flop from NAND Gate

Figure 3.20 Structural model of D Flip-flop from NAND Gate

module dff_from_nand(); wire Q,Q_BAR; reg D,CLK; nand U1 (X,D,CLK) ; nand U2 (Y,X,CLK) ; nand U3 (Q,Q_BAR,X); nand U4 (Q_BAR,Q,Y); // Test bench initial begin $monitor("CLK = %b D = %b Q = %b Q_BAR = %b",CLK, D, Q, Q_BAR); CLK = 0; D = 0; #3 D = 1;



#3 D = 0; #3 $finish; end

always #2 CLK = ~CLK;

endmodule

Simulation Output

CLK = 0 D = 0 Q = x Q_BAR = x CLK = 1 D = 0 Q = 0 Q_BAR = 1 CLK = 1 D = 1 Q = 1 Q_BAR = 0 CLK = 0 D = 1 Q = 1 Q_BAR = 0 CLK = 1 D = 0 Q = 0 Q_BAR = 1 CLK = 0 D = 0 Q = 0 Q_BAR = 1 Multiplex from Primitives

Figure 3.20 Structural model of Multiplex from Primitives

module mux_from_gates (); reg c0,c1,c2,c3,A,B; wire Y; //Invert the sel signals not (a_inv, A); not (b_inv, B); // 3-input AND gate and (y0,c0,a_inv,b_inv); and (y1,c1,a_inv,B); and (y2,c2,A,b_inv); and (y3,c3,A,B); // 4-input OR gate



or (Y, y0,y1,y2,y3); // Test bench

initial begin $monitor ( "c0 = %b c1 = %b c2 = %b c3 = %b A = %b B = %b Y = %b", c0, c1, c2, c3, A, B, Y); c0 = 0; c1 = 0; c2 = 0; c3 = 0; A = 0; B = 0; #1 A = 1; #2 B = 1; #4 A = 0; #8 $finish; end always #1 c0 = ~c0; always #2 c1 = ~c1; always #3 c2 = ~c2; always #4 c3 = ~c3; endmodule Simulation Output

c0 = 0 c1 = 0 c2 = 0 c3 = 0 A = 0 B = 0 Y = 0 c0 = 1 c1 = 0 c2 = 0 c3 = 0 A = 1 B = 0 Y = 0 c0 = 0 c1 = 1 c2 = 0 c3 = 0 A = 1 B = 0 Y = 0 c0 = 1 c1 = 1 c2 = 1 c3 = 0 A = 1 B = 1 Y = 0 c0 = 0 c1 = 0 c2 = 1 c3 = 1 A = 1 B = 1 Y = 1 c0 = 1 c1 = 0 c2 = 1 c3 = 1 A = 1 B = 1 Y = 1 c0 = 0 c1 = 1 c2 = 0 c3 = 1 A = 1 B = 1 Y = 1 c0 = 1 c1 = 1 c2 = 0 c3 = 1 A = 0 B = 1 Y = 1 c0 = 0 c1 = 0 c2 = 0 c3 = 0 A = 0 B = 1 Y = 0 c0 = 1 c1 = 0 c2 = 1 c3 = 0 A = 0 B = 1 Y = 0 c0 = 0 c1 = 1 c2 = 1 c3 = 0 A = 0 B = 1 Y = 1 c0 = 1 c1 = 1 c2 = 1 c3 = 0 A = 0 B = 1 Y = 1 c0 = 0 c1 = 0 c2 = 0 c3 = 1 A = 0 B = 1 Y = 0 c0 = 1 c1 = 0 c2 = 0 c3 = 1 A = 0 B = 1 Y = 0 c0 = 0 c1 = 1 c2 = 0 c3 = 1 A = 0 B = 1 Y = 1



3.10 BEHAVIORAL AND RTL MODELING

Verilog provides designers the ability to describe design functionality in an algorithmic manner. In other words, the designer describes the behavior of the circuit. Thus, behavioral modeling represents the circuit at a very high level of abstraction. Design at this level resembles C programming more than it resembles digital circuit design. Behavioral Verilog constructs are similar to C language constructs in many ways. Verilog is rich in behavioral constructs that provide the designer with a great amount of flexibility. 3.10.1 Operators

Verilog provided many different operators types. Operators can be, • Arithmetic Operators • Relational Operators • Bit-wise Operators • Logical Operators • Reduction Operators • Shift Operators • Concatenation Operator • Replication Operator • Conditional Operator • Equality Operator

3.10.1.1 Arithmetic Operators

• These perform arithmetic operations. The + and - can be used as either unary (-z) or binary (x-y) operators.

• Binary: +, -, *, /, % (the modulus operator) • Unary: +, - (This is used to specify the sign) • Integer division truncates any fractional part • The result of a modulus operation takes the sign of the first operand • If any operand bit value is the unknown value x, then the entire result

value is x • Register data types are used as unsigned values (Negative numbers are

stored in two's complement form) Example

module arithmetic_operators();

initial begin $display (" 5 + 10 = %d", 5 + 10);



$display (" 5 - 10 = %d", 5 - 10); $display (" 10 - 5 = %d", 10 - 5); $display (" 10 * 5 = %d", 10 * 5); $display (" 10 / 5 = %d", 10 / 5); $display (" 10 / -5 = %d", 10 / -5); $display (" 10 %s 3 = %d","%", 10 % 3); $display (" +5 = %d", +5); $display (" -5 = %d", -5); #10 $finish; end endmodule

Simulation Output

5 + 10 = 15 5 - 10 = -5 10 - 5 = 5 10 * 5 = 50 10 / 5 = 2 10 / -5 = -2 10 % 3 = 1 +5 = 5 -5 = -5

3.10.1.2 Relational Operators

Relational operators compare two operands and return a single bit 1or 0. These operators synthesize into comparators. Wire and reg variables are positive Thus (-3’b001) = = 3’b111 and (-3d001)>3d1 10, however for integers -1< 6

Operator Description a < b a less than b a > b a greater than b a <= b a less than or equal to b a >= b a greater than or equal to b

• The result is a scalar value (example a < b) • 0 if the relation is false (a is bigger than b) • 1 if the relation is true ( a is smaller than b) • x if any of the operands has unknown x bits (if a or b contains X)

Note: If any operand is x or z, then the result of that test is treated as false (0)

Example

module relational_operators();



initial begin

$display (" 5 <= 10 = %b", (5 <= 10)); $display (" 5 >= 10 = %b", (5 >= 10)); $display (" 1'bx <= 10 = %b", (1'bx <= 10)); $display (" 1'bz <= 10 = %b", (1'bz <= 10)); #10 $finish; end

endmodule

Simulation Output

5 <= 10 = 1 5 >= 10 = 0 1'bx <= 10 = x 1'bz <= 10 = x 3.10.1.3 Bit-wise Operators Bitwise operators perform a bit wise operation on two operands. This take each bit in one operand and perform the operation with the corresponding bit in the other operand. If one operand is shorter than the other, it will be extended on the left side with zeroes to match the length of the longer operand.

Operator Description ~ negation & and | inclusive or ^ exclusive or ^~ or ~^ exclusive nor (equivalence)

• Computations include unknown bits, in the following way: • -> ~x = x • -> 0&x = 0 • -> 1&x = x&x = x • -> 1|x = 1 • -> 0|x = x|x = x • -> 0^x = 1^x = x^x = x • -> 0^~x = 1^~x = x^~x = x • When operands are of unequal bit length, the shorter operand is zero-filled

in the most significant bit positions.



Example module bitwise_operators();

initial begin // Bit Wise Negation $display (" ~4'b0001 = %b", (~4'b0001)); $display (" ~4'bx001 = %b", (~4'bx001)); $display (" ~4'bz001 = %b", (~4'bz001)); // Bit Wise AND $display (" 4'b0001 & 4'b1001 = %b", (4'b0001 & 4'b1001)); $display (" 4'b1001 & 4'bx001 = %b", (4'b1001 & 4'bx001)); $display (" 4'b1001 & 4'bz001 = %b", (4'b1001 & 4'bz001)); // Bit Wise OR $display (" 4'b0001 | 4'b1001 = %b", (4'b0001 | 4'b1001)); $display (" 4'b0001 | 4'bx001 = %b", (4'b0001 | 4'bx001)); $display (" 4'b0001 | 4'bz001 = %b", (4'b0001 | 4'bz001)); // Bit Wise XOR $display (" 4'b0001 ^ 4'b1001 = %b", (4'b0001 ^ 4'b1001)); $display (" 4'b0001 ^ 4'bx001 = %b", (4'b0001 ^ 4'bx001)); $display (" 4'b0001 ^ 4'bz001 = %b", (4'b0001 ^ 4'bz001)); // Bit Wise XNOR $display (" 4'b0001 ~^ 4'b1001 = %b", (4'b0001 ~^ 4'b1001)); $display (" 4'b0001 ~^ 4'bx001 = %b", (4'b0001 ~^ 4'bx001)); $display (" 4'b0001 ~^ 4'bz001 = %b", (4'b0001 ~^ 4'bz001)); #10 $finish; end

endmodule

Simulation Output

~4'b0001 = 1110 ~4'bx001 = x110 ~4'bz001 = x110 4'b0001 & 4'b1001 = 0001 4'b1001 & 4'bx001 = x001 4'b1001 & 4'bz001 = x001 4'b0001 | 4'b1001 = 1001 4'b0001 | 4'bx001 = x001 4'b0001 | 4'bz001 = x001 4'b0001 ^ 4'b1001 = 1000 4'b0001 ^ 4'bx001 = x000 4'b0001 ^ 4'bz001 = z000 4'b0001 ~^ 4'b1001 = 0111 4'b0001 ~^ 4'bx001 = x111 4'b0001 ~^ 4'bz001 = x111



3.10.1.4 Logical Operators

Logical operators return a single bit 1 or 0. They are the same as bit-wise operators only for single bit operands. They can work on expressions, integers or groups of bits, and treat all values that are nonzero as “1”. Logical operators are typically used in conditional (if ... else) statements since they work with expressions. Operator Description ! logic negation && logical and || logical or

• Expressions connected by && and || are evaluated from left to right • Evaluation stops as soon as the result is known • The result is a scalar value: • -> 0 if the relation is false • -> 1 if the relation is true • -> x if any of the operands has x (unknown) bits

Example

module logical_operators();

initial begin // Logical AND $display ("1'b1 && 1'b1 = %b", (1'b1 && 1'b1)); $display ("1'b1 && 1'b0 = %b", (1'b1 && 1'b0)); $display ("1'b1 && 1'bx = %b", (1'b1 && 1'bx)); // Logical OR $display ("1'b1 || 1'b0 = %b", (1'b1 || 1'b0)); $display ("1'b0 || 1'b0 = %b", (1'b0 || 1'b0)); $display ("1'b0 || 1'bx = %b", (1'b0 || 1'bx)); // Logical Negation $display ("! 1'b1 = %b", (! 1'b1)); $display ("! 1'b0 = %b", (! 1'b0)); #10 $finish; end endmodule

Simulation Output

1'b1 && 1'b1 = 1 1'b1 && 1'b0 = 0 1'b1 && 1'bx = x 1'b1 || 1'b0 = 1 1'b0 || 1'b0 = 0



1'b0 || 1'bx = x ! 1'b1 = 0 ! 1'b0 = 1

3.10.1.5 Reduction Operators

Reduction operators operate on all the bits of an operand vector and return a single-bit value. These are the unary (one argument) form of the bit-wise operators. Operator Description & and ~& nand | or ~| nor ^ xor ^~ or ~^ xnor

• Reduction operators are unary. • They perform a bit-wise operation on a single operand to produce a single

bit result. • Reduction unary NAND and NOR operators operate as AND and OR

respectively, but with their outputs negated. • -> Unknown bits are treated as described before

Example

module reduction_operators();

initial begin // Bit Wise AND reduction $display (" & 4'b1001 = %b", (& 4'b1001)); $display (" & 4'bx111 = %b", (& 4'bx111)); $display (" & 4'bz111 = %b", (& 4'bz111)); // Bit Wise NAND reduction $display (" ~& 4'b1001 = %b", (~& 4'b1001)); $display (" ~& 4'bx001 = %b", (~& 4'bx001)); $display (" ~& 4'bz001 = %b", (~& 4'bz001)); // Bit Wise OR reduction $display (" | 4'b1001 = %b", (| 4'b1001)); $display (" | 4'bx000 = %b", (| 4'bx000)); $display (" | 4'bz000 = %b", (| 4'bz000)); // Bit Wise OR reduction $display (" ~| 4'b1001 = %b", (~| 4'b1001)); $display (" ~| 4'bx001 = %b", (~| 4'bx001)); $display (" ~| 4'bz001 = %b", (~| 4'bz001));



// Bit Wise XOR reduction $display (" ^ 4'b1001 = %b", (^ 4'b1001)); $display (" ^ 4'bx001 = %b", (^ 4'bx001)); $display (" ^ 4'bz001 = %b", (^ 4'bz001)); // Bit Wise XNOR $display (" ~^ 4'b1001 = %b", (~^ 4'b1001)); $display (" ~^ 4'bx001 = %b", (~^ 4'bx001)); $display (" ~^ 4'bz001 = %b", (~^ 4'bz001)); #10 $finish; end endmodule

Simulation Output

& 4'b1001 = 0 & 4'bx111 = x & 4'bz111 = x ~& 4'b1001 = 1 ~& 4'bx001 = 1 ~& 4'bz001 = 1 | 4'b1001 = 1 | 4'bx000 = x | 4'bz000 = x ~| 4'b1001 = 0 ~| 4'bx001 = 0 ~| 4'bz001 = 0 ^ 4'b1001 = 0 ^ 4'bx001 = x ^ 4'bz001 = x ~^ 4'b1001 = 1 ~^ 4'bx001 = x ~^ 4'bz001 = x 3.10.1.6 Shift Operators

Shift operators shift the first operand by the number of bits specified by the second operand. Vacated positions are filled with zeros for both left and right shifts (There is no sign extension). Operator Description << left shift >> right shift

• The left operand is shifted by the number of bit positions given by the right operand.



• The vacated bit positions are filled with zeroes

Example

module shift_operators();

initial begin // Left Shift $display (" 4'b1001 << 1 = %b", (4'b1001 << 1)); $display (" 4'b10x1 << 1 = %b", (4'b10x1 << 1)); $display (" 4'b10z1 << 1 = %b", (4'b10z1 << 1)); // Right Shift $display (" 4'b1001 >> 1 = %b", (4'b1001 >> 1)); $display (" 4'b10x1 >> 1 = %b", (4'b10x1 >> 1)); $display (" 4'b10z1 >> 1 = %b", (4'b10z1 >> 1)); #10 $finish; end endmodule

Simulation Output

4'b1001 << 1 = 0010 4'b10x1 << 1 = 0x10 4'b10z1 << 1 = 0z10 4'b1001 >> 1 = 0100 4'b10x1 >> 1 = 010x 4'b10z1 >> 1 = 010z 3.10.1.7 Concatenation Operator

The concatenation operator combines two or more operands to form a larger vector.

• Concatenations are expressed using the brace characters and , with commas separating the expressions within.

• ->Example: + a, b[3:0], c, 4'b1001 // if a and c are 8-bit numbers, the results has 24 bits

• Unsized constant numbers are not allowed in concatenations.

Example

module concatenation_operator();

initial begin // concatenation $display (" 4'b1001,4'b10x1 = %b", 4'b1001,4'b10x1); #10 $finish; end



endmodule Simulation Output

4'b1001,4'b10x1 = 100110x1

3.10.1.8 Replication Operator

Replication operator is used to replicate a group of bits n times. Say you have a 4 bit variable and you want to replicate it 4 times to get a 16 bit variable: then we can use the replication operator. Operator Description nm Replicate value m, n times

• Repetition multipliers (must be constants) can be used: • -> 3a // this is equivalent to a, a, a • Nested concatenations and replication operator are possible: • -> b, 3c, d // this is equivalent to b, c, d, c, d, c, d

Example

module replication_operator();

initial begin // replication $display (" 44'b1001 = %b", 44'b1001); // replication and concatenation $display (" 44'b1001,1'bz = %b", 44'b1001,1'bz); #10 $finish; end endmodule

Simulation Output

44'b1001 = 1001100110011001 44'b1001,1'bz = 1001z1001z1001z1001z

3.10.1.9 Conditional Operator

Conditional operator is like those in C/C++. They evaluate one of the two expressions based on a condition. It will synthesize to a multiplexer (MUX).

• ->cond_expr ? true_expr : false_expr • The true_expr or the false_expr is evaluated and used as a result

depending on what cond_expr evaluates to (true or false).

Example



module conditional_operator(); wire out;

reg enable,data; // Tri state buffer assign out = (enable) ? data : 1'bz; initial begin $display ("time\t enable data out"); $monitor ("%g\t %b %b %b",$time,enable,data,out); enable = 0; data = 0; #1 data = 1; #1 data = 0; #1 enable = 1; #1 data = 1; #1 data = 0; #1 enable = 0; #10 $finish; end endmodule

Simulation Output

time enable data out 0 0 0 z 1 0 1 z 2 0 0 z 3 1 0 0 4 1 1 1 5 1 0 0 6 0 0 z

3.10.1.10 Equality Operators

There are two types of Equality operators. Case Equality and Logical Equality.

Operator Description a === b a equal to b, including x and z (Case equality) a !== b a not equal to b, including x and z (Case inequality) a == b a equal to b, result may be unknown (logical equality)

a != b a not equal to b, result may be unknown (logical equality)



• Operands are compared bit by bit, with zero filling if the two operands do not have the same length

• Result is 0 (false) or 1 (true) • For the == and != operators, the result is x, if either operand contains an x

or a z • For the === and !== operators, bits with x and z are included in the

comparison and must match for the result to be true Note: The result is always 0 or 1 Example

module equality_operators(); initial begin // Case Equality $display (" 4'bx001 === 4'bx001 = %b", (4'bx001 === 4'bx001)); $display (" 4'bx0x1 === 4'bx001 = %b", (4'bx0x1 === 4'bx001)); $display (" 4'bz0x1 === 4'bz0x1 = %b", (4'bz0x1 === 4'bz0x1)); $display (" 4'bz0x1 === 4'bz001 = %b", (4'bz0x1 === 4'bz001)); // Case Inequality $display (" 4'bx0x1 !== 4'bx001 = %b", (4'bx0x1 !== 4'bx001)); $display (" 4'bz0x1 !== 4'bz001 = %b", (4'bz0x1 !== 4'bz001)); // Logical Equality $display (" 5 == 10 = %b", (5 == 10)); $display (" 5 == 5 = %b", (5 == 5)); // Logical Inequality $display (" 5 != 5 = %b", (5 != 5)); $display (" 5 != 6 = %b", (5 != 6)); #10 $finish; end endmodule

Simulation Output

4'bx001 === 4'bx001 = 1 4'bx0x1 === 4'bx001 = 0 4'bz0x1 === 4'bz0x1 = 1 4'bz0x1 === 4'bz001 = 0 4'bx0x1 !== 4'bx001 = 1 4'bz0x1 !== 4'bz001 = 1 5 == 10 = 0 5 == 5 = 1 5 != 5 = 0 5 != 6 = 1



3.10.2 Operator Precedence Operator Symbols Unary, Multiply, Divide, Modulus !, ~, *, /, %

Add, Subtract, Shift +, - , <<, >> Relation, Equality <,>,<=,>=,==,!=,===,!== Reduction &, !&,^,^~,|,~| Logic &&, || Conditional ? : 3.10.3 Timing controls

Various behavioral timing control constructs are available in Verilog. In Verilog, if there are no timing control statements, the simulation time does not advance. Timing controls provide a way to specify the simulation time at which procedural statements will execute. There are three methods of timing control:

• Delay based timing control • Event based timing control • Level-sensitive timing control

3.10.3.1 Delay-based timing control

Delay-based timing control in an expression specifies the time duration between when the statement is encountered and when it is executed. Delays are specified by the symbol #. Syntax for the delay-based timing control statement is shown below

<delay> ::= #<NUMBER> ||= #<identifier> ||= #<mintypmax_expression> <,<mintypmax_expression>>*) Delay-based timing control can be specified by a number, identifier, or a mintypmax_expression. There are three types of delay control for procedural assignments

• Regular delay control • Intra-assignment delay control • Zero delay control



Regular delay control

Regular delay control is used when a non-zero delay is specified to the left of a procedural assignment. Usage of regular delay control is shown below example,

module clk_gen (); reg clk, reset;

initial begin $monitor ("TIME = %g RESET = %b CLOCK = %b", $time, reset, clk); clk = 0; reset = 0; #2 reset = 1; #5 reset = 0; #10 $finish; end always #1 clk = !clk; endmodule

Simulation Output

TIME = 0 RESET = 0 CLOCK = 0 TIME = 1 RESET = 0 CLOCK = 1 TIME = 2 RESET = 1 CLOCK = 0 TIME = 3 RESET = 1 CLOCK = 1 TIME = 4 RESET = 1 CLOCK = 0 TIME = 5 RESET = 1 CLOCK = 1 TIME = 6 RESET = 1 CLOCK = 0 TIME = 7 RESET = 0 CLOCK = 1 TIME = 8 RESET = 0 CLOCK = 0 TIME = 9 RESET = 0 CLOCK = 1 TIME = 10 RESET = 0 CLOCK = 0 TIME = 11 RESET = 0 CLOCK = 1 TIME = 12 RESET = 0 CLOCK = 0 TIME = 13 RESET = 0 CLOCK = 1 TIME = 14 RESET = 0 CLOCK = 0 TIME = 15 RESET = 0 CLOCK = 1 TIME = 16 RESET = 0 CLOCK = 0 Intra-assignment delay control

Instead of specifying delay control to the left of the assignment, it is possible to assign a delay to the right of the assignment operator. Usage of intra-assignment delay control is shown in below example,



module intra_assign(); reg a, b; initial begin $monitor("TIME = %g A = %b B = %b",$time, a , b); a = 1; b = 0; a = #10 0; b = a; #20 $display("TIME = %g A = %b B = %b",$time, a , b); $finish; end endmodule Simulation output

TIME = 0 A = 1 B = 0 TIME = 10 A = 0 B = 0 TIME = 30 A = 0 B = 0 Difference between the intra-assignment delay and regular delay

Regular delays defer the execution of the entire assignment. Intra-assignment delays compute the right-hand-side expression at the current time and defer the assignment of the computed value to the left-hand-side variable. Intra-assignment delays are like using regular delays with a temporary variable to store the current value of a right-hand-side expression.

Zero delay control

Zero delay control is a method to ensure that a statement is executed last, after all other statements in that simulation in that simulation time are executed. This is used to eliminate race conditions. However, if there are multiple zero delay statements, the order between them is nondeterministic. Usage of zero delay control is shown in below example, initial begin x=0; y=0; end initial begin #0 x=1; #0 y=1; End



Above four statements x=0,y=0,x=1,y=1 are to be executed at simulation time 0. However since x=1 and y=1 have #0, they will be executed last. Thus, at the end of time 0,x will have value 1 and y will have value 1.

3.10.3.2 Event based timing control An event is the change in the value on a register or a net. Events can be utilized to trigger execution of a statement or a block of statements. There are four types of event-based timing control

• Regular event control • Named event control • Event OR control • Level-sensitive timing control

Regular event control

The @ symbol is used to specify an event control. Statements can be executed on changes in signal value or at a positive or negative transition of the signal value. The keyword posedge is used for a negative transition as shown in below example,

module edge_wait_example(); reg enable, clk, trigger;

always @ (posedge enable) begin trigger = 0; // Wait for 5 clock cycles repeat (5) begin @ (posedge clk) ; end trigger = 1; end //Test bench

initial begin $monitor ("TIME : %g CLK : %b ENABLE : %b TRIGGER : %b", $time, clk,enable,trigger); clk = 0; enable = 0; #5 enable = 1; #1 enable = 0; #10 enable = 1; #1 enable = 0; #10 $finish;



end

always #1 clk = ~clk;

endmodule

Simulation Output

TIME : 0 CLK : 0 ENABLE : 0 TRIGGER : x TIME : 1 CLK : 1 ENABLE : 0 TRIGGER : x TIME : 2 CLK : 0 ENABLE : 0 TRIGGER : x TIME : 3 CLK : 1 ENABLE : 0 TRIGGER : x TIME : 4 CLK : 0 ENABLE : 0 TRIGGER : x TIME : 5 CLK : 1 ENABLE : 1 TRIGGER : 0 TIME : 6 CLK : 0 ENABLE : 0 TRIGGER : 0 TIME : 7 CLK : 1 ENABLE : 0 TRIGGER : 0 TIME : 8 CLK : 0 ENABLE : 0 TRIGGER : 0 TIME : 9 CLK : 1 ENABLE : 0 TRIGGER : 0 TIME : 10 CLK : 0 ENABLE : 0 TRIGGER : 0 TIME : 11 CLK : 1 ENABLE : 0 TRIGGER : 0 TIME : 12 CLK : 0 ENABLE : 0 TRIGGER : 0 TIME : 13 CLK : 1 ENABLE : 0 TRIGGER : 0 TIME : 14 CLK : 0 ENABLE : 0 TRIGGER : 0 TIME : 15 CLK : 1 ENABLE : 0 TRIGGER : 1 TIME : 16 CLK : 0 ENABLE : 1 TRIGGER : 0 TIME : 17 CLK : 1 ENABLE : 0 TRIGGER : 0 TIME : 18 CLK : 0 ENABLE : 0 TRIGGER : 0 TIME : 19 CLK : 1 ENABLE : 0 TRIGGER : 0 TIME : 20 CLK : 0 ENABLE : 0 TRIGGER : 0 Named event control

Verilog provides the capability to declare an event and then trigger and recognize the occurrence of that event. The event does not hold any data. A named event is declared by the keyword event. An event is triggered by the symbol . The triggering of the event is recognized by the symbol @.

Example

event received_data; always @(posedge clock) begin if (last_data_packet) received_data; end always @(received_data) data_buf=data_pkt[0],data_pkt[1];



Event OR control

Sometimes a transition on any one of multiple signals or events can trigger the execution of a statement or a block of statements. This is expressed as an OR of events or signals. The list of events or signals expressed as an OR is also known as a sensitivity list. The keyword or is used to specify multiple triggers as shown in below example, always @(reset or clock or d) begin if(reset) q=1’b0; else if (clock) q=d; end 3.10.3.3 Level-Sensitive Timing Control

Verilog allows a level-sensitive timing control, that is, the ability to wait for a certain condition to be true before a statement or a block of statements is executed. The keyword wait is used for level-sensitive constructs.

Example

always wait (count_enable) #20 count=count+1; From the above example, the value of count_enable is monitored continuously. If count_enable is 0, the statement is not entered. If it is logical 1, the statement count=count+1 is executed after 20 time units. If count_enable stays at 1, count will be incremented every 20 time units. 3.10.4 Procedural Blocks

Verilog behavioral code is inside procedure blocks, but there is an exception: some behavioral code also exist outside procedure blocks. We can see this in detail as we make progress.

There are two types of procedural blocks in Verilog:

• initial: initial blocks execute only once at time zero (start execution at time zero).

• always: always blocks loop to execute over and over again; in other words, as the name suggests, it executes always.



Example – initial

module initial_example(); reg clk,reset,enable,data;

initial begin clk = 0; reset = 0; enable = 0; data = 0; end

endmodule

In the above example, the initial block execution and always block execution starts at time 0. Always block waits for the event, here positive edge of clock, whereas initial block just executed all the statements within begin and end statement, without waiting.

Example – always

module always_example(); reg clk,reset,enable,q_in,data;

always @ (posedge clk) if (reset) begin data <= 0; end else if (enable) begin data <= q_in; end

endmodule

In an always block, when the trigger event occurs, the code inside begin and end is executed; then once again the always block waits for next event triggering. This process of waiting and executing on event is repeated till simulation stops.

3.10.5 Procedural Assignment Statements

• Procedural assignment statements assign values to reg, integer, real, or time variables and cannot assign values to nets (wire data types)

• You can assign to a register (reg data type) the value of a net (wire), constant, another register, or a specific value.

Example - Bad procedural assignment

module initial_bad(); reg clk,reset; wire enable,data;



initial begin clk = 0; reset = 0; enable = 0; data = 0; end

endmodule

Example - Good procedural assignment

module initial_bad(); reg clk,reset; wire enable,data;

initial begin clk = 0; reset = 0; enable = 0; data = 0; end endmodule 3.10.6 Procedural Assignment Groups

If a procedure block contains more than one statement, those statements must be enclosed within

• Sequential begin - end block • Parallel fork - join block

When using begin-end, we can give name to that group. This is called named blocks.

Example - "begin-end"

module initial_begin_end(); reg clk,reset,enable,data; initial begin $monitor( "%g clk=%b reset=%b enable=%b data=%b", $time, clk, reset, enable, data); #1 clk = 0; #10 reset = 0; #5 enable = 0; #3 data = 0; #1 $finish; end endmodule



Begin : clk gets 0 after 1 time unit, reset gets 0 after 11 time units, enable after 16 time units, data after 19 units. All the statements are executed sequentially.

Simulator Output

0 clk=x reset=x enable=x data=x 1 clk=0 reset=x enable=x data=x 11 clk=0 reset=0 enable=x data=x 16 clk=0 reset=0 enable=0 data=x 19 clk=0 reset=0 enable=0 data=0

Example - "fork-join"

module initial_fork_join(); reg clk,reset,enable,data;

initial begin $monitor("%g clk=%b reset=%b enable=%b data=%b", $time, clk, reset, enable, data); fork #1 clk = 0; #10 reset = 0; #5 enable = 0; #3 data = 0; join #1 $display ("%g Terminating simulation", $time); $finish; end endmodule

Fork: clk gets its value after 1 time unit, reset after 10 time units, enable after 5 time units, data after 3 time units. All the statements are executed in parallel.

Simulator Output 0 clk=x reset=x enable=x data=x 1 clk=0 reset=x enable=x data=x 3 clk=0 reset=x enable=x data=0 5 clk=0 reset=x enable=0 data=0 10 clk=0 reset=0 enable=0 data=0 11 Terminating simulation 3.10.7 Sequential Statement Groups The begin - end keywords:

• Group several statements together. • Cause the statements to be evaluated sequentially (one at a time)



• -> Any timing within the sequential groups is relative to the previous statement.

• -> Delays in the sequence accumulate (each delay is added to the previous delay)

• -> Block finishes after the last statement in the block.

Example – sequential

module sequential();

reg a;

initial begin $monitor ("%g a = %b", $time, a); #10 a = 0; #11 a = 1; #12 a = 0; #13 a = 1; #14 $finish; end

endmodule

Simulator Output 0 a = x 10 a = 0 21 a = 1 33 a = 0 46 a = 1

3.10.8 Parallel Statement Groups

The fork - join keywords:

• Group several statements together. • Cause the statements to be evaluated in parallel (all at the same time). • -> Timing within parallel group is absolute to the beginning of the group. • -> Block finishes after the last statement completes (Statement with

highest delay, it can be the first statement in the block).

Example – Parallel

module parallel();

reg a;

initial fork $monitor ("%g a = %b", $time, a);



#10 a = 0; #11 a = 1; #12 a = 0; #13 a = 1; #14 $finish; join

endmodule

Simulator Output 0 a = x 10 a = 0 11 a = 1 12 a = 0 13 a = 1

3.10.9 Blocking and Nonblocking assignment

Blocking assignments are executed in the order they are coded, hence they are sequential. Since they block the execution of next statement, till the current statement is executed, they are called blocking assignments. Assignment are made with "=" symbol. Example a = b; Nonblocking assignments are executed in parallel. Since the execution of next statement is not blocked due to execution of current statement, they are called nonblocking statement. Assignments are made with "<=" symbol. Example a <= b; Note: Correct way to spell 'nonblocking' is 'nonblocking' and not 'non-blocking'. Example - blocking and nonblocking

module blocking_nonblocking();

reg a,b,c,d; // Blocking Assignment initial begin #10 a = 0; #11 a = 1; #12 a = 0; #13 a = 1; end

initial begin #10 b <= 0; #11 b <= 1; #12 b <= 0; #13 b <= 1; end



initial begin c = #10 0; c = #11 1; c = #12 0; c = #13 1; end

initial begin d <= #10 0; d <= #11 1; d <= #12 0; d <= #13 1; end

initial begin $monitor("TIME = %g A = %b B = %b C = %b D = %b",$time, a, b, c, d); #50 $finish; end

endmodule

Simulator Output TIME = 0 A = x B = x C = x D = x TIME = 10 A = 0 B = 0 C = 0 D = 0 TIME = 11 A = 0 B = 0 C = 0 D = 1 TIME = 12 A = 0 B = 0 C = 0 D = 0 TIME = 13 A = 0 B = 0 C = 0 D = 1 TIME = 21 A = 1 B = 1 C = 1 D = 1 TIME = 33 A = 0 B = 0 C = 0 D = 1 TIME = 46 A = 1 B = 1 C = 1 D = 1 3.10.10 assign and deassign

The assign and deassign procedural assignment statements allow continuous assignments to be placed onto registers for controlled periods of time. The assign procedural statement overrides procedural assignments to a register. The deassign procedural statement ends a continuous assignment to a register.

3.10.11 force and release

Another form of procedural continuous assignment is provided by the force and release procedural statements. These statements have a similar effect on the assign-deassign pair, but a force can be applied to nets as well as to registers. One can use force and release while doing gate level simulation to work around reset connectivity problems. Also can be used insert single and double bit errors on data read from memory.



3.10.12 Conditional Statements 3.10.12.1 The Conditional Statement if-else if - else statement controls the execution of other statements. In programming language like c, if - else controls the flow of program. When more than one statement needs to be executed for an if condition, then we need to use begin and end as seen in earlier examples. Syntax : if if (condition) statements; Syntax : if-else if (condition) statements; else statements; Syntax : nested if-else-if if (condition) statements; else if (condition) statements; ................ ................ else statements;

Example- simple if

module simple_if();

reg latch; wire enable,din; always @ (enable or din) if (enable) begin latch <= din; end

endmodule

Example- if-else

module if_else(); reg dff; wire clk,din,reset;

always @ (posedge clk) if (reset) begin



dff <= 0; end else begin dff <= din; end endmodule

Example- nested-if-else-if

module nested_if();

reg [3:0] counter; reg clk,reset,enable, up_en, down_en; always @ (posedge clk) // If reset is asserted if (reset == 1'b0) begin counter <= 4'b0000; // If counter is enable and up count is asserted end else if (enable == 1'b1 && up_en == 1'b1) begin counter <= counter + 1'b1; // If counter is enable and down count is asserted end else if (enable == 1'b1 && down_en == 1'b1) begin counter <= counter - 1'b1; // If counting is disabled end else begin counter <= counter; // Redundant code end

3.10.12.2 The Case Statement

The case statement compares an expression to a series of cases and executes the statement or statement group associated with the first matching case:

• case statement supports single or multiple statements. • Group multiple statements using begin and end keywords.

Syntax of a case statement is given below, case () < case1 > : < statement > < case2 > : < statement > ..... default : < statement > endcase



Example- case module mux (a,b,c,d,sel,y);

input a, b, c, d; input [1:0] sel; output y; reg y; always @ (a or b or c or d or sel) case (sel) 0 : y = a; 1 : y = b; 2 : y = c; 3 : y = d; default : $display("Error in SEL"); endcase endmodule

Example- case without default

module mux_without_default (a,b,c,d,sel,y); input a, b, c, d; input [1:0] sel; output y; reg y; always @ (a or b or c or d or sel) case (sel) 0 : y = a; 1 : y = b; 2 : y = c; 3 : y = d; 2'bxx,2'bx0,2'bx1,2'b0x,2'b1x, 2'bzz,2'bz0,2'bz1,2'b0z,2'b1z : $display("Error in SEL"); endcase

endmodule

Example- case with x and z

module case_xz(enable); input enable; always @ (enable) case(enable)



1'bz : $display ("enable is floating"); 1'bx : $display ("enable is unknown"); default : $display ("enable is %b",enable); endcase endmodule

3.10.12.3 The casez and casex statement

Special versions of the case statement allow the x ad z logic values to be used as "don't care":

• casez : Treats z as don't care. • casex : Treats x and z as don't care. •

Example- casez

module casez_example(); reg [3:0] opcode; reg [1:0] a,b,c; reg [1:0] out; always @ (opcode or a or b or c) casez(opcode) 4'b1zzx : begin // Don't care about lower 2:1 bit, bit 0 match with x out = a; $display("@%0dns 4'b1zzx is selected, opcode %b",$time,opcode); end 4'b01?? : begin out = b; // bit 1:0 is don't care $display("@%0dns 4'b01?? is selected, opcode %b",$time,opcode); end 4'b001? : begin // bit 0 is don't care out = c; $display("@%0dns 4'b001? is selected, opcode %b",$time,opcode); end default : begin $display("@%0dns default is selected, opcode %b",$time,opcode); end endcase

Simulation Output - casez

@0ns default is selected, opcode 0000 @2ns 4'b1zzx is selected, opcode 101x @4ns 4'b01?? is selected, opcode 0101 @6ns 4'b001? is selected, opcode 0010 @8ns default is selected, opcode 0000



Example- casex

module casex_example();

reg [3:0] opcode; reg [1:0] a,b,c; reg [1:0] out; always @ (opcode or a or b or c) casex(opcode) 4'b1zzx : begin // Don't care 2:0 bits out = a; $display("@%0dns 4'b1zzx is selected, opcode %b",$time,opcode); end 4'b01?? : begin // bit 1:0 is don't care out = b; $display("@%0dns 4'b01?? is selected, opcode %b",$time,opcode); end 4'b001? : begin // bit 0 is don't care out = c; $display("@%0dns 4'b001? is selected, opcode %b",$time,opcode); end default : begin $display("@%0dns default is selected, opcode %b",$time,opcode); end endcase

Simulation Output - casex

@0ns default is selected, opcode 0000 @2ns 4'b1zzx is selected, opcode 101x @4ns 4'b01?? is selected, opcode 0101 @6ns 4'b001? is selected, opcode 0010 @8ns default is selected, opcode 0000

3.10.13 Looping Statements

Looping statements appear inside procedural blocks only; Verilog has four looping statements like any other programming language.

• forever • repeat • while • for



3.10.13.1 The forever statement The keyword forever is used to express this loop. The loop does not contain any expression and executes continually, until the $finish task is encountered. Normally we use forever statements in initial blocks.

Syntax: forever < statement >

One should be very careful in using a forever statement: if no timing construct is present in the forever statement, simulation could hang. The code below is one such application, where a timing construct is included inside a forever statement.

Example module forever_example ();

reg clk;

initial begin #1 clk = 0; forever begin #5 clk = !clk; end end initial begin $monitor ("Time = %d clk = %b",$time, clk); #100 $finish; end endmodule

3.10.13.2 The repeat statement

The keyword repeat is used for thi sloop. The repeat construct executes the loop a fixed number of times. A repeat construct cannot be used to loop on a general logical expression. The repeat loop executes < statement > a fixed < number > of times. Syntax: repeat (< number >) < statement >

Example- repeat

module repeat_example(); reg [3:0] opcode; reg [15:0] data; reg temp; always @ (opcode or data)



begin if (opcode == 10) begin // Perform rotate repeat (8) begin #1 temp = data[15]; data = data << 1; data[0] = temp; end end end // Simple test code initial begin $display (" TEMP DATA"); $monitor (" %b %b ",temp, data); #1 data = 18'hF0; #1 opcode = 10; #10 opcode = 0; #1 $finish; end endmodule 3.10.13.3 The while loop statement

The keyword while is used to specify this loop. The while loop executes as long as an < expression > evaluates as true. If the loop is entered when the while-expression is false, the loop is not executed at all. This is the same as in any other programming language. Syntax: while (< expression >) < statement >

Example- while

module while_example(); reg [5:0] loc; reg [7:0] data; always @ (data or loc) begin loc = 0; // If Data is 0, then loc is 32 (invalid value) if (data == 0) begin loc = 32; end else begin while (data[0] == 0) begin loc = loc + 1; data = data >> 1; end



end $display ("DATA = %b LOCATION = %d",data,loc); end initial begin #1 data = 8'b11; #1 data = 8'b100; #1 data = 8'b1000; #1 data = 8'b1000_0000; #1 data = 8'b0; #1 $finish; end endmodule 3.10.13.4 The for loop statement

The for loop is the same as the for loop used in any other programming language.

• Executes an < initial assignment > once at the start of the loop. • Executes the loop as long as an < expression > evaluates as true. • Executes a < step assignment > at the end of each pass through the loop.

Syntax: for (< initial assignment >; < expression >, < step assignment >) < statement > Note: Verilog does not have ++ operator as in the case of C language.

Example – For

module for_example(); integer i; reg [7:0] ram [0:255]; initial begin for (i = 0; i < 256; i = i + 1) begin #1 $display(" Address = %g Data = %h",i,ram[i]); ram[i] <= 0; // Initialize the RAM with 0 #1 $display(" Address = %g Data = %h",i,ram[i]); end #1 $finish; end endmodule



3.11 DATA FLOW MODELING AND RTL

For small circuits, the gate-level modeling approach works very well because the number of gates is limited and the designer can instantiate and connects every gate individually. Also, gate-level modeling is very intuitive to a designer with a basic knowledge of digital logic design. However, in complex designs the number of gates is very large. Thus, designers can design more effectively if they concentrate on implementing the function at a level of abstraction higher than gate level. Dataflow modeling provides a powerful way to implement a design. Verilog allows a circuit to be designed in terms of the data flow between registers and how a design processes data rather than instantiation of individual gates. Dataflow modeling has become a popular design approach as logic synthesis tools have become sophisticated. This approach allows the designer to concentrate on optimizing the circuit in terms of data flow. For maximum flexibility in the design process, designers typically use a Verilog description style that combines the concepts of gate-level, dataflow, and behavioral design. In the digital design community, the term RTL (Register Transfer Level) design is commonly used for a combination of dataflow modeling and behavioral modeling.

3.11.1 Continuous Assignment Statements

A continuous assignment is the most basic statement in dataflow modeling, used to drive a value onto a net. A continuous assignment replaces gates in the description of the circuit and describes the circuit at a higher level abstraction. A continuous assignment statement starts with the keyword assign. They represent structural connections.

• They are used for modeling Tri-State buffers. • They can be used for modeling combinational logic. • They are outside the procedural blocks (always and initial blocks). • The continuous assign overrides any procedural assignments. • The left-hand side of a continuous assignment must be net data type.

Syntax: assign (strength, strength) #(delay) net = expression;

Example - One bit Adder design using continuous assignment statement module adder_using_assign (); reg a, b; wire sum, carry; assign #5 carry,sum = a+b; initial begin $monitor (" A = %b B = %b CARRY = %b SUM = %b",a,b,carry,sum); #10 a = 0; b = 0; #10 a = 1; #10 b = 1;



#10 a = 0; #10 b = 0; #10 $finish; end endmodule Example - Tri-state buffer using continuous assignment statement

module tri_buf_using_assign();

reg data_in, enable;

wire pad; assign pad = (enable) ? data_in : 1'bz;

initial begin $monitor ("TIME = %g ENABLE = %b DATA : %b PAD %b", $time, enable, data_in, pad); #1 enable = 0; #1 data_in = 1; #1 enable = 1; #1 data_in = 0; #1 enable = 0; #1 $finish; end endmodule 3.11.2 Propagation Delay Continuous Assignments may have a delay specified; only one delay for all transitions may be specified. A minimum: typical: maximum delay range may be specified.

Example - Tri-state buffer

module tri_buf_using_assign_delays(); reg data_in, enable; wire pad; assign #(1:2:3) pad = (enable) ? data_in : 1'bz;

initial begin $monitor ("ENABLE = %b DATA : %b PAD %b",enable, data_in,pad); #10 enable = 0; #10 data_in = 1; #10 enable = 1; #10 data_in = 0; #10 enable = 0;



#10 $finish; end

endmodule 3.12 STRUCTURAL GATE LEVEL DESCRIPTION

3.12.1 2 to 4 Decoder

Digital circuit for 2 to 4 decoder

RTL Coding for 2 to 4 decoder

module decoder_2to4_gates (A0,A1,D0,D1,D2,D3);

input A0,A1; output D0,D1,D2,D3;

wire n1,n2; not i1 (n1,A0); not i2 (n2,A1); and a1 (D0,n1,n2); and a2 (D1,n1,A1); and a3 (D2,A0,n2); and a4 (D3,A0,A1);

endmodule



3.12.2 Comparator

Digital circuit for Comparator

RTL Coding for Comparator

module comparator(A,B,AEQB,ALTB,AGTB);

input [3:0]A,B; output AEQB,ALTB,AGTB;

wire A3,A2,A1,A0; wire B3,B2,B1,B0;

assign A3=A[3]; assign A2=A[2]; assign A1=A[1]; assign A0=A[0]; assign B3=B[3]; assign B2=B[2]; assign B1=B[1]; assign B0=B[0];

wire N1,N2,N3,N4,N5,N6,N7,N8; wire N11,N12,N13,N14,N15,N16,N17,N18;

wire M1,M2,M3,M4; wire M11,M12,M13,M14,M15,M16,M17,M18;

not A11(N1,A3); not A12(N2,B3); not A13(N3,A2); not A14(N4,B2); not A15(N5,A1); not A16(N6,B1);



not A17(N7,A0); not A18(N8,B0);

and A19(N11,B3,N1); and A20(N12,A3,N2); and A21(N13,B2,N3); and A22(N14,A2,N4); and A23(N15,B1,N5); and A24(N16,A1,N6); and A25(N17,B0,N7); and A26(N18,A0,N8);

nor A27(M1,N11,N12); nor A28(M2,N13,N14); nor A29(M3,N15,N16); nor A30(M4,N17,N18);

and A31(M11,M1,N13); and A32(M12,M1,N14); and A33(M13,M1,M2,N15); and A34(M14,M1,M2,N16); and A35(M15,M1,M2,M3,N17); and A36(M16,M1,M2,M3,N18); and A37(AEQB,M1,M2,M3,M4);

or A38(ALTB,N11,M11,M13,M15); or A39(AGTB,N12,M12,M14,M16);

endmodule 3.12.3 D-latch

Digital circuit for D-latch

RTL coding for D-latch

module d_latch( D,E,Q,Qbar);

input D,E; output Q,Qbar;



wire n1,n2,n3,n4; not a11(n1,D);

and a12(n2,E,n1); and a13(n3,E,D);

nor a14(Q,n2,Qbar); nor a15(Qbar,n3,Q);

endmodule 3.12.4 D Flip Flop

Digital circuit for D Flip Flop

RTL coding for D Flip Flop

module d_flipflop( D,C,Q,Qbar);

input D,E; output Q,Qbar;

wire n1,n2,n3,n4;

not a11(n1,D); nand a12(n2,C,n1); nand a13(n3,C,D); nor a14(Qbar,n2,Q); nor a15(Q,n3,Qbar);

endmodule



3.12.5 Half Adder

Digital circuit for half adder

RTL coding for Half Adder

module half_adder(A,B,S,C);

input A,B; output S,C;

xor a11(S,A,B); and a12(C,A,B);

endmodule 3.12.6 Full Adder

Digital circuit for Full Adder

RTL coding for Full Adder

module full_adder(A,B,CIN,S,COUT);

input A,B,CIN; output S,COUT;

wire n1,n2,n3,n4,n5,n6; xor a11(n1,A,B); xor a12(S,n1,CIN);

and a13(n2,n1,CIN);



and a14(n3,A,B);

or a15(COUT,n2,n3);

endmodule 3.12.7 Ripple Carry Adder

Digital circuit for Ripple Carry Adder

RTL coding for Ripple Carry Adder Circuit

module ripple_carry_adder(ai,bi,cin,sum,carry);

input ai,bi,cin; output sum,carry; wire n1,n2,n3; wire m1,m2,m3,m4,m5,m6,m7; not a11(n1,ai); not a12(n2,bi); not a13(n3,cin);



nand a14(m1,bi,cin); nand a15(m2,cin,ai); nand a16(m3,ai,bi); nand a17(m4,ai,bi,cin); nand a18(m5,n3,n2,ai); nand a19(m6,n3,bin,n1); nand a20(m7,cin,n2,n1); nand a21(carry,m1,m2,m3); nand a22(sum,m4,m5,m6,m7); endmodule

VLSI DESIGN CMOS CHIP DESIGN


UNIT –4

4.1 INTRODUCTION TO CMOS Over the past decade, Complementary Metal Oxide Semiconductor (CMOS) technology has played an increasingly important role in the global integrated circuit industry. Not that CMOS technology is that new. In fact, the basic principle behind the MOS filed-transistor was proposed by J.Lilienfeld as early as 1925, and a similar structure closely resembling a modern MOS transistor was proposed by O.Heli in 1935. In VLSI design, a logic function is implemented by means of a circuit consisting of one or more basic cells, such as NAND or NOR gates. The implementation can be used as a library cell in the design phase. In CMOS circuits, it is possible to implement complex Boolean functions by means of NMOS and PMOS transistors. A cell is an interconnection of CMOS transistors. A CMOS cell is depicted in Figure 4 .1. It consists of a row of PMOS transistors and a row of NMOS transistors corresponding to the PMOS and NMOS sides of the circuit, respectively. The automatic generation of standard CMOS logic cells has been studied intensively during the last decade. The continuous progress in VLSI technology presents new challenges in developing efficient algorithms for the layout of standard CMOS logic cells. The cell generation techniques are classified into random generation and regular style. A random cell generation technique does not exploit any particular structure to produce the cells. It uses a general technique, such as a hierarchical place-route algorithm. Random generation methods produce compact and (if desired) high-performance layouts. However, it takes a long time to design a cell. Regular generation techniques employ a predefined structure to design a cell. Cells generated in this manner occupy more area but can be designed faster. Traditional cell structures based on regularity are, for example, PLAs, ROMs, and RAMs. The disadvantage of the ROM-based cell is that it takes a lot of area, as it uses many redundant transistors.

Figure 4.1 CMOS



4.2 LOGIC DESIGN WITH CMOS 4.2.1 COMBITIONAL LOGIC

If two N_SWITCHES are placed in series ,the compsoite switch constructed by the action is closed (or ON) if both swicthes are closed(or ON)figure 4.2(a).This yeilds an AND function. The corrresponding strcuture for P-SWITCHES is shown in figure 4.2(b).The composite switch is closed if the onputes are set to ‘0’. When two N-SWITCHES are placced in parallel figure 4.2(c), the composite switch is closed (if either input is as’1’).Thus ans OR function is created. The switch shown in figure 4.2(d) is composed of two P-SWITCHES placed in parellel.Incontrast to previous case ,if either input is a ‘0’ the switch is closed. By using combinations of these contrucitons, CMOS combination gates may be constructed. s1=0 s1=0 s1=1 s1=1 s1=0 s1=0 s1=1 s1=1 s2=0 s2=1 s2=0 s2=1 s2=0 s2=1 s2=0 s2=1 a b s1=0 s1=0 s1=1 s1=1 s1=0 s1=0 s1=1 s1=1 s2=0 s2=1 s2=0 s2=1 s2=0 s2=1 s2=0 s2=1 c d Figure 4.2 Connection and behavior of series and parallel of N-SWITHCES

and P-SWITCHES



4.2.2 INVERTER

Table 4.1 outlines the truth table of required to implement a logical inverter.

INPUT OUTPUT 0 1 1 0

Table 4.1 Inverter Truth Table

Figure 4.3 CMOS Inverter

If Vin is high (at or near Vdd ) the NMOS transistor will be turned on. The voltage between the gate and substrate of the p-channel device is at or near zero. The gate is at Vdd and so is the moat.Hence the upper transistor will be turned off. The output will thus be low. If the input voltage is at or near ground (a “low”) then the n-channel device is turned off. The voltage between the gate and substrate of the p-channel device is now - Vdd (The gate is 0 and the substrate is at +Vdd ). If the PMOS transistor has a threshold voltage VT of, say, -2 V, then it will be turned on and the output will be high. Note however, that in either state, high or low, there is no static current flowing through the inverter.

The transfer characteristics for this circuit. Are a little more complicated. First, let’s make sure that the voltages and currents defined. From the figure 4.4, Vgs-n the n-channel gate-source voltage is just Vin. Vgs-p the gate-source voltage for the p-channel device is Vin-VddId-n=Id-p=IdVds-P the drain source voltage for the p-channel transistor can be written as Vds-n-Vdd.

Figure 4.4 Schematic of a CMOS inverter



4.2.3 The NAND Gate

The two transistors Q1 and Q3 resemble the series-connected as a complementary pair from the inverter circuit. Both are controlled by the same input signal (input A), the upper transistor turning off and the lower transistor turning on when the input is "high" (1), and vice versa. Transistors Q2 and Q4 from figure 4.5 are similarly controlled by the same input signal (input B), and how they will also exhibit the same on/off behavior for the same input logic levels. The upper transistors of both pairs (Q1 and Q2) have their source and drain terminals paralleled, while the lower transistors (Q3 and Q4) are series-connected. What this means is that the output will go "high" (1) if either top transistor saturates, and will go "low" (0) only if both lower transistors saturate.

Figure 4.5 Logic Schematic

The following sequence of illustrations shows the behavior of this NAND gate for all four possibilities of input logic levels (00, 01, 10, and 11): The pull-down tree is a series pair of N-switches with one end connected to Vdd and the other end connected to the output.



Figure 4.6 Behavior of NAND Gate

The output level of the structure ,given the logic levels on the control inputs shown in the Table 4.2.

CONTROL INPUT A CONTROL INPUT B OUTPUT 0 0 Z 0 1 Z 1 0 Z 1 1 0

Table 4.2 Nand Gate Pull-down Truth Table

The pull-up tree is a parallel connection pair of P-SWITHCES with one end connected to Vdd and the other end connected to NAND gate output.The level of the output of the combined switch is shown in Table 4.3. CONTROL INPUT A CONTROL INPUT B OUTPUT 0 0 1 0 1 1 1 0 1 1 1 Z

Table 4.3 Nand Gate Pull-up Truth Table

The combined state of the ouput depends on the combination of the pull-up states and the pull-down states 4.2.4 The NOR Gate

As compare with the NAND gate, transistors Q1 and Q3 work as a complementary pair, as do transistors Q2 and Q4. Each pair is controlled by a single input signal. If either input A or input B are "high" (1), at least one of the lower transistors (Q3 or Q4) will be saturated, thus making the output "low" (0). Only in the event of both inputs being "low" (0) will both lower transistors be in cutoff mode and both



upper transistors be saturated, the conditions necessary for the output to go "high" (1). This behavior, of course, defines the NOR logic function.

Figure 4.7 Logic Schematic

4.3 TRANSMISSION GATES

4.3.1 Multiplexers

Complementary switches may be used to select between a number of inputs, thus forming a multiplexer function. Figure 4.8(a) shows a connection diagram for a 2-input multiplexer. As the switches have to pass ‘0’s and ‘1’s equally well. Complementary switches with n-and p-transistors are used. Multiplexer are key components in CMOS memory elements and data manipulation structures. A Output A 0 S Output B B 1 S S a b

Figure 4.8 A 2-input CMOS multiplexer

4.3.2 Lathes

A structure called D latch using 2-input multiplexer and two inverter is shown in Figure 4.9(a). It consist of a data input, D, a clock input, CLK, and output Q and –Q. When CLK=’1’, Q set to D and –Q is set to –D (the logical NOT of D) Figure 4.9(b).There are number of ways are used to indicate the logical NOT of a signal. The form D is often used in texts. However, CAD systems due to the use of an ASCII a feedback path around the inverter pair is established figure 4.9(c). This causes the current state of Q to be stored. While CLK=’0’ the input D is ignored. This is known as level-sensitive latch. That is the state of the output is dependent on the level of the clock signal. The latch shown is a positive level-sensitive latch. By reversing the control connection to the multiplexer, a negative level-sensitive latch may be constructed.

C

C



D -Q a

CLK Q D -Q b CLK=1 Q -Q D CLK=0 c Q

Figure 4.9 A CMOS positive-level-sensitive D latch

4.4 CMOS CHIP DESIGN OPTIONS

4.4.1 ASIC

What is an ASIC?

Application Specific Integrated Circuit) Pronounced "a-sick." A chip that is custom designed for a specific application rather than a general-purpose chip such as a microprocessor. The use of ASICs improves performance over general-purpose CPUs, because ASICs are "hardwired" to do a specific job and do not incur the overhead of fetching and interpreting stored instructions. However, a standard cell ASIC may include one or more microprocessor cores and embedded software, in which case, it may be referred to as a "system on a chip" (SoC).

A full custom ASIC chip is the most costly, and like standard cell ASICs, uses a custom-designed mask for every layer in the chip. Unlike standard cells, designers of a full custom device have total control over the size of every transistor forming every logic gate, so they can "fine tune" each gate for optimum performance. Thus, a full custom ASIC performs electronic operations as fast as it is possible to do so, providing that the circuit design is efficiently architected.

C

C



Today, full custom ASICs represent a small percentage of the ASIC market because gate arrays, structured ASICs and standard cells turn circuit designs into working chips much faster and at much less cost. Such chips have greatly improved in speed over the years and provide the necessary performance for many applications. The speed advantage of a full custom ASIC is not as relevant as it was in the past. It is used primarily for devices such as microprocessors that must run as fast as possible and will be produced in huge quantities. Also promoting the decline of full custom ASICs are chip manufacturers that make generic chips containing all the necessary functions for specific mass market products such as DVDs, CDs, digital cameras, etc. An ASIC (Application Specific Integrated Circuit) is a semiconductor device designed especially for a particular customer (versus a Standard Product, which is designed for general use by any customer). The two major categories of ASIC Technology are Array- Based and Cell-Based (also referred to as Standard Cell). Array-Based ASICs configure a customer's design at the metal layers, whereas Cell-Based ASICs are uniquely fabricated at all layers of the silicon process including the diffusion layers. 4.4.2 Uses of ASICs

To save chip area, ASIC technology integrates the logic and much of the memory formerly distributed among multiple ICs, thus improving reliability, optimizing PC Board space, and reducing component costs. In addition, the higher integration and smaller size results in significantly better system performance. ASICs were originally used solely to replace or consolidate TTL “glue” logic and consisted of relatively low complexity logic. Improvements in design tools and implementation software, in process technology, and in large pin count packages, now integrate much more of the logic formerly distributed among numerous ICs onto a single, very large scale System-On-a-Chip (SOC) 4.4.3 Full Custom ASICs

In a Full-custom ASIC all mask layers are customized show in figure 4.10.Full Custom designs offer the highest performance and the smallest die size, with the disadvantage if increased design time, higher complexity and costs, together with the highest risks of failure .This design option only makes sense when either libraries nor IP cores are available, or when very high performances are required. Time after time, fewer projects are really “full-custom”, because of the very high cost and the prohibitively slow time to market.

Most of the full-customs works are related library cell generation or minor parts of a full design. Examples of full-custom IC specific parts are high voltage (automobiles-avionic), analog processing and analog/digital communication devices, sensors and transducers. Traditionally microprocessors and memories



were exclusively full-custom, but industry is increasingly turning semicustom ASIC techniques in these areas too.

Figure.4.10 In a Full-custom designs every layer must be designed

Figure 4.11 Application –Specific components

4.4.5 Semi-Custom ASICs

In order to reduce the unaffordable cost of full custom in most projects, a wide variety of design approaches have been developed to shorten design time , cut down the costs , and automate the processes. These approaches are commonly called semicustom. Semicustom designs are performed at the logic (gate) level. As such, they lose some of the flexibility available from a full-custom fashion that is the price paid for much easier design techniques. Semicustom solutions can be further categorized into gate array and standard cell.

4.4.6 Standard- Cell-Based ASIC

Standard cells are logic components (e.g., gates, multiplexer, adders, flip flops) previously designed and stored in a library. The standard cell areas (also called flexible blocks) in a cell-based (cell-based IC or CBIC- a common term in Japan, pronounced “sea-bick”) CBIC are built of rows of standard cells-like a wall built of bricks. The standard cell areas may be used in combination with larger predesigned cells, perhaps microcontroller or even microprocessor, known as



megacells. Megacells are also called mega functions, full-custom blocks, system-level macros (SLMs), fixed blocks, cores, or Functional Standard Blocks (FSBs).

The ASIC designer defines only the placement of the standard cells and the interconnect in a CBIC. However, the standard cells can be placed anywhere on the silicon, this means that all mask layers of a CBIC are customized and are unique to a particular customer. The advantage of CBIC is that designers save time, money, and reduce risk by using a predesigned, pretested and pre characterized standard-cell library. In addition each standard cell can be optimized individually. During the design of the cell library each and every transistor in every standard cell can be chosen to maximize speed or minimizing area, for example. The disadvantages are time or expense of designing or buying the standard-cell library and time needed to fabricate all layers of the ASIC for each new design.

Figure 4.12 Typical Standard cell layouts

A designed is created using the library cells as inputs to CAD system; logic schematic diagram or hardware description language (HDL) code description. Next, further CAD tool automatically converts the design into a chip layout. Standard –cell designs are typically organized on the chip, rows of constant height cells (figure 4.12). Together with logic –level components cells, standard –cell systems typically offer-higher-level function such as multipliers and memory arrays. This allows the use of predesigned (or automatically generated) high –level components to complete the design. 4.4.7 Gate Array Asic

Gate Arrays (Gas) are basically composed of continuous arrays of P- and n-type transistors. The silicon vendor provides master or base wafers, to be then personalized according to the interconnection information supplied by the customers. Therefore, the designers supply the personated information that defines the connections between transistors in the gate array. Although a gate array standardized the chip at the geometry level, user interaction to gates, is performed through an ad hoc CAD tool.



The gate array also called masked gate array (MGA), or prediffused array) used library components and macros that reduce the development time.

Three main types of gate array can be mentioned

• Channel • Channelless and • Structured.

4.4.8 Channeled Gate Array

Figure 4.13 shows a channeled gate array. The important features of this type of MGA are: • Only the interconnect is customized. • The interconnect uses predefined spaces between rows of base cells. • Manufacturing lead time is between two days and two weeks.

Figure 4 .13 Channel Gate Array

In channeled gate array, the interconnections are drawn within predefined spaces (channels) between rows of logic cells.

4.4.9 Channelless Gate Array

Figure 4.14 shows a channelless gate array (also known as a channel-free gate array, sea-of-gates array, or SOG array). The important features of this type of MGA are: • Only some (the top few) mask layers are customized – the interconnect. • Manufacturing lead time is between two days and two weeks.

Figure 4 .14 Channel Gate Array



In a channelless (channelfree gate array o r sea-of-gates), there are no connections channels, the connections are drawn with the upper metal layers, that is on the top of the logic cells. In both cases, only some mask layers (the upper ones) must be customized.

4.4.10 Structured Gate Array An embedded gate array or structured gate array (also known as masterslice or masterimage) combines some of the features of CBISCs and MGAs. One of the disadvantages of the MGA is the fixed gate-array base cell. This makes the implementation of memory, for example, difficult and inefficient. Figure 4.15 shows an structured gate array. The important features of this type of MGA are the following,

• Only the interconnect is customized. • Customs blocks (the same for each design) can be embedded. • Manufacturing lead time is between two days and two weeks.

Figure 4.15 Structured Gate Array

An embedded gate array gives the improved area efficiency and increased performance of a CBIC but with the lower cost and faster turnaround of an MGA. One disadvantage of an embedded gate array is that embedded function is fixed.

4.5 PROGRAMMABLE LOGIC

As the investment made in any chip designs is significant, designers search for ways in which to amortize the design effort over a large number of devices. This might results from a huge single market for one device or, more probably, multiple smaller markets for a more adaptable device. The larger the unit volume for a part lowest its cost will be the end user. Programmability is one way to achieve a wider use for a particular part. This is epitomized by the microprocessor. Often, though, the cost or speed of a microprocessor may not meet system goals and an alternative solution is required. In CMOS one may be divided this spectrum of Programmable devices into three areas,



• Programmable Logic Structures. • Programmable Interconnect. • Reprogrammable Gate Arrays.

4.5.1 Programmable Logic Structures

The first broad classes of programmable CMOS devices are represented by the programmable logic devices referred to as PALs (Programmable Array Logic, Advanced Micro Devices, Inc.) or PLDs (Programmable Logic Devices). Generally these devices are implemented as AND-OR plane devices as shown in Figure 4.16.In the design shown a number of inputs feed vertical wires , which are selectively connected to an AND-OG gate. Each AND-OR gate has a variable number of products terms that feed the gate. This gate in turn feed an I/O cell, which allows result into the AND-OR plane.PAL devices come in a large range of sizes with a variable of inputs, outputs, product terms, and I/O –cell complexity.

Figure 4.16 PAL architecture



4.5.2 Programmable of PALs The Programming of PALs is done in three ways,

• Fusible links • UV-erasable EPROM • EEPROM-Electrically Erasable Programmable ROM

4.5.3 Fusible Links

A fusible links use a metal such platininium silicide or titatinium tungsten to form links that are blown when a certain current is exceeded in the fuse. This is normally accomplished by using a higher than normal programming voltage applied to the device. This technology is normally used in conjunction with a bipolar process (as opposed to a CMOS process) where the small devices can readily sink the current needed to blow the fuses. Programming is a one- time operation. As an alternative to current, a laser can be used to cut aluminum fuses in normal CMOS technologies. Frequently this is used in redundant memory technique where a spare column may be switched in to replace a failing one. 4.5.4 UV-erasable EPROM

UV-erasable memories typically use a floating gate structure. Here a floating gate is interposed between the regular MOS transistor gate and the channel. To program the cell, a voltage around 13-14 volts is applied to the control gate while the drain of the transistor to be programmed is held at around 12 volts. This results in the floating gate becoming charged negatively. This increases the threshold of the transistor (to around 7 volts), thus rendering it permanently “off” for all normal circuit voltage 9 maximum 5-6 volts). The process can be reversed by illuminating the gate with UV light.

4.5.5 EEPROM EEPROM (also written E2PROM and pronounced e-e-prom or simply e-squared),

which stands for Electrically Erasable Programmable Read-Only Memory, is a type of non-volatile memory used in computers and other electronic devices to store small amounts of data that must be saved when power is removed, e.g.,

calibration tables or device configuration. When larger amounts of static data are to be stored (such as in USB flash drives) a specific type of EEPROM such as

Figure 4.17 EEPROM Structure



flash memory is more economical than traditional EEPROM devices. EEPROM technology allows the electrical programming and erasure of CMOS ROM cells. This type of programming forms the most popular in use today for CMOS and is the most likely encountered by the IC system designers in today’s foundry possesses.

4.5.6 Programmable Interconnect

In a PAL the device is programmed by changing the characteristics of the switching element. An alternative would be to program the routing. This has been demonstrated via a number of techniques including Laser Pantography, where a laser lays down paths of metal under computer control. Commercially, programmable routing approaches are represented by products from Actel, QuickLogic, and other companies. Metal Polysilicon 100 MΩ Dielectric ONO Layer n+ diffusion Metal Polysilicon 200-500 MΩ Dielectric n+ diffusion after breakdown

Figure 4.18(a) Antifuse

The Actel Field Programmable Gate Array are based on an element called a PLICE (Programmable Low-Impedance Circuit Element) or anti-fuse. An antifuse is normally high resistance (>100 MΩ). On application of appropriate programming voltages, the antifuse is changed permanently to a low-resistance structure (200-500Ω). The structure of an antifuse is shown in figure 4.18(a). It consists of an ONO (oxide-nitride-oxide) layer sandwiched between a polysilicon layer on top and n+ diffusion on the bottom. The QuickLogic array is based on a structure called a ViaLink, which consist of a sandwich of material between metal1 and metal2.This is illustrated in figure 4.18(b).The “on” resistance of this structure is somewhat lower than that in figure 4.18(a). One chip architecture that uses the antifuse in shown in figure 4.19.Logic elements are arranged in rows separated by horizontal interconnect. Interconnect permanently connected to the logic elements passes vertically. Both horizontally and vertical segments are segmented into variety of lengths. Segments may be joined by programming



antifuse. Certain special logic elements are surrounded by I/O pads and programming and diagnostic logic Metal Via Sio2 metal link via Sio2

Figure 4.18(b) ViaLink

Figure 4.19 Programmable Interconnect Model

4.6 ASIC DESIGN FLOW

Figure 4.20 shows the sequence of the steps to design an ASIC; we call this a Design Flow. The steps are listed below with a brief description of the function of each steps. Design Entry: Enter the design into an ASIC design system, either using a Hardware Description Language (HDL) or schematic entry Logic Synthesis: Use an HDL (Verilog/VHDL) and a logic synthesis tool to produce a netlist-a description of the logic cells and their constraints. System Partitioning: Divide a large system into ASIC –sized pieces. Prelayout Simulation: Check to see if the design functions correctly. Floorplanning: Arrange the blocks of the netlist on the chip. Placement: Decide the collations of the cells and block.



Routing: Make the connections between cells and blocks. Extraction: Determine the resistance and capacitance of the interconnect.

Postlayout Simulation: Check to see the design still works with the added loads of the interconnect. Steps 1-4 are part of Logical Design. And steps 5-9 are part of Physical Design. There is some overlap. For example, system portioning might be considered as either logical or physical design. To put it another way, when performing system partitioning the main factor to be consider include both logical and physical factors.

Figure 4.20 ASIC Design Flow

VLSI DESIGN CMOS TESTING


UNIT-5

5.1 THE NEED FOR TESTING

While in real estate the refrain is “Location!” the comparable advice in IC design should be “Testing! Testing! Testing!” .While most problems in VLSI design has been reduced to algorithm in readily available software, the responsibilities for various levels of testing and testing methodology can be significant burden on the designer.

The yield of a particular IC was the number of good die divided by the total number of die per wafer. Due to the complexity of the manufacturing process not all die on a wafer correctly operate. Small imperfections in starting material, processing steps, or in photomasking may result in bridged connections or missing features. It is the aim of a test procedure to determine which die are good and should be used in end systems.

• Testing a die can occur: • At the wafer level • At the packaged level • At the board level • At the system level • In the field

By detecting a malfunctioning chip at an earlier level, the manufacturing cost may be kept low. For instance, the approximate cost to a company of detecting a fault at the above level is:

• Wafer $0.01- $.1 • Packaged-chip $0.10-$1 • Board $1-$10 • System $10-$100 • Field $100-$1000

Obviously, if faults can be detected at the wafer level, the cost of manufacturing is kept the lowest. In some circumstances, the cost to develop adequate tests at the wafer level, mixed signal requirements or speed considerations may require that further testing be done at the packaged-chip level or the board level. A component vendor can only test the wafer or chip level. Special systems, such as satellite-borne electronics, might be tested exhaustively at the system level. Tests may fall into two main categories. The first set of tests verifies that the chip performs its intended function; that is, that it performs a digital filtering function, acts as a microprocessor, or communicates using a particular protocol. In other



words, these tests assert that all the gates in the chip, acting in concert, achieve a desired function. These tests are usually used early in the design cycle to verify the functionality of the circuit. These will be called functionality tests. The second set of tests verifies that every gate and register in the chip functions correctly. These tests are used after the chip is manufactured to verify that the silicon in intact. They will be called manufacturing tests. In many cases these two sets of tests may be one and the same, although the natural flow of design usually has a designer considering function before manufacturing concerns.

5.1.1 Functionality Tests

Functionality tests are usually the first tests a designer might construct as part of the design process. For most systems, functionality tests involve proving that the circuit is functionally equivalent to some specification. That specification might be a verbal description, a plain-language textual specification, a description in some high-level computer language such as C, FORTRAN, Pascal, or Lisp or in a hardware-description language such as VHDL, ELLA, or Verilog or simply a table of inputs and required outputs. Functional equivalence involves running a simulator at some level on the two descriptions of the chip and ensuring for all inputs applied that the outputs are equivalent at some convenient checkpoints in time. The most detailed check might be on a cycle-by-cycle basis.

Functional equivalence may be carried out at various levels of the design hierarchy. If the description is in a behavior language, the behavior at the system level may be verifiable. For instance, in the case of a microprocessor, the operating system might be booted and key programs might be run for the behavioral description. However, this might be impractical for a gate-level model and extremely impractical for a transistor-level model. The way out of this impasse is to use the hierarchy inherent within a system to verify chips and modules within chips.

5.2 MANUFACTURING TEST PRINCIPLES

A critical factor in all LSI and VLSI design is the need to incorporate methods of testing circuits. This task should proceed concurrently with any architectural considerations and not be left until fabricated parts are available.

Figure 5.1(a) shows a combinational circuit with n-inputs. To test this circuit exhaustively a sequence of 2^n inputs must be applied and observed to fully exercise the circuit. This combinational circuit is converted to a sequential circuit with addition of m-storage registers, as shown in figure 5.1b the state of the circuit is determined by the inputs and the previous state. A minimum of 2^ (n+m) test vectors must be applied to exhaustively test the circuit. Clearly, this is an important area of design that has to be well understood.



n n

2^n inputs required to exhaustively test circuit (a)

n n m m Clk

2^(n+m) inputs required to exhaustively test circuit For n=25 m=50 1micro seconds/test, the test time is oner 1 billion years

(b) Figure 5.1 The combinational explosion in test vectors

5.2.1 FAULT MODELS 5.2.1.1 Stuck-At-Faults In order to deal with the existence of good and bad parts it is necessary to propose a “fault model”, that is, a model for how faults occur and their impact on circuits. The most popular model is called the “stuck-at” model. With this model, a faulty gate input is modeled as a “stuck at zero”(stuck-at-0,S-A-0,SA0) or “stuck at one” (Stuck-At-1,S-A-1,SA1). This model dates from board level designs where this was determined to be an adequate set of models for modeling faults. Figure 5.2 illustrates how an S-A-0 or S-A-1 fault might occur. These faults most frequently occur due to thin oxide shorts or metal-to-metal shorts.

Combinational

Logic

Combinational

Logic

Register



Figure 5.2 CMOS stuck-at faults

5.2.1.2 Short-Circuit and Open-Circuit Faults

Other models include “stuck-open” or “shorted” models. Two shorted faults are shown in figure 5.3. Considering the faults shown in figure5.3, the short s1 is modeled by an S-A-0 fault at input A, while short S2 modifies the function of the gate. What becomes evident is that to ensure the most accurate modeling, faults should be modeled at the transistor level, because it is only at this level that the complete circuit structure is known. For instance, in the case of a simple NAND gate, the intermediate node in the series n-pair is “hidden” by the schematic. What this implies is that test generations must be done in such a way as to take account of possible shorts and open circuits at the switch level. Although the switch level may be the most appropriate level, expediency dictates that most existing systems rely on Boolean logic representation of circuits and S-A-0 and S-A-1 fault modeling. A particular problem that arises with CMOS is that it is possible for a fault to convert a combinational circuit into a sequential circuit. This is illustrated for the case of a 2-input NOR gate in which one of the transistors is rendered ineffective (stuck open or stuck closed) in figure 5.4.This might be due to a missing source, drain or gate connection. If one of the n-transistors is stuck open, then the function displayed by the gate will be

F= (not (A+B)) + (A. (not B).Fn)

Where Fn is the previous state of the gate. Similarly if the B n-transistors drain connection is missing, the function is

F= (not (A+B)) + ((not A).B.Fn)

If either p-transistor is open, the node would be arbitrarily charged until one of the n-transistors discharged the node. Thereafter it would remain at zero, bar charge



leakage effects. This problem has caused researchers to search for new methods of test generation to detect such behavior.

Figure 5.3 CMOS bridging faults

Figure 5.4 A CMOS open fault causes sequential faults

Currently debate ranges over whether an SA0/SA1 approach to testing is adequate for testing CMOS. It is also possible to have switches exhibit a “stuck-open” or “stuck-closed” state. Stuck closed states can be detected by observing the static VDD current (IDD ) while applying test vectors. Consider the gate fault shown in Figure 5.5 where a p-transistor in a 2-input NAND gate is shorted. This could physically occur if stray



metal overlapped the source and drain connection or if the source and drain diffusions shorted. If the test vector 11 is applied to the A and B input and measure the static IDD

, the change to notice that it rises to some value determined by the β ratios of the n and p transistors. While the debate continues and test cycles are at a premium, the SA0/SA1 model will suffice for some time to come.

Figure 5.5 A defect that Causes Static IDD current

5.2.2 Observability

The observability of a particular internal circuit node is the degree to which one can observe that node at the outputs of an integrated circuit. This measure is of importance when a designer/tester desires to measure the output of a gate within a larger circuit to check that it operates correctly. Given a limited number of nodes that may be directly observed, it is the aim of well-designed chips to have easily observed gate outputs, and the adaption of some basic test design techniques can aid tremendously in this respect. Ideally, one should be able to observe directly or with moderate indirection every gate output within integrated circuit. While at one time this aim was hindered by limited gate-count processes and a lack of design methodology, current design practices and processes allow one to approach this ideal.

5.2.3 Controllability

The controllability of an internal circuit node within a chip is measure of the case of setting the node to a 1 or 0 state. This measure is of importance when assessing the degree of difficulty of testing a particular signal within a circuit. An easily controllable node would be directly settable via an input pad. A node with little



controllability might require many hundreds or thousands of cycles to get it to the right state. Often one finds it impossible to generate a test sequence to set a number of poorly controllable nodes into the right state. It should be the aim of a well-designed circuit to have all nodes easily controllable. In common with observability, the adoption of some simple design for test techniques can aid tremendously in this respect.

5.2.4 Fault Coverage

A measure of goodness of a test program is the amount of fault coverage it archives; that is, for the vectors applied, what percentages of the chip’s internal nodes were checked. Conceptually, the way in which the fault coverage is calculated is as follows. Each circuit node is taken in sequence and held to 0(S-A-0), and the circuit is simulated. Comparing the chip outputs with a known “good machines”-a circuit with no nodes artificially set to 0 (or 1).When a discrepancy is detected between the “faulty machine” and the good machine. The fault is marked as detected and the simulation is stopped. This is repeated for setting the node to 1 (S-A-1).In turn. Every node is stuck at 1 and 0, sequentially. The total number of nodes that, when set to 0 or 1, do result in the detection of the fault, divided by the total number of nodes in the circuit, is called the percentage-fault coverage.

The above method of the fault analysis is called sequential fault grading. While this might be practical for small circuits, or by using hardware simulation accelerators on medium circuits, the time to complete the fault grading may be very long. On average KN cycles assuming that, on average N/2 cycles are needed to detect each fault need to be simulated, where K is the number of nodes in the circuit and N is the length of test sequence. For K=1000 and N=12000, 12 million cycles are required. At 1ms per cycle, this yields 12,000 seconds or 3hrs 20 minutes. For K=100000 and `N=360,000, 3.6x10^9 cycles are required. At 1s per cycle, 1040 years would be required to do sequential fault grading. To overcome these long simulation times many ingenious techniques have been invented to deal with fault simulation.

5.2.5 Automatic Test Pattern Generation (Atpg)

In IC industry, designers designed circuits, layout drafts people completed the layout, and the test engineer wrote the tests . In many ways, the test engineers were the Sherlock Holmes of the industry, reverse engineering circuits and devising tests that would test the circuits in an adequate manner. For the longest time, test engineers implored circuit designers to include extra circuitry to ease the burden of test generation. As processes have increased in density and chips increased in complexity, the inclusion of test circuitry has become less of an overhead for both the designer and the manager worried about the cost of the die. In addition, as tools have improved, more of the burden for generating tests has



fallen on the circuit/logic designer. To deal with this burden, methods for automatically generating tests have been invented. Collectively these are known as ATPG, for Automatic Test Pattern Generation. In practice, one may find that ATPG is of great use in the generation of test vectors or that for a variety of reasons it is not applicable.

Most ATPG approaches have been based on simulation. A five-valued logic form is commonly used to implement test generate algorithms. This consists of the states 1, 0, D, D and X. 0 and 1 represent logical zero and logical one respectively. X represents the unknown or DON’T-CARE state. D represents a logic 1 in a good machine and a logic 0 in a faulty machine while D represents a logic 0 in a good machine and logic 1 in a faulty machine. The truth tables for inverters, AND, and OR gates are shown in tables 5.1, 5.2, and 5.3.

Table 5.1 Inverter Z=NOT A With the use of this five-valued logic by considering the circuit shown in fig5.6 where an S-A-0 fault is to be detected at node h. Alternatively call a circuit machine, which is customary in test momentclature. Thus node h would have value D. There are two objectives. The first is to propagate the D on node h to one or more primary outputs (Pos). A primary output is a directly observable signal, such as a pad or a scan output. This path to the primary output (or outputs) is called the sensitized path. The second objective is to set node h to state D via a set of primary inputs (PIs). A primary input is one that can be directly set via a pad or some other means. The gate driving node h is the Gate Under test or GUT. From node h we backtrack to the primary inputs (a, b, c, d, e) to find the necessary input vector required to set node to a 1. Because the gate driving node h is an AND gate from the above definition (a D is a 1 in a good machine), both inputs (f, g) have to be set to 1 to set h to 1. Proceeding further toward the inputs, to assert node f as a 1, both nodes a and b have to be set to a 1. Because node g is driven by an OR gate, either node c or node d need to be set to a 1 to assert node g.

A Z

0 1

1 0

X X

D D

D D



Table 5.2 2-input AND gate Z=A AND B

A B 0

1 X D D

0 0

1 X D D

1 1

1 1 1 1

X X

1 X X X

D D

1 X D 1

D D

1 X 1 D

Table 5.3 2-input OR gate Z=A OR B Thus a vector a,b,c,d of 1,1,1,0 or 1,1,0,1 is required to control node h. To observe that node g has been set to a D, input node e has to be set to a 1. Thus the resultant test vector is a,b,c,d,e=1,1,0,1,1 or 1,1,1,0,1. If we are checking for an S-A-1 fault at node h, we must be able to set it to 0. By similar reasoning to that for the S-A-0 case the test vector would be a,b,c,d,e=0,1,X,X,1 or 1,0,X,X,1 or 0,0,X,X,1 or [1,1,0,0,1]. Similarly, for other nodes a summary of the vectors is as in table 5.4

A B 0

1 X D D

0 0

0 0 0 0

1 0

1 X D D

X 0

X X X X

D 0

D X D 0

D 0

D X 0 D



NODE TEST VECTORa,b,c,d,e h S-A-0 1,1,0,1,1,1,1,1,0,1 h S-A-1 0,1,X,X,1,1,0,X,X,1,0,0,X,X,1,1,1,0,0,1 f S-A-0 1,1,0,1,1,1,1,1,0,1 f S-A-1 0,0,0,1,1,0,0,1,0,1 g S-A-0 1,1,0,1,1,1,1,1,0,1,1,1,1,1,1 g S-A-1 1,1,0,0,1

Table 5.4 Node-vector Summary of D-algorithm

Figure 5.6 The D-algorithm –sensitization step

Figure 5.7 Reconvergent fan-out with D notation

The next step is to collapse the vectors into least set that covers all nodes. A possible set is [1,1,0,1,1],[0,0,1,0,1],[1,1,0,0,1].

The reason for using a five-valued logic is shown in figure5.7. Here an additional AND gate and INVERT gate have been added to the circuit. From the figure 5.7 that a fault at node h is essentially unobservable. This circuit suffers from what is called reconvergent fan-out. The usual basis for manual generation of tests by test engineers and many current automatic test-pattern generation programs is the D-algorithm(DALG), PODEM and PODEM-X are improved algorithms that are more efficient than the original DALG and in addition treat error-correcting circuits composed of XOR gates with reconvergent fan-out. Another ATPG algorithm is called FAN and an improved efficiency algorithm dealing with



tristate drivers called ZALG has been developed. Other work has concentrated on dealing at a module level rather than the gate level. In basis these algorithms start by propagating the D value on an internal node to a primary output. This is called the D-propagation phase. The selection of which gates to pass through to the output is guided by observability indexes assigned to gates. At any particular gate input, the gate with the highest observability is selected. Once the D value is observable at a primary output, the next step is to determine the primary input values that are required to enable the fault to be observed and tested. This proceeds by backtracking from the faulted signal and sensitized path-enables toward the primary inputs. The selection of which path to proceed along toward the inputs is aided by controllability indices assigned to nodes. This is known as the backtrace step.

Controllabilities and observabilities can be assigned statically or dynamically. The SCOAP algorithm is one method of assigning controllabilities and observabilites. In the SCOAP system the following six testability measures are defined for each circuit node:

• CC0(n)-combinatorial 0 controllability of node 0 • CC1 (n)-combinatorial 1 controllability of node n. • C0(n) –combinatorial observability of node n • SC0(n)-sequential 0 controllability of node n • SC1(n) –sequential 1 controllability of node n • S0 (n) –sequential observability of node n.

The combinatorial measures are applied to the outputs of logic gates, while the sequential measures apply to registers and other “sequential” modules. As an example, for the AND gate shown in figure 5.8 the CC1 value is

CC1 (z) =CC1 (a) +CC1 (b) +1

Figure 5.8 NAND gate That is, the 1-controllability of the output of the AND gate is the sum of the 1-controllabilities of each input because each input has to be set to 1 to set the output to 1. the 1 is added at the end because the AND gate represents one stage of combinatorial logic. The sequential 1-controllability is given by

SC1 (z) =SC1 (a) +SC1 (b)



The combinatorial 0-controllability is given by

CC0 (z) =min [CC0 (a), CC0 (b)] +1

This arises due to the fact that either a 0 on a or b forces a 0 at the output. Therefore the easiest controllable input may be used. The sequential controllability is given by

SC0 (z) =min [SC0 (a), SC0 (b)],

The combinatorial observability of a is given by

C0 (a) =C0 (z) +CC1 (b) +1; That is, the observability of z added to the combinatorial 1-controllability of b. This occurs because b has to be forced to a 1 to make a observable. The sequential observability of a is given by

S0 (a) =S0 (z) +S0 (b)

Similar equations may be derived for other gate types. The SCOAP algorithm proceeds by first calculating the circuit controllabilities by propagating controllabilities from logic inputs. Following this, the observabilities are propagated from the logic outputs. Figure 5.9(a) shows a logic circuit with the 1-controllabities annotated. Figure 5.9(b) shows the observabilities.

Figure 5.9 SCOAP testability measure example (a) controllabilities (b) observabilities

In cases of multiple fan-out, the minimum observability measure is used. The presence of high controllability numbers a node that is difficult to control, while the presence of high observability numbers indicates nodes that are difficult to observe. As mentioned above, the testability measures are used to guide the



selection of paths in the D-propagation and backtrace phase of the D-algorithm based ATPG procedures. Other testability measures, such as COP and LEVEL are also used. COP testability measures are probabilistic in nature.

5.2.6 Fault Grading And Fault Simulation

Fault grading consists of two steps. First, the node to be faulted is selected. Normally global nodes such as reset lines and clock lines are excluded because faulting them can lead to unnecessary simulation. A simulation is run with no faults inserted, and the results of this simulation are saved. Following this process, in principle, each node or line to be faulted is set to 0 and then 1 and the test vector is applied. If, and when, a discrepancy is detected between the faulted circuit response and good circuit response, the fault is said to be detected and the simulation is stopped, and the process is repeated for the next node to be faulted. If the numbers of nodes to be faulted is K, and the average number of test vectors is N, the number of simulation cycles Sk is approximately

SK =(2*(N/2)*k)+N = K(N+1)=KN

This serial fault simulation process is therefore running K sets of the test vector set. With a small vector set, simple circuit, or very fast simulator, this approach is feasible. However, for large test sets and circuits, it is highly impractical.

To deal with this problem, a number of ideas have been developed to increase the speed of fault simulation.

Parallel simulation is one method for speeding up simulation of multiple machines. In this method m words in an n-bit computer are used to encode the state of n “machines” for a 2^m state machine. Two n-bit words may be used to encode n machines for a three-state simulation. More computer words may be used to encode simulators with more states. Moreover this principle has been extended to special-purpose hardware where the computer word length could be optimized to deal with substantially more circuits in parallel. Now if M circuits can be simulated in parallel, then

SK =KN/M

Concurrent simulation is currently the most popular method for software based fault simulation. The technique uses a nonfaulted version of the circuit to create a “good” machine model. Each fault creates a new faulty machine that is simulated parallel with the good machine. Thus N+1 simulations mat have to be completed, where N is the number of faults. Concurrent simulators rely on a number of heuristics to reduce the amount of simulation. For instance, when a difference is noted between a faulted machine and a good machine at an externally observable



point, the faulty machine is dropped from the simulation queue and the fault is “detected”. If the bad machines has an X or Z compared to a 1 or 0 for the good machine, the fault is “possible detect”. Obviously, the more externally observable nodes a circuit has, the quicker bad machines get dropped from the simulations. Normally, only the good machine state is stored, with each node listing the fault machines that differ with the good machine. The different state is often small, which implies that there is a small amount of extra simulation to be done. In other words most simulation for a faulty machine is exactly the same as the good machine. This is what concurrent simulation exploits. Fault collapsing occurs when two different faults result in the same faulty machine. This is noted, and one of the faulty machines may be dropped. Some machines performs static fault collapsing prior to simulation. For instance, an SA0 fault on the input of an inverter is the same as an SA1 fault at the output of the same inverter. With some fault simulators it is possible to create a fault dictionary.

This is a cross reference that maps an observed fault to a set of possible internal faults. It is of use when the tester wishes to track down the actual internal failure rather than just call the part. Apart from software based simulations, hardware-fault simulation accelerators that can provide a speedup over software based simulators are also available.

5.2.7 Delay Fault Testing

The fault models we have dealt with to this point have neglected timing. Failures that occur in CMOS could leave the functionality of the circuit untouched, but affect the timing.

Figure 5.10 An example of a delay fault For instance, consider the layout shown in figure5.10 for a high-power NAND gate composed of paralleled n and p transistors. If the link illustrated was opened, the gate would still function, but with increased pull-down time. In addition, the



fault now becomes sequential because the detection of the fault depends on the previous state of the gate and the simulation clock speed.

5.2.8 Statistical Fault Analysis

Conventional fault analysis can consume large CPU resources and take a long time. An alternative to this is what is called statistical fault analysis. This method of fault analysis relies on estimating the probability that a fault will be detected. In summary, a fault free simulation is performed on a circuit in which some extra statistics are gathered by a modified simulator on a per-input vector basis. These are as follows

• Zero-counter—the 0 count on each gate input when a 1->0 change of the output is detected.

• One –counter – the 1 count on each gate input when a 0->1 change of the output detected.

• Sensitization –counter – incremented if the input changes cause the output to be sensitized.

• Loop-counter – used to detect and deal with feedback. GATE TYPE B1 (l) B0(l) AND B1 (m).(C1(m)/C1(l)) B0(m).(S(l)-1(m))/C0(l) OR B1 (m).(S(l)-C0(m))/C1(l) B0(m).C0(m)/C0(l) NAND B0 (m).C0(m)/C1(l) B1(m).(S(l)-1(m))/C0(l) OR B0 (m).(S(l)-C1(m))/C1(l) B1(m).(C1(m)/C0(l)) NOT B0 (m) B1(m)

Table 5.5 Statistical Fault Analysis 1 And 0 Observations The one-controllability of line l is given by

C1 (l) = one-count/N

Where N is the number of vectors.

The zero-controllability is given by

C0 (l) =zero-count/N

The one-level sensitization probability is

S (l) = sensitization-count/N



The observabilities are calculated by propagating from gate outputs to gate inputs. For common gates, Jain and Agrawal derive the one-observabilites (B1) and zero-observabilities (B0) for common gates as shown in table 5.5

Methods also exist to deal with fan-out where two observabilities must be combined. Once these observability and controllability measures have been determined, the probability of fault detection may be calculated as follows

D1 (l) =B0 (l).C0 (l)

Where D1 (l) is the probability of detection that line l is SA1 D0 (l) =B1 (l). C1 (l)

Where D0 (l) is the probability of detection that line l is SA0

From these values the fault coverage of the circuit mat be calculated. The results of using this technique follow very closely the results generated by conventional fault simulation.

5.2.9 Fault Sampling

Another approach to fault analysis is known as fault sampling. This is used in circuits where it is impossible to fault every node in the circuit. Nodes are randomly selected and faulted. The resulting fault-detection rate may be statistically inferred from the number of faults that are detected in the fault set and the size of the set. As with all probabilistic methods it is important that the randomly selected faults be unbiased. Although this approach does not yield a specific level of fault coverage, it will determine whether the fault coverage exceeds a desired level. The level of confidence may be increased by increasing the no of samples.

5.3 DESIGN STRATEGIES FOR TEST

5.3.1 Design for Testability The key to designing circuits that are testable are the two concepts that introduced called controllability and observability. Restated, controllability is the ability to set (to 1) and reset (to 0) every node internal to the circuit. Observability is the ability to observe either directly or indirectly the state of any node in the circuit.

The three main approaches to what is commonly called design for testability. These may be categorized as:

• Ad-hoc testing

• Scan-based approaches

• Self-test and built-in testing



Following this, the application of these techniques to particular types of circuits. In this treatment we will look at:

• Random logic(multilevel standard-cell, two-level PLA)

• Regular logic arrays (data paths)

• Memories (RAM, ROM)

5.3.2 Ad-Hoc Testing

Ad-Hoc test techniques, as their name suggests, are collections of ideas aimed at reducing the combinational explosion of testing. Common techniques involve:

• Partitioning large sequential circuit

• Adding test points

• Adding multiplexers

• Providing for easy state reset

Long counters are good examples of circuits that can be tested by ad-hoc techniques. For instance imagine you have designed an 8-bit counter and want to test it. Figure 5.11(a) shows a naïve implementation in which the counter only has a RESET and a CLOCK input, with the terminal count (TC) being observable. The designer probably thought that a reset and 256 clock cycles, followed by the observation of TC, would be adequate for testing purposes. Apart from the nonobservability of the count value (Q<7:0>), the main problem is the number of cycles required to test a single counter. Possible ad-hoc test techniques are shown in figure 5.11(b) and figure 5.11(c). In figure 5.11(b), a parallel-load feature is added to the counter. This enables the counter to be preloaded with appropriate values to check the carry propagation within the counter. Another technique is to reduce the length of each counter to, say, 4 bits, as shown in figure 5.11(c). This is achieved by having the test signal block the carry propagate at every 4-bit boundary. With this method 16 vectors exhaustively can test each 4-bit section. The carry propagate between 4-bit sections may be tested with a few additional vectors.

Another technique classified in this category is the use of the bus in a bus-oriented system for test purposes. This is shown on figure 5.12(a) for a very simple accumulator. Each registers has been made loadable from the bus and capable of being driven onto the bus for testing purposes. A more general scheme is illustrated in figure 5.12(b), where the normally inaccessible inputs are set and the outputs are observed via the bus. Frequently, multiplexers may be used to provide alternative signal paths during testing. In CMOS transmission gate multiplexers provide low area and speed overhead. Figure 5.13(a) shows a scheme called a design for autonomous test,



which uses multiplexers. Figure 5.13(b) shows the circuit configured for normal use, while figure 5.13(c) shows the circuit configuration to test module A.

Figure 5.11 Ad-hoc test techniques applied to a counter

Any design should always have a method of resetting the internal state of the chip within a single cycle or at most a few cycles. Apart from making testing easier, this also makes simulation faster because a few cycles are required to initialize the chip.

In general, ad-hoc testing techniques represent a bag of tricks developed over the years by designers to avoid the overhead of a systems approach to testing, which will be described in the next section. While these general approaches are still quite valid, process densities and chip complexities necessitate a structured approach to testing.



Figure 5.12 Bus-oriented test techniques 5.3.3 Scan-Based Test Techniques

A collection of approaches have evolved for testing that lead to a structured approach to testability. The approaches stem from the basic tenets of controllability and observability.



Figure 5.13 Multiplexer based testing

5.3.3.1 Level Sensitive Scan Design (LSSD)

A popular approach is called Level Sensitive Scan Design or the LSSD approach, introduced by IBM. This is based on two tenets. First, that the circuit is level sensitive. The second principle of LSSD is that each register may be converted to a serial shift register.



The basic building block in LSSD is that Shift Register Latch or SRL. A block level implementation of a polarity hold SRL is shown in figure 5.14(a). It consists of two latches L1 and L2. L1 has a serial port data, I, and an enable, A. It also has a data port, D, and an enable, C. when A is high, the value of L1 (T1) is set by the value of I, while when C is high, L1 is set by D. A and C cannot be simultaneously high. When signal B in L2 is high, T1 is passed to T2. A gate-level implementation of the SRL is shown in figure 5.14(b) and 5.14(c). In normal operation, the D input is the normal input to the register, while the T2 signal is the output. L1 is the master while L2 is the slave. SRLs may be connected in series by using the T2 output and the I input of successive latches. During normal system operation, A is held low and C and B may be thought of as a two-phase nonoverlapping clock. When data is to be loaded into the SRL or dumped out of SRL, A and B are used as a two-phase shift clock.

Figure 5.14 A shift register latch

Figure 5.15(a) shows a typical LSSD scan system. An expanded view as shown in figure 5.15(b). The first rank of SRLs have inputs driven from a preceding stage and have outputs QA1, QA2 and QA3. These outputs feed a block of combinational logic. The output of this logic block feeds a second rank of SRLs with outputs QB1, QB2 and QB3. Figure 5.15(c) shows a typical clocking sequence. Initially the shift-clk and C2 are clocked three times to shift data into the first rank of SRLs (QA1-3). C1 is asserted, and then c2 is asserted, clocking the output of the logic block into the second rank of SRLs (QB1-3), shift-clk and



C2 are then clocked three times to shift QB1, QB2 and QB3 out via the serial-data-out line.

Figure 5.15 An LSSD scan chain: (a) basic architecture (b) example circuit

(c) example timing Testing proceeds in this manner of serially clocking the data through the SRLs to the right point in the manner of serially clocking the data through the SRLs to the right point in the circuit, running a single system clock cycle and serially clocking the data out for observation. In this scheme, every input to the combinational block may be controlled and every output may be observed. In addition, running a serial sequence of 1’s and 0’s through the SRLs can test them.



5.3.3.2 Serial Scan

Level Sensitive Scan went to great pains to provide a hazard-free latching scheme. Faster clock speeds and design for smaller overhead in the registers has to led to simplifications in the SRL that give up a little on the hazard front but retain the scan principle mentioned above. The hazard is moved inside the register, which with careful design can be guaranteed to be race free for a particular process and environment characteristics.

A schematic for a commonly used CMOS edge sensitive scan register is shown in figure5.16. A MUX is added before the master latch in a conventional D register. TE is the Test Enable pin and TI is the Test Input pin. When TE is enabled, T1 is closed into the register by rising edge of CLK. Figure 5.17 shows some circuit level diagram of CMOS SRL implementation. Figure 5.17(a) shows a frequently used implementation, which uses transmission gates to implement the multiplexers. The layout density overhead for this latch is minimal. In addition, because the addition of the testability MUX places two transmissions gates in series, the increase in delay is minimized. Two further implementation of the input MUX are shown in figure 5.17(b) and 5.17(c). Figure 5.17(b) shows the addition of only two transistors and s single control line. A register so implemented does have the normal problems associated with used single-polarity transmission gates. Alternatively, the checks may be gated, as shown in figure 5.17(c). While this minimizes transistors, it may lead to unacceptable hold-time constraints on the register. Because the signals applied to the master latch are delayed with respect to the main clock, the data has to be held for a longer time at the input.

Figure 5.16 A typical CMOS scan-register



Figure 5.17 Various CMOS scan-latch options

5.3.3.3 Partial Serial Scan

Quite often in a design, one may not find it area and speed-efficient to implement scan registers in every location where a register is used. This occurs for instance in signal-processing circuits where many pipeline registers might be used to achieve high speed. If these are in the data-flow section of the chip, then one can think of the logic that has to be tested as the logic with the pipeline registers removed. In this case only the input and output registers need to be scannable. This technique of testing is known as partial scan, and depends on the designer making decisions about which registers need to be made scannable.



Consider the design shown in figure 5.18. in a fault scan test strategy all registers would have to be scannable. A partial scan design is shown in figure 5.18(a) where only two registers have been made scannable (R6 and R3). In addition, these registers have the ability to hold their state dependent on a HOLD control. The part of the circuit that is being tested and monitored by the scan registers is shown in figure 5.18(b). it may be proven that, by holding the vectors at the input of the kernel for three clock cycles, the kernel may be represented by the combinational-equivalent circuit shown in figure 5.18(c). This circuit may be used by an ATPG program to generate test vectors.

Figure 5.18 The application of scan techniques to employ partial scan (a) pipeline circuit (b) kernal of pipeline circuit (c) combinational equivalent of

kernal



5.3.3.4 Parallel Scan

One can imagine that serial-scan chains can become quite long, and the loading and unloading sequence can dominate testing time. An extension of serial scan is called random-access or parallel scan.

The basic idea is shown in figure 5.19. Each register in the design is arranged on an imaginary grid where registers on common rows receive common data lines and registers in common columns receive common read and write signals. In the figure, an array of 2-by-2 registers as shown. The D and Q signals of the registers are connected to the normal circuit connections. Any registers output may be observed by enabling the appropriate column read line and setting the appropriate address on an output data multiplexer. Similarly, data may be written to any register.

Figure 5.20 shows a D-register implementation called a Cross-Controlled Latch. It consists of a normal CMOS master-slave edge-triggered register augmented by two small n-transistors, N1 and N2. When test-write-enable is high, and clk is low, the value of node Y (D) may be sensed on sense[i] via transistor N2. When test-write-enable is low, probe[j] is high, and clk is high, the value on sense[i] can be driven onto node Y. This is seen immediately at the output of the register. The net effect on the register-timing parameters of the extra transistors is to slightly increase the minimum clock-pulse width. The area impact for an ASIC based register is around 3%.

The large amount of observable outputs are compressed using signature analysis. The large number of observable outputs leads to very efficient concurrent-fault simulation.

Figure 5.19 Parallel scan –basic structure



Figure 5.20 Parallel scan register (a cross-controlled latch)

5.3.4 Self-Test Techniques Self-test and built-in test techniques, as their names suggest, rely on augmenting circuits to allow them to perform operations on themselves that prove correct operation.

5.3.4.1 Signature Analysis and BILBO

One method of incorporating a built-in test module is to use signature analysis or cyclic-redundancy checking. This involves the use of a pseudo random sequence generator (PRSG) to generate the input signals for a section of combinational circuitry and then using signature analyzer to observe the output signals.

A PRSG implements a polynomial of some length N. It is connected from a linear feedback shift register (LFSR), which is constructed, in turn from a number of 1-bit registers connected in serial fashion as shown in figure 5.21. The outputs of certain shift bits are XORed and fed back to the input of the LFSR to calculate the required polynomial. For instance, in figure 5.21 the 3-bit shift register is computing the polynomial f(x) =1+x+x^3. For an n-bit LFSR, the output will cycle through 2^n-1 states before repeating the sequence. Tables for determining suitable shift register (CFSR) includes the zero state, which may be required in some test situations.



Figure 5.21 Pseudo-random sequence generator (PRSG)

A signature analyzer is constructed by cyclically adding the outputs of a circuit to a register or an LFSR if successive logic blocks are to be tested in a like manner. A typical circuit is shown in fugre5.22 (a). As each test vector is run, the incoming data is XORed with the contents of the LFSR. At the end of a test sequence, the LFSR contains a number, known as the syndrome, which is a function of the current output and all previous outputs. This can be compared with the correct syndrome to determine whether the circuit is good or bad.

Signature analysis can be merged with the scan technique to create a structure known as BIBLO – for Built-In Logic Block Observation

Figure 5.22 Built-in logic block observation (BIBLO) (a) individual register

(b) use in a system



A 3-bit register is shown with the associated circuitry. In mode D (C0=C1=1), the registers act as conventional parallel registers. In mode A (C0=C1=0), the registers act as scan registers. In mode C (C0=1, C1=0) the registers act as a signature analyzer or pseudo-random sequence generator (PRSG). The registers are reset if C0=0 and C1=1. Thus a complete test-generation and observation arrangement can be implemented, as shown in Figure 5.22(b). In this case two sets of registers have been added in addition to some random logic to effect the test structure.

A chip set for FFT application was designed with local testing based on pseudo-random pattern generation and signature analysis. With a 28-bit pattern generator and a 17-bit signature at 10MHz it took 26 seconds to test the part.

5.3.4.2 Memory Self-Test

Embedding self-test circuits for memories in higher-speed circuits not only may be the way of testing the structures at speed but can save on the number of external test vectors that have to be run. A typical read/write memory (RAM) test program for an M-bit address memory might be as follows. FOR i=0 to M-1 write (data) FOR i=0 to M-1 read (data) then write (data) FOR i=0 to M-1 read (data) then write (data) FOR i=M-1 to 0 read (data) then write (data) FOR i=M-1 to 0 read (data) then write (data)

data is 1 and data is 0 for a 1-bit memory or a selected set of patterns for an n-bit word. For an 8-bit memory data might be x00, x55, x33, and x0F. An address counter, some multiplexers, and a simple-state machine result in a fairly low overhead self-test structure for read/write memories. The self-test consists of 256K cycles that input a checkerboard pattern to test for cell-to-cell interference. This is followed by 256K cycles in which the data is read out. Then a complemented checkerboard is written and read. A total of 1 million cycles provide a test sufficient for system maintenance.

ROM memories may be tested by placing a signature analyzer at the output of the ROM and incorporating a test mode that cycles through the contents of the ROM. A significant advantage of all self-test methods is that testing mat be completed when the part is in the field. With care, self-test may be performed during normal system operation.



5.3.4.3 Iterative logic array testing

Arrays of logic present an interesting problem to the test architect because the replication can be used to advantage in reducing the number of tests. In addition, by augmenting the logic extremely high fault coverage rates are possible. An iterative logic array (ILA) is a collection of identical logic modules (such as an n-bit adder). An ILA is C-testable if it can be tested with a constant number of input vectors independent of the iteration count. An ILA is I-testable if a particular fault that occurs in any module as a result of an applied input vector is identical for all modules in the ILA. Assuming that only one module is faulty, the detection of a fault may be made by using an equality test on the ILA outputs. 5.3.5 IDDQ testing

An increasingly popular method of testing for bridging faults is called is called IDDQ or current-supply monitoring. This relies on the fact that when a complementary CMOS logic gate is not switching, it draws no DC current. When a bridging fault occurs, for some combination of input conditions a measurable DC IDD will flow. Testing consists of applying the normal vectors, allowing the signals to settle, and then measuring IDD. To be effective any circuits that draw DC power such as pseudo-nMOS gates or analog circuits have to be disabled. Because many circuits now require SLEEP modes to reduce power, this may not be a substantial additional overhead. Because current measuring is slow, the tests must be run slower than normal, thus increasing the test time,. However, this technique gives a form of indirect massive observability at little circuit overhead.

5.4 CHIP-LEVEL TEST TECHNIQUES

In the past the design process was frequently divided between a designer who designed the circuit and a test engineer who designed the test to apply to that circuit. The advent of the ASIC, small design teams, the desire for reliable ICs, and rapid times to market have all forced the “test problem” earlier in the design cycle. In fact, the designer who is only thinking about what functionality has to be implemented and not about how to test the circuit will quite likely cause product deadlines to be slipped and in extreme cases precuts to be stillborn. In this section some practical methods of incorporating test requirements into a design. This discussion is structured around the main types of circuit structure that will be encountered in a digital CMOS chip.

5.4.1 Regular Logic Array

Partial serial scan or parallel scan is probably the best approach for structures such as datapaths. One approach that has been used in a Lisp microprocessor is shown in figure 5.23. Here the input busses may be driven by a serially loaded register. These in turn may be sourced onto a bus, and this bus may be loaded into



a register that may be serially accessed. All of the control signals to the datapath are also made scannable. 5.4.2 Memories

Memories may use the self-testing techniques mentioned in section 5.3.4.2. alternatively, the provision of multiplexers on data inputs and addresses and convenient external access to data outputs enables the testing of embedded memories. It is a mistake to have memories indirectly accessible (i.e., data is written by passing through logic, data is observed after passing through logic, addresses cannot be conveniently sequenced). Because memories have to be tested exhaustively, any overhead on writing and reading the memories can substantially increase the test time and, probably more significantly, turn the testing task into an effort inscrutability

Figure 5.23 Datapath test scheme



5.4.3 Random Logic Random logic is probably best tested via full serial scan or parallel scan.

5.5 SYSTEM-LEVEL TEST TECHNIQUES

Traditionally at the board level, “bed-of-nails” testers have been used to test boards. In this type of a tester, the board under test is lowered onto a set of test points that probe points of interest on the board. These may be sensed and driven to test the complete board. At the chassis level, software programs are frequently used to test a complete board set. For instance, when a computer boots, it might run a memory test on the installed memory to detect possible faults. The increasing complexity of boards and the movement to technologies like Multichip Modules (MCMs) and surface-mount technologies resulted in system designers agreeing on a unified scan-based methodology for testing chips at the board(and system level). This is called Boundary Scan.

5.5.1 Boundary Scan

5.5.1.1 Introduction

The IEEE 1149 Boundary Scan architecture is shown in figure 5.24. In essence it provides a standardized serial scan path through the I/O pins of an IC. At the board level, ICs obeying the standard may be connected in a variety of series and parallel combinations to enable testing of a complete board or, possibly, collection of boards. The description here is a précis of the published standard. The standard allows for the following types of tests to be run in a unified testing framework;

Figure 5.24 Boundary scan architecture



• Connectivity tests between components

• Sampling and setting chip I/Os

• Distribution and collection of self-test or built-in-test results

5.5.1.2 The Test Access Port (TAP)

The Test Access Port (TAP) is a definition of the interface that needs to be included in an IC to make it capable of being included in a Boundary-Scan architecture. The port has four or five single-bit connections, as follows

• TCK (The Test Clock Input) – used to clock tests into and out of chips. • TMS (The Test Mode Select) –used to control test operations. • TDI (The Test data Input) – used to input test data to a chip. • TDO (the Test Data Output) –used to output test data from a chip.

It also has an optional signal

• TRST (The Test Reset Signal) used to asynchronously reset the TAP controller, also used if a power-up reset signal is not available in the chip being tested.

The TDO signal is defined as a tristate signal that is only driven when the TAP controller is outputting test data.

5.5.1.3 The Test Architecture

The basic test architecture that must be implemented on a chip is shown in figure 5.25 it consists of:

• the TAP interface pins • a set of test-data registers to collect data from the chip • an instruction register to enable test inputs to be applied to the chip • a TAP controller, which interprets test instructions and controls the flow

of data onto and out of the TAP.

Data that is input via the TDI port may be fed to one or more test dat registers or an instruction register. An output MUX selects between the instruction register and the data registers to be output to the tristate TDO pin.



Figure 5.25 TAP architecture

5.5.1.4 The TAP controller

The TAP controller is a 16-state FSM that proceeds from state to state based on the TCK and TMS signals. It provides signals that control the test data registers, and the instruction register. These include serial-shift-clocks and update clocks. The state diagram shown in figure 5.26. The state adjacent to each state transition is that of the TMS signal at the rising edge of TCK. The reader is referred to the standard for complete descriptions of these states. It is probably best to understand them by examining a typical test sequence. Starting initially in the Test-Logic-Reset state, a low on TMS transitions the FSM to the Run-Test/Idle mode. Holding TMS high for the next three TCK cycles places the FSM in the Select-DR-Scan, Select-IR-Scan, and finally Capture-IR mode. In this mode two bits are input to the TDI Port and shifted into the instruction register. Asserting TMS for a cycle allows the instruction register to pause while serially loading to allow tests to be carried out. Asserting TMS for two cycles, allows the FSM to enter the Exit2-IR mode on exit from the Pause-IR state and then to enter the Update-IR mode where the Instruction Register is updated with the new IR value. Similar sequencing is used to load the data registers.

5.5.1.5 The Instruction Register (IR)

The instruction register has to be at least two bits long, and logic detecting the state of the instruction register has to decode at least three instructions, which are as follows;

• BYPASS – This instruction is represented by an IR having zeroes in it. It is used to bypass any serial-data registers in a chip with a 1-bit register. This allows specific chips to be tested in a serial-scan chain without having to shift through the accumulated SR stages in all the chips

• EXTEST –This instruction allows for the testing of off-chip circuitry and is represented by all ones in the IR.



• SAMPLE/PRELOAD –This instruction places the boundary-scan registers (i.e., at the chips I/O pins) in the DR chain, and samples or preloads the chips I/Os

In addition to these instructions, the following are also recommended:

• INTEST – This instruction allows for single-step testing of internal circuitry via the boundary-scan registers.

• RUNBIST –This instruction is used to run internal self-testing procedures within a chip.

Further instructions may be defined as needed to provide other testing functions. A typical IR bit is shown in figure 5.27

Figure 5.27 Instruction register bit implementation

5.5.1.6 Test-Data Registers

The test-data registers are used to set the inputs of modules to be tested, and to collect the results of running tests. The simplest data-register configuration would be a boundary-scan register (passing through all I/O pads) and a bypass register (1-bit long). Figure 5.28 shows a generalized view of the data registers where one internal data register has been added. A multiplexer under the control of the Tap controller selects which particular data register is routed to the TDO pin.

5.5.1.7 Boundary Scan Registers

The boundary scan register is a special case of a data register. It allows circuit-board interconnections to be tested, external components tested, and the state of chip digital I/Os to be sampled. Apart from the bypass register, it is the only data register required in a Boundary Scan compliant part.



Figure 5.28 TAP data registers

Figure 5.29 Boundary scan (a) input and (b) output cells

A single structure can be used for all I/O pad types, depending on the connections made to the cell. It consists of two multiplexers and two edge-triggered registers. Figure 5.29(a) shows this cell used as an input pad. Two register bits allow the serial shifting of data through the boundary-scan chain and the local storage of a data bit. This data bit may be directed to internal circuitry in the INTEST or RUNBIST modes (Mode=1). When Mode=0, the cell is in EXTEST or SAMPLE/PRELOAD mode. A further multiplexer under the control of shift DR controls the serial/parallel nature of the cell. The signal ClockDR and UpdateDR generated by the Tap Controller load the serial and parallel register, respectively. An output cell is shown in figure 5.29(b). When Mode=1, the cell is in EXTEST, INTEST, or RUNBIST modes, communicating the internal data to the output pad. When Mode=0, the cell is in the SAMPLE/PRELOAD mode.



Figure 5.30 Boundary scan tristate cell

Figure 5.31 Boundary scan bidirectional cell

Two output cells may be combined to form a tristate boundary-scan cell, as shown in figure 5.30. The output signal and tristate-enable each have their own muxes and registers. The Mode control is the same for the output-cell example. A bidirectional pin combines an input and tristate cell, as shown in figure 5.31

Download - Easy Learn to Verilog HDL

Top Related