Current Sources And Voltage References A Design Reference For Electronics Engineers Linden T Harrison

Current Sources And Voltage References A Design
Reference For Electronics Engineers Linden T
Harrison download
https://ebookbell.com/product/current-sources-and-voltage-
references-a-design-reference-for-electronics-engineers-linden-t-
harrison-2290534
Explore and download more ebooks at ebookbell.com

Here are some recommended products that we believe you will be
interested in. You can click the link to download.
Electrical Energy Generation In Europe The Current Situation And
Perspectives In The Use Of Renewable Energy Sources And Nuclear Power
For Regional Electricity Generation 1st Edition Jorge Morales Pedraza
Auth
https://ebookbell.com/product/electrical-energy-generation-in-europe-
the-current-situation-and-perspectives-in-the-use-of-renewable-energy-
sources-and-nuclear-power-for-regional-electricity-generation-1st-
edition-jorge-morales-pedraza-auth-5140626
Retirement Security In The United States Current Sources Future
Prospects And Likely Outcomes Of Current Trends Jack L Vanderhei
https://ebookbell.com/product/retirement-security-in-the-united-
states-current-sources-future-prospects-and-likely-outcomes-of-
current-trends-jack-l-vanderhei-1623942
Electrical Energy Generation In Europe The Current And Future Role Of
Conventional Energy Sources In The Regional Generation Of Electricity
1st Edition Jorge Morales Pedraza Auth
https://ebookbell.com/product/electrical-energy-generation-in-europe-
the-current-and-future-role-of-conventional-energy-sources-in-the-
regional-generation-of-electricity-1st-edition-jorge-morales-pedraza-
auth-4950572
Paclitaxel Sources Chemistry Anticancer Actions And Current
Biotechnology Mallappa Kumara Swamy
https://ebookbell.com/product/paclitaxel-sources-chemistry-anticancer-
actions-and-current-biotechnology-mallappa-kumara-swamy-49432912

Plasma Source Mass Spectrometry Current Trends And Future Developments
Special Publications 1st Edition J G Holland D R Bandura
https://ebookbell.com/product/plasma-source-mass-spectrometry-current-
trends-and-future-developments-special-publications-1st-edition-j-g-
holland-d-r-bandura-2007354
Current Sources Voltage References 2005th Edition Linden Tharrison
https://ebookbell.com/product/current-sources-voltage-
references-2005th-edition-linden-tharrison-230910504
Active Power Decoupling Technology In Singlephase Currentsource
Converters Yonglu Liu
https://ebookbell.com/product/active-power-decoupling-technology-in-
singlephase-currentsource-converters-yonglu-liu-51148184
Current Issues In Knowledge Management Premier Reference Source 1st
Edition Murray E Jennex
https://ebookbell.com/product/current-issues-in-knowledge-management-
premier-reference-source-1st-edition-murray-e-jennex-1479122
Reenvisioning Theological Education Exploring A Missional Alternative
To Current Models Mr Robert J Banks
https://ebookbell.com/product/reenvisioning-theological-education-
exploring-a-missional-alternative-to-current-models-mr-robert-j-
banks-11123336

CURRENT SOURCES &
VOLTAGE REFERENCES

CURRENT SOURCES &
VOLTAGE REFERENCES
Linden T. Harrison
AMSTERDAM • BOSTON • HEIDELBERG • LONDON
NEW YORK • OXFORD • PARIS • SAN DIEGO
SAN FRANCISCO • SINGAPORE • SYDNEY • TOKYO
Newnes is an imprint of Elsevier Newnes

Newnes is an imprint of Elsevier
30 Corporate Drive, Suite 400, Burlington, MA 01803, USA
Linacre House, Jordan Hill, Oxford OX2 8DP, UK
Copyright © 2005, Elsevier Inc. All rights reserved.
No part of this publication may be reproduced, stored in a retrieval system, or
transmitted in any form or by any means, electronic, mechanical, photocopying,
recording, or otherwise, without the prior written permission of the publisher.
Permissions may be sought directly from Elsevier’s Science & Technology Rights
Department in Oxford, UK: phone: (+44) 1865 843830, fax: (+44) 1865 853333, e-
mail: permissions@elsevier.co.uk. You may also complete your request on-line via
the Elsevier homepage (http://elsevier.com), by selecting “Customer Support” and
then “Obtaining Permissions.”
Recognizing the importance of preserving what has been written, Elsevier prints its
books on acid-free paper whenever possible.
Library of Congress Cataloging-in-Publication Data
Application Submitted.
British Library Cataloguing-in-Publication Data
A catalogue record for this book is available from the British Library.
ISBN: 0-7506-7752-X
For information on all Newnes publications
visit our Web site at www.books.elsevier.com
05 06 07 08 09 10 10 9 8 7 6 5 4 3 2 1
Printed in the United States of America

DEDICATION
This book is dedicated to the memory of these eight outstanding pioneers, now passed away, but
whose contributions to the semiconductor industry have been immeasurable:
Robert J. Widlar
Legendary analog IC designer and engineer. Created the first monolithic op amps, bandgap voltage
references, current sources, and linear regulators. In 1981 co-founded Linear Technology Corp. with
friends Robert Dobkin, George Erdi, and Robert Swanson.
David Talbert
Widlar’s processing specialist at both Fairchild Semiconductor, then later at National Semiconductor.
He transformed Widlar’s revolutionary designs into real silicon devices, and created the first “super beta”
transistors that are used in bipolar analog ICs.
Dr. Robert Noyce
Physicist, and co-inventor of the IC (planar design) in 1964. Created several types of alloy transistors.
Co-founded Fairchild Semiconductor, Intel, Sematech, and the Semiconductor Industry Association.
Previously worked with Nobel Laureate, Dr. Shockley. Oversaw the development of Intel’s earliest
microprocessor and memory products.
Dr. Jean Hoerni
Swiss-born physicist, mathematician, and semiconductor researcher. Invented the all-important
PlanarTM process (1959), and created some of the first silicon BJTs and JFETs. Previously worked
with Nobel Laureate, Dr. Shockley. Co-founder of Fairchild Semiconductor, Teledyne Amelco, Union
Carbide Electronics, and Intersil.
Dr. Willis Adcock
Chemist, and semiconductor researcher. In 1954 he created the industry’s first silicon-based
semiconductor material at Texas Instruments Inc., who then introduced the world’s first silicon
transistors. Dr. Adcock later recruited Jack Kilby, who went on to co-invent the IC (mesa design), and
receive the Nobel Prize for physics in 2000.
Dr. Karl Lark-Horovitz
Austrian-born chemist, physicist, and materials researcher. Led the Physics Dept. at Purdue University
for several decades. It was his work with germanium during WWII, that enabled making the most
durable rectifiers available, for military radar systems.
Dr. Russell Ohl
Chemist, materials researcher, and HF radio pioneer at AT&T Bell Labs. Discovered the PN
junction, and created the first silicon diodes in 1940. Invented the modern solar cell.
Dr. Julius E. Lilienfeld
German-born physicist, researcher, and US Patent-holder (1930), who laid the foundations for the
voltage-controlled FET. Also invented the electrolytic capacitor.
“Tall oaks from little acorns grow”
David Everett (1769 - 1813)

vii
Contents
List of Figures and Photos xiii
List of Tables xxv
Acknowledgements xxix
1. A Short History of References 1
1.1 Introduction 1
1.2 The ﬁrst JFETs and op amps 3
1.3 The ﬁrst bandgaps 5
1.4 The buried-zener debuts 7
1.5 Advancements in technology 8
1.6 Other topologies emerge 9
2. An Introduction to Current Sources 13
2.1 An overview 13
2.2 Precision resistors, networks, and trimmers 21
2.3 Essential development equipment 29
2.3.1 Bench power supply unit 29
2.3.2 A thermostatically controlled oven 30
2.3.3 A calibrated, precision ammeter 30
3. The P-N Junction 31
3.1 Characteristics of the P-N junction 31
4. Using BJTs to Create Current Sources 47
4.1 Characteristics of the BJT 49
4.2 Using the BJT as a current source 67
4.3 Widlar current sources 74
4.4 Wilson current mirrors 81
4.5 Wyatt current source 87
4.6 Multiple current mirrors 89
4.7 Cascode current mirrors 91
4.8 Current scaling 95

viii
Contents
4.9 Modified current sources and example applications 102
4.9.1 Running the current source from split
power supplies 102
4.9.2 Improving power supply rejection 103
4.9.3 Alternative current divider 104
4.9.4 Modified three-transistor mirror-source 104
4.9.5 Current source linearly charges capacitor in VCO 105
4.9.6 Current source in a high-frequency laser transmitter 107
4.9.7 Temperature-compensated current sink 108
4.9.8 Compound current mirrors 109
4.9.9 Current mirrors help DACs control oscillator
frequency and duty cycle 110
4.9.10 Using current sources as active loads 111
4.9.11 Modified current source squares the
reference current 113
4.9.12 Digitally controlled variable current source 116
4.9.13 High-pass filter’s response is set by compound
current mirrors 116
4.9.14 Simple LED current sources 118
4.9.15 A low-noise AC amplifier biased by an LED
current source 120
4.9.16 A composite BJT-JFET current source with very
high output impedance 121
4.9.17 A composite BJT-MOSFET high-power
current source 121
4.9.18 A DAC-controlled current pump uses a Wilson
current source 122
5. Using Precision Matched-Pairs, Duals, and Quads 125
5.1 Precision BJT matched-pairs 125
5.2 Quality dual transistors 129
5.2.1 General-purpose BJT duals and quads 133
6. Using JFETs and CRDs to Create Current Sources 137
6.1 The JFET paved the way 137
6.2 Characteristics of the JFET 140
6.3 Using the JFET as a current source 151
6.4 The JFET cascode current source 163
6.5 JFET current regulator diodes 166
6.5.1 Characteristics of the CRD 168
6.5.2 A design guide 175

Contents
ix
6.5.3 An overview of various popular CRD families 175
6.6 Using JFETs to create ultra-low-leakage diodes 180
7. Creating Medium-Power Current Sources with DMOS FETs 185
7.1 Depletion-mode DMOS FETs 186
7.2 The importance of silicon-gate 188
7.3 Characteristics of depletion-mode DMOS FETs 190
7.4 Depletion-mode DMOS current sources 195
7.5 The cascode DMOS current source 197
7.6 The JFET-DMOS cascode current source 199
7.7 Lateral depletion-mode DMOS FETs 200
8. Creating Current Sources with Power MOSFETs 203
8.1 Characteristics of enhancement-mode MOSFETs 204
8.2 Using the enhancement-mode MOSFET as a
current source 218
8.3 Using “smart” power MOSFETS 228
8.4 IXYS power MOS current sources 229
8.5 Lateral enhancement-mode MOSFETs 230
9. Using Analog CMOS Arrays to Create Current Sources 231
9.1 RCA pioneered CMOS 231
9.2 Characteristics of CMOS FETs 235
9.3 Using CMOS linear arrays to create current sources 241
9.3.1 CMOS cascode current sources 245
9.4 Using ALD’s programmable EPADs® to create
precision current sources 252
9.5 ALD breaks the gate-threshold barrier 253
10. Using Monolithic IC Current Sources and Mirrors 261
10.1 National’s LM134—a monolithic IC current source 261
10.2 Current source applications for the LM134 266
10.3 Using the LM134 as a temperature sensor 271
10.4 TI/Burr-Brown’s REF-200 monolithic current source 273
11. Creating Precision Current Sources with Op Amps and
Voltage References 281
11.1 How op amps evolved 281
11.2 Some op amp characteristics 286
11.3 Op amp supply bypassing and input protection 294
11.4 Creating current sources with op amps 296
11.5 Creating precision current regulators with op amps 311

x
Contents
12. An Introduction to Voltage References 319
12.1 Introduction and history 319
12.2 Understanding voltage reference speciﬁcations 324
12.2.1 Initial accuracy (initial error) 326
12.2.2 Temperature drift (tempco, TCVo) 326
12.2.3 Long-term drift 332
12.2.4 Noise 333
12.2.5 Thermal hysteresis 335
12.2.6 Line regulation 336
12.2.7 Load regulation 336
12.2.8 Maximum output current rating (IOUT; mA) 336
12.2.9 Supply voltage range 336
12.2.10 Supply current (IS) or quiescent current (IQ) 337
12.2.11 Dropout voltage 338
12.2.12 Turn-on settling time (ton; µS) 338
12.2.13 Turn-on drift (dV/T) 338
12.2.14 Transient response 338
12.2.15 Sleep/Enable 338
12.2.16 Power dissipation 338
12.3 Enhancing the voltage reference design 339
12.3.1 Input and output bypassing 339
12.3.2 Noise reduction 341
12.3.3 Trimming 346
12.4 Unused terminals on the package 354
12.5 Package types 355
12.6 PCB layout 356
12.7 Why not do it yourself? 356
12.8 Comparing precision 359
13. The zener Diode and the TC zener Reference 363
13.1 Introduction 363
13.2 Characteristics of the zener diode 365
13.3 Some simple zener applications 382
13.4 Temperature-compensated zeners 393
14. Characteristics of Monolithic Voltage References 403
14.1 Bandgap voltage references 404
14.2 Buried-zener voltage references 415
14.3 The XFET® voltage reference 421
14.4 The Intersil/Xicor FGA™ voltage reference 424
14.5 Low-voltage considerations 426

Contents
xi
14.6 Comparing the different topologies 433
15. A Review of Some Outstanding Monolithic Voltage References
and Their Applications 437
Introduction 437
15.1 Applying the bandgap shunt reference 438
Analog Devices ADR510 441
National Semiconductor LM4051-ADJ 444
Maxim MAX6138 446
Maxim MAX6006 448
Linear Technology LT1634 449
15.2 Applying ﬁxed-series bandgap references 451
National Semiconductor LM4140 455
National Semiconductor LM4130 457
Analog Devices AD780 461
Maxim MAX6129 463
Maxim MAX6126 465
15.3 Applying adjustable-series bandgaps 466
Maxim MAX6325 469
Maxim MAX6037-ADJ 470
Intersil/Xicor x60250 472
Digital compensation 476
15.4 Using the Analog Devices’ XFET® reference 478
Analog Devices ADR431A 479
15.5 Applying buried-zener references 483
Texas Instruments REF102 488
Linear Technology LTZ1000 491
15.6 Applying the Intersil/Xicor FGA™ X60008 494
Intersil/Xicor x60008 495
Creating precision current source circuits 498
ISL60002 498

xii
Contents
ISL60007 498
15.7 Multiple voltage references and multiple loads 501
Digitally selectable 5-reference calibrator 502
Same voltage to multiple loads 503
15.8 Monolithic voltage references—a look to the future 506
A. References and Tables 513
Power of 10 and Equivalents 513
Temperature scale conversion 515
Precision resistor comparisons 515
Gain vs. dB conversion Table 516
Reactance chart for filter design 517
Small signal model for the BJT 518
Simplified BJT models 518
BJT operating modes Table 519
JFET models 519
MOS transistor models 520
PPM to % converter 521
Allowable noise levels for n-bit systems 521
Different voltage reference configurations 522
Allowable tempco requirements 523
B. Glossary 525
C. Bibliography 535
C.1 PART I—Current Sources 535
C.2 PART II—Voltage References 540
D. Contact Information 543
D.1 SEMICONDUCTOR MANUFACTURERS 543
D.2 ELECTRONICS DISTRIBUTORS 549
D.3 PRECISION PASSIVES MANUFACTURERS 550
D.4 INSTRUMENTATION MANUFACTURERS 555
D.5 Magazines & Periodicals 557
Index 559

xiii
Figures
List of Figures and Photos
Figure 1.1 Pioneers of the industry’s earliest JFETs and CRDs 3
Figure 1.2 Examples of Widlar current sources 4
Figure 1.3 Precision reference circuits from the mid-1970s 5
Figure 1.4 National Semiconductor’s LM109 regulator 6
Figure 1.5 Robert Dobkin’s novel buried-zener structure 7
Figure 1.6 LM134 monolithic adjustable current source 8
Figure 1.7 Analog Devices’ XFET®
voltage reference 10
Photo 1.1 A photomicrograph of Intersil’s x60008 reference 12
Figure 2.1 A typical transistor amplifier’s Q-point 14
Figure 2.2 Common current source symbols 15
Figure 2.3 Examples of different types of current sources 16
Figure 2.4 How current sources are used in op amps 17
Figure 2.5 Super-gain transistors first used in National’s LM108 18
Figure 2.6 National Semiconductor’s LM6361 Fast VIP®
op amp 19
Figure 2.7 The input stage for a typical C-B RRI op amp 20
Figure 2.8 National Semiconductor’s LMC6482/4 CMOS op amps 21
Photo 2.1 Caddock Electronics’ thin-film resistor arrays 27
Figure 2.9 Adding series resistors maintains high accuracy 28
Figure 2.10 A simple low-noise regulated power supply 29
Figure 3.1 Bohr models of the silicon and aluminum atoms 32
Figure 3.2 Part of the Periodic Table of Elements 33
Figure 3.3 The bandgap 33
Figure 3.4 The diode symbol 34
Figure 3.5 Forward-biasing the diode 35
Figure 3.6 A simple bias and TC compensation scheme 37

xiv
Figure 3.7 Vfwd vs Ifwd vs Tamb 38
Figure 3.8 Reverse-biasing the diode 38
Figure 3.9 A diode’s characteristic nonlinear V/I curve 39
Figure 3.10 Reverse current vs reverse voltage for 1N4148 40
Figure 3.11 Comparison of a diode and BJT’s characteristics 42
Figure 3.12 FET, and CMOS input/output protection networks 44
Figure 3.13 Creating diodes from BJT and FET transistors 45
Figure 4.1 Comparing Germanium and Silicon’s V-I curves 48
Figure 4.2 Showing the structure and operation of the BJT 50
Figure 4.3 Examples of popular transistor packages 51
Figure 4.4 Some BJT characteristics 53
Figure 4.5 Current gain (hFE) test circuit 54
Figure 4.6 Showing how temperature affects the BJT’s hFE 56
Figure 4.7 Showing the mechanisms of alpha and beta in the BJT 57
Figure 4.8 Various V(BR)CE breakdown voltage test circuits 59
Figure 4.9 ICO -(ICBO) the BJT’s reverse leakage current 60
Figure 4.10 ICEO reverse leakage current versus temperature 61
Figure 4.11 IEBO: the BJT’s reverse-biased leakage current 62
Figure 4.12 Test circuit measures ICEO, ICEX, ICER, and ICES 62
Figure 4.13 Showing the variation of VBE with the IE (PTAT) 65
Figure 4.14 VBE(sat) vs IC for 2N3904 and 2N3906 66
Figure 4.15 A BJT’s VCE(sat) measurement circuit 66
Figure 4.16 Early BJT current sources - 1960s, and 1970s 69
Figure 4.17 The importance of transistor matching 69
Figure 4.18. Basic BJT current sources 71
Figure 4.19 A simple 500 µA current source in NPN and PNP 73
Figure 4.20 Widlar’s original current sink 74
Figure 4.21 Widlar’s Leap - transforms the current source 75
Figure 4.22 Examples of Widlar’s compensation techniques 77
Figure 4.23 Widlar’s improved current sink 78
Figure 4.24 A basic three-transistor mirror-sink 78
Figure 4.25 Basic Wilson current mirrors 82
Figure 4.26 Low-voltage basic Wilson current mirrors 84

xv
Figure 4.27 The Full Wilson PNP and NPN current mirrors 86
Figure 4.28 The Wyatt Cascode Peaking Current Source 87
Figure 4.29 Examples of multiple current mirrors 90
Figure 4.30 Emitter degeneration compensates for mismatches 91
Figure 4.31 The Cascode boosts frequency response and Zout 92
Figure 4.32 Multiple cascode current mirrors 94
Figure 4.33 Current sources used in op amps as active loads 95
Figure 4.34 Examples of PNP and NPN current multipliers 96
Figure 4.35 Examples of modified Wilson current multipliers 98
Figure 4.36 Examples of NPN and PNP current dividers 100
Figure 4.37 Examples of modified Wilson current dividers 102
Figure 4.38 Adding an emitter resistor allows division 102
Figure 4.39 Simple current sources running on split (±) supplies 103
Figure 4.40 Current mirror with enhanced power-supply rejection 104
Figure 4.41 A novel BJT current divider 105
Figure 4.42 A modified 3-transistor mirror-source 106
Figure 4.43 Current mirror controls oscillator charge time 107
Figure 4.44 Current source sets IQ in a laser transmitter 109
Figure 4.45 Temperature-compensated BJT current sources 110
Figure 4.46 Compound current mirrors 111
Figure 4.47 Current sources control a frequency generator 112
Figure 4.48 Current sources used as amplifier active loads 113
Figure 4.49 Current mirrors provide exponential output current 114
Figure 4.50 Digital pot controls Full Wilson current source 116
Figure 4.51 Compound current mirrors tailor a filter’s response 117
Figure 4.52 LED temperature-compensates transistor’s VBE 119
Figure 4.53 LED-referenced current source enhances amplifier 120
Figure 4.54 Composite cascoded BJT-JFET current source 122
Figure 4.55 A high-power BJT/Power MOSFET current sink 123
Figure 4.56 A digitally controlled bipolar current pump 124
Figure 5.1 Excellent matching using precision matched pairs 126
Figure 5.2 Pin-out for LM194/LM394 and MAT-02 precision duals 127
Figure 5.3 Improving matched duals for use as current sources 129

xvi
Figure 5.4 LM3046M, an NPN transistor monolithic array 131
Figure 5.5 LM3046M current source applications 132
Figure 5.6 Packages used for the popular 2N3906 BJT 133
Figure 6.1 The FET family tree 138
Figure 6.2 Typical JFET applications 139
Figure 6.3 The structure and operation of the JFET 141
Figure 6.4 Typical output characteristics for an N-channel JFET 143
Figure 6.5 The Q-point for a typical N-channel JFET ampliﬁer 144
Figure 6.6 A JFET’s leakage current vs. temperature 145
Figure 6.7 IDSS measurement for an N-channel JFET 146
Figure 6.8 VGS(off) measurement for an N-channel JFET 147
Figure 6.9 Typical JFET transfer versus temperature curves 148
Figure 6.10 The JFET’s small-signal transconductance model 149
Figure 6.11 A JFET’s transconductance (gfs) characteristics 150
Figure 6.12 Calculating the N-channel JFET’s output impedance 151
Figure 6.13 JFET output conductance (gos) characteristics 152
Figure 6.14 The simplest JFET current source - the IDSS value 153
Figure 6.15 Basic JFET current sources 154
Figure 6.16 Examples of a JFET’s zero TC point 156
Figure 6.17 Example of a JFET’s zero TC point and load-lines 157
Figure 6.18 Comparing single and cascode JFET current sources 165
Figure 6.19 Examples of JFET current sources 167
Figure 6.20 JFET current regulator diode, its symbol and model 168
Figure 6.21 CRD equivalent circuits 169
Figure 6.22 Ultra-low-voltage operation with CRDs 169
Figure 6.23 The CRD’s V-I characteristic curve 170
Figure 6.24 CRDs in series for high-voltage applications 171
Figure 6.25 Reverse-bias equivalent of the CRD 171
Figure 6.26 Paralleling CRDs for higher current operation 172
Figure 6.27 Popular packages used for CRDs 174
Figure 6.28 Some typical CRD applications 180
Figure 6.29 JFET Pico Amp diodes 181
Figure 6.30 JFET/Pico-amp diode applications 182

xvii
Figure 7.1 The DMOS family tree 187
Figure 7.2 The structure and operation of the DMOS FET 187
Figure 7.3 Comparison of DMOS vertical and lateral structures 188
Figure 7.4 Depletion-mode FETs provide ultra-stable output levels 191
Figure 7.5 BVDSS measurement for an N-channel DMOS FET 192
Figure 7.6 IDSS measurement for an N-channel DMOS FET 192
Figure 7.7 Transfer characteristics for a depletion-mode FET 193
Figure 7.8 VGS(off) measurement for an N-channel DMOS FET 194
Figure 7.9 N-channel depletion-mode DMOS current source 196
Figure 7.10 A depletion-mode DMOS cascode current-source 198
Figure 7.11 A JFET - DMOS FET cascode current source 199
Figure 7.12 A high-voltage ramp generator 201
Figure 7.13 High voltage protection using DMOS current sources 201
Figure 8.1 Structure and operation of enhancement-mode FETs 206
Figure 8.2 Cross-sectional view of a vertical DMOS FET cell 207
Figure 8.3 Intrinsic parasitic elements present in MOSFETs 208
Figure 8.4 BVDSS measurement for an N-channel power MOSFET 209
Figure 8.5. Enhancement-mode MOSFETs IDSS differences 209
Figure 8.6 Output characteristics for a power MOSFET 210
Figure 8.7 ID(on) measurement for an N-channel power MOSFET 211
Figure 8.8 An enhancement-mode FET’s on-state resistance 212
Figure 8.9 The ID vs. VGS transfer characteristics 213
Figure 8.10 VGS(th) measurement circuit 213
Figure 8.11 The N-channel MOSFET’s capacitance model 214
Figure 8.12 The Safe Operating Area curves for a power MOSFET 217
Figure 8.13 Power dissipation in a power MOSFET 218
Figure 8.14 A Zener-based, power MOSFET current source/sink 219
Figure 8.15 A dual MOSFET current-source/sink 220
Figure 8.16 A combination CRD+BJT+MOSFET current source 221
Figure 8.17 A MOSFET+ Bandgap voltage reference current sink 222
Figure 8.18 A CMOS voltage-reference+MOSFET current-source 223
Figure 8.19 Important design considerations for a power MOSFET 224
Figure 8.20 Designing with the Maxim MAX6160 reference 226

xviii
Figure 8.21 Creating a high-power (>10-amp) current-sink 227
Figure 8.22 A “smart” enhancement-mode power MOSFET 228
Figure 8.23 This “smart” MOS device protects itself 229
Figure 8.24 IXYS’ high-voltage integrated current source 230
Figure 9.1 RCA introduced the ﬁrst commercial CMOS devices 232
Figure 9.2 CMOS input and output protection networks 235
Figure 9.3 The structure, operation, and models of CMOS FETs 236
Figure 9.4 Typical output characteristics for CMOS FETs 239
Photo 9.1. Advanced Linear Devices’ CMOS transistor arrays 243
Figure 9.5 Simple current mirrors using CMOS matched pairs 244
Figure 9.6 Different types of MOSFET cascode current sources 246
Figure 9.7 Implementing current sinks with ALD’s CMOS arrays 248
Figure 9.8 A multiple current source using a P-channel array 249
Figure 9.9 A MOS cascode multiple current mirror 249
Figure 9.10 A CMOS current multiplier 250
Figure 9.11 Current and voltage biasing in a CMOS op amp 251
Figure 9.12 Current mirrors set bias levels in an ampliﬁer 252
Figure 9.13 ALD’s EPAD®
programming system 253
Figure 9.14 Programming an EPAD®
matched transistor 254
Figure 9.15 Characteristics vs. ambient temperature 255
Figure 9.16 A mirror source using an EPAD®
matched pair 255
Figure 9.17 A mirror current sink using an ALD ETRIM™ array 256
Figure 9.18 A CMOS multiple current sink using an ALD ETRIM™ 257
Figure 9.19 A mirror current source using an ALD ETRIM™ array 258
Figure 9.20 A current sink multiplier using an ALD ETRIM™ array 258
Figure 9.21 A near-zero TC current sink using two ETRIM™ pairs 259
Figure 10.1 National Semiconductor’s LM134 - internal circuit 263
Figure 10.2 The LM134 used as a two-terminal current source 264
Figure 10.3 Methods for improving the LM134’s tempco 265
Figure 10.4 Setting the bias current for a micropower op amp 266
Figure 10.5 Creating a precision 10-nA current source 267
Figure 10.6 Setting the operating current for a reference stack 268

xix
Figure 10.7 Combining the LM134 with a JFET cascode 269
Figure 10.8 Combining the LM134 with a JFET/DMOS cascode 270
Figure 10.9 A precision ramp-generator using the LM334 271
Figure 10.10 A window comparator with a zero tempco reference 272
Figure 10.11 A simple Kelvin thermometer using an LM234-Z3 273
Figure 10.12 Texas Instruments’ REF-200 current source IC 274
Figure 10.13 REF-200 reverse polarity protection 275
Figure 10.14 A 300µA floating current source 275
Figure 10.15 A 200µA precision current sink 276
Figure 10.16 Combining a P-channel JFET with the REF200 277
Figure 10.17 A 200µA cascoded current source using JFETs 278
Figure 10.18 A super-cascoded 229µA current source 279
Photo 11.1 Robert J. Widlar, inventor of the IC op amp 283
Figure 11.1 Part of the op amp family tree 286
Figure 11.2 A CMOS op amp is made up of several stages 287
Figure 11.3 A bipolar op amp’s front-end amplifier and biasing 289
Figure 11.4 The importance of decoupling op amp power supplies 295
Figure 11.5 Op amp input-protection using JFET pico-amp diodes 295
Figure 11.6 Adding a trimmer changes the TCR drift 302
Figure 11.7 Simple but reliable op amp current sources 304
Figure 11.8 Op amp current sources with added JFET cascodes 307
Figure 11.9 Ultra high-precision op amp current sources 308
Figure 11.10 A single-ended, bilateral op amp current source 310
Figure 11.11 The Howland current pump 311
Figure 11.12 A fast, precision op amp current pump 312
Figure 11.13 A precision current sink with a “smart” MOSFET 312
Figure 11.14 A 100-mA precision current regulator 315
Figure 11.15 Low power op amp current sources 316
Figure 12.1 The voltage reference family tree 320
Figure 12.2 The voltage reference sets the DAC’s output range 321
Figure 12.3 Examples of early semiconductor references 322
Figure 12.4 LM109 regulator - first to use a bandgap reference 323

xx
Figure 12.5 Symbols used to depict voltage references 324
Figure 12.6 Configuring different types of voltage references 325
Figure 12.7 Typical output TCVo, graphed with the Box method 328
Figure 12.8 Typical output TCVo, graphed w/Butterfly method 329
Figure 12.9 Required tempco vs. temperature range (@1/2 LSB) 332
Figure 12.10 Examples of noise density characteristics 335
Figure 12.11 Adding a BJT to the output of a voltage reference 337
Figure 12.12 A voltage reference with input/output bypassing 340
Figure 12.13 Series references with noise reduction terminals 342
Figure 12.14 Noise filtering the voltage reference 343
Figure 12.15 Kelvin connections can improve performance 344
Figure 12.16 Low-pass filter characteristics 345
Figure 12.17 Modified Sallen-Key, two-pole, low-pass filter 345
Figure 12.18 Three-pole, 10-kHz, low-pass Chebyshev filter 346
Figure 12.19 Two-pole, 100-Hz low-pass filter with -40dB roll-off 347
Figure 12.20 The effect of trimming on reference accuracy 348
Figure 12.21 Series reference with a dedicated trim terminal 349
Figure 12.22 Op amps provide adjustment of the output voltage 350
Figure 12.23 Basic methods of temperature compensation 353
Figure 12.24 Using digital correction to compensate the tempco 354
Figure 12.25 Examples of voltage reference IC packages 355
Figure 12.26 Layout can be critical in some reference circuits 357
Figure 12.27 A quality bandgap circuit built with discrete parts 358
Figure 13.1 The zener diode family tree 364
Figure 13.2 Common zener applications and spin-offs 365
Figure 13.3 The zener diode symbol and polarity 366
Figure 13.4 A zener diode’s typical V-I curve 367
Figure 13.5 Zener diodes are available in various packages 368
Figure 13.6 Zener impedance decreases with increasing current 371
Figure 13.7 Dependance of breakdown voltage on temperature 372
Figure 13.8 Illustrating the zener’s zero TC0 current point 375
Figure 13.9 A zener voltage reference with a near-zero tempco 376
Figure 13.10 Design for worst-case power dissipation levels 377

xxi
Figure 13.11 A zener’s power vs. temperature derating curves 379
Figure 13.12 Basic thermal models for zener diodes 380
Figure 13.13 Series-connected zeners form a shunt regulator 384
Figure 13.14 A zener shunt regulator with an added BJT 386
Figure 13.15 A series regulator, using a zener, BJT, and a diode 387
Figure 13.16 A zener-MOSFET high-power current source 387
Figure 13.17 A zener-MOSFET-BJT power current source 388
Figure 13.18 A zener-protected ignition circuit 389
Figure 13.19 Zener protection of power MOSFETs and IGBTs 391
Figure 13.20 Over-voltage protection triggers an SCR crowbar 392
Figure 13.21 Zener voltage detector triggers a 555 monostable 392
Figure 13.22 The temperature-compensated zener - symbol 393
Figure 13.23 Typical construction of axial-leaded zener products 394
Figure 13.24 A schematic of the TC zener diode 395
Figure 13.25 A TC zener reference using the LM134/334 396
Figure 13.26 A TC zener-DMOS FET-op amp voltage reference 398
Figure 13.27 A high-voltage, low TC zener reference 400
Figure 14.1 Types of monolithic voltage references 403
Figure 14.2 The Widlar bandgap cell (a.k.a. the ∆VBE reference) 408
Figure 14.3 How bandgap compensation is achieved 410
Figure 14.4 The Brokaw cell 411
Figure 14.5 A practical Brokaw cell bandgap voltage reference 413
Figure 14.6 Creating zener diodes in monolithic IC form 416
Figure 14.7 The structures of surface- and buried-zeners 417
Figure 14.8 The versatility of the LM199 buried-zener reference 418
Figure 14.9 The LM199 creates a buffered 12V reference 419
Figure 14.10 Analog Devices’ AD588 ultra-precision buried-zener 419
Figure 14.11 Analog Devices’ ADR431 XFET®
schematic 422
Figure 14.12 Adjusting the output voltage of the XFET®
424
Figure 14.13 Intersil’s x60008 FGA™ voltage reference 427
Figure 14.14 Adding a Kelvin output to an Intersil FGA™ 428
Figure 14.15 The LM10 pinout and functional diagram 430
Figure 14.16 Some low-voltage LM10 reference applications 431

xxii
Figure 15.1 The two-terminal (shunt) bandgap voltage reference 439
Figure 15.2 ADR510 shunt sets the range for a 10-bit DAC 442
Figure 15.3 ADR525 shunt provides 2.5-volt to a dual 8-bit DAC 443
Figure 15.4 The LM4051-ADJ adjustable micropower shunt 445
Figure 15.5 Combining a MAX6138 shunt with a comparator 448
Figure 15.6 LT1634 provides both reference and power to a DAC 450
Figure 15.7 Creating a 3 µA current source with an LT1389 shunt 452
Figure 15.8 The basic 3-terminal series voltage reference 453
Figure 15.9 LT1461A series reference sets the range for an ADC 455
Figure 15.10 Adding a BJT to the output of an LM4140 reference 456
Figure 15.11 LM4130 provides a ﬁltered, bipolar reference 458
Figure 15.12 ADR390 provides outstanding specs in a 5-pin TSOT 459
Figure 15.13 A triple voltage reference, using ADR280 reference 461
Figure 15.14 AD780 doubles as temp. sensor or shunt reference 463
Figure 15.15 AD780 reference drives large capacitive loads 464
Figure 15.16 The bipolar range for ADC provided by MAX6129 465
Figure 15.17 MAX6126 provides ultra-low noise reference voltage 467
Figure 15.18 Some LT6650 low-voltage reference applications 468
Figure 15.19 MAX6325 series bandgap provides 16-bit accuracy 470
Figure 15.20 Create custom reference voltages with MAX6037 472
Figure 15.21. Xicor x60250 - digitally programmable reference 473
Figure 15.22 The ADR01 also provides a linear VPTAT voltage 476
Figure 15.23 Using EEPROM-based temperature-compensation 478
Figure 15.24 XFET®
provides low-noise reference to 16-bit ADC 480
Figure 15.25 Creating a Howland current pump using the XFET®
481
Figure 15.26 AD688 buried-zener reference has ±10V outputs 485
Figure 15.27 Enhancing the AD688 reference for 16-bit systems 486
Figure 15.28 AD586 buried-zener reference - schematic 487
Figure 15.29 Stacking AD586s provides three reference voltages 488
Figure 15.30 Texas Instruments’ REF102 buried-zener reference 490
Figure 15.31 Adding a BJT to LT1021A boosts load current 492
Figure 15.32 LTZ1000A buried-zener reference - schematic 493
Figure 15.33 LTZ1000 needs careful layout - delivers best specs 495
Figure 15.34 The ultra-high precision x60008 reference 497

xxiii
Figure 15.35 Creating precision current sources with the x60008 500
Figure 15.36 A digitally selectable, 5-reference system calibrator 503
Figure 15.37 One reference - multiple loads 505
Figure 15.38 P-channel MOSFETs provide very low-dropout 509
Figure 15.39 Pinout for high-performance voltage references 510
Appendix A1.1 Temperature scale conversion 515
Appendix A1.2 Precision resistor comparisons 515
Appendix A1.3 Gain (Av) vs. dB conversion Table 516
Appendix A1.4 Reactance chart for filter design 517
Appendix A1.5 Small-signal model for the BJT 518
Appendix A1.6 Simplified BJT models for PNP and NPN 518
Appendix A1.7 Transistor operating modes (CB, CE, CC) 519
Appendix A1.8 JFET models 519
Appendix A1.9 MOSFET models 520
Appendix A1.10 PPM to % converter 521
Appendix A1.11 Allowable noise levels for n-bit systems 521
Appendix A1.12 Voltage reference configurations 522
Appendix A1.13 Allowable tempco drift requirements 523

xxv
Tables
List of Tables
Table 2.1 Percentage (%) to parts-per-million (ppm) converter 23
Table 2.2 Comparison of various types of precision resistors 25
Table 4.1 Transistor operating modes 52
Table 4.2 The Relationship between VBE versus IC 64
Table 4.3 ∆VBE vs. current errors percentage 70
Table 4.4 LED color vs. forward voltage 119
Table 5.1 List of BJT precision matched-pairs and quads 128
Table 5.2 List of closely matched dual NPN and PNP transistors 130
Table 5.3 Manufacturers of dual and quad BJT transistors 134
Table 5.4 List of common transistors available as dual and quads 135
Table 6.1 JFET Vgs(OTC) and IDZ calculator 158
Table 6.2 List of low Gos and low Vgs(off) discrete JFETs 160
Table 6.3 JFET Current Source Design Check List 161
Table 6.4 Comparison of different CRD Series: Zd and Zk values 173
Table 6.5 US Manufacturers of Current Regulator Diodes 179
Table 7.1 List of depletion-mode vertical DMOS FETs 197
Table 9.1 Advanced Linear Devices’ matched CMOS transistors 242
Table10.1 LM134 monolithic current source family options 262
Table 11.1 Op amp input bias current requirements 291
Table 11.2 Some Precision Op Amps from U.S. Manufacturers 298

xxvi
List of Tables
Table 11.3 Maximum allowable drift for n-bit systems 300
Table 11.4 Performance comparisons of different references 300
Table 11.5 PPM and % converter for 1V and 2.5V references 305
Table 12.1 PPM to % converter for popular references 327
Table 12.2 Allowable tempco requirements for n-bit systems 331
Table 12.3 Allowable noise requirements ( µVpk-pk) 334
Table 12.4 Allowable noise requirements (nV/√Hz) 334
Table 12.5 Bit-accuracy vs. allowable drift and noise levels 352
Table 12.6 Performance comparisons - discrete vs. monolithic 360
Table 13.1 Percent to PPM converter for zener references 373
Table 13.2 Packages, wattage ratings, and derating factors 382
Table 13.3 Examples of 1N Series zeners 383
Table 13.4 Examples of other popular zeners 385
Table 13.5 Listing of 400-mW, axial DO-35 TC zeners 399
Table 13.6 US manufacturers of zeners, TC zeners and TVS 401
Table 14.1 Some popular bandgaps from the past 407
Table 14.2 Bandgap ∆ VBE vs. current-density ratios 411
Table 14.3 Comparing the different series reference topologies 434
Table 14.4 Checklist for voltage reference design 435
Table 15.1 Analog Devices’ ADR510, 1V shunt bandgap 441
Table 15.2 Analog Devices’ ADR520 family of shunt bandgaps 443
Table 15.3 National Semiconductor LM4051-ADJ shunt bandgap 444
Table 15.4 Maxim MAX6138 family of shunt bandgaps 446
Table 15.5 Maxim MAX6006 low power shunt bandgaps 449
Table 15.6 Linear Technology LT1634 family of shunt bandgaps 450
Table 15.7 Linear Technology LT1389 low power shunt bandgaps 451
Table 15.8 Linear Technology LT1461 series bandgaps 454
Table 15.9 National Semiconductor LM4140 series bandgaps 456
Table 15.10 National Semiconductor LM4130 series bandgaps 457
Table 15.11 Analog Devices’ ADR390 family of series bandgaps 459

List of Tables
xxvii
Table 15.12 Analog Devices’ ADR280 series bandgap 460
Table 15.13 Analog Devices’ AD780 family of series bandgaps 462
Table 15.14 Maxim MAX6129 family of series bandgaps 464
Table 15.15 Maxim MAX6126 family of series bandgaps 466
Table 15.16 Linear Technology LT6650 adj. series bandgap 467
Table 15.17 Maxim MAX6325 trimmable series bandgaps 469
Table 15.18 Maxim MAX6037 adjustable series bandgaps 471
Table 15.19 Intersil/Xicor x60250 programmable bandgap 472
Table 15.20 Analog Devices’ ADR01 trimmable series bandgaps 475
Table 15.21 Analog Devices’ ADR43xB family of series XFET®
479
Table 15.22 Analog Devices’ AD688 ±10-volt buried-zener 485
Table 15.23 Analog Devices’ AD586 +5-volt buried-zener 487
Table 15.24 Texas Instruments’ REF102 +10V buried-zener 489
Table 15.25 Linear Technology LT1021 family of buried-zeners 491
Table 15.26 Linear Technology LTZ1000/A 7V buried-zener 493
Table 15.27 Intersil/Xicor x60008 family of FGATM
references 496

xxix
Acknowledgements
The author would like to acknowledge the following individuals for their help and
advice in supplying information for this book:
Robert Dobkin:
Co-founder and CTO—Linear Technology Corporation
Inventor of the buried-zener voltage reference
Inventor of the monolithic temperature sensor
Co-inventor of the first commercial bandgap voltage reference
Creator of the first high-speed monolithic op amp
Creator of the first adjustable linear voltage regulator
Professor Thomas H. Lee:
Stanford Microwave ICs Laboratory
Department of Electrical Engineering
Stanford University
Robert A. Pease—Staff Scientist, and renowned columnist and guru
Naomi Mitchell—Senior Manager—Marketing Communications
National Semiconductor Corporation
Lisa Wong—Marketing Manager—Voltage References
Scott Wayne—Technical Communications Manager
Analog Devices, Inc.
Joe Neubauer—Technical Business Manager SP & C
John Van Zand—Publicity Manager
Maxim Integrated Products, Inc.

xxx
Acknowledgements
Carlos Laber—VP, Engineering
John M. Caruso—VP, Engineering
Andy Jenkins —Director of Marketing, Precision Analog Products
Jules Farago—Product Marketing Manager
Intersil Corporation
Robert Chao—President and Chief Engineer
John Skurla—Director of Marketing
Advanced Linear Devices, Inc.
Dr. Michael Wyatt
Honeywell Fellow and researcher
Space Systems Divn.
Honeywell Corporation
Don Howland—Director of Marketing
Linear Integrated Systems, Inc.
Dr. Mike Chudobiak
Avtech Electrosystems Co.
David L. Anderson
Snr. Applications Engineer
Caddock Electronics, Inc.
Vernon Bluhm
Former transistor researcher
(Syracuse, NY)
General Electric Company

Acknowledgements
xxxi
In addition, the author would like to thank the following individuals for their great
help and guidance with this book:
Harry Helms—my Editor at Elsevier Science & Technology
Don Snodgrass—who created the book’s original master CD
George Morrison—Senior Project Manager, Elsevier Science & Technology
Tara Isaacs—Senior Marketing Manager, Elsevier Science & Technology
Lori Koch—Promotions Coordinator, Elsevier Science & Technology
Mona Buehler—Pre-Production, Elsevier Science & Technology
Very special thanks go to:
Alan Rose, Tim Donar, and Ginjer Clarke at Multiscience Press, Inc., New
York for transforming my original Mac manuscript into this ﬁne book.
Thank you all so much !
Linden Harrison
lindenh248@aol.com

1
Chapter 1
A Short History of References
1.1 Introduction
Current sources and voltage references both depend on inherent characteristics of the
transistor, either the bipolar junction transistor (BJT) or the field-effect transistor (FET),
in order to operate properly. It seems only fitting then, in beginning this book, that we
first take a brief trip back in time, and trace a history of how and when some of these
products originated.
Actually, current sources and voltage references both predate the integrated circuit by
several decades. Current sources were originally created by simply using resistors or
by using resistors together with vacuum tubes. However, their combination did not pro-
vide much accuracy or stability. In the 1940s and ’50s, it was common to see vacuum
tube voltmeters (VTVMs) and other instrumentation based on the tube. Before being
used within integrated circuits, or designed as commercial IC products, voltage refer-
ences took the form of bulky and expensive laboratory standards. These included the
Weston cell, the Clark cell, as well as some types of batteries. Those types of labora-
tory standards were used for decades. The best-known standard cell was the Weston
cell, which produced a constant voltage output of 1.019 volts and was virtually inde-
pendent of temperature change. The Weston cell utilized an H-shaped glass container
and was powered by unique chemical actions. Great care was needed not to tip or
knock over the cell or to load the output with any appreciable current, because it would
temporarily cease operation. Recovery time could take days or even weeks. Both
types of cells produced stable voltages, though, accurate to a few parts per million
(ppm) or better. Modern Weston cells require being held in a temperature-controlled
bath. Mercury cells, originally developed during World War II, were also used as volt-
age references, mostly because they were small, cheap, and immune to the physical
problems of the fragile glass standard cells. Mercury cells provided an output of 1.35
volts at several milliamps (mA) and for more than 1,000 hours, but had much lower
accuracy and a higher temperature coefficient. All of these types have been mostly
superseded by semiconductor voltage references.
In modern times, the first milestone reached for a semiconductor-based reference was
the introduction of the zener diode, which serves as a voltage reference. This was cre-
ated in the late 1950s by its inventor Dr. Clarence Zener, a researcher at
Westinghouse Electric Company. Operation of the zener diode relies on some unique

2
Introduction
characteristics of a reverse-biased P-N junction and is usually operated so that it is in
parallel (shunt) with the load. It can be produced by one of several techniques, the
most common being a planar epitaxial method. Each zener diode’s particular specifi-
cations are mostly created by small differences in processing. Because it is a small
single component, it is still a popular workhorse today in many industrial and commer-
cial designs, because of its reasonable accuracy, ease of use, and low cost. The
downside of the zener was that it was temperature-sensitive, drifted, was noisy, and
above all required more than 6 volts to function. Low-voltage zeners were deemed to
be too unpredictable to use in any precision circuitry at the time.
As electronic instrumentation gradually became more sophisticated in the early 1960s,
some instrument designs called for an internal calibrated reference, which would
enable the instrument to calibrate itself. At first this took the form of using mercury cell
batteries, but then gradually, small discrete transistor-diode-zener circuits were used,
often housed in sealed plug-in modules. Along the way, engineers discovered that the
normal positive temperature coefficient of the zener could be compensated by the neg-
ative temperature coefficient of one or more series-connected rectifier diodes. They
found that by burning in, characterizing, and selecting zeners with a nominal value of
5.6 volts, and adding a rectifier diode (having a forward voltage of approximately 0.7
volts) in series, it would create a temperature-compensated (TC) zener reference.
However, it had to be run at a particular current of 7.5 mA, for which it was specified.
Burn-in time could be hundreds of hours or more. The longer the burn-in time and the
more characterization, the higher the cost. At the time, these specially created TC zen-
ers were the most accurate devices available. Attempts at using zeners of different volt-
ages did not produce as low a tempco as this TC zener combination. While the
temperature coefficient of the TC zener was very low (i.e., 1N827, 0.001%/˚C), the
combined voltage was typically around 6.3 volts, and so required a power supply of at
least 7 volts or higher to function properly.That was a serious limitation for any low-volt-
age designs.
The year 1958 was memorable for many reasons. This was the year that National
Semiconductor was originally founded, that photolithography was invented at Bell Labs,
that Dr. Jean Hoerni started developing his planar transistor prototypes at Fairchild
Semiconductor, and that a small semiconductor manufacturer called Crystalonics Inc.
was founded in Cambridge, Massachusetts, by two former Raytheon engineers, and
soon began making transistors. A year or two later they began to make JFETs, which at
the time had only been made at Fairchild Semiconductor’s development lab in Califor-
nia (again by Dr. Jean Hoerni). Crystalonics was first in the industry to put N-channel
JFETs into production and pioneered many of the original JFET designs and applica-
tions, including current regulator diodes (CRDs), as shown in Figure 1.1B.
The CRD is a small single component, made by adding (or diffusing) a resistor inter-
nally between a JFET’s gate and source terminals. This establishes a fixed voltage
between the gate and source, so that a constant current flows between the drain and
the source, irrespective of changes in supply voltage. For many engineers, the CRD is

The first JFETs and op amps
A Short History of References 3
considered to be equivalent to the zener diode, except that it provides a constant cur-
rent rather than a constant voltage.
1.2 The first JFETs and op amps
Another major contributor to both JFET and CRD technology was Siliconix, which
was founded in 1962 by several former Texas Instruments’ engineers, and a former
Bell Labs researcher. During the first few years of its existence, Siliconix created and
developed JFETs, in which it became the undisputed market leader. Siliconix also
developed several product families of CRDs, as well as later pioneering the power
MOSFET.
Another important milestone was reached in 1964, when the first commercial mono-
lithic op amp (Fairchild Semiconductor’s µA702) was introduced. It had been designed
by the legendary Robert J. Widlar. (This analog breakthrough was partly a result of
Widlar’s use of his precision current source references, which he used for biasing the
various amplifier stages, and five years later would lead to the creation of the mono-
lithic voltage reference.) The µA702 was followed about a year later by the µA709,
which Fairchild introduced in November 1965. The µA709 became an industry stan-
dard, although it required a resistor and two capacitors for compensation, and cost
around $50 apiece. The other side of the story was that it was hard to compensate
and harder still to manufacture, because it had a poor yield. Fairchild could barely
keep up with demand for this op amp, though, and had back orders for several years
after it was introduced. Two people at Fairchild Semiconductor were teamed with the
26-year-old Widlar: David Talbert, his processing specialist who actually made Wid-
lar’s designs in silicon and who invented the super-beta transistors that the designs
often required, and Jack Gifford, the 24-year-old product manager. He was one of the
few people who understood Widlar, amplifiers, feedback and control theory, and whom
Widlar got along with. Gifford helped make these early op amps such a huge success,
introduced the DIL package to the industry, and laid the foundations for today’s analog
Figure 1.1. Crysalonics and Siliconix pioneered some of the industry’s earliest JFETs in the
early 1960s, as well as making current regulator diodes (CRDs), as shown here.

4
The first JFETs and op amps
IC marketplace. (Today Jack Gifford is CEO of Maxim Integrated Products, one of the
world’s major suppliers of precision op amps and voltage references.)
In the meantime, Widlar started presenting technical papers at various industry forums
that dealt with improving accuracy by compensating for differences in beta, VBE, and
temperature changes, which otherwise made circuits drift. Widlar was one of the first
people to ever address these topics publicly, and he devised several unique current
sources (covered in Chapter 4, and depicted here in Figure 1.2). He also brought to
the attention of circuit designers many otherwise overlooked factors, such as the
importance of circuit board layout, drift and temperature coefficient, the use of preci-
sion passive components, and using Kelvin connections to avoid voltage drops in PC
boards, sockets, and cabling. Widlar included current sources in all his IC designs as a
means of biasing and compensation.
In 1968 another Fairchild designer, Dave Fullagar (originally from the United Kingdom,
with a degree from Cambridge University), while at Fairchild’s Mountain View facility,
designed the legendary bipolar op amp—the µA741. It was reliable, and better still, it
required no compensation. Above all, it was much easier to manufacture and had
good yields. In the early 1970s, designers began using it to buffer the TC zener diode,
to provide a more stable reference voltage, as shown in Figure 1.3A. Equally impor-
tant, the op amp also provided an easy way for circuit designers to create precision
current sources with low values. It was found that by combining a zener diode with
precision resistors and an op amp, one could create a stable current source, as shown
in Figure 1.3B. More about this topic is covered in Chapter 11.
Figure 1.2. Some examples of Widlar’s current sources.These were designed to compensate
for differences in the betas and VBE’s of the transistors.

The first bandgaps
1.3 The first bandgaps
Another milestone was reached in 1969 when Bob Widlar, then at National Semicon-
ductor, became the first analog IC designer to create an integrated voltage reference
(based on the bandgap principle he had conceived), as part of a regulator IC design,
the LM109, shown in Figure 1.4. This was the first high-power monolithic linear regula-
tor. It proved that it was possible to build such a device, which was temperature-com-
pensated by a precision reference, on one monolithic chip, despite significant changes
in chip temperature. Like all of Widlar’s (bipolar) designs, it included current sources in
the circuitry in order to establish correct bias levels. Yet another milestone was
reached in 1971, when Widlar, together with his friend and fellow designer Bob
Dobkin, co-designed the industry’s first commercially available monolithic bandgap
shunt reference—National Semiconductor’s LM113.
With the advent of the first monolithic A/D and D/A converters in the early 1970s,
much development focused on creating monolithic voltage references, as well as on
improving overall precision. Besides National Semiconductor, other early innovators
included Texas Instruments, Fairchild, RCA, Analog Devices, and Intersil. Some of the
early pioneering work was shared on an industry-wide basis in the form of published
technical papers. Technical articles also appeared in the electronics industry’s leading
Figure 1.3. Precision reference circuits from the mid-1970s.

6
The first bandgaps
magazines, as well as in manufacturers’ own in-house application notes. Up until this
point, these first-generation voltage reference ICs were marketed and referred to as
though they were (improved) zener diodes, which were the established industry stan-
dard at the time.
At about this time, a process called zener zapping was introduced, which would
impact all future precision monolithic analog circuitry, including today’s voltage refer-
ences. It was designed by George Erdi at Precision Monolithics, while he was design-
ing the first precision op amp, the OP07. It coupled his op amp design with a
technique he developed for adjusting the op amp’s input offset voltage (Vos) to a very
low level, by using a computer-controlled laser to short a string of parallel-connected
resistors and zeners. This trimming technique effectively shorts the zener to obtain a
precise voltage level. Erdi’s ultra-stable OP07 op amp brought with it such a major
improvement in precision that it became the industry standard and is still a steady pro-
duction item to this day. George Erdi later left Precision Monolithics and in 1981 co-
founded Linear Technology Inc. The technique he created has since been modified
and refined and applied to many other products besides op amps, but it is still used
throughout the analog semiconductor industry. Many of today’s voltage references are
laser-trimmed at the wafer-sort level, then trimmed again after the die has been
mounted into its package. Precision trimming is applied to certain internal monolithic
thin-film resistors, which helps enhance certain features of the device, such as its ini-
tial accuracy, drift, and tempco curve-correction circuitry.
Figure 1.4. National Semiconductor’s LM109 regulator was designed by the legendary Bob
Widlar, and the first monolithic analog IC product to use an on-board voltage refer-
ence. In this case the linear regulator was a 1-Amp power type in a steel TO-3
case. The bandgap reference helped compensate the regulator over its operating
temperature range.

The buried-zener debuts
1.4 The buried-zener debuts
In the mid-1970s, another milestone was reached when National Semiconductor
introduced the LM199, the first buried-zener monolithic voltage reference, shown in
Figure 1.5. This legendary product, designed by Robert Dobkin, offered a 6.95-volt
reference, with a 0.3-ppm/˚C temperature drift and a noise spec of about 7µVrms
(10Hz to 10KHz). This was better than anything at the time and still better than most
devices are even today. Uniquely, the device included an on-chip substrate heater for
stabilizing the chip’s temperature, which helped provide it with great accuracy and
the ultra-low tempco. Another important milestone was reached in the late 1970s,
when Paul Brokaw at Analog Devices created the first series (three-terminal) preci-
sion bandgap reference (the AD580), based on what is referred to today as the
Brokaw cell. This product was destined to become one of the most successful volt-
age reference ICs ever introduced and ushered in a new level of precision for voltage
reference ICs. Up until then, voltage references were either 6.9-volt or 10-volt (bur-
ied-zener) types, and the remainder were 1.2-volt shunt bandgap devices. The intro-
duction of a 5-volt voltage reference that had a separate input and output terminal
like a linear regulator, as well as not requiring any input resistor, was another remark-
able achievement.
In the late 1970s, National Semiconductor addressed the needs of instrumentation
designers by introducing the LM134 family of current source ICs. These were dedi-
cated three-terminal adjustable bipolar devices and had been designed by another
outstanding National designer, Carl Nelson. For the first time, a monolithic current
source IC was available, and it quickly became a favorite design-in of instrumentation
and other designers of precision circuitry. The versatile LM134 provided a current
source that was adjustable over a practical range of 1 µA to 5 mA and with an operat-
ing voltage range of between 1 to 40 volts. It offered excellent current regulation and
the ability to create a true floating current source, so that it could be used as either
Figure 1.5. Robert Dobkin’s novel buried-zener structure was revolutionary, because it pro-
vided ultra-low noise, a 0.3ppm/°C tempco, and greatly improved stability.

8
Advancements in technology
current source or a current sink. Figure 1.6 shows a simpliﬁed view of the LM134’s
internal circuitry, which incorporates both NPN and PNP bipolar transistors, as well as
some very-low-voltage P-channel JFETs, and an integrated capacitor. The JFETs are
used for start-up biasing and enable the internal BJT current mirrors. Uniquely, the
device also doubled as a linear temperature sensor, which made it even more attrac-
tive at the time. Carl Nelson later joined Linear Technology, once again working for his
former boss, Bob Dobkin. The LM134 family is still in production today and available
from both National Semiconductor and Linear Technology. More on this product family
can be found in Chapter 10.
1.5 Advancements in technology
Over the next few years, other analog semiconductor companies looked at getting into
the voltage reference market and followed National Semiconductor and Analog
Devices by introducing some outstanding products of their own. Most of them also
incorporated the laser-trimming technique developed at Precision Monolithics, which
enabled them to provide devices with enhanced precision. These included Precision
Monolithics (now part of Analog Devices), with its 10-volt REF01 and 5-volt REF02
series bandgap products; Burr-Brown (now part of Texas Instruments), with its 10-volt
buried-zener REF102; Linear Technology’s LTZ1000 (an exceptional 7-volt buried-
zener designed by Robert Dobkin) and its LT1021 (a buried-zener family designed by
Carl Nelson); Maxim’s innovative MAX676 family of bandgap products, with an on-
board temperature sensor, a temperature-correction ROM, and output Kelvin connec-
Figure 1.6. LM134 monolithic adjustable current source. A simpliﬁed view of its internal
circuitry.

Other topologies emerge
tions that provided a tempco of 0.6-ppm/˚C; and National Semiconductor again with its
low-cost bandgap devices LM185 family (previously designed by Bob Dobkin).
While current source references used either the LM134, dual/quad matched BJTs, or
JFET CRDs, in 1990 Burr-Brown introduced its remarkable REF200. This unique
device contained two ﬁxed 100-µA current sources, as well as a precision current mir-
ror, all packaged in an eight-pin surface-mount SOIC package. It used a proprietary
Burr-Brown dielectrically isolated (Difet®
) process, which completely isolated the three
circuits, making them independent of one another. Each of the 100-µA current sources
used a precision bandgap cell to provide a near-zero tempco, while the current mirror
used the reliable Full Wilson architecture. Each of the circuits was laser-trimmed at the
wafer level to provide the highest accuracy. The REF-200 could be pin-strapped to
provide currents of 50 µA, 100 µA, 200 µA, 300 µA, and 400 µA. Or by adding external
circuitry, one could create virtually any current, either smaller or greater than these.
(This is covered fully in Chapter 10.)
During the 1990s, the surface-mount package became widely used, so that most ana-
log products, including the current source ICs mentioned, most voltage references, as
well as discretes like transistors, JFETs, CRDs, and zener diodes were available in
this form of package. Pioneers of this were National Semiconductor, Maxim, TI, Ana-
log Devices, and others. Until then, references were available in the dual-in-line
ceramic or plastic package. Today most references, including many mature products,
are only available in surface-mount packages. The most commonly used surface-
mount package for a voltage reference is the SOIC-8.
Another factor that started to emerge in the late 1980s and throughout the ’90s was
that as A/D and D/A converters evolved, some of those products included their own on-
board reference. However, in every case, the on-board reference was a bandgap type,
which is inherently limited to about 12-bit resolution, even though the converter may be
capable of much higher resolution. It is fairly straightforward for the manufacturer to
build the bandgap on the same die as the converter, but it is mostly a user convenience
enabling a design to get off the ground faster. To support higher bit resolutions (e.g.,
14-, 16-, 20-bit) and lower noise levels, an external precision series voltage reference is
recommended. Of course, the chosen converter must have the option of using an
external reference, and of having its own internal reference turned off.
1.6 Other topologies emerge
By the mid-1990s, the existing types of voltage reference available were shunt and
series bandgaps and variations of the buried-zener. In the late ‘90s, Analog Devices
developed and introduced the ﬁrst generation of its XFET™ voltage reference, a topol-
ogy (type) based partly on the characteristics of the JFET. It had been about two
decades since a completely new voltage reference topology had been introduced. In
fact, Analog Devices had introduced the series bandgap in the form of its AD580.

10
The XFET, shown in Figure 1.7, was purposely designed to get around many of the
limitations of the bandgap and buried-zener types. These had noticeably begun to
impact A/D and D/A converter systems, whose operating voltages were increasingly
headed below 5 volts and whose increased resolutions depended on low-noise preci-
sion references. The main limitations of those references included operating voltage,
quiescent current/power dissipation, noise, and nonlinear temperature coefﬁcients.
Even the best bandgaps and buried-zener products still have nonlinear tempcos
(something that is inherent in their design), particularly at the extremes of their tem-
perature ranges. (Laser trimming the reference’s internal resistors helps compensate
for that.) By contrast, the inexpensive XFET provided a low linear tempco and allowed
for lower voltage operation, over a wider operating temperature range (automotive ver-
sus industrial and commercial) than the others. It also featured a lower noise level, a
lower quiescent current, lower thermal hysteresis, and a very low long-term drift. Two
further generations of the XFET have since followed, as the product evolves with the
ever-changing marketplace.
Early in the 21st century, a new type of bandgap emerged and is referred to in this
book as the super-bandgap. The main difference is that it is built with more advanced
technology, including processing, which all results in a series bandgap product with
many of the best features of the buried-zener. Super-bandgaps can be seen to have
three key features: (1) excellent initial accuracy, (2) a very low temperature coefﬁcient
(tempco), and (3) an ultra-low noise level. Several manufacturers make the super-
bandgap type of product. Some examples include National Semiconductor’s LM4140
family, built with a proprietary CMOS process (±0.1% initial accuracy, 3 ppm/˚C
Figure 1.7. Analog Devices’ XFET™ technology provides a low-cost, but very high-
performance reference.

tempco, and less than 2.5-µV pk-pk noise voltage); Linear Technology’s LT1461
(±0.04% initial accuracy, less than 3 ppm/˚C tempco, and a typical 8-µV pk-pk noise
voltage); Analog Devices’ AD780 (±0.04% initial accuracy, less than 3 ppm/˚C tempco,
and a 4-µV pk-pk noise voltage); Maxim’s BiCMOS MAX6126 family (±0.02% initial
accuracy, 3 ppm/˚C tempco, and a 1.3-µV pk-pk noise voltage), and also Maxim’s
MAX6325 family (±0.02% initial accuracy, 1 ppm/˚C maximum tempco, and a 1.5-µV
pk-pk noise voltage). The super-bandgap type is certainly a major step forward for this
particular topology, and no doubt more of these products will be introduced soon.
More recently, in 2003, Xicor Corporation (now part of Intersil Corporation) announced
a completely new type of series voltage reference, using its proprietary Floating Gate
Analog (FGA™) technology. This new CMOS-based topology offers the lowest quies-
cent current of any voltage reference—less than 0.8 µA. It also has other characteris-
tics that challenge some of those of the buried-zener, as well as having voltage
options available of between 1.25 and 5 volts. It is a cousin of the EEPROM memory,
which provides nonvolatile storage of digital data, but Xicor’s FGA is a far more com-
plex component, storing analog voltage levels for more than 10 years. The unique
characteristics of the FGA translates into an exceptional voltage reference, which can
be used in up to 24-bit systems. Other analog products based on this exciting technol-
ogy are future possibilities.
In summary, the monolithic current source and the monolithic voltage reference can
both be traced back to the designs of Bob Widlar, who first implemented them. For
many current source designs at the circuit board level, it is often necessary to create
one’s own, unless the desired current level is within the range of the few dedicated
IC products available. Several chapters in this book will be helpful in that regard and
deal with creating good current sources using different types of transistors (e.g.,
BJTs, JFETS, MOSFETs) or combinations of them. One area that appears to have
great potential for the development of small precision current sources is with the
new EPAD®
and ETRIM™ families of matched MOS transistors from Advanced Lin-
ear Devices, Inc. These are both exciting new technologies that can be directly
applied to current sources.
As for the voltage reference IC, today there is a convergence in the marketplace as the
best of each topology attempts to compete with some of the most exotic specifications
first established by the buried-zener, a topology invented by Bob Dobkin, one of the
world’s greatest analog IC designers. Generally, the other topologies are becoming
more accurate, irrespective of whether they are shunt or series types. Over the past
decade, the constant need to support higher-resolution A/D and D/A converters in a
myriad of digitally based products has fueled the growth and improvement of the pre-
cision monolithic voltage reference. The 1 ppm/˚C tempco specification over a prod-
uct’s temperature range has become an industry-wide goal for series references.
Besides a few buried-zener products, only one or two super-bandgap products and a
5-volt version of the FGA™ have yet broken through that challenging barrier. While the
buried-zener has already reached beyond that barrier and peaked as a topology, fur-

12
ther advances in the other topologies will come in time. Designing voltage references
is not an easy matter, though, because characteristics are often interlinked. For exam-
ple, as one reduces the device’s current drain, its noise level usually increases or its
operating voltage range is adversely affected. Improvements will happen, but as a
combination of advancements in processing, design, packaging, and thermal man-
agement. During the past ﬁve years, the industry has witnessed the arrival of two
completely new topologies (XFET™ and FGA™) and an existing one has been
enhanced (super-bandgap). This is an exciting time in the evolution of electronic refer-
ences and seeing how they will be applied in the products of tomorrow’s world.
Photo 1.1. A photomicrograph of the Intersil x60008 die. This is the world’s newest and most
technically advanced voltage reference. (Photo courtesy of Intersil Corporation).

13
Chapter 2
An Introduction to Current Sources
2.1 An overview
Current sources are basic electronic building blocks that are used extensively in the
architectures of analog ICs, as well as in OEM circuit board designs. In both cases,
current sources are created by combining diodes, resistors, and transistors (BJTs or
FETs). They can also be created at the circuit board level by using discretes, matched
pairs, transistor arrays, or by combining op amps with precision voltage references.
The various techniques for doing so will be reviewed in Part 1 of this book. Although a
few dedicated monolithic current source ICs are commercially available, it is often nec-
essary to create one’s own circuit to match the particular needs of the application.
Although most forms of today’s instrumentation use either voltage or current refer-
ences, the former are far more available. As a result, designers frequently use voltage
references together with precision resistors, so that a stable reference voltage is con-
verted into a precise current. Applications for current sources range from biasing and
stabilization to reference and linearizing. For example, in the design of an op amp, the
IC designer will use current sources to create active loads for the amplifier stages and
to establish precise bias levels. By providing a constant current, this forces amplifier
stages to stay at the Q-points within their active linear regions (see Figure 2.1). In a
circuit board design, a current source may be used for linearly charging a capacitor
with a constant current, as in a precision timing circuit or in a peak detector. In a med-
ical instrument application, a sensor and a low-noise, front-end amplifier could be
biased using precision current sources to assist in recovering very-low-level signals.
The advantages of using these building blocks is their inherent constant current out-
puts, which are mostly independent of changes in supply voltage, temperature, load
resistance, or load voltage. One could liken the current source to that of a precision
current regulator. These advantages, when compared with using a simple fixed resis-
tor load, include the following:
Greater precision
Better repeatability
Improved temperature stability

14
An overview
Lower long-term drift
Higher output impedance
Increased bandwidth
Larger signal range
Current sources are not a new innovation. They predate the integrated circuit by at
least a couple of decades. Before their implementation in integrated circuits, they were
used in vacuum tube–based circuits (triodes and pentodes). When both NPN and
PNP silicon transistors became readily available during the 1960s, analog designers
were able to build current sources that connected to either a positive or negative sup-
ply rail. Then when the silicon bipolar IC became a practical reality in the late 1960s,
the current source had already become an integral part of the internal architecture, for
the purposes of biasing and stability. Much development focused on creating current
sources for various types of monolithic analog ICs between the mid-60s and the mid-
80s, although it still continues today, but at a slower pace. Early innovators included
Philbrick, Texas Instruments, Fairchild Semiconductor, National Semiconductor, GE,
Figure 2.1. A typical transistor ampliﬁer’s characteristics showing the Q-point, in the center of
its active region.

An overview
An Introduction to Current Sources 15
Analog Devices, RCA, and Motorola Semiconductors, among many others. Some of
the early pioneering work was shared on an industry-wide basis (a small fledgling
industry at the time), in the form of published technical papers such as the IEEE’s
Journal of Solid State Circuits. Technical articles also appeared in some of the elec-
tronics industry’s leading-edge magazines of that era, including Electronics, Electronic
Design, and EDN, as well as in manufacturers’ own in-house application notes. Over
the past decade, however, the subject of current sources has not otherwise been a
much-publicized topic.
Although most engineers have seen a current source’s symbol depicted in an analog
IC’s circuit schematic (see Figure 2.2), in most cases even if its circuit is shown, it may
only be a simplified version because the actual circuit is proprietary and may be pat-
ented. Few IC manufacturers, if any, give out specifics about how their circuits are built
and biased, other than on a need-to-know basis and how it applies to the end user.
However, the current source is an essential element in most analog circuits’ architec-
ture and operation, irrespective of the IC’s function (e.g., op amp, A/D converter, volt-
age regulator, RF receiver, video amplifier, D/A converter).
While loosely referred to as current sources, there are actually three specific types.
Typically, a current source (usually comprising P-type devices) connects between the
positive supply rail (+V) and the load, while a current sink (usually comprising N-type
devices) connects between the load and a negative (-V) or ground (0V) potential. A
current mirror (also known as a current reflector) can connect to either rail and usually
provides multiple current sources/sinks that either mirror (1:1 match) or are arranged
in preset current ratios (e.g., 1:2, 1:4, 1:8). This is sometimes shown in a circuit sche-
matic as a multicollector or multiemitter structure, commonly created by analog IC
designers. Figure 2.3 shows some examples of current sources, current sinks, and
current mirrors, created in different technologies. We will focus on the specifics of
these different types in subsequent chapters.
Figure 2.2. Common current source symbols.

16
An overview
Over the years, current sources have been applied in subsequently newer technolo-
gies and processes, such as high-speed bipolar, complementary-bipolar, bipolar-FET,
analog metal-gate CMOS, and more recently in very-low-voltage processes. The
examples in Figure 2.4 show part of the internal circuitry for two typical op amp input
stages (one is a bipolar-FET device in Figure 2.4A and the other is a CMOS device in
Figure 2.4B). As you can see, current sources are employed throughout.
Figure 2.3. Examples of different types of current sources.

An overview
An Introduction to Current Sources 17
Two major advances with the op amp came in those early days from National Semi-
conductor. The first was Bob Widlar’s implementation of what are referred to as super-
gain transistors. These were specially processed, very-low-voltage NPN transistors,
which had very high hFEs (typically of around 5000, at a very low collector current of
about 1 µA). For the first time, it uniquely enabled a monolithic bipolar op amp to have
ultra-low-input bias currents similar to those of a JFET, but over the full military tem-
perature range. They were a concept devised and implemented by Widlar for some of
his designs and created at the wafer level by his processing specialist, Dave Talbert.
Super-gain transistors were first used in the LM108 precision op amp, which National
introduced in 1969. In a typical op amp schematic, they are depicted as a regular NPN
transistor, but with a hollow base junction, as shown in Figure 2.5. Today super-gain
Figure 2.4. These examples show how current sources, sinks, and mirrors are used as active
loads, and for biasing in the front-end amplifiers of typical bipolar-FET (left) and
CMOS (right)voltage-feedback op amps.

Exploring the Variety of Random
Documents with Different Content

He wanted her now because he knew that he might never have her.
He wanted her with the fierce hunger of a pirate for a prize; but the
very sharpness of his desire made him see that it was sheer
selfishness to press his point. He overlooked the fact that it would
have been perfectly useless. No pressure would have changed
Marian. Pressure had done what it could for her already: it had
moved her to tears. She dried them now, and suggested that they
had stayed on the downs long enough.

CHAPTER X
It sometimes seemed to Stella as if Chaliapine had brought on the
war. Those last long golden summer days were filled with his music,
and then suddenly out of them flashed the tents in the park, the
processions of soldiers and bands, the grim stir that swept over
London like a squall striking the surface of a summer sea.
The town hall did not collapse, but it shook. It was a place where, as
a rule, the usual things took place, and even unusual things
happened usually; but there were several weeks at the beginning of
the war when all day long strange things happened strangely. Offices
were changed, the routine of years was swept up like dust into a
dust-pan, and a new routine, subject to further waves of change,
took its place. Workers voluntarily offered to do work that they were
unaccustomed to do. The council hall became a recruiting office. No.
8, the peculiar sanctum of the sanitary inspector, was given up to an
army surveyor. Tramps asked the cashier questions. It was like the
first act of Boris Goudonoff. Even food was carried about on trays,
and as for proclamations, somebody or other was proclaiming
something all day long.
There was no religion and no dancing, but there was the same sense
of brooding, implacable fate; it took the place of music, and seemed,
without hurry and without pause, to be carrying them all along in a
secret rhythm of its own toward an unseen goal.
Mr. Leslie Travers ruled most of the town hall committees, and he
required innumerable statistics to be compiled and ready to be
launched intimidatingly at the first sign of any opposition to his
ruling.
Stella, to whom the work of compiling fell, had very little time to
consider the war.

When she got home she usually went to sleep. From time to time
she heard Mrs. Waring announcing that there was no such thing as
war and Eurydice reciting battle-odes to Belgium.
For the first time in her life Eurydice shared a common cause. She
was inclined to believe that England was fighting for liberty. She
knew that France was, partly because France was on the other side
of the channel and partly because of the French Revolution. The
destruction of Louvain settled the question of Belgium. To Eurydice,
whatever was destroyed was holy. Later on she became a violent
pacifist because Mr. Bolt said that we ourselves were Prussian; but
for the moment nobody, not even Mr. Bolt, had traced this evasive
parallel.
Professor Waring wrote several letters to the papers, asking what
precautions the Belgians were taking about Sanskrit manuscript. He
had a feeling that King Albert, though doubtless an estimable young
man and useful in the trenches, might, like most kings, have been
insufficiently educated to appreciate the importance of Sanskrit. That
men should die in large numbers to protect their country was an
unfortunate incident frequent in history, but that a Sanskrit
manuscript should be destroyed was a national calamity, for the
manuscript could never be replaced.
He made an abortive effort to reach Belgium and see about it
himself, but at the Foreign Office he was stopped by a young man
with a single eyeglass, from whom the professor had demanded a
passport. The exact expression used by this ignorant young person
was, I'm awfully sorry, sir, but I'm afraid just at present Sanskrit
manuscript will have to rip.
Professor Waring promptly addressed letters of remonstrance and
advice to several German professors upon the subject. They were
returned to him after three weeks, with a brief intimation that he
was not to communicate with the enemy. Professor Waring had
considered German professors to be his natural enemies all his life;
this had been his chief reason for communicating with them. He was

fitted, as few officials in the Foreign Office can ever have been fitted,
to point out to the German professors the joints in their armor.
They had a great deal of armor and very few joints, and it
discouraged Professor Waring to leave these unpierced spots to the
perhaps less-practised hands of neutrals.
But it was not until the destruction of Louvain that he grasped to the
full the reaction of his former antagonists. When Professor Waring
read a signed letter from some of the German professors agreeing to
the destruction of the famous Belgian library he acquiesced in the
war. He stood in front of his wife and woke Stella up in order to
make his declaration.
Henrietta, there is a war, he announced. It is useless for you to
assert that there is not. Not only is there a war, but there should be
one; and if I were twenty years younger, though wholly
unaccustomed to the noisy mechanisms of physical destruction, I
should join in it. As it is, I propose to write a treatise upon the
German mind. It is not one of my subjects, and I shall probably have
to neglect valuable work in order to undertake it; still, my researches
into the rough Stone Age will no doubt greatly assist me. Many just
parallels have already occurred to me. I hope that no one in this
house will be guilty of so uneducated a frame of mind as to
sympathize with the Teutonic iconoclasts even to the extent of
asserting, as I believe I heard you assert just now, Henrietta, that
none of them exist.
Mrs. Waring murmured gently that she thought an intense
hopefulness might refine degraded natures, but the next day she
bought wool and began to knit a muffler. She had capitulated to the
fact of the war. While she knitted she patiently asserted that there
was no life, truth, intelligence, or force in matter; and Stella, when
she came home in the evening, picked up the dropped stitches.
It was strange to Stella that her only personal link with the war was
a man whom she had seen only once and might never see again.

She thought persistently of Julian. She thought of him for Marian's
sake, because Marian was half frozen with misery. She thought of
him because unconsciously he stood in her mind for England. He
was an adventurer, half-god, half-child, who had the habit of winning
without the application of fear. She thought of him because he was
the only young, good-looking man of her own class with whom she
had ever talked.
Marian was afraid that Stella might think she had been
unsympathetic to Julian about his mission. She told Stella, with her
usual direct honesty, how angry she had been with him.
I know I was nasty to him, she said. I can't bear to have any one
involve me first and tell me about it afterward.
Of course you can't, agreed Stella, flaming up with a gust of
annoyance more vivid than Marian's own. How like him! How
exactly like him to be so high-handed! Fancy whirling you along
behind him as if you were a sack of potatoes! Of course you were
annoyed, and I hope you gave him a good sharp quarrel. One only
has to look at Julian to see that he ought to be quarreled with at
regular intervals in an agreeable way for the rest of his life.
I don't like quarrels, Marian said slowly. They don't seem to me to
be at all agreeable; but I don't think Julian will act without
consulting me again.
Stella looked at Marian curiously. What was this power that Marian
had, which moved with every fold of her dress, and stood at guard
behind her quiet eyes? How had she made Julian understand
without quarreling that he must never repeat his independences?
Stella was sure Marian had made him understand it. It would be of
no use to ask Marian how she had done it, because Marian would
only laugh and say: Nonsense! It was perfectly easy. She probably
did not know herself what was the secret of her power; she would
merely in every circumstance in life composedly and effectively use
it. Was it perhaps that though Julian had involved her actions, he

had never involved Marian? Was love a game in which the weakest
lover always wins?
Of course I've never been in love, Stella said slowly, and I haven't
the slightest idea how it's done or what happens to you; but I fancy
quarreling might be made very agreeable. Love is so tremendous,
isn't it, that there must be room for concealed batteries and cavalry
charges; and yet of course you know all the time that you are loving
the person more and more outrageously, so that nothing gets
wasted or destroyed except the edges you are knocking off for
readjustments.
I don't think I do love Julian outrageously, Marian objected. I
didn't, you see, do what he wanted: he had a mad idea of getting a
special license and having a whirlwind wedding, leaving me directly
afterward. Of course I couldn't consent to that.
Couldn't you? asked Stella, wonderingly. I don't see that it matters
much, you know, when you give that kind of thing to a person you
love. If you do love them, I suppose it shows you're willing to marry
them, doesn't it? But how, when, or where is like the sound of the
dinner-bell. You don't owe your dinner to the dinner-bell; it's simply
an arrangement for bringing you to the table. Marriage always
seems to me just like that. I should have married Julian in a second
if I'd been you; but I should have made him understand that I
wasn't a sack of potatoes, if I'd had to box his ears regularly every
few minutes for twenty-four hours at a stretch.
Surely marriage is sacred, said Marian, gravely. Stella's point of
view was so odd that Marian thought it rather coarse.
But it needn't be long, objected Stella; you can be short and
sacred simultaneously. In fact, I think I could be more sacred if I
was quick about it; I should only get bored if I was long.
You have such a funny way of putting things, said Marian, a little
impatiently. Of course I know what you mean, but I don't like being
hurried. I love Julian dearly, and I will marry him when there is time

for us to do it quietly and properly. Meanwhile it's quite awful not
hearing from him. I have never been so miserable in my life.
Stella sat on the floor at Marian's feet with Marian's misery. She
entered into it so deeply that after a time Marian felt surprised as
well as comforted. She had not thought grief so pictorial. She felt
herself placed on a pinnacle and lifted above the ranks of happier
lovers. She thought it was her love for Julian that held her there; she
did not know that it was Stella's love for her. Stella for a time saw
only Marian—Marian frozen in a vast suspense, Marian racked with
silences and tortured with imagined dangers. She did not see Julian
until Marian had gone, and then suddenly she put her hands to her
throat, as if she could not bear the sharp pulsation of fear that
assailed her. If all this time they were only fearing half enough and
Julian should be dead?
She whispered, Julian dead! Then she knew that she was not
feeling any more for Marian. She was feeling for herself. Fortunately,
she knew this didn't matter. Feeling for oneself was sharp and
abominable, but it could be controlled. It did not count; and she
could keep this much of Julian—the fear that he might be dead. It
would not interfere with Marian or with Julian. Hopes interfere: but
Stella had no personal hopes; she did not even envisage them. She
claimed only the freedom of her fears.

CHAPTER XI
It is disconcerting to believe that you are the possessor of one kind
of temper—a cold, deadly, on-the-spot temper—which cuts through
the insignificant flurries of other people like a knife through butter,
and then to find a sloppy explosiveness burst from you unaware.
Mr. Travers had never dreamed that in the town hall itself he could
ever be led to lose a thing he had in such entire control as his
temper. He did not lose it when the blushing Mr. Belk had the
audacity to stop him in mid-career, on his way to his sanctum
through No. 7, the outer office of his assistant clerks, though they
were, as a body, strictly forbidden to address him while passing to
and fro. Mr. Belk was so ill advised as to say:
If you please, sir, it's four o'clock, and Miss Waring hasn't been out
to lunch yet. Mr. Travers merely ran his eye over Mr. Belk as a
fishmonger runs his eyes over vulnerable portions of cod laid out for
cutting, and brought down his chopper at an expert angle.
Since when, Mr. Belk, he asked, with weary irony, has Miss
Waring's lunch been on your list of duties?
Then he passed swiftly into his office and faced Stella, closing the
door behind him. Temper shook him as a rough wind shakes an
insignificant obstacle. He could not hold it; it was gone. It blew
inside out like a deranged umbrella. He glared at Miss Waring. There
was nothing in her slight, bent figure, with its heavy, brown hair
neatly plaited in a crown about her head, which should have roused
any town clerk to sudden fury.
It's abominable, Mr. Travers exclaimed, bringing his trembling hand
down with a bang upon Stella's table, how women behave!

Stella said out loud, One hundred pounds, ten shillings, and
sixpence, and then looked up at her employer. She asked very
quietly who had vexed him. There might have been a fugitive gleam
of laughter at the back of her eyes, but there were shadows under
them that made her look too tired for laughter.
You, of course, he cried. How are we ever to get through with our
work if you won't eat? It's so silly! It's so tiresome! It's so uncalled
for! Why are you doing these wretched lists now?
Because, said Stella—and now the laughter ran out at him
unexpectedly and tripped him up—the town clerk has a meeting at
five o'clock at which these statistics must be at hand to justify him in
having his own way!
Put them down! said Mr. Travers savagely. Stella laid down her pen
with the ready obedience which can be made so baffling when it
proceeds from an unconsenting will. Now go out and get something
to eat, he went on, while I do the wretched things. And don't let
this occur again. If you have too much to do,—and I know the
correspondence gets more and more every day,—mention it. We
must get some help in.
She was gone before he had finished his sentence—gone with that
absurd dimple in the corner of her cheek and the sliding laughter of
her eyes.
She had left behind her a curious, restless emptiness, as if the very
room itself waited impatiently for her return. It was half an hour
before she came back. The town clerk had had to answer three
telephone messages and four telegrams. If the outer office had not
known that he was there and Miss Waring wasn't, he would have
had more interruptions. Nevertheless, the figures had helped Mr.
Travers to recover his temper.
He was an expert accountant, and you can take figures upon their
face-value. They are not like women; they have no dimples.

Mr. Travers was prepared to be the stern, but just, employer again.
He remained seated, and Stella leaned over his shoulder. He had not
expected that she would do this.
What have you had to eat? he asked. It was not at all what he had
intended to say to Stella.
A cup of tea, two ham sandwiches, and a bun:—such a magnificent
spread for seven-pence! replied Stella, cheerfully. You've forgotten
to put in what the insurance will be—there at the bottom of the
page.
Mr. Travers rose to his feet. He was taller than Stella, and he
considered that he had a commanding presence. Stella slid back into
her seat.
You ought to have had, said Mr. Travers, with labored quietness,
beefsteak and a glass of port.
Anybody could tell, said Stella, tranquilly, that you are an
abstemious man, Mr. Travers. Port! Port and steak! You mean porter.
All real drinkers know that port is sacred. Bottles of it covered with
exquisite cobwebs are kept for choice occasions; they are brought in
softly by stately butlers, walking delicately like Agag. It is drunk in
companionable splendor, tenderly ministered to by nothing more
solid than a walnut, and it follows the courses of the sun. There, you
did quite a lot while I was away, and if you don't mind just looking
through those landlords' repairing leases on your desk, I dare say I
shall have finished this before five.
Mr. Travers opened his mouth, shut it again, and returned to his
repairing leases. He was not an employer any more. He was not an
icy, mysterious tyrant ruling over a trembling and docile universe: his
own secretary had literally told him to run away and play!
But it was in the night watches that the worst truth struck him. He
had been furious with Miss Waring for not spending more upon her
lunch, he had upbraided her for it, and she had never turned round

and said, Look what I earn! The opportunity was made to her
hand. How can women secretaries earning a hundred a year eat
three-and-sixpenny lunches? That ought to have been her answer.
Why wasn't it? She hadn't been too stupid to see it. She had seen it,
and she had instantly, before he had had time to see it himself,
covered it up and hidden it under that uncalled-for eulogy on port. It
was not fear. She hadn't been afraid to stand up to him (uncalled-for
eulogies were standing up to him); besides she had previously called
him unfair to his face. It was just something that Miss Waring was—
something that made the color spring into Mr. Traver's face in the
dark till his cheeks burned; something that had made Mr. Belk dare
his chief's displeasure to get her lunch; something that wasn't
business.
She wouldn't take an advantage, because I'd given it to her, he
said to himself. I thought everybody took an advantage when they
had the sense to see it; but she doesn't, though she has plenty of
sense. But the world couldn't go on like that.
This brilliant idea reassured Mr. Travers; he stopped blushing. He
was relieved to think that the world couldn't go on like Stella; but
there was something in him, a faint contradictory something, that
made him glad that Stella didn't go on like the world.
He went to sleep with these two points unreconciled.

CHAPTER XII
Stella had always known that it would come; she had spent two
months far-seeing it. It had usually taken the form of a telegram
falling out of Mrs. Waring's wool, or Eurydice standing upon the
steps, Cassandra-like, to greet her with a message from Marian.
Marian would come to give her the message, but she wouldn't wait;
she would drive swiftly away in a motor, and leave the broken
universe behind her. But disasters do not come as we have planned
their coming.
It was a dull November day, the streets were full of dying leaves,
and at the end of all the cross-roads surrounding the town hall a
blue mist hung like a curtain. Marian, in black velvet and furs, with
old Spanish ear-rings gleaming from her shell-like ears, stood in
disgust upon the steps of the town hall. Her small face was frozen
with unexpected pain, but she could still feel annoyed with the
porter. She stood in the thronged corridor and asked decisively for
Miss Waring.
The porter told her that Miss Waring worked in No. 7, or, at any rate,
No. 7 would know where she was working.
Marian stared slightly over the porter's head.
My good man, she said, how am I to know where No. 7 is? Go
and tell her to come to me. Here is my card.
All the way to No. 7 the porter concocted brilliant retorts to this
order. He would tell her he was not a footman and that this wasn't
Buckingham Palace. He would say roughly that, if she had eyes in
her head, she could find No. 7 for herself. But he was intimidated by
Marian's ear-rings. A secret fear that she might turn out to be the
lord mayor's daughter drove him to No. 7.

Stella was filing letters when he knocked, and when he saw the card
she knew the messenger had come; but she did not forget to say as
usual, Oh, thank you, Humphreys.
She finished filing the letters before she looked for Mr. Travers.
He was coming out of the council chamber at the top of a flight of
stairs. She stood there for a moment, holding him with her eyes, her
lips parted. She looked like a bird that has been caught in a room
and despairs of finding the way out.
Her face was strained and eager, and her sensitive eyebrows were
drawn together in a little tortured frown; but she spoke quietly as
soon as her breath came back to her.
Mr. Travers, a friend of mine is in trouble. May I go to her for the
afternoon? There is still a great deal to do,—I know I ought not to
ask you to let me go,—but Mr. Belk and Miss Flint are so kind that I
am sure they would help me. I—I should be very grateful if you
could spare me.
Certainly not, said Mr. Travers, sharply. I mean, of course, you can
go; but I won't have Mr. Belk or Miss Flint near me. I will do the
work myself.
Oh, she cried, aghast at this magnanimous humility on the part of
her employer, please don't! Do let me ask them! I'd so much rather
—
Mr. Travers waved her away. He wanted to do the work himself, and
he wanted her to be aghast. He descended the stairs rapidly beside
her.
You may leave immediately, Miss Waring, he said sternly as they
reached No. 7; and I will make my own arrangements about your
work.
Stella fled. Again he felt the sense of wings, as if he had opened a
window, and a bird had flown past him into liberty.

He did not want her to be grateful, but he thought she might have
looked back. She had noticed him only as a barrier unexpectedly
fallen. She had not seen how strange it was that a barrier of so
stubborn and erect a nature as Mr. Travers should have consented to
fall.
If any one else had asked him for an afternoon with a friend in
trouble, Mr. Travers knew that he would have said, Your friends'
troubles must take place outside office-hours. But when he had
seen Stella's face he had forgotten office-hours.
Marian was sitting on a chair in the corridor. Her expression implied
that there was no such thing as a town hall, and that the chair was a
mere concession to unnecessary space. She said, as she saw Stella:
Please be quick about putting your things on. Yes, it's bad news
about Julian.
Stella was quick. Marian said no more until they were seated
together in the motor; then she gave Stella a letter she had received
from Lady Verny. Lady Verny wrote:
My dear Marian: You must prepare yourself for a great distress.
Julian is in England, but he is very much injured. I want you to
go to him at once. Whenever he is conscious he asks for you.
My dear, if he recovers,—and they think that if he has an
incentive to live he will live,—he will be partially paralyzed. I
know that he will want to free you, and it will be right that you
should even now feel free; but till then—for a month—will you
give him all you can? All he needs to live? It is a great deal to
ask of you, but I think you are good and kind, and that I shall
not ask this of you in vain. His life is valuable, and will still be
so, for his brain is not affected. Before he relapsed into
unconsciousness he was able to give the Government the
information he acquired. I think it is not wrong to help him to
live; but of course I am his mother, and it is difficult for me to
judge. All this is very terrible for you, even the deciding of

whether you ought to help him to live or not. If I might suggest
anything to you, it would be to talk about it with that friend of
yours, Miss Waring.
Come to me when you have seen him. Do not think, whatever
your decision is, that I shall not realize what it costs you, or fail
to do all in my power to help you to carry it out.
Yours affectionately,
Helen Verny.
Stella dropped the letter and looked at Marian. Marian sat erect, and
her eyes burned. She was tearless and outraged by sorrow. There
are people who take joy as a personal virtue and sorrow as a
personal insult, and Marian was one of these people. Happiness had
softened and uplifted her; pain struck her down and humiliated her
solid sense of pride.
Why wasn't he killed? she asked bitterly, meeting Stella's
questioning eyes. I could have borne his being killed. Value! What
does Lady Verny mean by value? His career is smashed; his life is to
all intents and purposes over. And mine with it! It is very kind of her
to say he will release me. I do not need his mother to tell me that.
She seems to have overlooked the fact that I have given him my
word! Is it likely that I should fail him or that I could consent to be
released? I do not need any one to tell me my duty. But I hate life! I
hate it! I think it all stupid, vile, senseless! Why did I ever meet him?
What good has love been to me? A few hours' happiness, and then
this martyrdom set like a trap to catch us! And I don't like invalids. I
have never seen any one very ill. I sha'n't know what to say to him.
Oh, yes, you will, when you see him, said Stella; it was all that for
a while she could say.
She had always believed that Marian had a deep, but close-locked,
nature. Love presumably would be the key.

It was unlocked now. Pain had unlocked it, instead of love, and
Stella shivered at the tearless hardness, the sharp, shallow sense of
personal privation that occupied Marian's heart. She had not yet
thought of Julian.
Stella told herself that Marian's was only the blindness of the
unimaginative. The moment Marian saw Julian it would pass, and
yield before the directer illumination of the heart. Marian's nature
was perhaps one of those that yields very slowly to pain. When she
saw Julian she would forget everything else. She would not think of
her losses and sacrifices any more, or her duties. Stella felt curiously
stung and wasted by Marian's use of the word duty. Was that all
there was for the woman whom Julian loved? Was that all there was
for Julian!
But she could deal only with what Marian had; so, when she spoke
again, Stella said all she could to comfort Marian. She spoke of
Julian's courage; she said no life in Julian could be useless that left
his brain free to act. She suggested that he would find a new career
for himself, and she pictured his future successes. Beneath her lips
and her quick outer mind she thought only of Julian, broken.
They stopped in a large, quiet square, at the door of a private
hospital. There was no sound but the half-notes of birds stirring at
twilight in the small square garden, and far off the muffled murmur
of distant streets.
A nurse opened the door.
You are Miss Young? she said to Marian. Yes, of course, we were
expecting you. Sister would like to see you first.
They stood for a moment in a small neat office. The sister rose from
an old Dutch bureau, one of the traces of the house's former
occupants, and held out her hand to Marian. Her eyes rested with
intentness upon the girl's face.

Sir Julian is almost certain to know you, she said gently, but you
mustn't talk much to him. He has been much weakened by
exposure. He lay in a wood for three days without food or water.
There is every hope of his partial recovery, Miss Young; but he needs
rest and reassurance. We can give him the rest here, but we must
look to you to help us to bring back to him the love of life.
Marian stood with her beautiful head raised proudly. She waited for a
moment to control her voice; then she asked quietly:
Is the paralysis likely to be permanent?
The sister moved a chair toward her, but Marian shook her head.
It is a state of partial paralysis. He will be able to get about on
crutches, the sister replied. Won't you rest for a few moments
before going up to him, Miss Young?
No, thank you, said Marian; I will go up to him at once.
She turned quickly toward the door, and meeting Stella's eyes, she
took and held her arm tightly for a moment, and then, loosing it,
walked quickly toward the stairs. Stella followed her as if she had no
being. She had lost all consciousness of herself. She was a thought
that clung to Julian, an unbodied idea fixed upon the cross of
Julian's pain. She did not see the staircase up which she passed; she
walked through the wood in which Julian had lain three days.
He was in a large, airy room with two other men. Stella did not know
which was Julian until he opened his eyes. There was no color in his
face, and very little substance. The other men were raised in bed
and looked alive, but Julian lay like something made of wax and run
into a mold. Only his eyes lived—lived and flickered, and held on to
his drifting consciousness.
The nurse guided Marian to his bed, and, drawing a chair forward,
placed it close to him. Marian leaned down and kissed his forehead.
She had determined to do that, whatever he looked like; and she did
it.

His lips moved. She bent down, and a whisper reached her: I said
I'd come back to you, and I have. Then he closed his eyes. He had
nothing further to say.
Marian did not cry. After the first moment she did not look at Julian;
she looked away from him out of the window. She did not feel that it
was Julian who lay there like a broken toy. It was her duty. She had
submitted to it; but nothing in her responded to this submission
except her iron will.
The nurse had forgotten to bring a chair for Stella. She leaned
against the door until a red-haired boy with a bandaged arm, on the
bed nearest to her, exclaimed earnestly:
Do take my chair! You look awfully done.
She was able to take his chair because her hands were less blind
than any other part of her, and she smiled at him because she had
the habit of smiling when she thanked people. Then her eyes went
back to Julian. Her heart had never left him; and she knew now that
it never would leave him again.
She did not know how long or short it was before Marian rose
gracefully, and said in her clear, sweet voice, I shall come again to-
morrow, Julian.
Marian stopped at each of the other bedsides before she joined
Stella. She said little, friendly, inclusive words to the other two men,
which made them feel as if they would like to sweep the floor under
her feet.
All the same, the red-haired man explained after the door closed,
it was the untidy little one, piled up against the door, that minded
most. I dare say she was his sister.
He had no need to lower his voice, though he did lower it, for fear of
its reaching Julian.

Julian had been reassured, and now he was resting. Consciousness
had altogether receded from him, perhaps that it might give him a
better chance of resting.

CHAPTER XIII
Julian roused himself with the feeling that he had said only half of
what he had intended to say to Marian. It had been in his mind a
long time. It was while he was lying out under the pine-trees that he
had realized what he had got to say to Marian if he ever got back.
There was a complicated cipher message for the Government, which
he had kept quite clear in his mind, and eventually given to an
intelligent doctor to send off; and there was the message to Marian,
which he himself would have to say when he saw her.
I've come back, as I promised; but I can't marry you now, of
course. I'm a crock.
The first time he saw Marian he had got through only the first part
of the sentence. There was no hurry about the rest of it. The doctor
and the sister had both assured him that there was no hurry. They
had been very kind, and quite as honest as their profession
permitted. They said Marian would come back, and he could tell her
then.
They admitted, when he cross-questioned them with all the
sharpness of which he was capable, that he would be a cripple. They
did not bother him with futile commiserations. They gave him quietly
and kindly the facts he asked for. He would never be able to walk
again, but he could get about easily on crutches.
Julian did not want to live very much, but his mother's eyes hurt him
when he tried not to; and then Marian came again, and he got
through the rest of his sentence.
You see, he explained in a low whisper which sounded in his head
like a gong, marriage is quite out of the question.

Marian was there with smiles and flowers, just as he had so often
pictured her; but she sat down with a curious solidity, and her voice
sounded clearer than it had sounded in his dreams.
Nothing alters our engagement, Julian, she said. Nothing can.
She spoke with a finality that stopped his thinking. He had finished
his sentence, and it seemed hardly fair to be expected to start
another on the spur of the moment. He gave himself up to a feeling
of intense relief: he had got off his cipher to the Government and he
had released Marian.
He had known these were going to be difficult things to do. The
cipher had been the worst. The French doctor had taken some time
to understand that Julian must neither die nor be attended to until
he had sent the cipher off; and now the business about Marian was
over, too. He had only to lie there and look at her day by day coming
in with roses. They did not talk much. Julian never spoke of his
symptoms, but they were too radical to free him. He lay under them
like a creature pinned under the wreckage of a railway accident.
Slowly, day by day, his strength came back to him; and as it came
back, peace receded. His eyes lost their old adoring indulgence; they
seemed to be watching Marian covertly, anxious for some gift that
she was withholding from him. He did not demand this as a right, as
the old Julian would have done, breaking down the barriers of her
pride to reach it. He pleaded for it with shamed eyes that met hers
only to glance away. Something in her that was not cruelty as much
as a baffling desire to escape him made her refuse to give him what
his eyes asked.
Julian had loved her for her elusiveness, and the uncaptured does
not yield readily to any appeal from the hunter. The prize is to the
strong.
She would not have withstood a spoken wish of his; but there is
something in speechless suffering from which light sympathies
shrink away. Pity lay in Marian a tepid, quickly roused feeling,

blowing neither hot nor cold. She cried easily over sad books, but
she had none of the maternal instinct which seizes upon the faintest
indication of pain with a combative passion for its alleviation. She
became antagonistic when she was personally disturbed by
suffering.
She was keeping her word to Julian while her heart was drifting
away from him; and he, while he desired her to be free, instinctively
tried to hold her back. They had both put their theories before their
instincts, and they expected their instincts to stand aside until their
theories had been carried out.
Perhaps if Julian could have told her his experiences he might have
recaptured her imagination; but when she asked him to tell her
about them, he said quickly, I can't, and turned away his head. He
was afraid to trust himself. He wanted to tell her everything. He was
afraid that if he began, his reticence would break down, and he
would tell her things which must never pass his lips. He longed for
her to know that every day, and nearly every hour, he had fought
and conquered intricate abnormal obstacles. He had slipped across
imminent death as a steady climber grips and passes across the face
of a precipice.
He had never faltered. All that he had gone to find he had found,
and more. At each step he had seen a fresh opportunity, and taken
it. He had been like a bicyclist in heavy traffic assailed on every side
by converging vehicles, and yet seeing only the one wavering ribbon
of his way out. And he had won his way out with knowledge that
was worth a king's ransom. He could have borne anything if Marian
would realize that what he had borne had been worth while. But
after her first unanswered question, Marian never referred again to
what he had done. She behaved as if his services had been a
regrettable mistake.
She talked with real feeling about the sufferings of those who fought
in the war. Her eyes seemed to tell him what her lips refrained from
uttering, that she could have been more sorry for him if he had been

wounded in a trench, and not shot at and abandoned by a nervous
sentry firing in the dark. He could not remember the exact moment
when out of the vague turmoil of his weakened mind he gripped this
cold truth: Marian was not tender.
When she was not there he could pretend. He could make up all the
beautiful, loving little things she had not said, and sometimes he
would not remember that he had made them up. Those were the
best moments of all. He believed then that she had given him what
his heart hungered for. He was too much ashamed of his ruined
strength to feel resentment at Marian's coldness. It struck him as
natural that she should care less for a broken man.
His mind traveled slowly, knocking against the edges of his old
dreams.
He thought perhaps a nursing home wasn't the kind of place in
which people could really understand one another, all mixed up with
screens and medicine bottles, and nurses bringing things in on trays.
If he could see Marian once at Amberley for the last time, so that he
could keep the picture of her moving about the dark wainscoted
rooms, or looking out from the terrace above the water meadows,
he would have something precious to remember for the rest of his
life; and she mightn't mind him so much there, surrounded by the
dignity of the old background of his race. One day he said to her:
I want to go to Amberley as soon as I can be moved. I want to see
it again with you.
In December? asked Marian, with lifted, disapproving brows. It
would be horribly damp, my dear Julian, all water-meadows and
mist. You would be much more comfortable here.
Julian frowned. He hated the word comfort in connection with
himself.
You don't understand, he said, a little impatiently. I know every
inch of it, and it's quite jolly in the winter. We are above the water. I

want to see the downs. One gets tired of milk-carts and barrel
organs, and the brown tank on the roof across the way. You
remember the downs, Marian?
His eyes met hers again with that new, curiously weak look of his.
Marian turned her head away. How could Julian bear to speak of the
downs?
She saw for a moment the old Julian springing up the hillside
assured and eager, the fine, strong lover who had taken her heart by
storm. She spoke coldly to this weaker Julian.
Yes, she said, I am not likely to forget the downs. I spent the last
happy hours of my life there; but I cannot say I ever wish to see
them again.
Julian's eyes fell, so that she could not see if he had even noticed
how bitterly she remembered Amberley.
The next day she found him sitting up for the first time. He was
propped up by cushions, but it made him look as if he had gained
some of his old incisive strength.
The other two men had been moved, and they had the large, bare
room to themselves.
No sound came from the square beneath them; in the house itself
there were passing footsteps and the occasional persistent buzzing
of an electric bell.
Look here, said Julian in a queer, dry voice, I've got an awful lot
to say to you—d'you mind drawing your chair nearer? I meant to say
it at Amberley. I'd have liked it better there. I rather hate this kind of
disinfected, sloppy place for talk. You must loathe it, too. But here or
there it's got to be said. You said something or other when I first put
it to you—about our engagement never being broken. It was awfully
good of you, of course. I couldn't see through it at the time. I
wanted to let things slide. But it's all nonsense my dear girl. Women
like you can't marry logs of wood.

Welcome to our website – the perfect destination for book lovers and
knowledge seekers. We believe that every book holds a new world,
offering opportunities for learning, discovery, and personal growth.
That’s why we are dedicated to bringing you a diverse collection of
books, ranging from classic literature and specialized publications to
self-development guides and children's books.
More than just a book-buying platform, we strive to be a bridge
connecting you with timeless cultural and intellectual values. With an
elegant, user-friendly interface and a smart search system, you can
quickly find the books that best suit your interests. Additionally,
our special promotions and home delivery services help you save time
and fully enjoy the joy of reading.
Join us on a journey of knowledge exploration, passion nurturing, and
personal growth every day!
ebookbell.com

Current Sources And Voltage References A Design Reference For Electronics Engineers Linden T Harrison

More Related Content

Similar to Current Sources And Voltage References A Design Reference For Electronics Engineers Linden T Harrison (20)

Recently uploaded (20)

Current Sources And Voltage References A Design Reference For Electronics Engineers Linden T Harrison