Huijsing Operational Amplifiers

THE KLUWER INTERNATIONAL SERIES IN ENGINEERING AND COMPUTER SCIENCE
ANALOG CIRCUITS AND SIGNAL PROCESSING Consulting Editor: Mohammed Ismail. Ohio State University

Related Titles:

HIGH SPEED AID CONVERTERS: Understanding Data Converters Through SPICE A. Moscovici

ISBN: 0-7923-7276-X

ANALOG TEST SIGNAL GENERATION USING PERIODIC -ENCODED DATA STREAMS B. Dufort, G.W. Roberts

ISBN: 0-7923-7211-5

HIGH-ACCURACY CMOS SMART TEMPERATURE SENSORS A. Bakker, J. Huijsing

ISBN: 0-7923-7217-4

DESIGN, SIMULATION AND APPLICATIONS OF INDUCTORS AND TRANSFORMERS FORSiRFIcs

A.M. Niknejad, R.G. Meyer ISBN: 0-7923-7986-1 DESIGN AND IMPLEMENTATION B.E. Jonsson

ISBN: 0-7923-7871-7

RESEARCH PERSPECTIVES ON DYNAMIC TRANSLINEAR AND LOG-DOMAIN CIRCUITS

W.A. Serdijn, J. Mulder ISBN: 0-7923-7811-3

CMOS DATA CONVERTERS FOR COMMUNICATIONS M. Gustavsson, J. Wikner, N. Tan

ISBN: 0-7923-7780-X

DESIGN AND ANALYSIS OF INTEGRATOR-BASED LOG -DOMAIN FILTER CIRCUITS G.W. Roberts, V. W. Leung

ISBN: 0-7923-8699-X

VISION CHIP

A. Moini

ISBN: 0-7923-8664-7

COMPACT LOW-VOLTAGE AND HIGH-SPEED CMOS, BiCMOS AND BIPOLAR OPERATIONAL AMPLIFIERS

K-J. de Langen, J. Huijsing

ISBN: 0-7923-8623-X

CONTINUOUS-TIME DELTA-SIGMA MODULATORS FOR HIGH-SPEED AID CONVERTERS: Theory, Practice and Fundamental Performance Limits

J.A. Cherry, W. M. Snelgrove

ISBN: 0-7923-8625-6

LEARNING ON SILICON: Adaptive VLSI Neural Systems G. Cauwenberghs, M.A. Bayoumi

ISBN: 0-7923-8555-1

ANALOG LAYOUT GENERATION FOR PERFORMANCE AND MANUFACTURABILITY K. Lampaert, G. Gielen, W. Sansen

ISBN: 0-7923-8479-2

CMOS CURRENT AMPLIFIERS

G. Palmisano, G. Palumbo, S. Pennisi ISBN: 0-7923-8469-5

OPERATIONAL AMPLIFIERS

Theory and Design

by Johan Huijsing

Delft University of Technology, The Netherlands

KLUWER ACADEMIC PUBLISHERS BOSTON I DORDRECHTI LONDON

A C.I.P. Catalogue record for this book is available from the Library of Congress.

ISBN 0-7923-7284-0

Published by Kluwer Academic Publishers, P.O. Box 17, 3300 AA Dordrecht, The Netherlands.

Sold and distributed in North, Central and South America by Kluwer Academic Publishers,

101 Philip Drive, Norwell, MA 02061, U.S.A.

In all other countries, sold and distributed by Kluwer Academic Publishers,

P.O. Box 322. 3300 AH Dordrecht, The Netherlands.

Printed on acid-free paper

All Rights Reserved

© 200 I Kluwer Academic Publishers, Boston

No part of the material protected by this copyright notice may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without written permission from the copyright owner.

Printed in the Netherlands.

To my dear wife Willeke and children Hans, Adriaan, Mirjam, and Reineke, who have given me love and support.

ACKNOWLEDGEMENTS

Simon Middelhoek stimulated me to write this overview book. Wim van Nimwegen drew all the pictures in a very clear way. Anja de Koning and Mary van den Berg typed out my manuscript. Maureen Meekel provided computer support.

Wendy Sturrock and many students helped to correct this book. Thanks to all.

Above all, thanks to God through Jesus Christ, who is the Lord of my life.

CONTENTS:

SUMMARY xix

INTRODUCTION xxi

NOTATION xxiii

1. DEFINITION OF OPERATIONAL AMPLIFIERS 1

Nullor Concept. 1

Classification based on number of floating ports 2

1.1 Operational Inverting Amplifier (OIA) 3

Current-to- Voltage Converter 3

1.2 Operational Voltage Amplifier (OVA) 4

Non-inverting Voltage Amplifier 4

Voltage Follower (VF) 5

1.3 Operational Current Amplifier (OCA) 6

Current Amplifier 6

Current Followers (CF) 7

1.4 Operational Floating Amplifier (OFA) 8

Voltage-to-Current Converter 9

Voltage and Current Follower (VCF) 10

1.5 Conclusion 10

1.6 References 12

2. MACROMODELS 15

2.1 Operational Inverting Amplifier (OIA) 15

Definition of: offset voltage and current, input and

output impedance, gain 16

2.2 Operational Voltage Amplifier (OVA) 17

Definition of: input bias current, input common-mode

rejection ratio 17

2.3 Operational Current Amplifier (OCA) 19

Definition of: output bias current, output common-mode

current rejection ratio 19

2.4 Operational Floating Amplifier (OFA) 20

x

OPERATIONAL AMPLIFIERS, THEORY AND DESIGN

U sing all definitions 21

2.5 Macromodels in Spice 22

Macromodel mathematical 22

Macromodel Miller-compensated 23

Macromodel nested-Miller-compensated 24

Conclusion 26

2.6 Measurement Techniques for Operational Amplifiers 26

Gain measurement of an OT A 26

Gain measurement of an OpAmp 28

Gain and offset measurements of an OpAmp 29

General measurement setup for an OpAmp 30

2.7 Problems and Simulation Exercises 33

2.8 References 38

3. APPLICATIONS

3.1 Operational Inverting Amplifier 39

Current-to-Voltage Converter 40

Inverting Voltage Amplifier 41

3.2 Operational Voltage Amplifier 42

Non-Inverting Voltage Amplifier 42

Voltage Follower 44

Bridge Instrumentation Amplifier 44

3.3 Operational Current Amplifier 46

Current Amplifier 46

3.4 Operational Floating Amplifier 48


Inverting Current Amplifier 49

Differential Voltage-to-Current Converter 50

Instrumentation Voltage Amplifier 52

Instrumentation Current Amplifier 53

Gyrator Floating 54

Conclusion 56

3.5 Dynamic range 56

Dynamic range over supply-power ratio 56


Inverting Voltage Amplifier 59

CONTENTS

Xl

Non-Inverting Voltage Amplifier 60

Inverting Voltage Integrator 61

Current Mirror 62

Conclusion Current Mirror 63

Non-Ideal Operational Amplifiers 64

Conclusion 66

3.6 Problems 67

3.7 References 72

4. INPUT STAGES 75

4.1 Offset, Bias, and Drift 75

Isolation techniques 76

Balancing techniques 77

4.2 Noise 84



Conclusion 88

4.3 Common-Mode Rejection 88



Combination of isolation and balancing 90

Common-mode cross-talk ratios (CMCR) 91

Parallel input impedance 92

Collector or drain impedance 94

Tail impedance 95

Collector-base impedance 95

Base impedance 96

Back-gate influence 96

Total CMCR 98

Conclusion 99

4.4 Rail-to-rail Input Stages 100

Constant g m by constant sum of tail-currents 102

Constant gm by constant sum of roots of tail currents .. 105

Constant gm by spill-over control 106

Constant S,« in CMOS by saturation control 112

Constant gm in CMOS by multiple input stages 114

xu OPERATIONAL AMPLIFIERS, THEORY AND DESIGN

Constant gm in CMOS by constant sum of VGS 114

Rail-to-rail in CMOS by back-gate driving 118

Extension of the common-mode input range 118

Conclusion 118


4.6 References 128

5. OUTPUT STAGES 131

5.1 Power Efficiency of Output Stages 131

5.2 Classification of Output Stages 137

5.3 Feedforward Class-AB Biasing (FFB) 140

FFB Voltage Follower Output Stages 140

FFB Compound Output Stages 146

FFB Rail-to-Rail General-Amplifier Output Stages 149

Conclusion 160

5.4 Feedback Class-AB Biasing (FBB) 161

FBB Voltage-Follower Output Stages 162

FBB Compound Output Stages 163

FBB Rail-to-Rail General Amplifier Output Stages 169

Conclusion 176

5.5 Saturation Protection and Current Limitation 176

Output Saturation Protection Circuits 177

Output Current Limitation Circuits 179


5.7 References 191

6. OVERALL DESIGN 197

6.1 Classification of Overall Topologies 197

Nine overall Topologies 198

Voltage and current gain boosting 203

Input voltage and current compensation 204

6.2 Frequency Compensation 205

One-GA-Stage Frequency Compensation 207

No internal poles 210

Two-GA-Stage Frequency Compensation 211

Two-GA-stage Parallel Compensation (PC) 212

CONTENTS Xlll

Two-GA-stage Miller Compensation (MC) 215

Three-GA-Stage Frequency Compensation 224

Three-GA-Stage Nested Miller Compensation (NMC) .225 Three-GA-Stage Multipath Nested Miller

Compensation (MNMC) 228

Four-GA-Stage Frequency Compensation 232

Four-GA-Stage Hybrid Nested Miller Compensation

(HNMC) 232

Four-GA-Stage Multipath Hybrid Nested Miller

Compensation (MHNMC) 235

Four-GA-Stage conditionally stable MHNMC 237

Mu1ti-GA-Stage compensations 238

Reverse Nested Miller Compensation (RNMC) 238

Conclusion 238

6.3 Slew Rate 239

6.4 Non-Linear Distortion 242

Conclusion 249


6.6 References 258

7. DESIGN EXAMPLES 261

Nine overall topologies 261

7.1 GA-CF Configuration 262

Operational Transconductance Amplifier (OTA) 262

Folded-Cascode Operational Amplifier 265

Telescopic-Cascode Operational Amplifier 268

Feedforward HF compensation 270

Input voltage compensation 271

Input class-AB boosting 273

Voltage-gain boosting 275

Conclusion 276

7.2 GA-GA Configuration 277

Basic bipolar R-R-out class-A Operational Amplifier .. 277 Improved basic bipolar R-R-out class-A Operational

Amplifier 279

Basic CMOS R-R-out class-A Operational Amplifier .. 280

XIV OPERATIONAL AMPLIFIERS, THEORY AND DESIGN

Improved basic CMOS R-R-out class-A Operational

Amplifier 281

Conclusion 282

7.3 GA-CF-VF Configuration 283

High-Speed bipolar class-AB Operational Amplifier .. 283 High-slew-rate bipolar class-AB Voltage-Follower

Buffer 287

Conclusion 288

7.4 GA-GA-VF Configuration 288

General bipolar class-AB Operational Amplifier with

Miller Compensation 289

IlA 741 Operational Amplifier with Miller

Compensation (MC) 292

Conclusion 293

7.5 GA-CF-VF/GA Configuration 294

High-Frequency all-NPN Operational Amplifier with

mixed PC and MC 294

Conclusion 297

7.6 GA-GA-VF/GA Configuration 298

LMI0l class-AB all-NPN Operational Amplifier with

MC 298

NE5534 class-AB Operational Amplifier with

bypassed NMC 300

Precision all-NPN class-AB Operational Amplifier

with NMC 302

Precision HF all-NPN class-AB Operational Amplifier

with MNMC 305

IGHz, all-NPN class-AB Operational Amplifier with

MNMC 308

2Volt Power-efficient all-NPN class-AB Operational

Amplifier with MDNMC 308

Conclusion 310

7.7. GA-CF-GA Configuration 311

Compact 1.2Volt R-R-out CMOS class-A OpAmp with

MC 311

Compact 2Volt R-R-out CMOS class-AB OpAmp with

CONTENTS

xv

MC 315

Compact 2Volt R-R-in/out CMOS class-AB OpAmp

with MC 318

Compact 1.2Volt R-R-out CMOS class-AB OpAmp

with MC 322

Conclusion 324

7.8 GA-GA-GA Configuration 324

1 Volt R-R-out CMOS class-AB OpAmp with MNMC 324 Compact 1.2 Volt R-R-out BiCMOS class-AB OpAmp

with MNMC 329

ESD Protection 332

1.8 Volt R-R-in/out bipolar Class-AB OpAmp (NE5234)

with NMC 332

Conclusion 337

7.9 GA-GA-GA-GA Configuration 337

1 Volt R-R-in/out Bipolar c1ass-AB OpAmp with

MNMC 338

1.2 Volt R-R-out CMOS class-AB OpAmp with

MHNMC 344

Conclusion 348


7.11 References 358

8. FULLY DIFFERENTIAL OPERATIONAL

AMPLIFIERS 365

8.1 Fully Differential GA-CF Configuration 365

Fully differential CMOS OpAmp with linear-mode

CM-out control 366

Fully differential telescopic CMOS OpAmp with

linear-mode CM-out control 368

Fully differential CMOS OpAmp with LTP CM-out

control 369

Fully differential GA-CF CMOS OpAmp with

input-CM feedback CM-out control 371

Fully differential CMOS OpAmp with R-R buffered

resistive CM-out control 372

XVI OPERATIONAL AMPLIFIERS, THEORY AND DESIGN

8.2 Fully Differential GA-CF-GA Configuration 375

Fully differential CMOS OpAmp with R-R resistive

CM-out control 375

Conclusion 377

8.3 Fully Differential GA-GA-GA-GA Configuration .. 378 Fully differential CMOS OpAmp with switched-

capacitor CM-out control 378

Conclusion 379


8.5 References 385

9. OPERATIONAL FLOATING AMPLIFIERS (OFA) 387

9.1 Introduction 387

9.2 Unipolar Voltage-to-Current converter 389

Unipolar single-transistor V-I converter 391

Unipolar OpAmp-gain-boosted accurate V-I

converter 392

Unipolar CMOS accurate V -I converter 393

Unipolar bipolar accurate V-I converter 394

Unipolar OpAmp accurate V-I converter 395

Conclusion 396

9.3 Differential Voltage-to-Current converters 396

Differential simple V -I converter 396

Differential accurate V-I converter 397

Differential CMOS accurate V -I converter 398

9.4 Instrumentation Amplifiers 400

Instrumentation Amplifier (semi) with three OpAmps .400 Instrumentation Amplifier with a differential

V-I converter for input sensing 401

Instrumentation Amplifier with differential

V-I converters for input and output sensing 403

Instrumentation Amplifier with simple differential V-I

converters for input and output sensing 404

Instrumentation Amplifier (Bipolar) with common-

mode voltage range including negative rail voltage ... 406 Instrumentation Amplifier CMOS with common-mode

CONTENTS

XVll

voltage range including negative rail voltage 408

Instrumentation Amplifier simplified diagram and

general symbol 409

Conclusion 410

9.5 Universal class-AB voltage-to-current converter

design using an Instrumentation Amplifier 411

Universal V-I converter design with semi-

Instrumentation Amplifier 411

Universal V-I converter design with real

instrumentation amplifier 412

Conclusion 414

9.6 Universal class-A OFA design 414

Universal class-A OFA design with floating zener-

diodesupply 414

Universal class-A OFA design with supply current

followers 415

Universal class-A OFA design with long-tailed-pairs .. 417

Conclusion 422

9.7 Universal class-AB OFA realization with

power-supply isolation 423

Universal floating power supply design 424

Conclusion 424

9.8. Universal Class-AB OFA design 425

Universal class-AB OFA design with total-output-

supply-current equalization 426

Universal class-AB OFA design with current mirrors .. 429 Universal Class-AB OFA design with output-current

equalization 430

Universal class-AB voltage-to-current converter

with instrumentation amplifier 432

Conclusion 433

9.9 Problems 433

9.10 References 439

BIOGRAPHY 443

XVlll OPERATIONAL AMPLIFIERS, THEORY AND DESIGN

INDEX 445

SUMMARY

A systematic circuit design of operational amplifiers is presented. It is shown that the topology of all operational amplifiers can be divided in nine main overall configurations. These configurations range from one gain stage up to four or more gain stages. Many famous designs are completely evaluated.

High- frequency compensation techniques are presented for all nine configurations. Special focus is on low-power low-voltage architectures with rail-to-rail input and output ranges.

The design of fully differential operational amplifiers and operational floating amplifiers is being developed. Also, the characterization of operational amplifiers by macromodels and error matrices is presented, together with measurement techniques for their parameters.

Problems and simulation exercises have been supplied for self-evaluation.

INTRODUCTION

The goal of this book IS to equip the circuit designer with a proper understanding of the theory and design of operational amplifiers (OpAmps). The core of the book presents the systematically design of operational amplifiers. All operational amplifiers can be classified into a periodic system of nine main overall configurations. This division enables the designer to quickly recognise, understand, and choose optimal configurations.

Chapter 1 defines four basic types of operational amplifiers on the basis of the external ground connections of the input and output ports of generalized linear active network elements. Whether an input or output port needs to be isolated from ground has a big impact on the circuit design of the input and output stages, as will be shown in later chapters.

A complete set of linear parameters, by which each of the above four basic types of operational amplifiers can be quantified, is given in Chapter 2. This provides the reader with a sense of which parameters are most important. Chapter 2 also presents macromodels and measurement techniques for OpAmp parameters.

A systematic treatment of sources of errors in important applications of the above four basic types of operational amplifiers is presented in Chapter 3.

Input stages are evaluated in Chapter 4. Important aspects such as bias, offset, noise, and common-mode rejection are considered. Low-voltage input stages with a rail-to-rail input voltage range are extensively discussed.

A classification of push-pull output stages is presented in Chapter 5. Three possible topologies are explored: voltage follower stages, compound stages, and rail-to-rail general amplifier stages. Designs are presented with feedforward and feedback biasing class-AB techniques.

XXll OPERATIONAL AMPLIFIERS, THEORY AND DESIGN

Emphasis is on voltage and current efficiency.

A classification of operational amplifiers into nine main overall configurations is presented in Chapter 6. The classification consists of two two-stage OpAmps, six three-stage OpAmps, and one four- or multistage OpAmp. High-frequency compensation techniques are developed for all configurations ranging from one gain stage up to four or more gain stages. Methods are presented for obtaining a maximum bandwidth over power ratio for certain capacitive load conditions. Slew-rate and distortion are also considered.

Chapter 7 presents design examples of each of the nine main configurations. Many well-known OpAmps are fully elaborated. Among them are simple CMOS OpAmps, high-frequency bipolar OpAmps, Precision bipolar and BiCMOS OpAmps, low-voltage CMOS and bipolar OpAmps, and OpAmps with a high output drive capability in CMOS as well as in BiCMOS technology.

The design of fully differential operational amplifiers with common-mode feedback is developed in Chapter 8. Special focus is on low-voltage architectures.

When the output port as well as input port are designed such that they are both isolated from ground, the most universal linear active network element is created: the operational floating amplifier. The concept of this OpAmp gives the designer the freedom to work with current signals as well as voltage signals. Realizations of operational floating amplifiers are developed in Chapter 9 also in relation to instrumentation amplifiers.

Problems and simulation exercises have been supplied for most of the chapters to facilitate self-evaluation of the understanding and design skills of the user of this book.

NOTATION

OpAmp OA OIA OVA OCA OFA GA

VF

CF CM

IA

a Av Ava

f3

Bv C

Cox CM Cp D

f fT fa

gm i I IB Ie ID IE Is IQ

operational amplifier operational amplifier operational inverting amplifier operational voltage amplifier operational current amplifier operational floating amplifier general amplifier stage

voltage follower stage

current follower stage

current mirror stage instrumentation amplifier

temperature coefficient voltage gain

DC voltage gain

current gain of bipolar transistor voltage attenuation of feedback network capacitor value

specific capacitance of gate oxide Miller capacitor value

parallel capacitor value

distortion

frequency

transit frequency of a transistor zero-dB frequency transconductance of a transistor small-signal current

current

bias current collector current drain current emitter current supply current quiescent current

XXIV


k K L M R S Sr T

Q

v

V VB Vee VDD

VEE Ve Ves VeT Vs VSN

VSP Vss VT VTH W

""

Boltzman I S Constant = ""Cox WIL

length of gate in MOS transistors CMOS transistor

resistor value

signal

slew rate

generalized transistor bipolar transistor small-signal voltage voltage

bias voltage

positive supply voltage with bipolar transistors positive supply voltage with MOS transistors negative supply voltage with bipolar transistors generator voltage

gate-source voltage

active gate-source voltage (Ves- VTH) total-supply voltage

negative supply voltage

positive supply voltage

negative supply voltage with MOS transistors thermal voltage kT/q

threshold voltage of MOS device width of gate in MOS transistors mobility of change carriers

Extrinsic device parameters

RL

CL CM RDRe ReRB RsRE

NOTATION

xxv

Intrinsic Small-signal transistor parameters

r; r; ro r; r;

r, r,

Cds c:

Cgs Cbe

e; s; /LII /Lp

fJll fJp

1. DEFINITION OF OPERATIONAL AMPLIFIERS

Nullor Concept

In 1954 Tellegen introduced the concept of a universal active network element under the name of "ideal amplifier" [1. 1]. The name "nullor" , generally accepted now, was given to it by Carlin in 1964 [1.2]. The symbol of a nullor is shown in Fig. 1.1.

+ +

,---0-~ ~-0--~

Vi Ii) 0 nullor co (10 Vo

passive network

Fig. 1.1: A two-port network composed of a passrve network and a nullor

The nullor is defined as a two-port network element whose ports are called input and output ports and whose input voltage V; and input current I; are both zero, so:

(1.1)

The nullor concept only has significance if a passive network external to the nullor provides for a feedback from the output port into the input port [1.3]. The output voltage Vo and the output current 10 will be determined by the passive network elements in such a way that the input requirements V;=O, 1;=0 are satisfied.

An accurate signal transfer requires, firstly, accurate passive components and secondly, a practical nullor realization which approximates V; = 0, 1;=0.

2


This implies that the nullor realization should have a high gain, a low input noise, and low offset voltage and current (see Chap. 2.1). All linear and non-linear analog transfer functions can be implemented with nullor realizations and passive components.

Classification based on number of floating ports

We will now classify four nullor types on the basis of the number of ports which are floating, beginning with both ports grounded and ending with both ports floating. There are two main reasons for this kind of classification. Firstly, the larger the number of ports which are grounded the simpler the construction of the active device will be. Secondly, the larger the number of grounded ports the lower the number of possible feedback topologies will be allowed.

We will give each of the four nullor types a name which will be explained later. The first one with two grounded ports will be called operational inverting amplifier (OIA). The second one with the input port floating and output port grounded will be called operational voltage amplifier (OVA). The third one with the input port grounded and output port floating will be called operational current amplifier (OCA). Finally, the fourth one with both ports floating will be called operational floating amplifier (OF A). The adjective "operational" was coined by John R. Ragazzini and his colleagues in a paper [1.4] published in 1947. That paper described the basic properties of an OIA used with linear and nonlinear feedback. The adjectives "inverting" (I), "voltage" (V), "current" (C), or "floating" (F), are given by the present author to distinguish the four types of Operational Amplifiers according to their most striking attribute, as we will see in the next sections of this chapter. The most popular one, the OVA will be shortened to OpAmp in most parts of this book, where the distinction between the different types is not needed.


3

1.1 Operational Inverting Amplifier (OIA)

A practical approximation of a nullor having both ports grounded will be called an "operational inverting amplifier" (OIA). The grounded input port makes the construction of the input stage relatively easy, because it only needs to function at one voltage level [1.4] [1. 5]. Similarly, the grounded output port makes it relatively easy to construct an output stage having a high power efficiency, because the current return path can be directly connected to the grounded supply voltages. The negative sign (inverting) of the amplification factor makes it possible to obtain stable negative feedback with passive components connected directly from the output to the input port. The parallel connection of the feedback circuit at the input and output of the amplifier results in a low virtual entrance impedance (see Chap. 3.1), suitable for accurate current sensing at virtual zero input power, and a low exit impedance, suitable for obtaining an accurate output voltage.

Current-to- Voltage Converter

The most simple application of an OIA is the transimpedance amplifier or current-to-voltage transactor. This circuit is shown in Fig. 1. 1. La with a symbol of a nullor and in Fig. 1.1.1. b with a practical symbol of an OIA.

Fig. 1. 1. La: Transimpedance amplifier with a nullor symbol

Fig. 1. 1. 1.b: Transimpedance amplifier with an OIA symbol

The current-to-voltage transfer factor

(1.2.1)

4 OPERATIONAL AMPLIFIERS, THEORY AND DESIGN

(1.1.1)

can be accurately determined by -Z1 if the OIA satisfies two requirements: firstly, a high gain, and secondly, a low input offset voltage and offset current. A high gain also assures low entrance and exit impedances.

1.2 Operational Voltage Amplifier (OVA)

A practical nullor approach having only the output port grounded and the input port floating [l.6, 1.7] will be called "operational voltage amplifier" (OVA) or OpAmp. Currently, it is the most widely applied universal active device. The floating character of the input port imposes special demands on the construction of the input circuit, as will be discussed in Chapter 4.3 and 4.4. The floating input port allows series coupling of negative feedback.

This results in a high entrance impedance suitable for accurate voltage sensing at virtual zero input power. The parallel coupling of the feedback network with the grounded output port assures a low exit impedance.

Non-inverting Voltage Amplifier

The most essential application of the OVA is the non-inverting voltage amplifier or voltage-to-voltage transactor. The circuit is drawn in Fig. 1.2. La with a nullor symbol and in Fig. 1.2.1.b with a practical amplifier symbol for an OVA.

The voltage amplification factor

can be accurately determined by the impedance ratio (Z] + Z2)/Z2 if the OVA satisfies the requirements: a high gain, a low input offset voltage and current, independent of the common-mode voltage of the input port,


5

and a low input bias current. A high gain assures a high entrance impedance and a low exit impedance.

I2

--

+

I1 nullor

---

(I)--t--I 0 co

Fig. 1.2.1.a: Voltage amplifier with a nullor symbol

Voltage Follower (VF)

Fig. 1.2.1.b: Voltage amplifier with an OV A symbol

A special situation occurs if the OVA has its negative input terminal connected with the output terminal. We will call such a device a "voltage follower" (VF), because the exit voltage follows the entrance voltage. The construction of a universal active device with this connection may be simpler than without this connection, because no voltage shifting is required between the input and output. The VF circuit is given in Fig. 1.2.2.b with an OVA symbol.

I2
--
I1 ¢
--- +
co
+
v, v2 v1 v2 Fig. 1.2.2.a: Voltage follower with a nullor symbol

Fig. 1.2.2.b: Voltage follower with an OVA symbol

(1.2.2)

6


The voltage follower has the unique property that the voltage amplification factor

precisely equals plus unity, independently of any passive components, if the amplifier satisfies the three requirements: high gain, low input offset voltage and current, and a low input bias current. The accuracy of the plus-unity voltage transfer is not limited by the tolerances of any passive components. Note that the accuracy of the minus-unity voltage transfer of a voltage inverter does depend on the tolerance of a ratio of two impedances, as shown in Chapter 3.1, Fig. 3.1.2.

The voltage follower uses the most important attribute of a floating input port, viz. that the potential at one input terminal precisely follows the potential at the other input terminal.

1.3 Operational Current Amplifier (OCA)

A nullor approximation which has only the input port grounded and the output port floating [1.8] will be called an "operational current amplifier" (OCA). An output port with a floating character is difficult to construct, as we will see in Chapter 9. However, this labour is rewarded for applications requiring a high output impedance by using feedback in series coupling with the output port. This series feedback results in an exit with a current-source character, while the grounded input port with parallel feedback assures a low entrance impedance.

Current Amplifier

The most elementary application of the OCA is a current amplifier, whose circuit is shown in Fig. 1.3.l.b with an OCA symbol. The amplifier is the current dualogon of the voltage amplifier of Fig. 1.3.1.b. The amplification factor


7

(1.3.1)

is accurately determined by the admittance ratio (Yj + Y~/Yj if the amplifier satisfies: a high gain, a low input offset voltage and current, and an output port with a low output bias current, because this current is directly added to the output. Note, that the minus sign merely results from the choice of the opposite current notations of I, and 12, The low entrance impedance allows current sensing at a low entrance voltage Vj' The current source character at the exit yields an accurate current transfer independently of the load impedance.

I1

---

nullor I2

---

o co 1--1--0

+

Fig. 1.3.1.a: Current amplifier with a nullor symbol

Fig. 1.3.1.b: Current amplifier with an OCA symbol

Current Followers (CF)

A special situation occurs if the negative input terminal of the OCA is connected with the output terminal. We will call such a configuration a "current follower" (CF), because the exit current follows the entrance current. This circuit is the current dualogon of the voltage follower. The circuit is drawn in Fig. 1.3.2.a with a nullor symbol and in Fig. 1.3.2.b with an OCA symbol.

The current follower has the unique attribute that the current amplification factor

(1.3.2)

8


precisely equals plus unity, independently of any passive component values, if the gain is high, the input offset voltage and current is low, and if the output port has a low bias current. In contrast, the currentamplification factor of a current mirror, which nominally is minus unity, does depend on the matching of two passive elements (see Chap. 3.4).

Fig. 1.3.2.a: Current follower with a nullor symbol

Fig. 1.3.2.b: Current follower with an OCA symbol

Note that the minus sign in Eq. 0.3.2) is needed because the output current 12 is defined in the opposite direction regarding 120 when the current is being transferred through the CF.

The current-follower action reveals the most important attribute of a floating output port, namely that the current which flows into one output terminal is precisely followed by the current which flows out of the other output terminal. This attribute is the very dualogon of the voltage-follower action of a floating input port.

1.4 Operational Floating Amplifier (OF A)

A nullor approximation which has both the input and the output ports floating [1.9] will be called an "operational floating amplifier" (OFA) [1. 8, 1. 9]. Earlier it was called a "monolithic nullor" [1.10], or second generation current conveyer [1. 11]. The construction of such a universal active device combines the demands of both floating input and output


9

ports.

The OF A provides the maximum freedom for composing feedback configurations. With simple passive components it is possible to apply negative feedback in series with input and output ports, which results in both a high entrance and exit impedance.

Voltage-to-Current Converter

A specific application of the OF A is the voltage-to-current converter or trans admittance amplifier. Such a circuit is shown in Fig. l.4.1.a with a nullor symbol and in Fig. 1.4.1. b with a practical OF A symbol. The voltage-to-current transfer factor

(1.4.1)

will be accurately determined by one admittance - Yj if the amplifier satisfies four requirements: high gain, low input offset voltage and current, low input bias current, and low output bias current. The negative feedback in series with both ports ensures a high entrance impedance and a high exit impedance, which gives the transactor a voltage-sensing entrance and a current -source exit character.

nullor o

+

+

Fig. 1.4.1.a:Transadmittance amplifier with a nullor symbol

Fig. 1.4.1.6: Transadmittance amplifier with an OF A symbol


Voltage and Current Follower (VCF)

In fact, the trans admittance amplifier of Fig. 1.4.1.a/b does not apply all potentialities of the OF A. It is applied in the special case in which the lower terminal of the input port is connected with the lower terminal of the output port. This connection may simplify the construction of the OF A, because no voltage level shifter is needed between the input circuit and one output terminal, as we will see in Chapter 9.2. An OFA with this connection can be called a "voltage and current follower" (VCF).

A nullor which has this connection is also called a "three-terminal nullor" or a "unitor" [1.12]. In Fig. l.4.1.b the VCF firstly acts like a voltage follower, accurately transferring the entrance voltage Vj towards the voltage Vy on the upper side of the admittance Yj, and secondly like a current follower, accurately transferring the current I, through the admittance Yj towards the current 12 at the upper exit terminal.

1.5 Conclusion

A classification of universal active devices has been given on the basis of the number of ports which are connected to ground or to each other. The more ports not internally connected, the more freedom there is in the choice of the feedback configuration although this creates more complications with the construction of the device. Fig. 1. 5.1 presents an overview of the four types of active devices with different grounding schemes while Fig. 1.5.2 gives the three types of followers with one interconnection between the ports.

Fig. 1.5.1.a: Operational inverting amplifier (OIA)

Fig. 1.5.1.b: Operational voltage amplifier (OVA) or OpAmp


11

Fig. 1.5.1.c:, Operational current amplifier (OCA)

Fig. 1.5.1.d: Operational floating amplifier (OF A)

Fig. 1.5.2.a: Voltage follower (VF) with an OV A symbol

Fig. 1.5.2.b: Current follower (CF) with an OCA symbol

Fig. 1.5.2.c: Voltage and current follower (VCF) with an OFA symbol

Four transfer functions are particularly suited to accurate signal transfer. They can be implemented with the four basic types of active devices:

Current-to-voltage converter with an OIA Voltage-to-voltage converter with an OVA Current-to-current converter with an OCA Voltage-to-current converter with an OFA


In the ideal case, their signal transfer is independent of the source and load impedances. Moreover, the signal transfer depends on the theoretical minimum number of passive components.

Three transfer functions have the unique attribute that their accuracy is independent of any passive components. These types are:

Voltage follower (V F) with an OVA Current follower (CF) with an OCA

Voltage and current follower (VCF) with an OF A

1. high gain, with a stable high-frequency close-loop feedback

behavior,

2. low input offset and noise voltage and current,

3. when a floating input port is needed, a low input bias current,

4. when a floating output is needed, a low output bias current.

If each of these seven types of functions were realized with nullor approximations which do not have the right ports floating, a larger number of passive components and/or active devices is needed. Such realizations are less accurate and usually more expensive than the realizations with the right kind of active devices.

The requirements to be satisfied by the universal active devices are:

It is remarkable that the grades of perfection of the four requirements mentioned have no absolute limits. This means that there is a large area of technical exploration present for the design of active electronic building blocks.

1.6 References

[1.1] Tellegen, B.D.H.

"La recherche pour une sene complete d'elements de circuit ideaux non-lineaires"


13

Rendicouti Del Seminario Matermatico e Fisico di Milano, April, 1954, vol. 25, pp. 134-144.

[1.2] Carlin, H.J.

"Singular network elements"

IEEE Trans. Circuit Theory, March 1964, vol CT-ll, pp. 67-72.

[1.3] Tellegen, B.D.H.

"On nullators and norators"

IEEE Trans. Circuit Theory, Dec. 1966, vol CT-13, pp. 466-469.

[1.4] Ragazzini, K.R. et al.

"Analysis of problems in dynamics by electronic circuits"

Proc. IRE, May, 1947.

[1.5] Korn, G.A. and Korn, F.M.

"Electronic analog and hybrid computers" McGraw Hill Book Company, New York, 1964.

[1.6] Graeme, J.G. et al.

"Operational amplifiers, design and applications" McGraw Hill Book Company, New York, 1971.

[1.7] Graeme, J.G.

"Applications of operational amplifiers, third generation techniques"

McGraw Hill Book Company, New York, 1973.

[1.8] Huijsing, J.H.

"Operational Floating Amplifier"

IEE Proceedings, Vol. 137, Pt.G, No.2, April 1990, pp. 131- 136.



"Design and Applications of the Operational Floating Amplifier (OFA): The most Universal Operational Amplifier".

Journal of AICSP, Kluwer Ac. Publ., Boston, 4, 1993, pp. 115- 129.

[1.10] Huijsing, J.H. and Korte, J. de

"monolithic nullor - a universal active network element" IEEE Journal of Solid State Circuits, vol.SC-12,

Feb. 1977, pp. 59-64.

[1.11] Sedra, A.S. and Smith, K.C.

"A second generation current conveyer and its applications" IEEE Trans., 1970, CT-17, pp. 132-134.

[1.12] Keen, A.W.

"A topological nonreciprocal network element" Proc. IRE 47, 1959, pp. 1148-1150.

2. MACROMODELS

The qualities of the universal active devices mentioned in Chapter 1 can be specified by their macromodels or equivalent circuits and by transfer matrices. These representations should contain all elements for quantifying the four qualities of gain, offset, and if applicable, the bias current of input and output ports. Macromodels may also include the HF parameters and non-linear effects.

The first four sections of this chapter contain only a linear representation of the macromodels of the four types of OpAmps. The elements may include a complex description to include the HF behavior.

Non-linear behavior will be represented by SPICE macromodels for OpAmps in Section 2.5.

Measurement techniques for Operational Amplifiers are discussed III Section 2.6.

2.1 Operational Inverting Amplifier (OIA)

The operational inverting amplifier is a three-terminal network in which one terminal is grounded. The equivalent circuit of Fig. 2.1.1 contains all elements of a uni-lateral driven source

Fig. 2.1.1: Equivalent circuit of an operational inverting amplifier (OIA)


Definition of: offset voltage and current, input and output impedance, gain

For the purpose of standardization, all elements are chosen such that the currents at the terminals can be expressed in the voltages at the terminals. Thus, most elements are admittances. The main element IS a voltage-controlled current source with a transadmittance ~.

Together with the input admittance Y; and output admittance Yo> these three admittances form the essential ingredients of an active device. The admittances may be taken as a Laplace transform to represent the high-frequency response. No elements are added to represent internal feedback because of the reasonable simplification that a universal active device with more than one internal cascaded amplifier stage is unilateral. And if there is internal feedback, for instance by heat transfer on the chip, this should be considered separately.

An offset voltage source ~OffS and an offset current source lioffs have been added on the input side. They represent all additive DC errors of the device. In addition, they may include all noise quantities. The input voltage after subtraction of the offset voltage has been denoted by ~ * (with asterisk).

Vi lIYt Y)Yt 10 Vio/fs

+

Ii Y/Yt YlalYt Vo lio/fs

(2.1.1)

The errors of an active device application directly follow from the deviations from the input requirements ~=O and Ii=O. For that reason, we will choose a mathematical description of the device which directly produces the deviation from the ideal behavior. Such a description for each of the different kinds of devices will be called an error matrix. In the case of an OIA, being a two-port element, the error matrix is equal to the following chain matrix:

The error matrix clearly shows the requirements to be met by the OIA.

2. MACROMODELS

17

Firstly, a high transconductance 1'; is needed. This also includes a high voltage amplification factor Av=-1';IYo and a high current amplification factor Ai = 1';1Y;. Secondly, the input offset voltage ~OffS and current Iioffs should be low, and so should be the spectral input noise voltage ~n and current lin, which denote the noise components of the offset quantities.

2.2 Operational Voltage Amplifier (OV A)

The operational voltage amplifier is a four-terminal network of which one terminal is grounded. An equivalent circuit is drawn in Fig. 2.2.1. For an adequate description of the floating character of the input port we should distinguish the differential-mode (DM) input voltage ~d= ~l- ~2 and current lid = (Iil-Ir)12 from the common-mode (CM) input voltage ~c = (~l + ~)/2 and current lie = (11 + 1)12.

Ii1 Vi offs

--- + Vi1 0---+--+-,.---,,------,----,

+

Vid

Fig. 2.2.1: Equivalent circuit of an operational voltage amplifier (OVA)

Definition of: input bias current, input common-mode rejection ratio

The basic three elements, the admittances 1';, Y;dd and Yo, are the normal ones in each active device. They have already been discussed along with the equivalent circuit of the OIA. In addition, there are the commonmode input admittances 1';cl and 1';c2 and two common-mode input bias-current sources with an equal value Iibias. Furthermore, the CM input


voltage V:c * causes an output current V:c *y; IHi in addition to the current V:d*Y; of the main voltage-controlled current source. This is the result of a crosstalk of the CM input voltage V:c * on the DM input voltage V:d*' The factor Hi is called the "common-mode rejection ratio" (CMRR) Hi = ( oVi/ oVid to.v»- The reciprocal factor 1 /H, can be called the "common-mode crosstalk ratio" (CMCR). All these additional elements show the non-ideal floating character of the input port.

The main errors of an OVA are quantified by the following error matrix (2.2.1):

V .. Ff

IOJJS

(2.2.1)

V. * rb·

IC I laS

with: V:d= V:l- V:2

V:c = (V:l + V:~ 12 1id= (Ii]-li~/2

Iic = a, + Ii~ 12 .r;d=.r;dd

.r;c = (.r;cJ + .r;c~ /2 V:c * = V:c - V:off/2 = V:c

The four elements in the upper-left part of the matrix represent the basic description of any controlled source, as discussed along with the OIA. The right-hand part of the matrix contains all errors caused by the non-ideal floating character of the input port. An OVA should have a high value of the CMRR Hi and a low input admittance .r;c The dots in the matrix represent normally negligible effects.

Finally, the input bias current Iibias, which is also a result of the non-ideal isolating or floating character of the input port should be low and is placed in the separate column of additive error sources.

2. MACROMODELS

19

2.3 Operational Current Amplifier (OCA)

The operational current amplifier is a four-terminal device in which one input terminal is grounded. An equivalent circuit is shown in Fig. 2.3.1. The floating character of the output port can best be emphasized by distinguishing the differential-mode output voltage Vod= Vo1- V02 and current lod= (101-10;)12 from the common-mode output voltage Voc= (Vol + Vo;)12 and current loe= (101 +10;)12 .

V·* I

..---.-----...,----.--------<?5 V 01 +

Vod

L--l---,---...J...._--4------,----<?5 V 02

Fig. 2.3.1: Equivalent circuit of an operational current amplifier (OCA)

Definition of: output bias current, output common-mode current rejection ratio

In addition to the elements of an OIA, the equivalent circuit contains the elements which express the parasitic liaisons between the output port and ground. Firstly, there are the common-mode output admittances Yoe1 and Yoe2. Secondly, we have two output bias sources with an equal value lobiasThirdly, these bias currents are modulated as a function of the input voltage V*i with a transconductance ~ IHo.

The quantity H, will be defined as the output common-mode current rejection ratio (CMCRR) for the output currents. The reciprocal quantity 11Ho will be called the output common-mode current crosstalk ratio (CMCCR). It describes the current crosstalk of a DM output current lod= V*i~ on the CM output current loc= V;~ 1Ho at Vod= Voe=O. The


CMCRR for output currents is dual in regard to the CMRR for input voltages.

The main errors are quantified in the following error matrix (2.3.1):

V. lIYt «», lod Viojfs
I
t, =Y/Yt YloiY! -V + (ojfs
0
lIH r:». Y Voc obias
oc 0 oc (2.3.1)

with: Vod = Vol- V02 lod= (lol-lo~/2

Voc = (Vol + Vo~ 12 lac = (101 + 10~/2 Yod=YOdd

Yoc = (Yael + Yoc~ 12

Again, the four elements in the upper-left part of the matrix represent the errors of any controlled source, as described along with the OIA. The elements of the lower row represent the non-ideal floating character of the output port. An OCA should have a high CMCRR H; and a low CM output admittance YOC" Moreover, the output bias current lobias must be placed in the column of additive error sources. The dots in the matrix stand for negligible effects.

2.4 Operational Floating Amplifier (OFA)

The operational floating amplifier is a five-terminal network in which one terminal is grounded. An equivalent circuit is shown in Fig. 2.4.1. The floating character of the input and output ports can best be expressed by distinguishing DM and CM input and output voltages and currents, as was done with the OVA and OCA.

2. MACROMODELS

21

Using all definitions

The equivalent circuit contains all elements of a driven source:

Y;V*id}, Y;dd and Yodd' the offset sources ~offS and Iioffs; all elements which represent the liaisons of the input port with the common ground: Y;c}' Y;c2' Iibias (2x) and V*icY;IHi; and all elements which have the same function for the output port: Yoel' Yoc2' Iobias (2x) and V*idY;IHo (2x). These elements have been discussed with the OIA, OVA, and OCA.

+ 10 bias

Fig. 2.4.1: Equivalent circuit of an operational floating amplifier (OFA)

V. 11Yt Yo/Yt 11Hi lod v.;
I
lid YidlYt YidYodlYt YjHi -V liojjs
0
+ (2.4.1)
l. Y. V*. (bias
Ie Ie Ie
I uu; Yo/Ho Y V obias
oe oe oe The errors of the OFA are quantified by the error matrix (2.4.1):

The four elements in the upper-left part of the matrix represent the errors of any controlled source as discussed along with the OIA. The elements of the third column represent the non-ideal floating character of the input port, described along with the OV A, while the elements of the fourth row do the same with regard to the output port, described along with the OCA. The elements in the additional row represent all offset and bias


sources. The dots in the matrix stand for negligible effects.

2.5 Macromodels in Spice

In Chapters 2.1 to 2.4 we have presented linear circuit models for the four Operational Amplifier types OIA, OVA, OCA, and OFA. These are theoretically correct but not always practical. The main practical shortcoming is the lack of non-linear behavior description. It is desirable for shortening the simulation time in SPICE simulations of large systems with many Operational Amplifiers, to have relatively simple macromodels, which nevertheless do take into account the non-linear behavior, saturation effects, and slew rate.

Macromodel mathematical

In some cases, like the use of OpAmps in switched capacitance circuits, it is often sufficient to have only the non-linear behavior of the input stage modeled. In some SPICE programs this can be entered by a formula description. An example of such a description is given by Lin et al. in [2.1] and shown in Fig. 2.5.1.

Fig. 2.5.1:

Vo (s)

--

Avo 9m1

A two-pole model taking into account the slew rate limitation. The maximum available currents of the input stage are given by I' 0 and roo The transfer has the characteristic slope of gm1

The description consists of two blocks. The first block describes in a first-order approximation the non-linear behavior of the input stage. The

2. MACROMODELS

23

second block describes the frequency response with two-poles. With this model the slewing and settling behavior can be modeled in first-order approximation. For further details see [2.1].

Macromodel Miller-compensated

If we need to take a larger number of non-idealities and non-linearities into account, such as input offset and noise and saturation effects we need to use a more extended circuit.

Fig. 2.5.2:

SPICE macromodel of a Miller-compensated two-stage CMOS Operational Amplifier according to Boyle et al. The input stage is modeled by a differential transistor pair, while the output stage is built up by driven sources and saturation diodes

As an example the SPICE macromodel of Boyle et al. [2.2] is shown in Fig. 2.5.2. It models the input stage by using the differential pair M] and Mz, diodes D3 and D4 to model saturation effects at the output, and a Miller (Chap. 6) compensation with C2 across the output stage. The parasitic currents of the protection diodes at the input are modeled by means of current sources fBJ and fB2. The output current is limited by the diodes D] and D, together with the series resistor ROi and a reproduction of the output voltage across RcGcVout. The same model can also be used


with bipolar circuits. The input bias current sources may be left out in that case, because they are already built in the bipolar transistor models.

The macromodel can also be used if a voltage follower is added at the output as a third stage. But the model cannot be used with more complex Operational Amplifiers. More precise macromodels have been presented by Mark Alexander and Derek Bowers [2.3], and others.

Macromodel nested-Miller-compensated

The effects of common-mode-depending bias currents and saturation become even more serious in the application of low-voltage OpAmps with rail-to-rail input and output voltage ranges, such as the bipolar OpAmp NE 5234, as explained with Fig. 7.8.3. A SPICE macro-model should incorporate these effects. An example of such a model is given by Feyes et al. [2.4] using controlled sources and diodes. However, a simpler model is shaped if we just equip the input and output stages with transistors using strongly simplified transistor models, while we use controlled sources for the intermediate stage similar to Boyle's approach. This is shown in Fig. 2.5.3.a and b.

The input stage is composed of two complementary transistor pairs and a tail current selector, as explained with Fig. 4.4.4. Two different offset sources can be inserted, one for each pair. The output stage is modeled together with its R-R saturation properties by two complementary bipolar transistors QlJ and Q1b and a trans linear class-AB loop through D1h D1b and a floating supply source replica Vsp- VSN' as explained with Fig. 5.3.13.b. The diodes D13 and D14 prevent internal overdriving.

The intermediate stage is linearly modeled by a simplified transistor model. The total macromodel has three poles: one at the output determined by the load capacitance, one at the input of the output stage determined by the diffusion capacitors of the output transistors, and one at the input of the intermediate stage determined by R22 C22. These three poles are handled by nested-Miller-compensation through CM1 and CM2, as

2. MACROMODELS

25

explained with Fig. 6.2.16.

\tin +0--+---+--~r---+--------------+--~r-~

Fig. 2.5.3.a: A SPICE macromodel of a low-voltage bipolar R-R input stage for a three-stage bipolar OpAmp

Fig. 2.5.3.b: A SPICE macro model of a low-voltage-nested-Millercompensated intermediate and output stage of a three stage bipolar OpAmp with R-R output


With this model the change in input bias current and offset voltage and saturation effects are properly modeled when the common-mode input voltage passes from below the negative rail up even across the positive rail. The noise is also being properly modeled together with the slew rate and frequency characteristic. At the output a proper saturation behavior near the negative and positive rail voltage is modeled.

Conclusion

In conclusion, in SPICE simulations of large systems where many OpAmps and other components function together the simulation time can be much shortened by using simplified macromodels for the Operational Amplifiers. These macromodels often use transistor models at the input and at the output, where the largest non-linearities occur, while linear controlled current sources are used in the central part of the model.

2.6 Measurement Techniques for Operational Amplifiers

Gain measurement of an OT A

The measurement of Operational Amplifier characteristics is not easy because their parameters can seldom be directly found in an open circuit. The gain is so large that any offset and noise will drive the output in complete saturation. Hence the OpAmp has to be placed in a stable feedback measurement setup and the parameters to be measured have to be indirectly measured.

Exempt from the mentioned problem is the measuring of the transconductance ~=Gm of an OTA (Chap. 7.1). This value is so low, that it can be easily directly measured. A voltage source Vs can be directly connected to the input as shown in the first measurement approach of Fig. 2.6.1. The top-top value should not be larger than about 20 m V, in order that the input stage is not being overdriven in its non-linear region. At the output a relatively small resistor, i.e. RL =lkQ can be connected to ground. The

2. MACROMODELS

27

Gm can be found as the ratio of the output voltage and the input voltage divided by the load resistance RL• So:

(2.6.1)

The output load resistor RL may not be too large. Otherwise the parasitic open output conductance Yo in parallel with 1/RL will make the measurement inaccurate. This is particularly the case at high frequencies where the parallel output capacitance CL will attenuate the output voltage strongly. The situation is drawn in Fig. 2.6.1.

For the measurement of offset and noise of an OTA the same setup can still be used. At a grounded input the output voltage indicates the offset through a calculation by the value of the transconductance and the load resistance according to:

(2.6.2)

Fig. 2.6.1:

Measurement of the transconductance Y;=Gm of an OTA

However, if we want to measure the open voltage gain A v' the above setup does not fit anymore. The load resistance RL has to be chosen larger than the open output resistance Ro=1/Go and the offset would fully drive the output to one of the supply rails. Fiddling with a DC input bias voltage would help in some cases, but cannot be a standard measurement procedure. This can only be used in SPICE simulations as we may expect perfectly matched transistors and an inherent low offset.


Gain measurement of an OpAmp

The second closest approach is to use feedback for proper biasing in such a way that the negative input terminal is fed back by a resistor R] with a value larger than the open output resistance R, but capacitively grounded. The situation is depicted in Fig. 2.6.2.

I

I _LC -r- p

I

I

Fig. 2.6.2:

Measuring the voltage gain Av= V/V; of an Operational Amplifier

The external loop contains an extra pole at a time constant '[] =R] C]. This must be larger than the DC gain 2A v multiplied by the dominating time constant '[d=RoCp of the open-loop amplifier, to satisfy the stability of the loop. As the dominating pole frequency 112 nTd may easily be as low as 1.6kHz at Ro=lOMQ and Cp=lO pF, the external pole frequency '[] must be lower than O,008Hz at a voltage gain Av of 105• When we choose R] = lOOMQ, C] must have a value larger than O. 2f-tF. The voltage gain as a function of the frequency (Bode plot) can now be measured as Av= V/V; with a network analyzer with active probes (Cin < 3pF). If we do not take '[] large enough, the phase margin could become too low and the circuit may multivibrate at a low frequency.

The above solution may work well with most CMOS OTAs. But if the open output resistor R, becomes larger, a problem arises as the value of R] is limited.

If the input stage has bipolar transistors, the feedback resistor has to

2. MACROMODELS

29

supply the input bias current. This requires a much lower value of R, and a higher value of Cj•

A possible solution will then be to insert a single-transistor voltagefollower buffer between the output and the connection to the feedback resistor R, in order not to load the open output too much. In SPICE simulations we can easily solve the problem by replacing R, by a large inductance L]> so that there is no DC loss across this element.

With an OpAmp with two or more gain stages the open output impedance is normally so low that a much lower value of R, is allowed.

Gain and offset measurements of an OpAmp

A third, simple, and effective solution is to use an inverting amplifier feedback configuration as shown in Fig. 2.6.3.

The idea is to apply the amplifier in an inverting configuration with a minus unity gain at R2=10kD, Rj=10kD, or i.g. with a gain of -100 at R2=lkD, Rj=100kD inverting configuration. The voltage gain can be measured by a network analyzer by simply connecting one active probe to the output and the other active probe to the input of the amplifier.

+

Fig. 2.6.3:

Inverting feedback configuration for measuring the voltage gain Av= V/~ of an Operational Amplifier

The only care we must take is that the probe capacitance Cpi at the input does not cause a pole in the loop at a frequency where the loop gain is still larger than unity. This should not be a problem at a strong feedback


attenuation Bv=R/Rj• The frequency characteristic can be measured with this third method up to several hundred Megaherz.

If the open output impedance R, is high, then R, must also be chosen sufficiently high to avoid degradation of the measured voltage gain. This inverting-amplifier setup also allows us to simply measure the offset between the input terminals. The input voltage noise can also be measured at the input in a frequency band where the feedback is active, or Aflv> 1.

In conclusion, of these three simple measuring setups, the third measuring circuit is quite powerful.

General measurement setup for an OpAmp

The above methods do not satisfy a general measurement setup in which more parameters can be measured, such as common-mode rejection. The measurement of the CMRR is particularly difficult as we cannot apply feedback for biasing, as the feedback will destroy the CMRR by its ground connection (see Chap. 4.3). Alternatively we can superimpose on the supply voltages and the output reference voltage a common-mode voltage. For that purpose we need to place the device under test in a feedback loop with the aid of a second amplifier. As a fourth example, a more general low-frequency setup is given in Fig. 2.6.4 [2.5].

The amplifier under test, OA l» is supplied with positive and negative supply voltages, respectively Vsp and VSN' while a common-mode voltage source VCM moves the supply voltages and the output reference voltage in regard to the input voltages. A broadband helping amplifier OA2 compares the output voltage Vo of the amplifier under test with a source voltage ~ and amplifies the difference to the output measurement voltage VM. This voltage is being fed back to the input of the amplifier under test by an attenuation network Bv=R/R2 and bias current measurement resistors R3 and R4. The capacitor C2 provides phase lead to stabilize the long loop through two amplifiers. The low-frequency voltage gain can be measured by dividing Vs by (R/Rj) VM, so:

2. MACROMODELS

31

(2.6.3)

Amplifier under-test Dj

+

Fig. 2.6.4:

General low-frequency measurement setup for voltage gain Av, bias current Ibias, offset voltage Voffs' noise voltage """ and common-mode rejection CMRR of an Operational Amplifier OAl under test aided by a helping Operational Amplifier OAz

This can be done, for instance, by a network analyzer. Active probes need not be used, depending on the impedance levels of the source and output of OAz. Load conditions can be changed by applying ZL. The measurement of the frequency characteristic is correct up to about 1 MHz, which is the pole of the feedback network.

For an offset measurement the switches S, and S2 have to be closed. The output voltage VM represents the input offset voltage amplified by the inverse feedback attenuation.

(2.6.4)

The spectral input noise can also be read from VM.


(2.6.5)

The bias current Ibias can be found by opening only SI'

(2.6.6)

The offset current Ioffs similarly can be found by opening both switches SI and S2 and by using equal resistors R3=R4'

(2.6.7)

For the measurement of the common-mode rejection ratio a commonmode voltage source VCM has to be used which is able to drive the supplyvoltages Vsp and Vsp of the device under test. The result can be measured

as:

(2.6.8)

Note that the loop gain of the measurement setup is high (AOlAo2R/Rl). The high frequency behavior of the loop is determined by the two dominating time constants of both amplifiers. The stability of the loop is ensured if the help amplifier OA2 has a high dominating pole frequency, say IMHz, and if the feedforward capacitor C2 corrects the phase lag of OA2 above that frequency. The measurement setup may function up to frequencies of IMHz.

In conclusion, we have seen that it is not easy to measure OpAmp parameters for we have to take care of proper biasing simultaneously. For SPICE simulations, the second method can be used to measure frequency characteristics. The third method is powerful in practical situations. The measurement of the CMRR is particularly difficult, as we cannot apply feedback to properly bias the OpAmp as this destroys the CMRR. A proper method is to alternatively move the supply with a common-mode voltage. This is done in the fourth general measurement setup.

2. MACROMODELS

33

2.7 Problems and Simulation Exercises

Problem 2.1

Figure 2.5.2 shows a Boyle type macromodel developed for CMOS operational amplifiers. Starting from that picture, calculate the parameters of an operational amplifier with a differential gain aVD = 1 OOdB, commonmode rejection ratio CMRR=90dB, unity gain bandwidth iOdB=10MHz with a phase margin ({Jm = 76deg. The slew-rate specifications are Srp=10VII"'s for rising output voltage and Srn=9VIJ.ts for falling output voltage. A short-circuit current Isc= 7mA should be supplied at the output, with an output DC impedance Roul = 1 KQ. Supply voltages are Vsp- VsN=3V, and the macromodel should dissipate Pd=3mW. No capacitive or resistive load is present.

Solution

The gain in the first stage is chosen to be equal to unity:

(2.7.1)

As RDJ usually has values around 21gmi for convenient common mode input range,

(2.7.2)

In order to calculate gm]> weak inversion transistors will be considered for the input stage with an approximate gllli given by

Iss

gill] lOOmV (2.7.3)

The tail current Iss results from Srp value and a convenient Miller capacitor:

RDJ =21g1llJ =200mVIIss= lOKQ RSJ =RDJ-l/gIllJ= lIglllJ =5KQ

(2.7.5)


C2=2pF Iss=C2S rp =20JlA

(2.7.4)

Returning to RDJ and RSJ values:

Using these values, the unity gain bandwidth results:

(2.7.6)

(2.7.9)

which is close enough to the desired bandwidth. It can be increased by choosing a lower value for RDJ. The gain in the virtual intermediate stage of the model is larger than unity, given by its components G, and R2:

Ga= llRD] =O.lmS R2=lOOKQ

(2.7.7)

The output impedance will be equal to ROJ for frequencies above:

(2.7.8)

Below this corner frequency, the DC output impedance is

Because the short -circuit current will be given by the ratio of a diode voltage and ROJ> ROJ should be less than DC output impedance:

(2.7.10)

The rest of the gain up to aVD = 100dB is given by the third stage:

2. MACROMODELS

35

R02=Rolit -ROt =900.0 o; =avrfiD/R2R02 = 1l.IS

(2.7.11)

The gain of the ReGe voltage repeater should be equal to unity with an R, value much smaller than ROJ' so:

R; =10.0

G =1/R =O.IS

e c

(2.7.12)

This voltage repeater puts the difference between the voltages at the output and at the internal output on ROJ' such as if a difference exists the maximum current supplied is the short-circuit current:

Isc= V JROI =0.6V/IOO.Q=6mA

(2.7.13)

To correct the short circuit current value to the desired 7mA, the diode threshold voltage can be changed to 0.7V for simulation purposes. The Srn value is controlled by Cs:

(2.7.14)

The common-mode rejection ratio is given by Gem:

(2.7.15)

For the desired phase margin, CJ introduces a second pole

(2.7.16)

which produces:

(2.7.17)

The power dissipated by the whole macromodel is modeled by R; which takes the value:


(2.7.18)

Simulation Exercise 2.1

The Boyle macromodel depicted in Figure 2.7.1 can be simulated for AC analysis using a simulation setup as shown in Figure 2.7.2.

VSP
O.8V V3
D17
R14
Vout
100
D18 VSP
Boyle macro
Vout
R5
10k
VSN
R4
1meg l___--_l_-_L_------------------_l_--¢VSN

Fig. 2.7.1:

Boyle macromodel for a two-stage operational amplifier

Fig. 2.7.2:

Differential gain simulation for an OpAmp macromodel

2. MACROMODELS

37

The use of ideal high-valued capacitors and resistors allow the simulator to solve correctly the biasing point for the transistors inside the operational amplifier. At DC, the amplifier is basically a repeater with the positive input connected to ground. What is the low limit frequency for a DC voltage gain simulation, considering the effect of C4? What can be the additional benefits of replacing R4 with an inductor?

Simulation Exercise 2.2

The circuit shown in Figure 2.7.3 is used to simulate a Boyle macromodel of an operational amplifier for AC solution of commonmode to differential crosstalk.

Using this crosstalk gain and the differential gain, the common-mode rejection ratio can be calculated. At DC, the inductor keeps the operational amplifier in a repeater configuration, while above a limit frequency given by the L3C4 time constant, both inputs of the operational amplifier are tied together and the circuit is placed in an open-loop configuration. Simulate this circuit using the Boyle macromodel shown in Figure 2.7.1 and decrease the common-mode rejection ratio with 20dB by adjusting the G5 transconductor. What other circuit elements in Figure 2.7.1 affect the common-mode to differential crosstalk?

Boyle macro

L3

'---- ___J 1 e1 0'------'

Fig. 2.7.3:

Common-mode to differential crosstalk simulation


2.8 References

[2.1] J.C. Lin, J.H. Nevin

"A modified Time Domain model for nonlinear analysis of an operational amplifier"

IEEE Journal of Solid-State Circuits, Vol. SC-21, No.3, June 1986.

[2.2] Graeme R. Boyle, Barry M. Cohn, Donald O. Pederson, James E. Solomon

"MacromodeUing of integrated circuit operational amplifiers" IEEE Journal of Solid-State Circuits, Vol. SC-9, no. 6, December 1974.

[2.3] Mark Alexander, Derek F. Bowers

"New Spice compatible OpAmp model boosts AC simulation accuracy"

EDN, February 15, 1990.

[2.4] Jan. H.A. Feyes, Ron Hogervorst, Johan H. Huijsing "Macromodeling operational amplifiers"

ISCAS 93 Conference, London, Vol. 5, pp 681-684.

[2.5] Connelly, J.A.

"Analog integrated circuits"

John Wiley and Sons, New York, 1975.

3. APPLICATIONS

This chapter describes a number of general applications suitable for quantifying the requirements of universal active devices or Operational Amplifiers. The transfer of each example is described by a matrix containing, firstly, one or more nominal values, and secondly, error terms having low values. The nominal values are determined by the circuit configuration and by the gain-setting passive components in the circuit. The error terms are determined by the non-idealities of the active devices as discussed in Chapter 2. Each of the error terms can be relatively easily expressed as a sum of errors caused by:

firstly, the non-zero input voltage; secondly, the non-zero input current;

thirdly, if the input port is floating, the CM input current; and fourthly, if the output port is floating, the non-zero CM output current.

With the aid of these error terms one can estimate the required specifications of the active devices or Operational Amplifiers.

In the last section particularly the requirements of OpAmps are evaluated based on the desired dynamic range.

It appears that for an optimum ratio of dynamic range and supply power the output of an Operational Amplifier should have a current -efficient class-All biasing and a voltage-efficient rail-to-rail voltage range. In some cases the input should also have a voltage-efficient rail-to-rail commonmode voltage range.

3.1 Operational Inverting Amplifier

The operational inverting amplifier has both the input and output ports grounded which restricts the feedback to parallel connections. This results in applications with a low input and output impedance.

I] lIZtn + uz; 11Aie V2 I'offs

+

V] lIAve 11Yte -12 V'offs


Current-to- Voltage Converter

As we have already discussed in Chapter 1.1 , the trans impedance amplifier or current-to-voltage converter is the most basic application of the OIA. This amplifier configuration is drawn in Fig. 3.1.1 together with the source and load circuits.

Fig. 3.1.1: Transimpedance amplifier with source and load circuits

For a description of the transfer we choose a kind of matrix which clearly presents the nominal transfer and additionally presents all errors in relation to the input quantities ~ and l.. For a two-port the total matrix is equal to the chain matrix. The matrix of Fig. 3.1.1 is given in (3.1.1):

(3.1.1)

with:

1lZtn= -11Z2

1lZte= -YiY2+ Yo)/~-YJY2+ Yo)/~

1lAie= -Y/~ -Y/Yr

i/s;» -(Y2+ Yo)/~

1/~e= -ur,

I] offs = ~ off/Z2 + I, offs

V] offs = ~ offs

and with Y2 = 11Z2

3. APPLICA nONS

41

The nominal term 11Ztn represents the minus reciprocal value of the nominal impedance -1IZ2. All other terms are error terms caused by nonidealities as described by the error matrix 2. 1.1. Each error term can be written as a sum of partial errors which originate, firstly, from the nonzero OIA input voltage V; and secondly, from the non-zero input current Ii. The first error term 11Zte represents an entrance error current V/Zte as a function of the exit voltage V2 at -12=0. This error current is composed of two partial error currents: V2YiY2+ YJIYt flowing through Y2 as a consequence of the non-zero input voltage V;=- ViY2+ Yo)/~ and the nonzero input current I] =- V2Y;(Y2+ Yo)/~. The second error term lIAie represents an entrance error current I/Aie as a function of -12 at V2=0. This error current is composed of two partial error currents: 12Y/~ flowing through Y2 as a consequence of the input voltage V; = I/~, and the input currents Ii=I2Y;/~. The other error terms are built up in the same way. The input offset voltage V;Ojfs and current Iiojfs give rise to the entrance offset voltage V]ojfs and current I]OjfS' as presented. The spectral input noise voltage V;n and current lin can be thought of as superimposed on the relevant offset quantities [3.1, 3.2]. An extra entrance noise current, caused by the noise current IZ2n of the impedance Z2> is present.

Inverting Voltage Amplifier

Fig. 3.1.2: Inverting voltage amplifier with source and load circuits

Another important application of the OIA is the inverting voltage amplifier, a configuration widely used in analog computer circuits. The


circuit is shown in Fig. 3.1.2. The matrix is given in (3.1.2):

v] 11AvIl + lIAve 1/Yte V2 I10jfs +

I] uz; llAie -12 V10jfs

(3.1.2)

with:

1lAvn= -1Z/Z2

lIAve= -(Z1 +Z~(Y2+ YO)IYrZ2-Z1Y;(Y2+ Yo)IYr 1lZte= -11Z2

An important disadvantage of the inverting voltage amplifier is its large entrance current error V/Zte =- V/Z2• The non-inverting voltage amplifier using an OVA, which will be described in the next section, does not have this disadvantage.

3.2 Operational Voltage Amplifier

The operational voltage amplifier (OV A) has a floating input port that provides applications with a high input impedance by a serial feedback through the input port.

v] 11AvIl + lIAve l/Yte V2 I10jfs +

I] lIZte l/Aie -12 V10jfs

(3.2.1)

Non-Inverting Voltage Amplifier

The most basic application of the OVA is the non-inverting voltage amplifier (see Chap. 1. 1.2) as shown in Fig. 3.2.1

The transfer of the circuit can be described by one nominal value and error terms having low values, as described by matrix (3.2.1):

3 . APPLICATIONS

43

with:

us; =Z/(Zj+Z;J

1 lAve = (Yo + Y)IY; + Y;lYo + Ys)/~Y; + 1 IAVllIi- Y;c/Yr4vn

ur; =1 IY; + Y;/~ Y;

I/Z; = Y;lYo + Ys)/Y;

llAie= Y;/Y;

V;OffS = V;offs + t, off/Yp

Ijoffs =

+Iibia/~ +Iibias

Fig. 3.2.1: Voltage amplifier with source and load circuits

The reciprocal nominal value of the voltage amplification factor is JIAvn=Z/(Zj+Z;J. All other matrix elements are error terms caused by non-idealities. The first error term lIAve represents the entrance error voltage V/Ave as a function of the exit voltage V2 at -12=0. The error voltage is the sum of four partial error voltages. The first of these partial error voltages VlYo+ Y'sJIY; represents the differential input voltage V;d needed to generate the exit voltage V2•

The second partial error voltage V2Y;lYo+ Ys)IYpY; is the voltage loss across the parallel admittance ~ = Yj + Y2 of the feedback network as a result of the differential input current lid = V2Y;lYo+ Y)IY;. The third partial error voltage V/AVllIi is the equivalent differential input voltage V;d evoked by the crosstalk JIHi from the CM input voltage V;c= V/Avn" The


fourth partial error voltage - Vzl';c/Y r4vn is the voltage loss across the parallel admittance Yp = y/ + Yz of the feedback network as a result of the common-mode input current Iic= Vzl';c/Avn' The other error terms are built up in the same way. In addition to the offset and noise quantities of the OIA (see Chap. 3.1), there are the input bias current Iibias and the input bias noise current Iibn and their effects. An extra entrance noise voltage, caused by the noise voltage Vzpn of the parallel impedance Zp = l/Yp, is present.

Voltage Follower

If we choose Z/ = 00 and Zz = 0, the voltage amplifier becomes a voltage follower with a nominal voltage amplification factor Aun = 1 in that all error contributions with a Yp in the denominator disappear.

Bridge Instrumentation Amplifier

A bridge instrumentation amplifier can be built with an OVA and bridge of four impedances Z11 through Z22 (Fig. 3.2.2). The circuit can be thought to be composed partly from the inverting voltage amplifier (Fig.

Fig. 3.2.2:

Bridge instrumentation amplifier with source and load circuits

3. APPLICATIONS

45

3.1.2) and the non-inverting voltage amplifier (Fig. 3.2.1). If the bridge is well balanced, which means that LJZ/Zb=1-Z21Z1/Z22Z11«1, the circuit only amplifies the differential-mode entrance voltage V:d while the common-mode entrance voltage V:c is rejected. The circuit of a bridge instrumentation amplifier is given in Fig. 3.4 and the matrix with the error terms in Eq.3.4.

VI llA +lIA lIYte utt, V2 l.iff
vn ve to s
lId = lIZte 1IA. -I + Viojfs (3.2.2)
Ie 2
lIc lIZlcc V. IF
IC I [as
with: 1 IAvn = -Z11IZ21

I/A; = -LJZ/2Zb(Avn + 1)Avn

-(Zll+Z21)(YO+ Y*21)IY721-(Z11+Z12)Y;lYo+ Y*21)IY;

-1I2Hr4vn- Y;c/2YllAvn + Y;c/2YlzAvn

11He = - LJZ/Zb (A VII + 1)

-1 IY 721- (ZII + ZI zJ Y;/Y;(ZII + Z21)

-1 IHi- Y;/Y11- Y;/Y12

11Z1cc= 11(Z11 +Z21)

and with: LJZ/Zb = 1-Z21Z12IZ22Z11«l

Yll = 1IZ11, Y12 = 1/Z12' Y21 =1 IZ21, Y22 = 11Z22

The main disadvantage of this type of instrumentation amplifier is that the bridge resistors establish a connection between the input port of the OVA and the output and ground. This destroys the CM isolation barrier of the input port of the OV A and thus destroys the CMRR of the application. Therefore, the common-mode crosstalk ratio (CMCR) 11He is directly determined by the imbalance LJZ/Zb of the bridge and a factor lI(Avn + 1) depending on the nominal amplification factor AVIl" Another disadvantage is the relatively low CM entrance impedance 1IZ1cc=11(Z11+Z21) which is also caused by the lack of a CM isolation barrier.


The latter disadvantage can be overcome by connecting the input terminals in cascade with either two voltage followers, one for each terminal, or two voltage amplifiers of the type of Fig. 3.2.1, joined in a balanced configuration by connecting the bottom sides of Z1 to each other, instead of to ground.

A basically better way to build an instrumentation amplifier which does have a CM isolation barrier is to use two OFAs. An example of such an instrumentation will be given in Chapter 3.4.

3.3 Operational Current Amplifier

The operational current amplifier has a floating output port, which allows a high output impedance or current output by a serial feedback through the output port.

Current Amplifier

The most basic application of the OCA is the current amplifier (Chap. 1.3). This configuration is given in Fig. 3.3.1 together with its source and load circuits.

Fig. 3.3.1: Current amplifier with source and load circuits

The nominal amplification factor and error terms with small values can be described by the matrix of Eq.3.3.1.

3. APPLICATIONS

47

II lIAin + lIAie l/Zte -12 -I1offs +

Vj lIYte lIAve V2 V10ffs

(3.3.1)

with:

1IAin = Y/(Yj + Y j

1IAie= Y'JYp+Yo)IYpY; +YJ~ +Yo)/~Y; +2H/Ain-Yoc/Y~in

1lZte= YseYo/Y; + YXo/Y; + Yoc/Ain

11Y;e = (Yp + Yo)IYpY;

1 lAve = Yo/Y;

1jOf!S = V;Of!sY; + IiOf!S + 21obia/ Ain

Vj offs = V;Of!S

The reciprocal value of the nominal current amplification factor is 11 A = Y/ (Yj + Y j. All other terms are error terms caused by non-idealities. The first error term 11Aie can be written as the sum of four partial error terms.

The first partial error Y'JYp + Yo)/~Y; represents the entrance error current -I2~(~ + Yo)IYpY;. This results from the current V;Ys flowing through the series conductance Y's=11(1IYj+1IYj as a consequence of the input voltage V; = -I2(Yp + Yo)IYpY; needed to deliver the exit current -12. The second partial error YJ~ + Yo)/~Y; simply represents the input current 1i=-12Y;(Yp+Yo)IYpY; which is needed to deliver the exit current -12•

The third partial error 21HaAin is caused by the crosstalk 11Ho of the DM output current Iod=-I2 on the CM output current lac. Double this value (from both output terminals) must be counted.

The fourth partial error - Yoc/Y ~in is a result of the error current 12YOC/~ which flows into the CM output impedance Yael in parallel with the parallel admittance ~ = Yj + Y2 as a function of -12• The other error terms are made up in the same way. In addition to the offset quantities mentioned with the OIA, there are the output bias current Iobias and output bias noise current 1obn. There is an extra entrance noise current caused by the


noise current I YSN of the series admittance F', = 1 1(1 IY] + 1 IY2) .

3.4 Operational Floating Amplifier

The operational floating amplifier is the most versatile OpAmp. It permits series feedback through the input and output ports. This allows for applications with a high input and output impedance.


The most basic application of the OF A is the trans admittance amplifier or the voltage-to-current converter (Chap. 1.4). The circuit is shown in Fig. 3.4.1 with the source and load circuit.

Fig. 3.4.1: Voltage-to-current converter or trans admittance amplifier with source and load circuits

Vj llYtn + llYte lIAve -12 V10jfs +

Ij llAie stz; V2 I10jfs

(3.4.1)

The nominal trans admittance and error terms are presented in the matrix of Eq.3.4.1.

with:

11Yrn = -l!Y2

11Yre "" -(Yod+ Y~/Y2Yr-Y;iYod+ Y~/Y2YrY2-1IHiY2+ Y;c/Y2Y2-2IHoY2 + Yoc/Y2Y2

l/A; "" -Yo/Yr-Y;dYo/YrY2-2Yo/YPoY2-YoC/Y2

3. APPLICATIONS

49

1 I Aie = - ~d(Yod + Y z} IY)'';- ~c/Y2 11Zte = - ~dYo/J';

Vj offs = ~offs + Iioff/Y2 + Iibia/Y2 + 2Iobia/Y2 Ijoffs = Iioffs + Iibias

The reciprocal value of the nominal transadmittance is 11J';n=-1IY2. All other terms are error terms caused by non-idealities. Each error term is the sum of partial errors which can have six origins (see Chap. 2.4): firstly ~d=J(-I2' Vz}, secondly Iid=J(-I2, Vz}, thirdly V;d= J(~J, fourthly Iic= J(~J, fifthly Ioc= J(-Iz} , and lastly Ioc= J(VoJ·

The partial errors are placed in six columns according to the above sequence. The partial errors have already been explained in the preceding sections of this chapter. There is an extra entrance noise caused by the noise voltage VY2n of the admittance Y2•

Inverting Current Amplifier

An inverting current amplifier or current mirror can be made by adding an entrance admittance Yj to the trans admittance amplifier (Fig. 3.4.1). The entrance admittance Yj converts the entrance current I, into a voltage Vj which is in turn converted into the exit current 12 by the transadmittance amplifier containing Y2. A special case appears if Yj = Y2. In that case the circuit is known as the "current mirror". The circuit is drawn in Fig. 3.4.2 while the matrix is given by Eq.3.4.2.

II lIAin + lIAie 1/Zte -/2 V10ffs +

VI 11Yte lIAve V2 /loffs

(3.4.2)

with:

1IAin=-Y/Y2

1IAie=-YlYod+ Yz}IY2J';-~iYod+ Yz}IY2J';-1 IHr4inJ';c/Y2 + J';c/Y ~in

-2lHaAin + Yoc/Y ~in

11J';e =-1 IY2


Fig. 3.4.2: Inverting current amplifier with source and load circuits

The first term is the reciprocal nominal value of the current amplification factor. The other terms are error terms, as has been explained earlier. A disadvantage of the inverting current amplifier is the relatively high entrance voltage Vj = -I/~e = I/Y2 which is not present in the non-inverting current amplifier of Fig. 3.2.1.

Differential Voltage-to-Current Converter

Fig. 3.4.3: Instrumentation voltage-to-current or trans admittance amplifier with source and load circuits

3. APPLICATIONS

51

The floating character of the input of an OF A together with the current-source character of the output of an OFA having series feedback at its output can be applied to obtain the CM-voltage isolation needed for the entrance circuit of an instrumentation amplifier. This is shown by the instrumentation transadmittance amplifier of Fig. 3.4.3 [3.3]. Two OFAs, connected as voltage and current followers (VCFs), firstly, transfer the DM entrance voltage V;d at unity gain to the terminals of a conductance Yb and secondly, transfer the current -12d = - VI Y2 through Y2 towards the output terminals. The CM input voltage only affects the output current through the non-ideal properties of the OFAs, as described by the matrix Eq.3.4.3.

i. 1/Ziee

12e 1/Zteee 1/Zoee

-I Vloffs
2
V2d lloffs
+ (3.4.3)
VIe llbias
V2e i-: VI 11Ytn + lIYte lIAve uu,

lId lIAie uz;

with:

11Ytn = -11Y2

1 lYre ax: -2 (Yod + 2Y ~/2Y2Yr- Y;lYod+ 2Y ~/2Y2YrY2

-112Ri1Y2+ 112Ri2Y2+ Y;/2Y2Y2-1IRoIY2+ 1 IRo2Y2+ Yo/2Y2Y2 1 IRe = Yod/Yrl + YodiY;2 + Y;dl Yod/Yrl Y2 + Y;d2YodiYr2Y2

-t/u; + 11Ri2 + Y;c]2IY2- Y;c21IY2 + Yoc12IY[ Yoc2/Y2

1lZtcce= + Y;dYoiYr-Y;c-Yoe

and with: V1d= VjJ- V12, VIC = (VjJ + Vl~/2, t.: = (ljJ-Il~/2, 11c= (IJJ+/d/; idem for the exit voltages and currents; all OFA parameters without OFA number are average values.

The nominal value of the reciprocal admittance is 1IYrn=-1IY2. All other terms are error terms which have been explained in the preceding part of this chapter. The common-mode crosstalk ratio (CMCR), or the reciprocal value of the common-mode rejection ratio (CMRR), is liRe.

1lAvn= -1IY7-1d 1IH;= 11He+ 11FPiL

(3.4.4)


A complete instrumentation voltage amplifier arises if we include the current-to-voltage conversion function of the load impedances Zu and ZL2' In that case the reciprocal nominal overall voltage gain lIAvn and the overall CMCR 11H; are given by Eq.3.4.4.

The reciprocal value of the discrimination factor IS defined as: 11Fe = (12/V;J 112/V;J = 1 IZtcce Y;n'

11Avn =-1 IY2Z1s 11H; =1 IHe + 1 IF Pi

(3.4.5)

From the above expression we see that a low overall CMCR 11H; can only be obtained if the amplifier has a low CMCR 11He as well as a low reciprocal discrimination factor 11Fe. The effect of the latter term is further reduced by a low CMCR 11HiL of the load circuit.

Instrumentation Voltage Amplifier

An instrumentation voltage amplifier with a voltage-source character between the output terminals arises if we insert a balanced floating impedance amplifier with an OFA and two impedances Z11 and Z12 between the preceding example's instrumentation transadmittance amplifier and its load circuit [3.3, 3.4].

The complete instrumentation voltage amplifier circuit is drawn in Fig. 3.4.4 together with the source and load circuits.

The reciprocal nominal overall differential voltage gain and the CMCR of the complete amplifier are described by Eq.3.4.5.

3 . APPLICATIONS

53

Fig. 3.4.4: Instrumentation voltage amplifier with source and load circuits

The instrumentation amplifier has a floating entrance port and an independently floating exit port. It measures the differential entrance voltage at nominal zero entrance currents. The differential exit port has a voltage source character.

Instrumentation Current Amplifier

If we change the sequence of the trans admittance and trans impedance amplifiers of the preceding example we obtain an instrumentation current amplifier. This circuit is drawn in Fig. 3.4.5 together with the source and load circuits.

The overall reciprocal nominal current gain is given by Eq.3.4.6.

(3.4.6)

54


ZlC2

Fig. 3.4.5: Instrumentation current amplifier

The instrumentation current amplifier has a floating entrance port and an independently floating exit port. It measures the differential entrance current at nominal zero differential entrance voltage. The exit port has a current source character.

Gyrator Floating

From among the applications of the OF A the gyrator should not be left out. A fully floating gyrator can be composed of two instrumentation

trans admittance amplifiers as shown in Fig. 3.4.6 [3.5, 3.6]. The nominal and error terms of the transfer are given in Eq. 3.4.7.

1/Zt2111 + l/Zt21e

1/Zt2111 + l/Zt12e

V2
l/Zlce
VI (3.4.7)
Vic with:

3 . APPLICATIONS

55

1 IZt2J n = - Y2 lIZt2Jn=Yj

lIZt2Je=2YlYod+ 2Y;J12Y;+ Y;iYod+ 2Y;J12Y;-Y;/2+ 1 IHiY2- Yo/2+ 21HoY2 llZt2Je =-2YlYod+ 2Yj)12Y;-Y;iYod+ 2Yj)12Y;+ Y;/2-1IHiY2+ Yo/2-2IHoY2 lIZlde=2Y2Yo/2Y;+ Y;iYod+ 2 Yj)12 Y;+ Y;/2+ Yo/2

lIZ2de=2YjYo/2Y;+ Y;iYOd+ 2 Y;J 12 Y; + Y;/2+ Yo/2

and with average OFA parameters.

Fig. 3.4.6: Floating gyrator

The nominal resonant frequency CUo circuit which is composed of a gyrator with Yj = Y2 = llR and loaded on both sides with a capacitance C, = C2 = C is given in Eq.3.4.8.

cun=lIRC lIQ=2RIR[2C/C

(3.4.8)

with: lIRd=real(lIZde), cunCt=im. (lIZte), z; -z.; =Z2de, z; =Zt2Je =Zt21e

The second term in the expression of l/Q describes the phase-lag in admittance amplifiers. This phase-lag undamps the circuit at higher resonant frequencies.


Conclusion

The application examples given in this section have shown the relation between the specifications of the active devices and the accuracy of the applications mentioned. This is necessary for determining how far we have to go in improving the specifications of the active devices whose designs will be the subjects of the following chapters. The important overall specification of dynamic range brings about special requirements, as we will see in the next section

3.5 Dynamic range

The total amount of information that can be processed in an analog signalprocessing step is determined by the product of dynamic range (number of bits) and bandwidth. The dynamic range over power limitations will be evaluated in this section, while the bandwidth over power limitations will be extensively covered in Chapter 6.2.

The fundamental specification of low power is in contradiction to the fundamental specification of dynamic range. Therefore, it is important to see how these specifications relate in several OVA applications. An optimum for low power and dynamic range can be found if the output stage and input stage possess a rail-to-rail voltage range. This will be shown in this section, in which the OV A will simply be called Operational Amplifier (OpAmp).

Dynamic range over supply-power ratio

The trend towards smaller dimensions in VLSI circuits firstly leads to smaller break-down voltages across isolation barriers. The supply voltages will go down from 5V through 3V to 2V or even IV. Secondly, the high density of circuit cells on a chip limits the power that can be dissipated per circuit cell. Moreover, the increased use of batteries or solar power in wireless applications emphasise the above trend. As a consequence, the dynamic range (DR) of analog signals is squeezed down between a lower

3. APPLICATIONS

57

supply-voltage ceiling and a higher noise-voltage floor. The latter is a consequence of a lower supply current.

The maximum top value of a single-phase signal voltage is equal to half the supply voltage Vsst= Y~HI/2, as is shown in Fig. 3.5.l.a. Its RMS value is ~s = ~u/2 v:2. If this signal is present across a signal-processing resistor Rs' the supply power needed to drive this resistor in c1ass-B mode is PsuP= Vsu/av= V2su/2TrRs. The thermal noise voltage across this resistor equals: VN= (4kTB,Ry!2, in which k is Boltzman's constant, T the absolute temperature, and Be the effective bandwidth. The maximum dynamic range as a function of the supply power can now be calculated as

V2 P

DR =__::_: = 1t . sup

max 2 4 4kTB

VN e

(3.5.1)

from which we can find the dynamic-range over supply-power ratio as

DRmax _1t 1

(3.5.2)

Exactly the same expressions are found for the balanced case, where the top value of a balanced signal voltage is equal to the full supply voltage ~bt= Vsup, instead of half the supply voltage in the single case, see Fig. 3.5.l.b. To consume the same power, the value of the balanced resistor R, must be taken four times that of the single one: R; =4Rs•

Vsst 1 Vsbt 1
/1\ 2:Vsup,Isup (])() "2 Vsup, [sup
vst Vgr Vst Vgr
\(7
1 1
-"2Vsup -"2 Vsup
time- time- Fig. 3.5.l.a and b: Single and Balanced rail-to-rail voltage Vsst= Vsu/2 and Vsbt= ~up, respectively, across a c1ass-B driven signal-processing resistor has a maximum DR= (nI4)Psu/4kTBe (=89dB at Psup= 16jLW, Be= IMHz)



A simple example is shown in the single and balanced voltage-to-current converter shown in Fig. 3.5.2.a and Fig. 3.5.2.b with a single resistor of R, = 10kD or balanced resistor of R, = RbI + Rb2 = 40KD respectively, at a supply voltage ~up = Vsp- VSN of 1 Volt, in a bandwidth of IMHz. In this case the supply power Psup is 16"" W at a maximum sinusoidal signal. The result is a maximum dynamic range DRmax of 89dB. This maximum can only be obtained if the signal processing resistors can be driven in class-B and rail-to-rail, and when the amplifier is noise free.

If the output stage is biased in class-A instead of in class-B, the bias current must be equal to the maximum current and the DRIPsup ratio loses minimally a factor of n , or 5dB from its maximum value. This loss for class-A in regard to class-AB may easily be a factor of 100, or 40dB, in the many cases where the signals are much lower than their maximum values most of the time. This is the case in audio, telecommunications, hearing aids, etc. If the output voltage range is restricted to 113 of the supply voltage, for instance when a diode voltage VBE is lost at a supply voltage of 1 Volt, the DRIPsup ratio loses another factor 3, or 5dB.

RW" 20 kQ

Fig. 3.5.2.a and b: Single and balanced voltage-to-current converter with R, = 10kD and RbI = Rb2 = 20KD, respectively, with a DR=89dB at ~up=VsP-VsN=1V, Psup=16""W, R-R class-B output stage, and a bandwidth of IMHz

3. APPLICATIONS

59

Inverting Voltage Amplifier

The inverting voltage amplifier of Fig. 3.5.3.a, firstly, loses a factor of 2 or 3dB in its DRIPsup ratio because an additional input buffer is needed to supply the power in the resistor Rj• Otherwise, this power has to be supplied by the source. Secondly, another factor 2, or 3dB, is lost because of the noise of the two resistors. The resulting DRmax is 83dB at a supply voltage of 1 Volt and a bandwidth of 1MHz.

Fig. 3.5.3.a: Inverting amplifier with A =-R/Rj=-1, with a

DR = 89-3-3 = 83dB, at a supply voltage Vsup= Vsp- VSN of IV, Psup=16p.W, R-R, class B, 1MHz. If Rj=2kil, R2=20 kil, we obtain A =-10, with a DR =89-3-10= 76dB

When we would choose a gain of 10 in the inverting voltage amplifier of Fig. 3.5.3.a with Rj=2kil and R2=20kil, we firstly lose a factor of 2, or 3dB into the input buffer, and secondly another factor of 10, or lOdB because resistor R, only uses 1110 of the supply voltage range. This means that the 11'10 times larger current noise of resistor R, = 2kil will be reflected into the 10 times larger resistor R2 = 20kil, which gives rise to a 10 times larger noise power. The resulting DRmax is 76dB. The same result is obtained with the balanced version given in Fig. 3.5.3.b.


Fig. 3.5.3.b: Balanced inverting amplifier with A =-(R2+R4)/(Rj+R3) = -10, with a DR = 89-3-10= 76dB, at a supply voltage ~up=VSP-VSN of IV, Psup= 16/1W, R-R, Class-B, and lMHz

Non-Inverting Voltage Amplifier

The non-inverting voltage amplifiers shown in Fig. 3.5.4.a and b with a gain of 10 do better than the inverting one(s).We only lose a factor of 10, or 10dB, proportional to the gain because R, only uses 1110 of the supply voltage range. We do not lose the factor 2, or 3dB, because we do not need an additional input buffer. The DRmax is 79dB in a frequency band of lMHz and a supply voltage of IV.

Fig.3.5.4.a: Non-inverting voltage amplifier withA=(R2+Rj)/ R, = + 10, with a DR =89-10= 79dB, at a supply voltage ~up=VsP-VsNof lV,Psup= 16/1W, R-R, Class-B, and lMHz

3. APPLICA nONS

61

-lin=O

Fig. 3.5.4.b: Balanced non-inverting voltage amplifier with A = (R3+R2+R])IR] = + 10, with a DR=89-10= 79dB, at a supply voltage Vsup = Vsp - VSN of IV, PsuP = 16p., W, R-R, Class- B, and 1 MHz

Inverting Voltage Integrator

The balanced inverting voltage integrator shown in Fig. 3.5.5 only loses the factor 2, or 3dB, because of the use of input buffers. The capacitors do not add to the noise.

Fig. 3.5.5:

Balanced inverting voltage integrator with A = -(l/jwC]+l/jwC) (R]+R) with a DR = 89-3 = 86dB, at Vsup = Vsp- VsN=lV, Psup= 16p., W, R-R, Class-B, and 1MHz


Within the effective bandwidth of Be=1/2l1:RC, with R=R1 =R2=40kQ and C= C] = C], at a supply voltage of 1 V, the DRmax is 86dB. The resistor values have been chosen such that the supply power is again 16,..,., W at a maximum sinusoidal signal [3.7].

The dynamic range of an inverting voltage integrator is generally large:

DR=(n/4)P /4kTB =(n2/2)P RC/4kT

sup e sup

2

DR =n Vsup C/4kT

with: Be=1/(2l1:RC), R=R]=R2' C=C]=C2

(3.5.3)

Current Mirror

A very severe loss of the DR is found in current mirrors. The current mirror of Fig. 3.5.6 firstly, loses a factor n, or 5dB in the DRmax because the circuit operates in class-A, and, secondly, a factor 40, or 16dB with bipolar transistors because the signal is compressed in a voltage range of VTzkT/q=25mV across the gain-setting base-emitter resistors. These resistors are small in regard to Vsu/Isup and therefore their noise current is unnecessarily large. The resulting DR is only 68dB. This is a factor 30,

'-----_,__----0 VSN

Fig. 3.5.6:

Bipolar current mirror with Iou/Iin-n, with a DR = 89-5- I6 = 68dB, with n = 1, v'up = Vsp- VSN= 1 V, Pmp = 16,..,., W, classA, and IMHz

3. APPLICATIONS

63

or 15dB lower than the DR of the inverting voltage amplifier. Emitter degeneration resistors will help in decreasing the loss. A CMOS mirror will also do better, because a better use of the supply-voltage range IS made because of the larger intrinsic source resistances.

The dynamic range of a current mirror is generally:

DR =(n/(n + 1)2)(V IV )P /4kTB

T' sup sup e

DR=(1/4)(1/40)P /4kTB =89-5-16=68dB

sup e

(3.5.4)

with: Vr=kTlq=1140, n=L, Vsup=lV, Psup=16/J-W, class -A, and lMHz

Conclusion Current Mirror

From this equation it follows that in all amplifier realizations, in which the signal has to pass through a current mirror, the dynamic range over power ratio is quite low. This is the case in so-called "current-mode" amplifiers. Moreover, in these current-mode solutions, where the signal passes through one or more internal current mirrors, the accuracy of the transfer is limited to that of the matching of non-linear transistor characteristics of the internal current mirrors. Further, current-mode active network elements do not fit the basic and yet accurate nullor approach of Chapter 1. Instead, they have more complicated network models and cannot describe the overall transfer function more accuratetely than of the order of 0,5%.

Another approach, the so-called current-mode feedback does have a better high-frequency behavior than the normal approach, due to the elimination of a pole in the feedback network. However, this solution has higher noise and offset due to the feeding of an emitter or source bias current through the feedback network. And again it does not fit the basic nullor approach of Chapter 1. For these reasons, so-called "current-mode" solutions, like current-mode amplifiers, and current-mode feedback are not discussed in this book.


Non-Ideal Operational Amplifiers

If we take into account the non-idealities of the amplifiers, then the DR/Psup ratio is further reduced. Important non-idealities are caused by the input noise voltage and a restricted output voltage range.

The problem at the input is that we do not want to spill supply current in the input stage, while on the contrary, we need a large bias current for the active input devices in order to lower the input voltage noise. The input voltage noise can easily be estimated by the equivalent input series noise resistance Rneqs. For bipolar transistors Rneqs=r/2=kT/2q1e= Vr/2Ie' with Vr",,25mV at room temperature. For field-effect transistors we find Rneqs= r/s;» y/(2J1-Cox (W/2)ID)1!2, which is of the order of Rneqs=IOy/(ID)l!2 for transistors with a W/L ratio of 100, while y is of the order of 2. The W IL ratio has been chosen as large for analog applications to increase gm and lower noise and offset input voltages.

VSP
IB1+ IB1+ +IB2
R1 20 kQ Voul R2 20 kQ
R 10 kQ +
Voul
lin t T1 lin/2t

VSN r------,-------0 VSP

Fig. 3.5.7.a and b: Single and balanced current-to-voltage converter in a single-stage class-A realization with a DR = 89-5- 1 = 83dB for bipolar transistors, and DR =89-5- 2=82dB for CMOS transistors, R-R, Vsup=lV, Psup=16J1- W, IB=16J1-A, and IMHz.

An optimal solution would be to choose no separate input transistors, but to use one-stage amplifiers in which the input transistors are used as

3. APPLICATIONS

65

output transistors as well. This interesting realization will be shortly evaluated here. In fig. 3.5.7.a and 3.5.7.b, a single and balanced currentto-voltage application is shown with a one-stage single or balanced class-A transistor amplifier. The transistors T 1 and T 2 symbolize either bipolar or field -effect transistors.

Because of the class-A operation, we lose at least 5dB. With bipolar transistors the equivalent input noise resistor will be Rneqs z700Q for the single and 1500Q for the balanced version at a total supply current of 16,uA. The resulting extra noise is much lower than 1dB. With CMOS transistors, the equivalent input noise resistor will be Rneqs=5000Q for the single and 7000Q for the balanced version. The resulting extra noise is of the order of 2dB. At the output the signal cannot reach the rail within 100m V. This results in a loss of 1dB for bipolar as well as CMOS transistors.

We can avoid the -5dB loss of class-A biasing if we choose a class-AB push-pull configuration as shown in Fig. 3.5.8.a and 3.5.8.b.

102t rvsp I tl02 104t I ¢VSP
T2 T2 T4
R1 20 kQ YOU! R2 20 kQ
R10kQ +
you!
lin t T1 lin/2t T1 T3 t lin/2

VSN Fig. 3.5.8.a and b: Single and balanced current-to-voltage converter in a single-stage class-AB realization with a DR =89- 2=87dB for bipolar transistors and DR =89-5 = 84dB for CMOS transistors at V,upz1.2V, Psup= 20,u W, 1MHZ

The push-pull transistors have been connected with the emitters or sources


between the rails, while their bases or gates are connected. To ensure proper class-AB biasing the supply rail voltage has to be regulated at two diode voltages at a proper quiescent current IQ• The circuit may function at roughly 1.2V. While we have now avoided the -5dB loss of the class-A circuit with a class-AB approach, the input noise voltage has been increased because the quiescent current has been reduced strongly with no signal. If we choose the quiescent current at I/ IO of the maximum current, the extra noise reduces the DR by 2dB for bipolar transistors and 5dB for CMOS transistors. The result is still better than in class-A. However, we have to build an additional supply-voltage regulator which easily takes away another IdB from the dynamic range. Moreover, the solution is only valid for a fixed low voltage and for the inverting OpAmp type.

If we also have to take the DC offset into account, the dynamic range becomes even lower. At DC the dynamic range is DRmax DC= V2su/V2offr With bipolar transistors we may expect an offset of O.3mV, which results in DRmax Dc=70dB, at a supply voltage of IV. For field-effect transistors with an offset of 3mV the result is DR,nax Dc=50dB. Only chopping can elevate this limit.

Conclusion

We have shown that the thermal noise in the gain-setting resistors principally limits the dynamic-range over supply-power ratio of an analog signal operation to DR,nax/Psup = (4/ Jr)/4kTBe' At a supply voltage of I V and a gain setting resistor of 10kQ in a current-to-voltage converter, the supply power is 161'- W for sinusoidal signals and the dynamic range in a bandwidth of I MHz can never be better than DRmax=89dB, even for an ideal Operational Amplifier.

This maximum can only be obtained if the Operational Amplifier is able to, firstly, efficiently feed the full supply voltage range from rail-to-rail (R-R) to the load, and secondly, efficiently feed the supply current in a class-AB mode to the load or feedback resistor. In some cases, for

3. APPLICATIONS

67

instance in a voltage buffer input application, the Operational Amplifier must also be able, thirdly, to efficiently allow a common-mode signal from rail-to-rail. These three requirements impose the important requirements in the design of the input stages of Chapter 4 and output stages of Chapter 5.

In amplifiers where the above requirements can not be met, the dynamic range over power ratio is lower. For instance, when a diode prevents the output to go from rail-to-rail, but only allows an output voltage swing of O.3V at a supply voltage of IV, the dynamic range loses lOdB. When the biasing is not class-AB, but class-A, we lose at least 5dB or much more at low signal levels.

When we process analog signals through a bipolar current mirror without emitter degeneration, we lose l6dB of dynamic-range over power ratio. With degeneration or CMOS current mirrors the loss is still lOdB or more. Moreover, the accuracy and linearity of the overall transfer function is only as good as the matching of highly non-linear internal transistor-diode characteristics. For these reasons we will leave out current-mode amplifiers in this book.

3.6 Problems

Problem 3.1

The circuit in Figure 3.2.1 shows a non-inverting voltage amplifier built around the operational amplifier model explained with Figure 2.2.1. The impedances in the external circuit are Zs = IOKil, Z1 = IOKil and Zz=40Kil. Z, is the load impedance and has a value of 50KQ. The parameters for the operational amplifier macromodel are: ,,",offs = I Om V, I;offs=IOnA, I;b;Gs=50nA, Y;d=1120Mil, Y;c1=Y;c2=11100Mil, H;=80dB, Y;=IOOS and Yo=IIIKQ. Calculate the output voltage for an input voltage of Vs=IOmV, IOOmV, IV.

(3.6.1)


Solution

The matrix Equation (3.2.1) shows the relations between input/output voltages and currents for the application circuit. Solution starts by calculating the parameters Avll, Avo :r;o Zte, Aie·

These values can be used to calculate the nominal and error terms III Equation (3.2.1):

With these values replaced in the matrix Equation (3.2.1), and considering the voltage-current relations at the input and output

Vs-V,=Zi, Vz=Z/z

(3.6.2)

With these values replaced in the matrix Equation (3.2.1), and considering the voltage-current relations at the input and output

3. APPLICATIONS

69

VS-Vl=Z/1 V2=ZJ2

(3.6.3)

the following equation is obtained for V2:

1 1 1 Zs Zs

--+------+------

Avn Ave Zlte Zte Z~ie

(3.6.4)

Replacing the values for circuit parameters and signal source voltage, the three output voltage values result:

ViIOmV) =-4.9mV V2( 100m V) =445m V V2(1 V)=4.94V

(3.6.5)

Problem 3.2

Figure 3.4.1 shows a voltage-to-current converter built around the OFA macro model explained with Figure 2.4.1. The external components are Zs = IOKQ, Y2 = 1150kQ, 1'; = 1150kQ. The macromodel parameters have the following values: V;affs = 10m V, liaffs = lOnA, libias = lOOnA, Y;dd= 1110MQ, Y;el=Y;e2=11100MQ, Hi=80dB, Yr=O.IS, Yadd=IIIMQ, labias=lOOnA, Yael = Yae2=11IOOMQ, Ha=60dB. Calculate the output current 12 for a signal voltage Vs = 100m V, 1 V, 5V.

Solution

The matrix Equation (3.4.1) shows the relations between input/output voltages and currents for the application circuit. Solution starts by calculating the parameters Yrn' Yre' Ave' Aie, Zte'

70


(3.6.6)

VS-Z/l =V1 12=-YtV2

(3.6.7)

With these values replaced in the matrix Equation (3.4.1) and considering the voltage-current relations at the input and output

the following equation is obtained for [2:

(3.6.8)

Replacing the values for circuit parameters and signal source voltage, the three output voltage values result:

3. APPLICATIONS

71

12(100mv)= -1.47JLA 12(1v)=-19.4JLA 12(5 v) = -99.18JLA

(3.6.9)

Problem 3.3

The instrumentation amplifier shown in Figure 3.5.3. b relies on a differential input buffer and an inverting differential amplifier. Using the resistor nominal values R] =R3=4KQ, R2=R4=40KQ, a supply voltage Vsp=3V, VsN=OVand non-ideal operational amplifiers with input common mode voltage range LlVIN= (Vsp-1 V; Vs,vO. 5V), output voltage range LlVOUT= (Vsp-0.2V; VSN+0.2V) and ~offs=10mV, calculate the maximum input signal which can be amplified without distortion. Using this value, calculate the ratio of dynamic range over dissipated power which can be obtained with this amplifier over a bandwidth Llf=1MHz if each operational amplifier draws Id=100 JLA and has an input noise of Sf= 10n VI vHz. The resistors have a precision of LlR = 2 % . Boltzmann IS constant is K=1.38·1(f23, temperature T=300K.

Solution

The first limit is introduced at the input by the differential buffer, as both input operational amplifiers are connected as repeaters, thus not allowing all of the input voltage range to be used:

Vill1ill = Vioffs + V votrtoo» = VSN+0.2V + Vioffs =0.210V Vill1ax = V VINmax - Vioffs = Vsp -1.0V- Vioffs = 1.990V

(3.6.10)

Another factor of R/R] is lost in order not to saturate the output, which limits the input signal to

V. - v. . - (v VOUTlI1ax - V VO UTII1 ill) 0.26 V

unax tmtn

10

(3.6.11)


R2(1 +!1R)

A - =10.4

Vmax R1(1-M)

(V -V )

V. -V .. = VOUTmax VOUTmin =0.25V

unax tmtn A

Vmax

Considering also the maximum gain

(3.6.12)

this voltage range is reduced furthermore by the unwanted effect of the third operational amplifier input offset voltage:

(V -V )

V. -V .. - VOUTmax VOUTmin -2V. =0.248V

tmax mun A V ioffs

max

(3.6.13)

(3.6.16)

The dynamic range of the circuit is limited by the noise power and the maximum input signal. The input referred noise power is

(3.6.14)

and is composed from buffer noise and equivalent resistor noise (virtually the equivalent resistors are equal with R1, R3)' The dynamic range can now be calculated

(V. -V .. \2

DR=101oglo lmax !111m! -81.5dB

Pn

The ratio of dynamic range over dissipated power is

(3.6.15)

3.7 References

[3.1] Graeme, J.G. et al.

"Operational amplifiers design and applications"

3. APPLICATIONS

73

McGraw Hill Book Company, New York, 1971.

[3.2] Motchenbacher, C.D. and Fitchen, F.C.

"Low-noise electronic design"

John Wiley and Sons, New York, 1973.


"Instrumentation amplifiers: a comparative study on behalf of monolithic integration"

IEEE Transactions on Measurement and Instrumentation, Vol. IM-25, Sept. 1976, pp 227-231.

[3.4] Dool, Bernard J. and Huijsing, Johan H.

"Indirect Current Feedback Instrumentation Amplifier with a Common- Mode Input Range that includes the negative rail" IEEE Journal of Solid-State Circuits, Vol. 28, No.7, July 1993, pp 743-749.

[3.5] Voorman, J.D.

"The gyrator as a monolithic circuit in electronic systems" Thesis, Katholieke Universiteit Nijmegen, Gema B.V., Eindhoven, June 1977.

[3.6] Adams, K.M. and Deprettere, E.F.A. and Voorman, J.O.

"The gyrator in electronic systems"

Advances in Electronics and Electron. Physics, Vol. 37, Academic Press, Inc., San Francisco, 1975.

[3.7] Groenewold, G.

"Optimal dynamic range integrated continuous-time filters" , Ph D thesis, Delft University of Technology, Delft,

The Netherlands, 1992.

Huijsing Operational Amplifiers

Uploaded by

Copyright:

Available Formats

Huijsing Operational Amplifiers

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Huijsing Operational Amplifiers

Uploaded by

Copyright:

Available Formats

THE KLUWER INTERNATIONAL SERIES IN ENGINEERING AND COMPUTER SCIENCE

You might also like

Huijsing Operational Amplifiers

Uploaded by

Copyright:

Available Formats

Huijsing Operational Amplifiers

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Huijsing Operational Amplifiers

Uploaded by

Copyright:

Available Formats

﻿THE KLUWER INTERNATIONAL SERIES IN ENGINEERING AND COMPUTER SCIENCE

You might also like

THE KLUWER INTERNATIONAL SERIES IN ENGINEERING AND COMPUTER SCIENCE