Mechanical Mea SuGlielus o) re Pca atm eel eta rd Ce Perea rata enecr iui Een ot Greeti parry i‘The author and publisher of this book have used their best efforts in preparing this book. These efforts include the development, research, and testing of the theories and programs to determine their effectiveness. The author and publisher make no warranty of any kind, expressed or implied, with regard to these programs or the documentation contained in this book. The author and publisher shali not be liable in any event for incidental or consequential damages in connection with, or arising out of, the furnishing, performance, or use of these programs. Pearson Prentice Hall™ is a trademark of Pearson Education, Inc. Copyright © 2007 by Pearson Education, Inc. This edition is published by arrangement with Pearson Education, Inc. and Dorling Kindersley Publishing, Ine. This book is sold subject to the condition that it shall not, by way of trade or otherwise, be lent, resold, hired out, or otherwise circulated without the publisher's prior written consent in any form of binding or cover other than that in which it is published and without a similar condition including this condition being imposed on the subsequent purchaser and without limiting the rights under copyright reserved above, no part of this publication may be reproduced, stored in or introduced into a retrieval system, or transmitted in any form or by any means (electronic, mechanical, photocopying, recording or otherwise), without the prior written permission of both the copyright owner and the above-mentioned publisher of this book. ISBN 978-81-317-1718-9 First Impression, 2008 Second Impression, 2009 This edition is manufactured in India and is authorized for sate only in India, Bangladesh, Bhutan, Pakistan, Nepal, Sri Lanka and the Maldives. Circulation of this edition outside of these territories is UNAUTHORIZED. Published by Dorling Kindersley (India) Pvt. Ltd., licensees of Pearson Education in South Asia Head Office: 482, F.LE., Patparganj, Deli 110 092, India Registered Office: 14 Local Shopping Centre, Panchshee! Park, New Delhi 110 017, India. Printed in India by Taj Press.Contents Preface 15 1_Fundamentals of Mechanical Measurement 17 1 The Process of Measurement: An Overview 18 1.1 Introduction axe 18 12. The Significance of Mechanical Measurement. 19 13 Fundamental Methods of Measurement 20 13.1 Direct Comparison . . 20 13.2. Using a Calibrated System . 20 1.4 _ The Generalized Measuring System 21 L4t First, or Sensor-Transducer, Stage... .: . 21 14.2 Second, or Signal-Conditioning, Stage 21 14.3 Third, or Readout-Recording, Stage . ra 22 15. Typesof Input Quantities... . sce 24 15.1 Time Dependence... . . 24 15.2 Analog and Digital Signals. 24 16 MeasurementStandards 0.0. eee eevee eee 26 17 Calibration fect eee 26 18 Uncertainty: Accuracy of Results . 27 19 ReportingResults ........ occas 28 1.9.1 Laboratory Note or Technical Memo ............+. 28 1.9.2 Full Report 2 29 1.9.3 Technical Paper 29 110 Final Remarks 9 2 Standards and Dimensional Units of Measurement uM 2.1 Introduction ee ee ee ee eee a 22 Historical Background of Measurement in the United States... .... 31 23 The SI System nas mn 33 23.1 Establishment of the SIS: 33 23.2 Metric Conversion in the United States . . 35 24 = The Standard of Length... 6.0... 0-02 ee 36 241 elation? of the Meter: to the Inch . 36 25 The Standard of N 7 2.6 Time and Frequency Standards aay 27 ‘Temperature Standards... 0... s eve we vee vteeeeee 38 Electrical Standards AL 2.9 Conversions between Systems of Units . . . . eee .. 42 2.10 Summary... ....... 46 3 This one (NMA SECZ-X6C-3NJX4 Contents 3 Assessing and Presenting Experimental Data 50 3.1 Introduction... . . 50 3.2 Common Types of Error . . . Sl 3.2.1 Classification of Errors 52 3.2.2 Terms Used in Rating Tastrument | Batemans 3.3 Introduction to Uncertainty... 58 3.4 Estimation of Precision Uncertainty . . . . 59 3.4.1 Sample versus Population .. . . . mae 59, 3.4.2 Probability Distributions... . « . él 3.5 Theory Based on the Population... .. 2-0-0 +s +-eeeeeee 62 3.6 Theory Based on the Sample . . 67 3.6.1 An Example of Sampling. 68 3.6.2 Confidence Intervals for Large Samples 70 3.6.3 Confidence Intervals for Small Samples... . n 3.64 Hypothesis Testing for a Single Mean for a Small Sample Size =) eek: Oi EC 8s ES A OY FF CRONE OR 1S 3.6.5 __ Hypothesis Testing for a Single Mean for a Large Semple Size (n> 30). 7 3.6.6 ‘The ¢-Test Comparison of Sample ‘Means . 78 3.7 The Chi-Square (x?) Distribution. .... . . « wcimnee, 19) 3.7.1 Goodness of Fit Based on the Gaussian Distribution... . 83 3.8 Statistical Analysis by computer 87 3.9 Bias and Single-Sample Unceraimy ............- ea BF 3.9.1 Single-Sample Precision Uncertainty 89 3.10 _ Propagation of Uncertainty 89 3.11 Examples of Uncerainty Analysis 1 3.111 Rating Resistors... 0. oe 91 3.11.2 Expected Uncertainty for Flowmeter Calibration . . 94 3.12 Minimizing Error in Designing Experiments 98 3.13 Graphical Presentation of Data ....... cnet 3.13.1 General Rules for Making Graphs... . .. . ses 5D 3.13.2 Choosing Coordinates and Producing Straight Lives: ® 3.14 Line Fitting and the Method of Least Squares 108 3.14.1 Least Squares for Line Fits. . 108 3.14.2 Uncertainty in Line Fits... . « 110 3.14.3 Software for Curve Fitting . semaxcemen oe & MO 3.15 Summary fas eee £% . + 113 4 The Analog Measurand: Time-Dependent Characteristics 123 4.1 Introduction... . BEE GE HRS etre SEDTS 123, 4.2 Simple Harmonic Relations»... 0.00... . wee 123 4.3 Circular and Cyclic Frequene semen 125 44° Complex Relations... ...... 00.02 cece eee eeeev ees 126 4.4.1 Beat Frequency and Heterodyning 1 44.2 Special Waveiorms . scoaeeiemnaan 134 4.43 Nonperiodic or Transient Waveforms . . - 134 4.5 Amplitudes of Waveforms 138Contents 5 4.6 Frequency Spectrum... .. : 138 4.7 Harmonic, or Fourier, Analysis . ae oe ee . 140 ATL he ¢ Fourier Transform 41 4.1.2 Frequencies in Discretely Sampled Signals: Alain and Fre- quency Resolution . 3 143 4.7.3. AnExample of Discrete Fourier Analysis . . . 147 48 Summary... . 151 ‘The Response of Measuring Systems 156 5.1 Introduction... ... . Lee 156 5.2___ Amplitude Response 157 5.3 Frequency Response . . a . . 157 54 PhaseResponse .............-.- 158 5.5 Predicting Performance for Complex Waveforms . 158 5.6 _ Delay, Rise Time, and Slew Rate 161 5.7__ Response of Experimental System Elements 161 5.8 Simplified Physical Systems . 164 5.9 Mechanical Elements . . 164 5.9.1 Mass 2 Spring Force 59.3 Damping... . z 166 5.10 An Example of a Simple Mechanical System ' 167 5.11 The Importance of Damping.......... é 168 5.12 _Dynamic Characteristics of Simplified Mechanical Systems... 169 5.13 _ Single-Degree-of-Freedom Spring-Mass-Damper Systems 169 5:14 ‘The Zoro-Order Systom seve evs ve vs va ov CHOW OR AES 170 5.15 Characteristics of First-Order Systems : 170 5.15.1 The Step-Forced First-Order System é 171 5.15.2 The Harmonically Excited First-Order vstem 173 5.16 Characteristics of Second-Order Systems. 7 5.16.1 The Step-Excited Second-Order System 177 $.162 The Harmonically Excited Second-Order System 180 $.16.3 Generel Periodic Forcing 183, 5.17. Electrical Elements 3 186 5.18 First-Order Electrical ‘System . oe 188 5.19 Simple Second-Order Electrical System .. . 192 5.20 Calibration of System Response... . . 194 5.21 Summary... . 195 Semsnre 99 6.1__Intreduction 109 6.2 Loading of the Signal Source 200 6.3 The Secondary Transducer Hi 3a ai pad 200 6.4 Classification of First-Stage Devices . 202 66 Sliding-Contact Devices. . . 205 6.6.1 Potentiometer Resolution 207 6.6.2 Potentiometer Linearity... a 4 2076 Contents 6.7 _ The Resistance Strain Gage srorssettravetion MEL ses it sen 8h BE 6.8 ‘Themnistor 207 6.9 The Thermocouple see .. 208 6.10 _Variable-Inductance Transducers s z 208 610.1 Simple Self-Inductance Arrangements . 210 6.10.2 Two-Coil Mutual-Inductance Arrangements se .. 210 6.11 _‘The Differential Transformer 2B 6.12 Variable-Reluctance Transducers 213 6.13 Capacitive Transducers ee st fii os ws td Oe Gild Preeoclectiie Senge ee ee BND eS Sha a 6.15.1 Electrical Behavior of Semiconductors... 2.22.2... 219 6.15.2 pn-Junctions odes Soren, sie eh RA: PRON ee eet a as ee 6.16 Light-Detecting Transducers. . . 224 BilGs. Ti DORA a 360 Picton Debts ct oa oy ne oe be oe OR 6.16.3 Applications: ee 225 1 Hall-Fffect Sensors 6.18 Some Design-Related Problems... 1... s+... sees) 20 618.1 Manufacturing Tolerances... aioe oi 230 6.18.2 _ Some Temperature-Related Problems. _ 232 6.18.3 Methods for Limiting Temperature Ertors.......- 236 69) Samaimaiyss i ccc-wsemcavasseanewems ne bm On ae 238 7 Signal Conditioning 244 LL Introduction 244 7.2. Advantages of Electrical Signal Conditioning . . 245 7.3 Modulated and Unmodulated Signals . . 243 7.4 _ Input Circuitry 245 7.5, The Simple Current-Sensitive Circuit 247 7.6 The Ballast Circuit . 249 7.7 Voltage-Dividing Circuits . i 251 77.1 The Voltage-Dividing Potentiometer_ 251 77.2 LoadingEmor ..... +. 5-055 252 7.1.3 Use of EndResistors .. . . % 253 7.8 Small Changes in Transducer Resistance . . . 253 7.9 Resistance Bridges . . . 256 79.1 The Voltage-Sensitive Wheatstone Bridge. . . 79.2 The Current-Sensitive Wheatstone Bridge 7.9.3 The Constant-Current Bridge . 79.4 The AC Resistance Bridge . 795 — Compensation for Leads . . 79.6 Adjusting Bridge Sensitivity 7.10 Reactance or Impedance Bridges . . 7.11 Resonant Circuits ©... 2.6... 7.11.1 Undesirable Resonance ConditionsContents 7 7.12 Electronic Amplification or Gain 267 7.13. Electronic Amplifiers . 270 7.14 Operational Amplifiers... . . 271 714.1 "ypieal Op-Am Specicatons . 273 ns of the Op Amy 274 715 Spetid Amplifer iit - 279 7.15.1 Instrumentation Amplifiers 219 715.2 The Charge Amplifier . 29 716 Filters ......... 281 717 Some Filter Theory ....... 282 718 ActiveFilters... 2.20... 288 7.19 Differentiators and Integrators . 288 7.20 Shielding and Grounding : 290 7.20.1 Shielding . 290 7.20.2 Grounding . é LOVED ee ¥ ++ 298 7.21 Component Contig Methods. 292 7.22 Summary . 294 Digital Techniques in Mechanical Measurements 305 8.1 Introduction we ere 305 82 Why Use Digital Methods? . .. . 307 83 Digitizing Mechanical Inputs... . . « 308 84 Fundamental Digital Circuit Elements 308 84.1 Basic Logic Elements » eB *s @ . + 308 842 — Combination of Logic Elements: The Fi Fron eee es BID 84.3 IC Families. . . = es HS 312 84.4 1C Oscillators and Clock Signals... eee 315 84.5 Digital Displays 316 85 Number Systems... - ne sae oe . 317 86 BinaryCodes .... 319 86.1 Binary-Coded Decimal... . 319 8.6.2 Position Encoders and Gray Code... sss 320 8.6.3 Alphanumeric Codes - 2 ee 32 7 321 REA tiaras sep ere er &7__ Some Simple Digital Circuitry... 0... 325 CE ES PEEP EE | 87.2 Gating... . 327 8.7.3. Frequency Meter . 327 87.4 Wave Shaping eee 328 8.7.5 Inegrated-Circuit Counter and Frequency Meter. . . 328 8.7.6 — Multiplexing and Demultiplexing : 329 88 The Computer as a Measurement System . 331 89 The Microprocessor . . 332 8.10 The Microcomputer . . 334 8.10.1 Read-Only Memory (ROM) .....0. 0.0 s oe 334 8.10.2 Random Access Memory (RAM) . 334 8.10.3 Input and Output Ports (1/0 Ports) 335 8.10.4 External Buses . 3358 Contents I 10 8.11 Analog-to-Digital and Digital-to-Analog Conversion... .... . 8.11.1 A Digital-to-Analog Converter. 8.11.2 An Analog-to-Digital Converter... 1... 8.11.3 Analog-to-Digital Conversion Considerations... . B14 Digital Signal Processing nn : 8.12 Digital Images... . 8.13 Getting It All Together... 2... 8.14) SMIMMIREY!. occ sionsace oe oo woe seo RIOT eTRHe oe € Readout and Data Processing 9.1 Introduction 92 TheElectronic Counter 9.21 EventCounter ... . 9.22 Time-Interval Meter 9.23 — Events per Unit Time (EPUT) Meter. . . 9.24 CountEmor . 6... cece eee 9.3 Analog Blectric Meter Indicators. . 9.3.1 Voltmeter Sensitivity . 932 The Current Meter . . 933 ACMeters...... eat 9.34 The Multimeter and Resistance Measurement - 9.4 Meters with Electronic Amplification 9.5 Digital-Readout Multimeters ..... .. 9.6 The Cathode-Ray Oscilloscope (CRO) . . 9.6.1 Oscilloscope Amplifiers... . . 9.6.2 Sawtooth Oscillator or Time-Base Generator. 9.63 Synchronization or Triggering 9.1 Additional CRO Features . . . 97.1 Multiple Trace . . . 9.72 Magnification and Delayed Sweep 9.73 Digital Storage Oscilloscopes ...... . . 9.7.4 Single-Ended and Differential Inputs 5 98 XVPloters .. 2.0.0... 9.9 Digital Waveform Recorders 9.10 The Spectrum Analyzer . 9.11 LabVIEW . Applied Mechanical Measurements Measurement of Count, EPUT, Time Interval, and Frequency 10.1 Introduction Ba ese gee oa 10.2 Use of Counters ........ 10.2.1 Electronic Counters. 10.2.2. EPUT Meters 10.2.3 Time-Interval Meter . 335 337 340 341 343 345 350 350 351 351 351 354 354 355 356, 357 357 360 361 361 362 363 366 366 368 368 368 368 370 370 370 ah 373 379 380 380 381 381 382 38210.3 Stroboscopy and High-Speed Imaging... 0... 2.0... eee 10.3.1 The Stroboscope . 10.4 Frequency Standards . i 10.4.1 Global Positioning System Signals... . .. « 10.4.2 Radio Time and Frequency Transmission... . 10.4.3 Quartz-Crystal Oscillators ee I ‘Complex-Wave Oscillators and Function Generators . . . 10.5 Lissajous Figures for Finding Frequency and Phase Relations . . . 10.6 Heterodyne and Phase-Lock Measurements of Frequency 10.7 Measurement of Rotary Motion... 00.6.0 11 Displacement and Dimensional Measurement 11.1 Introduction see 11.2. A Problem in Dimensional Measurement... . 11.3 Gage Blocks 11.4 Assembling Gage-Block Stacks . . . 11S Surface Plates... .. 11.6 Temperature Problems . . . 11,7 Use of Comparators... . 11.8 Monochtomatic Light . . . 11.9 The Interferometer . . . 11.10 Measuring Microscopes v 1.10.1 Fixed-Scale Microscopes. 11.10.2 Filar Microscopes . . 11.103 ‘Traveling and Traveling-Stage Microscopes . 11.10.4. The Draw-Tube Microscope 11.105 Focusing wee 11.106 Digital Microscopes - see 11.11. Whole-Field Displacement Measurement . . . . 11,12 Displacement Transducers . . . . 11.13 The Differential Transformer or LVDT . 11.13.1 Input Power... 2. 1.13.2. Advantages of the LVDT 11.14 Surface Roughness... . . 12 Strain and Stress: Measurement and Analysis 12.1 Introduction ...........- 12.2 Strain Measurement 12.3. The Electrical Resistance Strain Gage 12.4 The Metallic Resistance Strain Gage 12.5. Selection and Installation Factors for Bonded Metallic Strain Gages . 12.6 Circuitry for the Metallic Strain Gage 12.7. The Strain-Gage Ballast Circuit 12.8 The Strain-Gage Bridge Circuit . 2 12.8.1 Bridges with Two and Four Arms Sensitive to Strain... 12.8.2. The Bridge Constant 12.8.3 Lead-Wire Error... .......-10 Contents 12.9 The Simple Constant-Current Strain-Gage Circuit 12.10 Temperature Compensation . . . 2.10.1 ‘The Adjacent-Arm Compensating Gage . 12.10.2 Self-Temperature Compensation 12.11 Calibration . . 12.12 Commercially Available Strain-Measuring Systems 12.121 The Basic Indicator bene eee 12.13 Strain-Gage Switching . 12.14 Use of Strain Gages on Rotating Shafts . 12.15. Stress-Strain Relationships... .. . . tee 12.151 The Simple Uniaxial Stress Situation . - 12.15.2 The Biaxial Stress Situation... . . . . 12.16 Gage Orientation and Interpreiation of Results . . 12.16.1 Gages Connected in Series . 12.17 Special Problems 12.17.1 Cross Sensitivity... . 12.172 Plastic Strains and the Postyield Gage . 12.17.3 Fatigue Applications of Resistance Strain Gages . 12.17.4 Cryogenic Temperature Applications 12.17.5 High-Temperature Applications 12.176 Creep... cece ae 12.17.7 Residual Stress Determination . 12.18 Final Remarks... . 0. ee ee 13 Measurement of Foree and Torque 13.1 Introduction . . . ns 13.2 Measuring Methods * 13.3 Mechanical Weighing Systems... 13.3.1 The Analytical Balance . . 13.32 Multiple-Lever Systems . . 13.33 ‘The Pendulum Force-Measuring Mechanism. - 13.4 Elastic Transducers 13.4.1 Calibration Adjustment 13.4.2 The Proving Ring 13.5 Strain-Gage Load Cells . . . 13.6 Piezoelectric Load Cells . 13.7 Ballistic Weighing . 13.8 Hydraulic and Pneumatic Systems 13.9 Torque Measurement 13.9.1 Mechanical and Hydraulic Dynamometers « 13.9.2 Electric Dynamometers 13.10 Transmission Dynamometers 14 Measurement of Pressure 14.1 Introduction 14.2. Static and Dynamic Pressures in Flaids . 14.3 Pressure-Measuring Transducers 461 461 461 461 462 463 463 463 473 473 475 476 476 477 479 479 483 484 486 490 490 492 494 495, 498 500 504 505 507Contents 11 144 Manometry.........0.00- 508 14.8 Bourdon-Tube Gages . . 513 145.1 Flat Metal Diaphragms... 514 14.5.2 Corrugated Metal Diaphragms . 515 14.53 Semiconductor Diaphragms 516 14.6 Additional Pressure Transducers... . 516 14.6.1 Strain Gages and Flat Diaphragms 516 14.62 Inductive Transducers . 517 14.6.3 Piezoelectric Transducers . S17 14.64 Capacitative Transducers 518 14.65. Strain-Gage Pressure Cells 519 14.7 Measurement of High Pressures . : 521 14.7.1 Electrical Resistance Pressure Gages. 21 14.8 Measurement of Low Pressures . 523 14.8.1 The McLeod Gage . . . 523 1482 Thermal Conductivity Gages wee 524 1483. onizationGages.............0-.- 525 149 Dynamic Characteristics of Pressure-! "Messing Systems 526 149.1 Gas-Filled Systems 3 527 14.9.2. Liquid-Filled Systems . 528 14.10 Calibration Methods... 2... « 530 14.10.1 Periodic Pressure Sources . 531 14.10.2 Aperiodic Pressure Sources 532 15 Measurement of Fluid Flow 542 15.1 Introduction... 6.6.0 eee eee ee 542 15.2 Flow Characteristics . . . 544 15.3. Obstruction Meters. . . Ai 546 15.3.1 Obstruction Meters for Incompressible Flow . . 346 15.3.2. Venturi Tube Characteristics . 548 15.3.3 Flow-Nozzle Characteristics 549 15.3.4 Orifice Characteristics . : 549 15.3.5 Relative Merits ofthe Venturi, Flow Nozzle, and Orifice... 556 15.4 Obstruction Meters for Compressible Fluids... . . 556 15.4.1 Choked Flow and the Critical Flow Venturi Meter. 559 15.5 Additional Flowmeters jw ee 15.5.1 Turbine Meters . a ae ee ae oe 15.5.2 Electromagnetic Flowmeters 562 15.5.3 Coriolis Flowmeters... . . 563 15.54 Vortex Shedding Flowmeters... . 564 15.55 Ultrasonic Flowmeters... .. . . 565 15.5.6 Positive Displacement Flowmeters 565 15.5.7 The Variable-Area Meter ..... . 566 15.6 Calibration of Flowmeters . 567 15.7 Measurements of Fluid Velocities . - 56912. Contents 16 15.8 15.9 15.10 Doppler-Shift Measurements 15.11 Flow Visualization Pressure Probes... . .. 15.8.1 Incompressible Fluids 15.8.2 Compressible Fluids 15.8.3 Total-Pressure Probes. 15.84 Static-Pressure Probes . . . . 15.85 Direction-Sensing Probes . . ‘Thermal Anemometry 15.101 Laser-Doppler Anemometry . 15.10.2 Ultrasonic-Doppler Accmoaichy ‘Temperature Measurements 16.1 16.2 16.3 16.4 16.5 16.6 16.7 16.8 16.9 Introduction... . Use of Thermal Expansion 16.2.1 Liquid-in-Glass Thermometers. 16.2.2 Calibration and Stem Correction 16.2.3 Bimetal Temperature-Sensing Elements . . Pressure Thermometers . ‘Thermoresistive Elements a 16.4.1 Resistance Thermometers aE 16.4.2 Instrumentation for Resistance Thermometry . . . 16.43 Thermistors 00222 e eee ee ‘Thermocouples: 16.5.1 Application Laws for Thermocouples 16.5.2. Thermocouple Materials and Construction 16.5.3 Values of the Thermocouple EMF 16.5.4 Measurement of Thermocouple EMF . 16.55. Electronic Instruments 16.5.6 Thermopiles and Thermocouples Connected in Parallel Semiconductor-Junction Temperature Sensors... . . . ‘The Linear Quartz Thermometer Pyrometry 16.8.1 Radiation Pyrometry Theory 16.82 Total-Radiation Pyrometry........ 16.8.3 Spectral-Band Pyrometry 16.8.4 Infrared Pyrometry . 16.85 Thermal Imaging . . Other Methods of Temperature Indication 16.10 Temperature Measurement Errors 16.11 Measurement of Heat Flux. . 16.12. Calibration of Temperature-Mé 16.10.1 Errors Associated with Convection, Radiation, and Conduction 16.10.2, Measurement of Temperature in Rapidly Moving Gas. . 16.10.3 Temperature Element Response . . . . 16.10.4 Compensation for Temperature Element Response 570 sm 572 572 573 575 o7 580 581 5&4 585 592 592 598 598 599 601 601 603 603 605 608 6ll 613 613 616 617 625 627 628 628 630 631 635 637 639 640 640 641 64 647 o49 652 653 65717 Measurement of Motion on 17.1 Introduction... 2.2... ..----- 671 17.2 Vibrometers and Accelerometers . . 672 173 Elementary Vibrometers and Vibration Detectors 672 174 Elementary Accelerometers 674 17.5 The Scismic Instrument . . 674 17.6 General Theory of the Seismic Instrument . 675 17.6.1 _ The Vibration Pickup ...... . 67 17.62 Phase Shift in the Seismic Vibrometer. 678 17.6.3 General Rule for Vibrometers 680 17.7 The Seismic Accelerometer ....... 680 17.7.1 Phase Lag in the Accelerometer... 0... 0. eee eee 682 17.7.2 Practical Design of Seismic Instruments 178 Calibration. ...... 4. . 179 Calibration of Vibrometers 17.10 Calibration of Accelerometers 17.10.1 Static Calibration . 17.10.2 Steady-State Periodic Calibration e 17.11 Response of the Seismic Instrument to Transients. . 17.12 Moasmemen of Velocity by the Laser Velociy Tanaducet 17.13 Vibration and Shock Testing... .. . . 17.14 Vibrational Exciter Systems .... . . . 17.14.1 Electromagnetic Systems. . . 17.14.2, Mechanical-Type Exciters ...... . ee 17.143 Hydraulic and Pneumatic Systems... 2... 17.14.4 Relative Merits and Limitations of Each System 17.15 Vibration Test Methods ....... Cheb etwas 17.15.1 The Brute-Force Method . 17.15.2 The Resonance Method . . . . 17.16 Shock Testing . 17.17 Shock Rigs . IAA ‘Ait Gun Shock: Producing Devices be 17.17.2 Spring-Loaded Test Rigs . 17.17.3 A Hydraulic-Pneumatic Rig . 17.17.4 Gravity Rigs . . . 17.17.53 Relative Merits and Limitations of Each Shock cg * 17.18 Practical Shock Testing . . wm 18 Acoustical Measurements 18.1 Inoduction ............0. 18.2 Characterization of Sound (Noise) . . 18.3 Basic Acoustical Parameters... . 18.3,1 Sound Pressure . ean 18.3.2 Sound Pressure Level . . 18.33 Power, Intensity, and Power Level. . . 18.3.4 Combination of Sound Pressure Levels 18.3.5 Attenuation with Distance14 Contents 18.4 Psychoacoustic Relationships ......... m5 185 Sound- Measuring Apparatus and Techniques nT 18.5.1 ic ‘ 17 18.5.2 ‘The Sound Level Meter. 79 18.5.3 Frequency Spectrum Analysis . . 7 18.5.4 The Discrete Fourier Transform . . 25 18.6 Applied Spectrum Analysis . ns 187 Measurement and Interpretation of Industrial and Environmental Noise | 727 18.7.1 Equivalent Sound Level, Leq 27 18.7.2 Sound Exposure Level (SEL) . . 728 18.7.3 Sound Intensity Measurement . . . . » 78 188 Notes on Same Practical Anes of Soun easuremen + BO 18.9 Calibration Methods . 732 18.10 Final Remarks... . 733 III Appendices 737 A. Standards and Conversion Equations 738 B Theoretical Basis for Fourier Analysis TAL C Number Systems 146 D Some Useful Data 751 E_ Stress and Strain Relationships 755 F Statistical Tests of Least Squares Fits 770 Answers to Selected Problems 714 IndexPART ONE FUNDAMENTALS OF MECHANICAL MEASUREMENTSection 1.4 The Generalized Measuring System 21 temperature measurement must be combined with a radio-frequency signal for transmission to the ground. In each of the various cases requiring amplification, or filtering, or remote record- ing, electrical methods suggest themselves. In fact, the majority of transducers in use, particularly for dynamic mechanical measurements, convert the mechanical input into an analogous electrical form for processing. 1.4 THE GENERALIZED MEASURING SYSTEM Most measuring systems fall within the framework of a general arrangement consisting of three phases or stages: Stage 1. A detection-transduction, or sensor-transducer, stage Stage 2. An intermediate stage, which we shall call the signal-conditioning stage Stage 3. A terminating, or readour-recording, stage Each stage consists of a distinct component or group of components that performs required and definite steps in the measurement. These are called basic elements; their scope is determined by their function rather than by their construction. Figure 1.2 and Table 1.1 outline the significance of each of these stages. 1.4.1 First, or Sensor-Transducer, Stage The primary function of the first stage is to detect or to sense the measurand, At the same ¢, ideally, this stage should be insensitive to every other possible input, For instance, if it isa pressure pickup, it should be insensitive to, say, acceleration; if it isa strain gage, it should be insensitive to temperature; if a linear accelerometer, it should be insensitive to angular acceleration; and so on. Unfortunately, itis rare indeed to find a detecting device that is completely selective. Unwanted sensitivity is a measuring error, called noise when it varies rapidly and drift when it varies very slowly. Frequently one finds more than a single transduction (change in signal character) in the first stage, particularly if the first-stage output is electrical (see Section 6.3). 1.4.2. Second, or \al-Conditioning, Stage The purpose of the second stage of the general system is to modify the transduced informa- tion so that it is acceptable to the third, or terminating, stage. In addition, it may perform ‘one or more basic operations, such as selective filtering to remove noise, integration, dif- Calibration Auxiliary power ‘Aundiary power input {not always requirea) (usually required) Le we J+] Recorder Measurand “S|- Sensor Signal >| >| >) Computer twaneducer [Traneauced”| conditioner | Anaiogous r signal diving |_| . (analogous signal Froceoo te input rou) L>} Controter FIGURE 1.2: Block diagram of the generalized measuring system.-warsks Suunsnour xojduuos Ajaane(ai e Jo umexFeIp Yo1g *$T TUNEL ‘yde6 zandwog ‘oun nod 17 SAU KyDOI@A, Buypiooat Ayoqen a Penowel —_,9S}0U, PALUEMLUN UN Je;nduioo 10} ——snBojeue eGeyon ——es}oU YM" ejoUIOIeJe008 Way Bayon posnaioy —_poreiBewy-out, reubis indino oBBHION, ; | seynduioo 1 NONI i weld he vowsinboe-212q) r+ Jowiduty be Sunes6enuy ‘ | | ‘wiershs inopees-Gujpiooey | wajshs Buruonipues-eubig | seonpevensosues seas —>y-——— zebeis To bees 25Section 1.10 Final Remarks 29 be sufficient, including a sentence or two stating the problem, a block diagram of the experimental setup, and some data presented cither in tabular form or as a ploited diagram. Any pertinent observations not directly evident from the data should also be included. Sufficient information should be included so that the experimenter can mentally reconstruct the situation and results 1 year or even 5 years hence. A date and signature should always be included and, if there is a possibility of important developments stemming from the work, such asa patent, a second witnessing signature should be included and dated. 1.9.2. Full Report The full report must relate all the facts pertinent to the project. It is even more important in this case to make the purpose of the project completely clear, for the report will be read by persons not closély associated with the work. The full report should also include enough detail to allow another professional to repeat the measurements and calculations. ‘One format that has much merit is to make the report proper—the main body—short and to the point, relegating to appendices the supporting materials, such as data, detailed descriptions of equipment, review of literature, sample calculations, and so on. Frequent reference to these materials can be made throughout the report proper, but the option to peruse the details is left to the reader. This scheme also provides a good basis for the technical paper, should it be planned. 1.9.3 Technical Paper A primary purpose of 4 technical paper is to make known (to advertise) the work of the writer. For this reason, two particularly important portions of the writing are the problem statement and the results. Adequately done, these two items will attract the attention of other workers interested in the particular field, who can then make direct contact with the writer(s) for additional details and discussion. Space, number of words, limits on illustrations, and perhaps time are all factors making the preparation of a technical paper particularly challenging. Once the problem statement and the primary results have been adequately established, the remaining available space may be used to summarize procedures, test setups, and the like, 1.10 FINAL REMARKS An attempt has been made in this chapter to provide an overall preview of the problems of mechanical measurement. In conformance with Section 1.9, we have tried to state the problem as fully as possible in only a few pages. In the remainder of the book, we will expand on the topics introduced in this chapter. Figure 1.5 illustrates the interrelation of these topics and their organization within this book. PROBLEMS LL. Write an executive summary of this chapter. 1.2, Consider a mercury-in-glass thermometer as a temmperature-measuring system. Discuss the various stages of this measuring system in detail 1.3, For the thermometer of Problem 1 obiained. specify how practical single point calibration may be 1.4, Setup test procedures you would use to estimate, with the aidonly of your present judgment and experience, the magnitudes of the common quantities listed. (a) Distance between the centerlines of two holes in a machined partSection 2.3 The Si System 33 called the General Conference on Weights and Measures and attended by representatives from most of the industrial countries of the world. In addition, numerous interim meetings are held to consider solutions to more specific problems, for later action by the General Conference, 23 THE SI SYSTEM 2.3.1 Establishment of the St System The International System of Units, or SI System, has its origins in the Decimal Metric System that was introduced at the time of the French Revolution, During the next two centuries, metric systems of measurement continued to evolve, and they came to encompass both mechanical and electrical dimensions. Finally, in 1960, the Eleventh General Conference on Weights and Measures formally established the SI System, consisting of dimensional standards for length, mass, time, electric current, thermodynamic temperature, andluminous intensity, The Fourteenth General Conference on Weights and Measures (1971) added the mole as the unit for amount of substance, completing the seven dimensional system in use today [7]. The seven base units of the SI System are listed in Table 2.1, Other dimensions can be derived from these base units by multiplying or dividing them. A few such derived units are assigned special names; others are not. For example, the unit of force, the newton, is obtained from the kilogram, the meter, and the second as newton = I kg/m s? In contrast, area is simply meters squared (m7). Work and energy are expressed in joules (kg. m*/s). The term heriz is used for frequency (s~') and the term pascal is used for pressure (N/m), Some derived units carrying special names are listed in Table 2.2, and some without special names are given in Table 2.3. Note that, whereas those assigned special ‘names that originate from proper names are not capitalized, the corresponding abbreviations are capitalized. It should be clear that all the various derived units can be expressed in terms of base units, Tn certain instances when a unit balance is attempted for a given equation, it may be desirable, or necessary, to convert all variables to base units. TABLE 2.1: Base Units in the SI System Uni Quanti eo es ny Name Symbol length meer m mass kilogram kg time seconds electric current ampere A temperature kelvin = K amount of subsiance mole mel luminous intensity candela. cd.36 Chapter2 Standards and jensional Units of Measurement 2.4 THE STANDARD OF LENGTH ‘The meter was originally intended to be one ten-millionth of the earth's quadrant. In 1889, the First General Conference on Weights and Measures defined the meter as the length of the International Prototype Mcter, the distance between two finely scribed lines on a platinum-iridium bar when subject to certain specified conditions. On October 14, 1960, the Eleventh General Conference on Weights and Measures adopted a new definition of the meter as 1,650,763.73 wavelengths in vacuum of the radiation corresponding to the transition between the levels 2pj9 and Sds of the krypton-86 atom. The National Bureau of Standards of the United States also adopted this standard, and the inch became 41,929.398 54 wavelengths of the krypton light. As it turned out, the wavelength of krypton light could only be determined to about 4 parts per billion, limiting the accuracy of the meter to a similar level. During the 1960s and early 1970s, laser-based measurements of frequency and watelength evolved to such accuracy that the uncertainty in the meter became the limiting uncertainty in determining the speed of light [10,11]. This limitation was of serious concern in both atomic and cosmological physics, and on October 20, 1983, the Seventeenth General Conference on ‘Weights and Measures redefined the meter directly in terms of the speed of light: ‘The meter is the length of the path travelled by light in vacuum during a time interval of 1/299,792,458 of a second. This definition has the profound effect of defining the speed of light to be 299,792,458 m/s, which had been the accepted experimental value since 1975 [12]. 2.4.1 Relationship of the Meter to the Inch The 1866 U.S. Statute had specified that | m = 39.37 inches, resulting in the relationship 1 in. = 2.540 00508 em (approximately) In 1959, the National Bureau of Standards made a small adjustment to this relationship to ensure international agreement on the definition of the inch [13]: Vin, = 2.54¢em (exactly) This simpler relationship had already been used as an approximation by engineers for years. ‘The difference between these two standards may be written as 2.54005 08/2.54 — 1 = 0.000002 or 0,0002%, which is about § in. per mile. We gain a scnsc of the significance of the difference by considering the following situation. In 1959, the work of the United States National Geodetic Survey was based on the 39.37 in./m relationship and a coordinate system with its origin located in Kansas. Changing the relationship from 39.37 in./m (exactly) to 2.54 in/em (exactly) would have caused discrepancies of almost 16 ftata distance of 1500 miles. Onecanonly imagine the confusion. over property lines if such a change had been made! This problem was resolved by defining separately the U.S, survey foot (12/39.37 m) and the international foot (12 x 2.54 em). The survey foot is still used with U.S. geodetic data and U.S. statute miles [14]Section 2.5 The Standardof Mass 37 2.5 THE STANDARD OF MASS The kilogram is defined as the mass of the International Prototype Kilogram, a platinum- iridium weight kept at the International Bureau of Weights and Measures near Paris. Of the basic standards, this remains the only one established by a prototype (by which is meant the original model or pattem, the unique example, to which all others are referred for comparison). Various National Prototype Kilogram masses have been calibrated by comparison to the Intemational Prototype Kilogram. ‘These masses are in turn used by the standards agencies of various countries to calibfate other standard masses, and so on, until one reaches masses or weights of day-to-day goods and services. Apart from the inconvenience of maintaining this chain of calibration, the defini- tion of the kilogram by an international prototype leads to several very fundamental prob- lems: ‘The prototype can be damaged or destroyed; the mass of thie prototype fluctuates by about one part in 10* owing to gas absorption and cleaning; and the prototype ages in an unknown manner, perhaps having resulted in 50 jg of variation during the past cen- tury [15]. In recent years, considerable effort has been given to developing a new mass stan- dard that can be reproduced in any suitably equipped lab, without the use of a prototype. One approach being considered is to precisely determine Avogadro's number by mass and density measurements of silicon crystals. This value of Avogadro's number could then be used with an atomic unit of mass to define the kilogram as the mass of a specific number of atoms [15,16]. An alternative approach, which promises somewhat better accu- racy, uses a “moving coil watt balance” to compare the mechanical and electrical power exerted on acurrent-carrying conductor that moves against gravity in a magnetic field, This technique leads to a definition of the kilogram in terms of fundamental physical quanti- ties [17.18]. The pound was defined in terms of the kilogram by the Mendenhall Order of 1893. In 1959, the definition was slightly adjusted [13], giving the relationship still in use today: 1 pound avoirdupois = 0.453 592 37 kilogram 2.6 TIME AND FREQUENCY STANDARDS Until 1956, the second was defined as 1/86,400 of the average period of revolution of the earth on its axis. Although this seems to be a relatively simple and straightforward definition, problems remained, There is a gradual slowing of the earth's rotation (about 0.001 second per century) [19], and, in addition, the rotation is irregular. ‘Therefore, in 1956, an improved standard was agreed on; the second was defined as 1/31,556,925.9747 of the time required by the earth to orbit the sun in the year 1900. This is called the ephemeris second. Although the unit is defined with a high degree of exactness, implementation of the definition was dependent on astronomical observation, which was incapable of realizing the implied precision. In the 1950s, atomic research led to the observation that the frequency of electro- magnetic radiation associated with certain atomic transitions may be measured with great38 Chapter 2 Standards and Dimensional Units of Measurement repeatability. One—the hyperfine transition of the cesium atom—was related to the ephem- eris second with an estimated accuracy of two parts in 109. On October 13, 1967, in Paris, the Thirteenth General Conference on Weights and Measures officially adopted the foliow- ing definition of the second as the unit of time in the SI System [7]: The second is the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the cesium 133 atom. Atomic apparatuses, commonly called “atomic clocks,” are used to produce the fre- quency of the mansition [20]. In a founain clock [21}, a gas of cesium atoms is inwoduced into a vacuum chamber, where a set of laser beams is used to slow the molecular motion, pushing a group of atoms into a ball and cooling them to a temperature near absolute zero. Another laser is then used to toss the ball of atoms upward into a microwave cavity, where some of the atoms are excited to higher energy levels. When the ball falls again, yet another laser is used to force the emission of radiation. This radiation is detected, yielding the desired frequency. The best cesium standards reproduce the second to an accuracy better than one part in 105. 2.7 TEMPERATURE STANDARDS The basic unit of temperature, the kelvin (K), is defined as the fraction 1/273.16 of the thermodynamic temperature of the triple point of water, the temperature at which the solid, liquid, and vapor phases of water coexist in equilibrium. The degree Celsius (°C) is defined by the relationship t=T-273.15 where f and T represent temperatures in degrees Celsius and in kelvins, respectively. In reality, two temperature scales are defined, a thermodynamic scale and a practical scale, The latter is the usual basis for measurement. The thermodynamic temperature scale is defined in terms of entropy and the properties of heat engines (22}. It can be implemented directly only with specialized thermometers that use media having a precisely known equation of state (a constant volume, ideal gas thermometer, for example). Such thermometers are difficult and time consuming to use ifaccuracy is desired, and, asa result,a corresponding scale which is more easily realizable is needed [7]. Thus, the thermodynamic scale iy normally approximated using a so-called practical scale. ‘A practical scale has two components. The first is a set of fixed reference temper- atures, defined by specific states of matter. The second is a procedure for interpolating between those reference points, for example, by measuring a temperature-dependent elec- trical resistance. Using the interpolation formulae and fixed points of the practical scale, one can calibrate any other temperature measuring device. The fixed reference temperatures must correspond to thermodynamic states that are very accurately reproducible. Zero degrees Celsius is the temperature of equilibrium between pure ice and air-saturated pure water at normal atmospheric pressure. However, a more precise datum, independent of both ambient pressure and possible contaminants, is the triple point temperature of water. As noted above, the value 273.16 K (or 0.0100°C) is assigned to this temperature. Relatively simple apparatus can be used to reproduce this temperature fixed point [23].Section 2.8 Electrical Standards 41 where E\(T), Ey (Trot) the radiant energy emitted by the black body per unit time, per unit area, and per unit wavelength at a wavelength A, and at a temperature 7’ or Tre, respectively Tet ‘The freezing point temperature of either silver (1234.93 K), gold (1337.33 K), or copper (1357.77 K) 0.014388 m- K Theradiant energy is typically measured by optical pyrometry (Section 16.8). The unknown temperature is then calculated by comparing the emission of a source at the unknown temperature to that from a source at the reference temperature. The International Temperature Scale of 1990 thus establishes means of determining any temperature from 0.65 K to more than 4000 K. In actual applications, the standardized pyrometer, the standardized resistance thermometer, or the standardized gas thermometers are used as secondary standards for calibration of working instruments (Section 16.12). Apart fom any uncertainties introduced in the calibration procedures, the major uncer- tainties in ITS-90 arise in realizing the fixed points. Ata lo level (Sections 3.53.6), the uncertainties in the fixed-point temperatures are =£0.5 to 1.5 mK for temperatures up to the melting point of gallium, increasing to +60 mK at the freezing point of copper [25]. The temperature units of the English Engineering System are defined in terms of the kelvin, Absolute temperature takes units of degrees Rankine (°R), which differ from the kelvin by a factor of 1 TPR) = 1.8 x T(K) ‘The degree Fahrenheit (°F)is defined by subtracting 459.67 from the temperature in degrees Rankine: TF) = TCR) — 459.67 2.8 ELECTRICAL STANDARDS In the SI System, all electrical units originate from the definition of the ampere. One ampere is defined as the current that produces a magnetic force of 2 x 10-7 N/mon a pair of thin parallel wires carrying that current and separated by one meter. The force on an appropriate pair of conductors can be measured directly, using a so-called current balance {26}. The current may be then calculated from the relations of electromagnetic theory. The remaining electrical units, such as volts and ohms, can all be derived from the value of the ampere and the mechanical units of mass, length, and time, again using the results of electromagnetic theory. ‘The measurement of current from the SI definition of the ampere is cumbersome, just as isthe measurement of temperature from the thermodynamic temperature scale. Obtaining the volt and the ohm from the SI definition is also difficult. Consequently, practical standards are normally used in place of the SI definitions in order to obtain the volt and the ohm, Traditionally, the practical realization of the volt was a so-called standard cell, an electrochemical cell of relatively high stability. National standards laboratories maintainedSection 2.9 Conversions between Systems of Units 45 EXAMPLE 2.1 Determine the conversion factor between pounds-force and newtons. Solution The conversion factors between inches and meters and between pounds-mass and kilograms were given in Sections 2.4 and 2.5. 1 Ibf = 32.174 Ibm - ft/s? = 32.174(0.453 592 37 kg)(12 x 0.0254 m)/s? 4.4482 kg - m/s? EXAMPLE 2.2 Water of density p and dynamic viscosity jx flows with velocity V through a pipe of diameter D. Calculate the Reynolds number, Re, from the data supplied, using (a) the English Engineering System of units and (b) the SI System, Before making the numerical calculations, check the balance of units. Referring to Section 15.2, we see that Re = pV D/s and, as discussed in that section, its value is unitless and hence is independent of the system of units used: thus, we should obtain the same numerical answers for both parts (a) and (b).. Data (see Appendix A for conversion factors): D =8.00 in, = 8/12 ft = 0.203 m p =62.3 Ibm/f? = 998 kg/m* (see Appendix D) V =4.00 ft/s = 1.22 m/s = 2.02 x 107 Ibf - s/ft? =9.67 x 1074 N-s/m? (see Appendix D) Solution 1. If we enter the units for each of the separate quantities appearing in the equation for Re, we have (Ibm f°) ft/s) (fx (EP ADE - )(1/¢e) or, entering the units for gc, (ibm /f? )(it/s) (YC? /Abf - s)UbF - s?/Ibm - ft) We see that the various units cancel, confirming the statement that the Reynolds number is unitless. In magnitude, Re = (62.3)(4.00)(8/12)/(2.02 10™*)(32.2) 255,000References 49 [15] Seyfried, P., and P. Becker. The role of N4 in the SI: an atomic path to the kilogram. Metrologia, 31:167-172, 1994, [16] Quinn, T. J. Conclusions of the International Workshop on the Avogadro Constant and the Representation of the Silicon Mole. Metrologia, 31:275-276, 1994, [17] Taylor, B. N. Determining the Avogadro constant from electrical measurements. Metrologia, 31:181-194, 1994. [18] Taylor, B. N., and P. J. Mohr. On the redefinition of the kilogram. Metrologia, 36:64— 65, 1999. [19] Clemence, G. M. Time and its measurement. Am. Scientist, 40(2):260, April 1952. [20] Sullivan, D. B., et al. Primary atomic frequency standards at NIST. J. Res. Natl. Inst. Stand. Technol., 106(1):47-63, 2001. [21] http//tf.nist gov/cesium/fountain.htm. This National Institute of Standards and Tech- nology Web page provides clear description of the NIST-F1 fountain clock, including. animation. [22] Bejan, A. Advanced Engineering Thermodynamics. New York: John Wiley, 1988. [23] Mangum, B. W, and G. T. Furukawa. Guidelines for realizing the International Tem- erature Scale of 1990 (ITS-90). Gaithersburg, Ma.: National Institute of Standards and Technology, Technical Note 1965, August 1990. [24] Preston-Thomas, H. The International Temperature Scale of 1990 (ITS-90). Metrolo- ‘gia, 27:3-10, 1990 (with corrections in Meirologia, 27:107. 1990). [25] Rusby, R. L., etal. Thermodynamic basis of the ITS-90. Metrologia, 28:9-18, 1991. [26] Driscoll, R.L.,andR. D. Cutkoskey. Measurement of current with the National Bureau of Standards current balance. Natl. Bur. Stand. J. Res., 60, April 1958. [27] Taylor, B. N. New measurements standards for 1990. Phys: Today, 23-26, August 1989, [28] Petley, B. W. Electrical units—the last thirty years. Metrologia, 32:495-S02, 1994/95. [29] Cook, A. H. The absolute determination of the acceleration due to gravity. Metrologia, 1(3):84, 1965.Section 3.2 Common Types of Error 53 Distribution of combined bias ‘and precision ‘error Total error Frequency of occurence | Precision anor A, Awe Measured value, x, (@) Total error i 5 Bias error 8 i z T— Precision error E A, %rue %n Measured value, Xp (b) FIGURE 3.1: Bias and precision errors: (a) bias error larger than the typical precision error, (©) typical precision error larger than the bias error.Section 3.2 Common Typesot Error 57 3 ® o * Number of observations ° 2 o low, ri | ae 299.7900 299.8000 Means 6 wenat thousands of kilometers per second FIGURE 3.4: Measured values of the speed of light, 1947-1967 (Data from Froome and Essen [2]). 3.2.2 Terms Used in Rating Instrument Performance The following terms are often employed to describe the quality of an instrument's readings. ‘They are related to the expected errors of the instrument. # Accuracy. The difference between the measured and true values. Typically, a manu- facturer will specify a maximum error as the accuracy; manufacturers often neglect to Teport the odds that an error will not exceed this maximum value. # Precision. The difference between the instrument's reported values during repeated ‘measurements of the same quantity. Typically, this value is determined by statistical analysis of repeated measurements. « Resolution, The smallest increment of change in the measured value that can be determined from the instrument's readout scale. The resolution is often on the same order as the precision; sometimes it is smaller. © Sensiti ‘The change of an instrument or transducer’s output per vnit change in the measured quantity. A more sensitive instrument’seading changes significantly in response to smaller changes in the measured quantity. Typically, an instrument with higher sensitivity will also have finer resolution, better precision, and higher accuracy. Reading error refers to erorintroduced when reading a number from the display scale of an instrument. This type of crror may sometimes be a bias error caused by truncation or rounding of the actual value to one within the resolution of the display. Reading error will also include error from inadequate instrument sensitivity if the instrument does not respond to the smallest fluctuations of the measured quantity. For example, a digital display may truncate an actual value of 10.4 toa displayed value of 10. The reading error of the digital3.4.2 Section 3.4 _ Estimation of Precision Uncertainty 61 2. Afinite number of items,n. is randomly drawn from whatis assumed to be a population of indefinite size. The properties of the assumed population are inferred from the sample. An important qualification underlies this discussion: The sample must be randomly selected from the population. If we select only the largest marbles from the bag, our sample will not accurately represent the whole population of marbles. Probability Distributions Probability is an expression of the likelihood of a particular event taking place, measured with reference to all possible events. Specifically, suppose that one of n equally likely cases, will occur and that m of these cases correspond to an event A. The probability that event A will occur is m/n. A penny is tossed. The total number of possible outcomes is two—heads and tails. If we choose heads (or tails) as event A, then the probability of A is | in 2, or 50%. A slightly more complex example is that of throwing a pair of dice. One possible outcome yields a sum of 2, six outcomes yield a sum of 7, and the remaining outcomes are as distributed in Fig. 3.6. The bar chart used here is termed a histogram. If we divide the ordinate of the chart by the total number of possible outcomes (36), we obtain a graph of the probability distribution. For example, the probability of rolling a 7 is 6 in 36, or 16.6%. Other distributions that will be considered in the following sections are these: 1. The Gaussian, or normal, probability distribution. When examining experimental data, this distribution is undoubtedly the first that is considered. The Gaussian dis- tribution describes the population of possible errors in a measurement when many independent sources of error contribute simultaneously to the total precision error in measurement. These sources of error must be unrelated, random, and of roughly the Number! possibilities yn os a o cy 23 4 5 6 7 8 § 10 11 12 ‘Sum of epots for a pair of tossed dice FIGURE 3.6: Distribution of results for a pair of thrown dice.Section 3.5 Theory Based on the Population 65 Probability (area) listed in table FIGURE 3.8: Standard normal distribution curve. Note that @ and b are inflection points. which is the standard curve shown at the top of Table 3.2 and in Fig. 3.8. The table lists the areas under the'curve between 0 and various values of z. Since the curve is symmetric about zero, the tabulation lists values for only half the curve. Bear in mind that the total area beneath the curve is equal to unity. This tabulation is sometimes called the z-distribution. ‘The following examples illustrate the nature and use of the tabulated data. EXAMPLE 3.2 (a) What is the area under the curve between z = —1.43 and (b) What is the significance of this area? 43? Solution (a) From Table 3.2, read 0.4236. This represents half the area sought. Therefore, the total area is 2 x 0.4236 = 0.8472, (b) The significance is that for data following the normal distribution, 84.72% of the population lies within the range -1.43 “tow (b) For given v Ho 1-a “te $ fc Kt at 0 Confidence interval “hey © bas FIGURE 3.11: Confidence intervals for the t-statistic: (a) one-sided, right, (b) one-sided, left, (c) two-sided. With a confidence of c% = (1 — a), the value of 1 lies in the unshaded interval.Section 3.6 Theory Based on the Sample 77 For the application of hypothesis testing, we must perform the following steps: 1, Define both the null and alternate hypothesis. 2. Define 2 level of confidence, c%. 3. Calculate the value of fexp from the data. 4, From Table 3.5 determine the proper value of i,» using the degrees of freedom v. 5. If texp falls in the reject Ho region, we reject Ho and accept the alternate hypothesis, Ha. 6. If fexp falls in the do not reject Ho region, we conclude that we do not have suffi- cient data to reject Ho at the level of confidence specified. (The strongest statistical statement occurs when we can reject Ho and accept Hy.) EXAMPLE 3.7 Using the data of Example 3.6, determine if the sample of 14 of the 1-oz weights comes from 2 population of weights whose true mean weight is greater than 1.00 o7, assuming a confidence level of 99%. Solution Ho: 2 = 1.00 07 Hai > 1.0002 fey = ss 0.04178 /-/14 v= (4-1) =13 fav = 10,01,13 = 2-650 Since this is a right tailed test and feyp falls in the do not reject Ho region, we conclude that the 99% confidence level that the population mean was not significantly different than 1.00 oz. 3.6.5 Hypothesis Testing for a Single Mean for a Large Sample Size (n > 30) Fora large sample size, the six steps listed in Section 3.6.4 are followed exactly except that fa,» is replaced by Zq and fexp is replaced by E=Ho Sa (3.28) Figure 3.12 can néw be interpreted as the similar criteria for rejecting or not rejecting the null hypothesis when the fa,» limits are replaced by za.Section 3.7 TABLE 3.6: x?-Distribution (values x2) The Chi-Square (x*) Distribution 81 a 0 Ge » | xm Moy Xs _M9s Hos _Xdoos _Hor_Xws |? 1] 0.000 0.000 0.001 0004 3.861 5.024 6.635. 7.879) 7 2} 0.010 0.020 0051 0103 5.991 7378 9.210 10.597 | 2 3] 0.072 O15 0.216 0.352 7.815 9.348 11.345 12.838 i 3 4 0.207 0.297 0.484 OTE 9.488 11.143 13.277 14.860 4 S| O412 0.554 0.831 1.145 11.070 12.832 15.086 16.750, | g 6) 0.676 0.872 1.237 1.635 12.592 14449 16.812 18.548) 6 7 | 0.989 1.239 1.690 2.167 14.067 16.013 18.475 20.278 | 7 8 1344 1.646 2.180 2.733 15.507 17.535 20.090 21.955 | 8 9 1.735 2.088 2.700 3325 16.919 19.023 21.666 23.589 g 10| 2.156 2558 3.247 3940 18307 20.483 23.209 25.188 | 10 11| 2.603 3.053 3.816 4575 19.675 21.920 24.725 26.757 | 11 12 | 3.074 3.571 4.404 = 5.226 21.026 23.337 26.217 28.300 | 12 13} 3565 4.107 $009 5.892 22.362 24.736 27.688 29.819 | 13 14| 4075 $660 5029 G57I 23.685 26119 21M 31319 | 14 15} 4.601 5.229 6.262 7.261 24.996 27.488 30.578 32.801 | 15 16 | 5.142 5.812 6.908 7.962 26.296 28.845 32.000 34.267 | 16 17| 5.697 6.408 7.564 8.672 27.587 30.191 33.409 35.718 | 17, | 6265 7.015 821 9380 31.526 34.805 37.158 | 18 19| 6844 7.63 8.907 10117 30.144 36191 38.582 | 19 20) 7434 8.260 8.591 10.851 31.410 37.566 39.997 | 20 21 | 8034 8.897 10.283 11591 2671 35479 38.932 41.401 2 22| 8.643 9542 10.982 12338 33924 36.781 40.289 42.796 22 23} 9.260 10.196 11,689 13,091 172, 38.076 41.638 44.181 | 23 24\ 9886 10¥Se 12401 13848 41s 0.361 42.980 4S's58 | 24 25 | 10.520 11.524 13.120 14.611 87.652 401646 44.314 46.928 | 25. 26 | 11.160 12.198 13.844 15.379 38.885 41.923 45.642 48.290 | 26 27 | 11.808 12.879 14.573 16.151 40.113 43.194 46.963 49.645 | 27 28 | 12461 13565 15.308 16928 41.337 44461 48.278 50.993 | 28 29) 13.121 14.256 16.047 17.708 42.557 45.722 49.588 52.336 | 29 30 | 13.787 18.953 16.791 _ 18493 33.773 46.979 50.892 30Section 3.7 The Chi Square (x?) Distribution 85 oor Of 1 510 30 5 70 90 98 998 9999 Cumulative % data items FIGURE 3.15: Normal probability plot of data listed in Table 3.7. ‘Skewed (nonequal Symmetrical but with too | distribution about great a proportion of ‘the mear) Points distributed: near the mean ‘Symmetrical but with too {Feat a proportion of Ponts distrouted at a distance from the mean AL 1 1 1% 50% 99% FIGURE3.16: Graphical effects of data skew and offset as displayed ona normal probability plot.86 Chapter3 Assessing and Presenting Experimental Data ‘TABLE 3.8: Pressure Data of Table 3.3 Arranged for a x?-Test Observed Expected Preseure, Frequency, Frequency, in MPa 01 Ei 3.965 1 0.80 3.975 1 4.14 3.985 12 12.57 3.995 25 20.06 4.005 33 3147 4.015 7 19.63 4.025 6 8.54 4.035 2 2.27 4.045 2. Frequency values for Q; in cach bin should be equal to or greater than to unity. There should be no unoccupied bins. 3. The use of x? is usually questioned if 20% of the values in either the O; or the E; cells or bins have counts less than 5. Often the cells or bins can be combined to eliminate the problem. For the data in Table 3.7 the “observed values,” Oj, are those in column B. To determine the “expected values,” E;, we must define a Gaussian distribution having a mean pressure, p, of 4.008 MPa anda standard deviation, 5), of 0.014 MPa, which are the mean pressure and the standard deviation of the data, Now using the values of p, Sp and n with Table 3,2 we can compute E; shown in Table 3.8. Finally, because several of the 8 bins have fewer than 5 members, we combine the first two and the last two before computing xZ,p, So that k = 6. Applying Eq. (3.30) and assuming a = 0.05 and v =k —3 = 3, 5 2 2 2 _ Ss (- By? _ @-4.94) ten = 494 (12 — 12.57)? (8 — 10.81)? : fp SEE 10.81 =415 2 2 Haw = Xo05,3 = 7-815 Since Zép < Xav+ With reference to Fig. 3.14, we conclude that the pressure data are normally distributed at the 95% confidence level,Section 38 Statistical Analysis by computer 87 Of special importance is the fact that v = k — 3. This is because the number of bins, ‘the mean, and the standard deviation were chosen from the experimental data. 3.8 STATISTICAL ANALYSIS BY COMPUTER Statistical analysis can often involve very large sets of data or require the application of a broad range of statistical tests. Consequently, a number of statistical software packages have been developed to assist in data analysis. Commercially marketed versions include MINITAB? and SPSS." Packages such as these are available for use on machines ranging from mainframes to personal computers. The statistical methods of the preceding sections were developed in the last century to reduce the laborious calculations that would otherwise be required in drawing statistical conclusions from samples. These methods are possible largely because the population's probability distribution has been assumed known. However, the digital computer makes detailed statistical computations easier. As a result, some current statistical research is directed toward using computer methods to relax the assumptions associated with classi- cal statistics. For example, can we determine small-sample confidence assuming that the population is Gaussian distributed? tervals without 3.9 BIAS AND SINGLE-SAMPLE UNCERTAINTY Precision error in repeat-sampled data reveals its own distribution, enabling us to bound its, magnitude using statistical methods. Bias error, by virtue of its systematic nature, provides no direct evidence of either its magnitude or its presence. The only direct method for uncovering bias error in a measurement is by comparison with measurements made using a separate, and presumably more accurate, apparatus. Unfortunately, a second set of apparatus is seldom used owing to cost and time constraints. Instead, we rely on knowledge of our own equipment to make estimates for the likely sizes of bias errors. Estimation of bias uncertainty relies heavily on experience and on an understanding of calibration accuracy and dimensional tolerances. Even with such experience and understanding, unexpected sources of bias error can be overlooked. Diligence, persistence, and careful examination of one’s results are essential in identifying and eliminating such errors. Estimates of bias uncertainty shouldbe a ied by odds ora confidence level [6]. Unlike statistical confidence levels, odds for bias uncertainty cannot be rigorously deter- mined. The level of confidence assigned is a product of our knowledge of the system, reflecting our assessment of the fraction of bias errors likely to land within the uncertainty interval. Sometimes the term coverage is used in place of confidence to reflect the empirical nature of these estimates in contrast to those derived using statistical methods. We have previously discussed general sources of bias error. Let's look at a few in more detail, Data reduction often requires knowledge of physical properties, system dimensions, or electrical characteristics. For example, a flowmeter measurement may depend on the density of water and a tube diameter; an amplifier gain may depend on the value of a resistor. Differences between the assumed values of these components and their actual values can systematically shift all data taken, creating a bias error. SMINITAB is registered to Minitab, Inc., 3081 Enterprise Drive, State College, PA, 16801. SPSS is « registered trademark of SPSS, Inc., 444 N. Michigan Avenue, Chicago, IL, 606188 Chapter3 Assessing and Presenting Experimental Data Reference standard Proper matchin soa tea | tet (may involve a number of ean —p| System being posing feaciut identifiaolo uncertanties) | [ommonly involves calibrated [to be compared with ancilary apparatus) the *known* input FIGURE 3.17: Block diagram showing calibration procedure. To find water's density, the temperature is specified and a tabulated handbook value is taken, giving, for example, 998 kg/m’. Inaccuracy or ambiguity in the specified temperature may cause a bias uncertainty in density of 2 kg/m?, for instance. This uncertainty may cover 95% of the temperature range we expect is possible for the system. ‘This inaccuracy will systematically affect all data reduced using the particular Value of density. Similarly, the nominal diameter of 2 pipe may differ from the production diameter by percent or so; and a manufacturer's rated value for a resistor can vary substantially from the resistor’s actual resistance. For carbon resistors, the manufacturing tolerance may be +5% or 10%; for higher-quality, metal-film resistors, the tolerance may be +0.01%. Typically, these tolerances might represent 95% coverage—that is, the variation of 95% of all resistors. Potential sources of error suchas these remain unchanged for each measurement made with the system. Ifthe uncertainty relatedto amanufacturing tolerance is unacceptably large, taking our own measurement of the specific part can usually reduce the uncertainty substantially. The uncertainty in a resistor's value can be reduced to the accuracy of the ohmmeter measuring it; or a pipe’s diameter can be measured to the accuracy of a set of calipers. Physical property data can also be measured, if need be, although it is more common to trust the carefully determined handbook values. Calibration uncertainty is another very common source of bias uncertainty, Cali- bration requires a reference or standard against which system response can be compared, The reference may be fixed or one-valued, such as the triple point of water or the other triple points and melting points used to define the practical temperature scale (Section 2.7). Alternatively, the standard may be capable of supplying a range of inputs comparable to the ange of the system, as do various commercially available voltage references. Naturally, the uncertainty of the standard should be considerably less than thet of the system being calibrated. A rule of thumb is that the uncertainty of the standard should be no more than. one-tenth that of the system being calibrated. Figure 3.17 shows a typical calibration arrangement. Normally, the indicated read- out is compared to the reference standard and a relation between the two is determined. Sometimes the readout scale can be adjusted until agreement with the standard is obtained; sometimes a line fit is used to relate the readout to the standard’s value. In either case, additional uncertainty appears in the comparison and adjustment process. Instrument manufacturers often supply calibration data with their products, which can assist in estimating the uncertainty of the instrument. For example, a particular position transducer is rated at 0.8 V output per millimeter of sensor displacement. The manufacturer has not specified the calibration uncertainty directly. However, we might assume that the uncertainty is roughly 0.05 Vimm, since this is the apparent resolution of the calibration. The coverage is also unknown, but our experience using the device may suggest that 90 toSection 3.10 Propagation of Uncertainty 89 95% coverage is a reasonable assumption. If necessary, we could reduce the uncertainty by conducting our own calibration. Examples of estimating bias uncertainty are given in Section 3.11. 3.9.1 . Single-Sample Precision Uncertainty When only one, two, or three repeat observations are made, the confidence intervals calcu- lated statistically can be quite large. In that circumstance, you may determine a narrower range for the mean value by treating precision errors like bias errors and estimating a standard deviation based on your knowledge of the instruments. For example, random variations in test conditions may cause a digital multimeter (DMM) reading to fluctuate; but if the reading is made only once, the random variation simply produces an overall range of uncertainty for the true value of this variable. The uncertainty (at 19 : 1 odds) is twice the standard deviation of the test condition. In other words, +1.96¢ ~ +20 will cover 95% (or 19 out of 20) of the readings made. If, on the other hand, the same measurement is made several times, the random variations can be averaged vut, and statistics can be used to place a narrower bound on the mean value. We can construct the single-sample estimate a little more formally. We begin by estimating o asa value (op, say) that is based on our knowledge of the experimental system. Thus, we are assuming that o of the population is known, and, with Eq. (3.18), the precision uncertainty can be estimated as Py une (3.31) With a single reading, n Taking a 95% confidence level, zo.95/ . s0 that no averaging is performed to reduce the uncertainty. 1.96 and P, © 1.960, (95%) (3.32) ‘The potential precision error underlying a single measurement of a random variable can usually be estimated from your knowledge of how finely an instrument will resolve, of how precisely an instrument may repeat a reading, or of how much the test condition fluctuates. Often, these estimates can be made in advance of performing the experiment, in order to gauge the expected uncertainty in the result. Section 3.11 includes an example of single-sample uncertainty analysis. 3.10 PROPAGATION OF UNCERTAINTY Often several quantities are measured, and the results of those measurements are used to calculate a desired result. For example, experimental values of density are usually determined by dividing the measured mass of a sample by the measured volume of that sample. Each measurement includes some uncertainty, and these uncertainties will create an uncertainty in the calculated result. What is that uncertainty? Finding the uncertainty in a result due to uncertainties in the independent variables i called finding the propagation of uncertainty. For uncertaintiesin the independent variables, the procedure rests on a statistical theorem that is exact for a linear function y of several independent variables x; with standard deviations g;; the theorem states that the standardSection 311 Examples of Uncertainty Analysis 93 The confidence is not given, but we shall assume it to be 95%. The full-scale reading of the meter is 20 k&, and after evaluating the terms and summing, the meter’s bias uncertainty can be estimated as By = £96.80 = 40.10k2 (95%) Notice that the reading error in the DMM scale is only 0.005 k®, which is much lower than the actual uncertainty in the DMM reading! This DMM has relatively high resolution and precision but much lower accuracy. The total uncertainty in the mean of the population is, from Eq. (3.4), 2 Un= (8% + 2)” . 31a = [10.107 + (0.7077) " ke =0.71kQ (95%) The uncertainty of the nominal value is Ux = 0.71 k@ (95%), or about 4%, The precision uncertainty in the mean is the major source of uncertainty. On the other han¢, if'a sample of 1000 resistors were used, the precision uncertainty would be reduced by a factor of ten (why?), and the bias uncertainty would be dominant. The tolerance of the resistors remains o be found. What we'd like is an estimate of the percentage deviation from the nominal value which includes, say, 95% of the resistors. One approach is to note that 95% of a Gaussian population lies within + 1.96c of the population mean jz (see Table 3.1). On that basis, we could approximate og © Sq and up © R, $0 that 1.96oR _ 1.965 Tolerance % = OR ~ Rk HR R that is, a tolerance of 11% (or about 10%. since that’s the nearest production tolerance). Ina manufacturing situation, engineers are usually more interested in estimating an interval that is c% certain to contain at least some percentage b of the population. For example, the manufacturer might wish to report, with c = 95% confidence, that b = 95% of resistors will have resistances within some specific range of resistances. As it turns out, the approach used in the preceding paragraph is a very poor way to estimate such tolerance limits, because it ignores the inaccuracy of Sg and R as estimates of the population's og and ite. Although 95% of a Gaussian population lies in the interval jre + 1.960, that is not truc of the interval & 4 1.96Sx. For example, our estimate of the mean has a 4% precision uncertainty; this means that the interval likely to contain 95% of the population should be broadened by something like an additional 44% of R beyond R + 1.96Sg. A proper estimate of tolerance must allow for this uncertainty as well as that in ¢g. More advanced statistical methods (3] show that, at a confidence of 95%, the 95% tolerance interval or the population is almost nvice as large as that estimated previously (i.e., the interval that is 95% certain to contain at least 4x + 1.960 turns out to be R + 3.532Se). After this extended discussion, it may interest you to learn that the resistors actually tested were nominally 18 k® with a tolerance of 10%.Section 3.11 Examples of Uncertainty Analysis 97 Likewise, the standard deviation of K is 12 ox _ (om)? y (2%)? 4 (2)? (Lee)* (i222)" #-(( +5) +(%) +(5% ADAP i 2 4 re = [om saxotears (3 0.002) + (; x 1) ] = 112% Our estimate for the single-sample precision uncertainty in K is, from Eq. (3.32). Pr _ 1.96 0K K Front Eq. (3.4), the total uncertainty in K is (ey (QT : Inspection of these results quickly reveals the parameters having the greatest con- tribution to the uncertainty. Improvement of the timing and weighing procedures would improve the results the most. Improvement of the pressure and diameter measurements would contribute significantly less improvement. Most of the total uncertainty is caused by precision uncertainty. We can reduce that uncertainty considerably by repeating the calibration experiment several times and averaging the results. Since Px will decrease as /n, taking n = 4 experiments will reduce Pr toabout 1.0%. Note that the density contributes almost nothing to the total uncertainty. Even if we don’t bother to measure the temperature, the contribution of density uncertainty remains = 1,96(0.0112) = 2.20% (95%) = [o 0108)" + (0.0220) = 245% (95%) negligible; that is, ( } 02)" (1)?, We conclude that careful temperature measurement, in this case, would be a waste of offort. Are the various uncertainty estimates simply good guesses? To a degree, they are. ‘but dismissing them as nothing but guesses would be flippant. The specific considerations leading to each estimate were not arbitrary; when properly made, such “guesses” have a strong foundation in the actual performance of the equipment and the method of taking the data, Even theestimated confidence percentages (usually 95%) are a quantitative assessment of our expectation for the variability of the data, although they are essentially just educited guesses. But if we admit to guessing, why not simply guess the overall uncertainty and skip all the intermediate steps? In answer, the detailed analysis provides a means for evaluating the relative effect of each identifiable source of error, thereby separating the more important ‘ones from the less important ones. Furthermore, one can evaluate the uncertainties of ench of the individual variables with considerably more assurance than one could judge the total.3.13.2 1 9. 10. 11. 12. 13. Section 3.13 Graphical Presentation of Data 101 along each axis, Avoid confusing labels such as “Pressure, Pa x 10°,” use units such as MPa instead. When plotting on semilog or full-log coordinates, use real logarithmic axes; do not plot the logarithm itself (e.g., plot 50, not 1.70).!? Logarithmic scales should have tick marks at powers of 10 and intermediate values, such as 10,20, 50, 100, 200, .... ‘The axes should usually include zero; if you wish to focus on a smaller range of data, include zero and break the axis, as shown in Fig. 3.4. ‘The choice of scales and proportions should be commensurate with the relative impor- tance of the variations shown in the results. If variations by increments of 10 are significant, the graph should not be scaled to emphasize variations by increments of 1. Use symbols such as ©, Ol, A, and © for data points. Do not use dots (.) for data. Open symbols should be used before filled symbols. You may place a legend defining symbols on the graph (if space permits) or in the figure caption. Place error bars on data points to indicate the estimated uncertainty of the measurement or else use symbols that are the same size as the range of uncertainty. ‘When several curves are plotted on one graph, different lines (solid, dashed, dash- dot, ...) should be used for each if the curves are closely spaced. The graph should include labels o 4 legend identifying each curve. Avoid using colors to differentiate curves, since colors are usually lost when the graph is photocopied. Theoretical curves should be plotted as lines, without showing calculated points. Curves fitted to data do not need to pass through every measurement like a dot-to-dot cartoon; however, if a data point lies far from the fitted curve, a discrepancy may be indicated {as for the first and the last three points in Fig. 3.20(b)]. Lettering on the graph should be held to the minimum necessary for clarity, Too much text (or too much data) creates crowding and confusion. Labels on the axes and curves should be oriented to be read from the bottom or from the right. Avoid forcing the reader to rotate the figure in order to read it. ‘The graph should have a descriptive but concise title. The title should appear as a caption to the figure rather than on the graph itself. Good graphing software can help produce graphs that adhere to these guidelines. However, some graphing packages violate even the simplest of these rules. Discretion is advised! Choosing Coordinates and Producing Straight Lines ‘The first step in making any graph is to decide which variables to plot and on what scale to plot them. Four basic graphical scales occur frequently in engineering work (Fig. 3.21). Linear coordinates have a linear variation of both the x- and y-scales. If a variable changes by several orders of magnitude or is exponentially related to another variable, then a loga- rithmically scaled axis may be preferable. Graphs having one logarithmically scaled axis and one linearly scaled axis are called semilogarithmic (ot semilog). Those for which both axes are logarithmic are called full logarithmic (full log, ot fog-log). When a quantity varies, with an angle, polar coordinates provide a physically suggestive format for the data '2an exception is made when the unit decibels is plotted.Section 3.13 Graphical Presentation of Data 105 0 40 80 120 160 200 240 Time, (8) ) FIGURE 3.22: Cooling data. (b) Semilogarithmic coordinates, (Note logarithmic variation of AT/ATo.) where AT is the measured temperature difference at any time, AZo is the temperature difference before the slug is immersed, and r is atime constant for the cooling. You may desire to find an experimental value for r. so that you can use Eq, (3.40) to estimate AT at values of ¢ where you have no measurements, That task is not straightforward using the linear sealing of Fig. 3.22(a). Instead, you could plot logiAT/ATy) as a function of ¢. Then the relationship between AT and ris igg aT 0g ( = which is the equation of a line with slope —0.4343/r and intercept zero.'> The graph is most easily made using semilogarithmic coordinates (Figure 3.22(b)], and r can be calculated from the slope of the line: ed Gan ~oayaae = ORLATUn/ATaL log [AT (2)/ AT] 1-2 (3.42) resulting in t = 98s. Note that while AT/Ap is plotted on the logarithmic coordinates, log(AT/ATo) must be used in calculating the slope, e 10 logarithmsare standard in graphical work: logyg ¢ = 0.4343.Section 3.14 Line Fittirig and the Method of Least Squares 109 data [3]. The method identifies the slope b and intercept a that minimize the sum of the squared deviations of the data from the fitted line, S?: Di - veo? (3.45) a Here, for the various measured values of xj, yj is the experimentally determined ordinate and y(x;) = a + bx; is the corresponding valic calculated from the fitted line; n is the number of experimental observations used. The result is LaDy (oxy? = rudy (Ex) Most scientific calculators incorporate programs for calculating least squares lines, a great convenience to the experimentalist. Consequently, least squares has become increas- ingly popular as the method for fitting lines. But to some degree this predominance has also promoted misuse of the method. The least squares method addresses only the precision error in yj; poor results are obtained if x; also includes large precision error. Least squares assumes, in effect, that the experimental x; are error free. To indicate the reliability of the fit, most pocket calculators and software packages Tetum the correlation coeficient, r, along with the least squares results: (3.46) explained squared variation about ym ————— in 3.47) total squared variation about yn where ym, is the mean of the measured yi: (3.48) The explained squared variation results from the straight-line change of y with x: YG - yn? iat The total squared variation also includes the precision error: = Im) = =F + Ly) — nl? fal Thus, ese = LL = yal (ou) © S40 iy — ym?Section 3.15 Summary 113 3.15 SUMMARY Every measurement includes some level of error, and this error can never be known exactly. However, a probable bound on the error can usually be estimated. This bound is called uncertainty. Uncertainty should always be accompanied by the odds (or confidence per- centage) that a particular error will fall within this bound, When presenting data either graphically or numerically, the uncertainty should also be shown. 1 Errors can usually be classified as either bias error or precision error. Bias (or system- atic) errors occur same way for each measurement made. Precision (or random) errors vary in size and sign with a zero average value. Bias uncertainty must be estimated from our knowledge of the measuring equipment or by comparison to other, more accurate systems, Precision uncertainty can be estimated statistically (Sections 3.1, 32). ‘The total uncertainty in a measurement includes both bias and precision error: = (8 + oir ‘The bias and precision uncertainty should have the same confidence level, typically 95% (Sections 3.3, 3.11). Random variables, such as precision ertor, may be characterized in terms of a pop- ulation and its distribution. The most common distribution for precision error is the Gaussian or normal distribution (Sections 3.4, 3.5). Properties of a population are estimated by taking a sample fromit. The sample mean, , and sample standard deviation, S,, are used to estimate the population mean and population standard deviation. The precision uncertainty in z is Se Prstaav 7 (%) with « = 1 — cand v =n — | (Sections 3.6, 3.11). ‘The x?-distribution may be used to compare the distribution of a sample to anexpected distribution. The accuracy with which a set of data fit the Gaussian distribution can be checked using either the x? “goodness-of-fit” test or a normal probability plot (Section 3.7). Bias uncertainty and single-sample precision uncertainty are estimated from our knowledge of the measuring system (Sections 3.9, 3.11). ‘When experimental data are used to compute a final result, the uncertainty of the data must be propagated to determine the uncertainty in the result: ay \ ay ox Ox,” Here u; is either a bias uncertainty, B;, or a precision uncertainty, P; (Sections 3.10, 3). ‘The accuracy of experiments can be improved before they are conducted by identifying and eliminating major sources of uncertainty (Sections 3.11.2, 3.12). uy3.18, 3.19. 3.20, 3.21. 3.22. Problems 117 Figure 3.9 shows a histogram based on the values listed in Table 3.3. As suggested in Section 3.6.1, prepare histograms representing the data, based on (a) seven bins, (b) eight bins, and (c) ten bins. ‘The manufacturer of inexpensive outdoor thermometers checks « sample of ten against a 68°F standard. The following results were obtained: 68.5 67.5 67 69 68 67 67.5 69 69. Using Student’s ¢-test, calculate the range within which the population mean may be expected to lie with a confidence level of 95%. Spacer blocks are manufactured in quantity to a nominal dimension of 125 mm, A sample of 12 blocks was selected and the following measurements were made. 128 132 1.29 1.23 126 126 1.20 1.29 124 123° 1.26 1.22 Using Student's t-test, determine the upper and lower tolerance values within which the population mean may be expected to fall with a significance level of 10%. In alaboratory itis suspected that the results from two different viscometers do not agree. Ten fluid samples were tested using apparatus A and corresponding samples were tested using apparatus B. The results are as follows: Viscosity (Dimensionless) Using Using Sample No. Apparatus Apparatus B 1 2 B 2 B 45 3 54 56 4 75 5 5 50 3 6 48 50 7 B 2 8 55 54 9 48 48 10 50 2 Determine whether there is a significant difference in the two systems at the 99% confi- dence level. Consider the equation y= 1.0— 0.24 001x7 o x=3) Determine the maximum uncertainty in y for 2% uncertainty in the variable x.Problems 121 (a) Make a linear plot of the data points. (b) Determine the constants for a linear least squares fit of the entire data set. Plot the fitted line on your graph of the data. (e) Because the resistivity is not a perfectly linear function of temperature, a more accurate fitcan be obtaified by limiting the range of temperature considered. Obtain the constraints for a linear least fit over the range of °C to 100°C only. Plot the result on your graph. 3.36, Show that y = a + 6x" will plot as a straight line on linear graph paper when y is plotted ‘as the ordinate and x” is plotted as the abscissa, Show that the intercept is equal to a and the slope is equal to b. 3.37. Show that if 1/y versus 1/x is plotted on linear paper, the functjon x/(ax + 2) (which may also be written 1/y =a + /x) will yield a straight line, with a as the intercept and bas the slope. 3.38. Show that y = ac’* will plot as a straight line on linear paper when log y is plotted as, the ordinate and x is plotted as the abscissa and that the intercept is equal to loga and the slope is equal to bloge. Note that with the slope known, b and c may be found by simultaneous solution of the slope equation and the original equation written for a selected (Gi, ¥i) point. 3.39, Select a range for x and make an x versus y plot of y = 12x*/3 on linear graph paper. ‘Now transform the data to log y and log x and plot on linear paper. The second set of data should plot as a straight line with a slope of 3 and an intercept of log 12. 3.40, From 1960 to 1983, the standard meter was defined as 1,650,763.73 wavelengths of the light emitted during the transition between the 2pip and the Sds levels of the krypton-86 atom. That emission line is stightly asymmetric, and its wavelength has a total uncertainty of about +2.4 x 1075 A (68%). By the early 1970s, laser technology permitted highly-precise determination of the speed of light, c, byusing the relation c = Af and measured values of laser wavelength, 4, and frequency, f. Laser frequency measurements had at that time reached relative uncertainties of uy/f = 6 x 107! (68%). (a) What was the uncertainty (95%) in the measured speed of light at that time? What factor limited the accuracy of this measurement? (b) In 1983, the meter was redefined as “the distance traveled by lightin vacuum during 1 time interval of 1/299792458 of a second.” How did this affect the uncertainty in the speed of light (in meters per second)? REFERENCES [1] ANSI/ASME 19.1-1985. ASME Performance Test Codes. Supplement on Instruments and Apparatus, Part 1, Measurement Uncertainty. New York, 1985. [2] Froome, K. D., and L. Essen. The Velocity of Light and Radio Waves, New York: Academic Press, 1969. [3} Miller, I. R., J. E. Freund, and R. Johnson. Probability and Statistics for Engineers. 4th ed. Englewood Cliffs, N.J.: Prentice Hall, 1990.Section 43 Circular and Cyclic Frequency 125 4.3. CIRCULAR AND CYCLIC FREQUENCY The frequency with which a process repeats itself is called cyclic frequency, f,and is typically measured in cycles per second, or hertz: 1 Hz = 1 cycle/s. However, the idea of circular frequency, «, is also useful in studying cyclic relations. Circular frequency has units of radians per second (radis). The connection between tie two frequencies is conveniently illustrated by the well-known Scotch-yoke mechanism. Figure 4.1(a) shows the elements of the Scotch yoke, consisting of a crank, OA, with aslider block driving the yoke-piston combination. If we measure the piston displacement from its midstroke position, the displacement amplitude will be ++0A. If the crank turns at @ radians per second, then the crank angle 6 may be written as wt. This, of course, is convenient because it introduces time 1 into the relationship, which is not directly apparent in the term 6. Piston displacement may now be written as = sosinot which is the same as Eq, (4.1). One cycle takes place when the crank turns through 27° rad, and, if f is the frequency in hertz, then w = 2nf 44) ‘Thus the displacement may instead be expressed in terms of cyclic frequency: 5 = so sin2aft (4.5) ale (@) ” FIGURE 4.1: (a) The Scotch-yoke mechanism provides a simple harmonic motion to the piston; (b) a spring-mass system that moves with simple harmonic motion.Section 4.4 Complex Relations 129 Yo =10sin wt +5 sin Seat Yro= 108i wt + 5 in Boot FIGURE 4.4: Examples of two-component waveforms with second term of various relative frequencies.Section 4.4 ComplexRelations 133 4.4.1 Beat Frequency and Heterodyning ‘The situation shown in Fig. 4.7(a) isthe basis of an important method of frequency measure- ment. Here two waves of equal amplitude and nearly equal frequency have been added. If ‘one wave has acyclic frequency of fo and the second wave has afrequency of fi = fo+f, then the resultant wave is y= Asin@z jor) + Asin[2n(fo+ 4/1] = 2Acos (anf ): sin (Sth, ) (49 slowly beating amplitude high-frequency wave This wave undergoes slow “beats” where the amplitude rises and falls. Although the cosine term in the amplitude has cyclic frequency of A f/2, we see that the amplitude itself has two minima per cycle of the cosine. Thus, beats occur at a frequency of Af. This kind of wave addition happens when a tuning fork is used to tune a musical instrument. The tuning fork and the musical instrument produce neatly equal tones, and, when the two sound waves are heard together, a lower beat frequency is also heard. The instrament is adjusted until the beat frequency is zero, so that the instrument’s frequency is identical to that of the tuning fork, When the difference frequency, Af, is much smaller than fo, addition of waves allows us to measure Af with less uncertainty than if we measured fp and f, separately and subtracted them. This technique for frequency measirement is called heterodyning (ection 10.6). It is very important in radio applications and in laser-doppler velocity measurements (Section 15.10). EXAMPLE 4.3 Helium-neon laser light has a frequency of 473.8 THz (473.8 x 10!? Hz). A helium-neon laser beam is reflected from a moving target. This creates « doppler shift in the beam, which increases its frequency by 3 MHz (3 x 10° Hz).! The reflected beam is “added to” an unshifted beam of equal intensity, by using mirrors to bring the beams together. What is the resulting signal? Solution FromEq,(4.9), with fo = 473.8THz and fy = fo-+Af = 473.8 THz-+3 MHz, y =2A.c0s (fy) sin (x4 = ‘) = 2A.00s [2(1.5 x 10°)1] sin [20(473.8 x 10"2)1] The amplitude has a cyclic frequency of 1.5 x 10 Hz = 1.5 MHz, causing zeros in amplitude twice per cycle for a beat frequency of 3-MHz. This would be manifested A doppler shift is an apparent change in the frequency of alight or sound wave that occurs when the wave source and receiver are in motion relative to one another. One typical example is the change in pitch of a passing train's whistle,yap1o smouLey amp wasaidas Kq1sessa22u 10U S20p suonENba asaqn mt pasn se 1 1 Eo — a [rosasouas 2] & wl “banca aes 2 ore as cm + uous) ent ae @er — [msn 2 2) [22 -( s+1mes00 22 + ws00) 2 = ¢ Oer 3 ve 1 3 [ies xo =( ~ 950 22 + cue 22 — ros) er mg so0 0 so) Wa + ames BE + son) Hf @or 2+ sogus® — sous) =< (ag pees la ay € z x £4 sozus? + sous) 2 = ¢ TET mus) 3 @6r te era $ x 101 = wz)ans Ze ( + rosusS + opus’ + sous) =k (oer [oa uns x4 (+ rose $ + ws) R= wwy uopEnby eunbig, 6° 314 UF UMoUS suHOsoAKyy srpoudg [eI9adg 30} suoKIENbA “Ay aTaWL 137Section 4.7 Harmonic, or Fourier, Analysis 141 ni Discrete sample ‘Analog signal FIGURE 4.15: Discrete sampling of a continuous analog signal. The value of the signal is recorded at intervals Ar apart for a period Practical harmonic analysis usually falls into one of the following four categories: 1, The waveform y(t) is known mathematical function. In this case, the integrals (4.13) and (4.14) can be evaluated analytically. These calculations are illustrated in Appendix B for two cases. 2. The waveform y(r) is an analog signal from a transducer. In this case, the waveform may be processed with an electronic spectrum analyzer to obtain the signal's spectrum (Sections 9.10, 18.5, and 18.6). 3. Alternatively, the analog waveform may be recorded by a digital computer, as dis- cussed in Chapter 8. The computer will store y(t) only at a series of discrete points, in time. Integrals (4.13) and (4.14) are replaced by sums and evaluated, as discussed in Section 4.7.1. 4, The waveform is known graphically, for instance, from a strip-chart recorder or the screen of an oscilloscope. In this case, y(1) may be read from the graph at a discrete series of points, and the integrals may again be evaluated as sums.? 4.7.1. The Discrete Fourier Transform The case when y(¢) is known only at discrete points in time is very important in practice because of the wide use of computers and microprocessors for recording signals. Normally, ‘a computer will read and store signal input at time intervals of At (Fig. 4.15). The computer records a total of N points over the time period T = N Ar.‘ Therefore, in the computer's memory, the analog signal y(1) has been reduced to a series of points measured st times t= At, 2At,..., NAt, specifically, y(At), y(QA0),..., ¥(N Ar). Wecan write this series ‘more compactly as y(t,) by setting , =rAtforr = 1,2....,.N. ‘To perform a Fourier analysis of a discrete time signal like this, the integrals in Eqs. (4.13) and (4.14) must be replaced by approximate numerical integration in the form of summations. Likewise, the continuous time ¢ is replaced by the discrete time t, = rr, 34an example of this approach is given in (1], Section 4.7 “Assume that the point at ¢ = 0 is not recorded and that N is even,Section 4.7 Harmonic, er Fourier, Analysis 145 Actual signal Apparent ()2to tor signal | Actual and (0 4=2t apparent ‘ signals FIGURE 4.17: Effect of varying the sample rate, f,, on the apparent signal obtained by discrete sampling. (Continued on page 146) T = N At =N/f,. the frequency of this wave is fawn dom nh T Nat Ww i} | HH es (4.21) Thus the fundamental frequency of the DFT, Af, is also that of the lowest-frequeney full wave that fits within the sampling period. No lower frequency (other than f = 0) is resolved.Section 4.7 Harmonic, or Fourier, Analysis TABLE 4,2: Experimental Data for Microphone Voltage TABLE 4.3: Calculated Harmonic Coefficients for the Data of Table 4.2 Harmonic Frequency, ndf, He Order, n eeudaununnre 1.08 _Time(ms) Voltage (mV) 0.11 2 0.33 0.44 0.56 0.67 0.78 0.89 1.00 Lal 1.22 1.33 1.44 1.56 1.67 1.78 1.89 2.00 0 500 1000 1500 2000 2500 3000 3500 10 30 70 65 15 —40 -50 Is 100 135 90 —40 —130 —140 110 30 -10 0 Bry mV 0.00 46.83 52.70 47.41 2.23 0.47 -0.51 2.93 =LAl 00 Coy mV, 1.04 55.14 72.02 52.25 3.56 2.03 1.09 2.93 117 149150 Chapter4 The Analog Measurand: Time-Dependent Characteristics 80 Ga (nV) é ° 1000 2060 3000 4000 Frequency, nt (Hz) FIGURE 4.20: Frequency spectrum for data of Table 4, and ceiling may have also contributed to some distortion of the original signals. Ideally, the test should have been run in an anechoic chamber. The sum of the measured harmonic components [Eq. (4.19)] is plotted together with the data in Fig. 4.21. Visually, the computed curve appears to fit the data perfectly. The DFT calculation reconstructs the original signal to within the accuracy of the original data, Amplitude, mV 180 Time, ms FIGURE 4.21; Comparison of the calculated Fourier series with the measured test data,100 psi 4.9. Using the data files in Table 4.4, if i Problems 153 Pressure Hs) FIGURE 4.23: Pressure-time record for Problem 4.7. in milliseconds and f(t) is in volts, determine the discrete Fourier transform for each set of digital data. (a) (b) te) (d) (e) (f) Use fio. Use f2(t). Use f(t). Use fat). Use f(t). Use fo(t). 4.10. A 500-Hz sine wave is sampled at a frequency of 4096 Hz. A total of 2048 points are taken, (a) (b) (ce) ‘What is the Nyquist frequency? ‘What is the frequency resolution? The student making the measurement suspects that the sampled waveform contains several harmonics of 500 Hz. Which of these can be accurately measured? What happens to the others? 4.11. A 150-Hz cosine wave is sampled at a rate of 200 Hz. (a) (b) (©) 412, (a) (b) (e) Draw the wave and show the temporal locations at which it is measured. ‘What apparent frequency is measured? Describe the relation of the measured frequency to aliasing. Give a numerical justification for your answer. Suppose that a $00-Hz sinusoidal signal is sampled at 750 Hz. Draw the discrete time signal found and determine the apparent frequency of the signal. If a 200-Hz component were present in the signal of part (a), would it be detected? Explain If 375-Hz component were present in the signal of part (a), would it be detected? ExplainSection 5.2 Amplitude Response 157 300 104 105 107 107 Input, ¢;(V) FIGURE 5.1: Gain versus input voltage for amplifier section of a commercially available strain measuring system for a frequency of | kHz (gain = output voltage/input voltage) 5.2 AMPLITUDE RESPONSE Amplitude response is governed by the system's ability to treat all input amplitudes uni- formly. If an input of 5 units is fed into a system and an output of 25 indicator divisions is obtained, we can generally expect that an input of 10 units will result in an output of 50 divisions, Although this is the most common case, other special nonlinear responses are also occasionally required. Whatever the arrangement, whether it be linear, exponential. or some other amplitude function, discrepancy between design expectations in this respect and actual performance results in poor amplitude response. Of course no system exists that is capable of responding faithfully over an unlim- ited range of amplitudes. All systems can be overdriven. Figure 5.1 shows the amplitude response of a voltage amplifier suitable for connecting a strain-gage bridge to an oscillo- scope. The usable range of the amplifier is restricted to the horizontal portion of the curve. The plot shows that for inputs above about 0.01 V the amplifier becomes overloaded and the amplification ceases to be linear. 5.3 FREQUENCY RESPONSE Good frequency response is obtained when a system reacts to all frequency components in the same way. If 100-Hz sine wave with an input amplitude of 5 units is fed intoa system and a peak-to-peak output of 25 cm results on an oscilloscope screen, we can expect that a 500-Hz sine-wave input of the same amplitude would also result in a 24 cm peak-to-peak output. Changing the frequency of the input signal should not alter the system's output magnitude so long as the input amplitude remains unchanged. Yet here again there must be a limit to the range over which good frequency response may be expected. This is true for any dynamic system, regardless of its quality. Figure 5.2 illustrates the frequency response relations for the same voltage amplifier used in Fig. 5.1. Frequencies above about 10 kHz are attenuated. Only inputs below this frequency limit are amplified in the correct relative proportion.Section 5.6 Delay, Rise Time, and Slew Rate 161 We can make similar calculations for any waveform for which a harmonic series can be written. In particular, to investigate a measuring system's response to a waveform of interest, we can make a Fourier analysis of that waveform and investigate the system's response characteristics using sine-wave test results, as before. 5.6 DELAY, RISE TIME, AND SLEW RATE Finally, a fourth type of response, which is actually another form of frequency response, is delay, or rise time. When a stepped or relatively instantaneous input is applied to a system, the output may lag, as shown in Fig. 5.5. The time delay after the step is applied, but before proper output magnitude is reached, is known as rise time. It is a measure of the system's ability to handle wansients. Sometimes rise time is defined specifically as the time, At, required for the system to pass from 10% to 90% of its final response. Altematively, transient response may be characterized by the seitling time required for the system response to remain within some small percentage of its final value. Slew rate is the maximum rate of change that the system can handle. In electrical terms, itis de/de, or volts per unit time (e.g., 25 V/s). When the voltages changes rapidly, the system can respond no faster than the slew rate. 5.7 RESPONSE OF EXPERIMENTAL SYSTEM ELEMENTS An experimentalist can usually avoid operating a measuring system under conditions when amplitude response or slew rate are limiting factors. For example, a solid-state amplifier ‘System input Input signal Settling time Output signal e FIGURE 5.5: Résponse of a typical system to a step input, showing rise time (At) and settling time.Section 5.9 Mechanical Elements 165, of a member is concentrated at a point. Depending on the geometry of the member and its application, the point of concentration may or may not be the center of gravity. In certain cases, the center of percussion may be the location of effective concentration. 5.9.2 Spring Force Many mechanical members deflect in direct proportion to the force exerted on them, that is, AF/As = k =a constant, where AF is an applied force increment and As is the resulting deftection increment. Most coil springs, beams, and tension/compression members abide by this relationship. It may be noted that the force is opposed to the deflection; that is, the resulting force always attempts to restore equilibrium. Torsional members commonly adhere to the relationship AT/Ad = k, = a constant, where AT is an applied torque increment and A@ is the resulting torsional deflection increment, The constants k and k, are called spring constants, or deflection constants. Elasticity is not always the source of the restoring force, however. In certain cases, suchas fora beam balance (see Section 5.10), the restoring force may be supplied by gravity. ‘When the motion of a concentrated mass is constrained by an equilibrium-seeking member Fig. 5.9(a)], simple vibration theory shows that the combination will have a natural frequency on =f (62) Vim where @q = circular frequency in radians per second (see Section 4.3) = 2nf, Ff = frequency of vibration in hertz, &c = the dimensional constant (see Table 2.6) A system of this sort is said to have a single degree of freedom; that is, itis assumed to be constrained in some way to oscillate in a single mode or manner, needing only one coordinate to fully describe its motion. @ (b) FIGURE 5.9: Elementary spring-mass systems: (a) without damping; (b) with viscous damping.Section 5.12 Dynamic Characteristics of Simplified Mechanical Systems 169 compromise, We would like equilibrium to be reached as quickly as possible in order to ger on with the job. It would seem that there might be an optimum value that should be used. Although this is not exactly the case because of other factors involved, damping of the order of 60% to 75% of critical is provided in many instruments of this type (see Section 5.16.2 for further clarification of this point). Although damping will tend to decrease the frequency of oscillation, it does not change the inherent sensitivity of the device, which is related to the undamped natural frequency. However, somecompromises must still he made in regard tosensitivity. Sensitivity increases in proportion to the undamped natural period, as shown in the previous section. Because a high natural period (or low natural frequency) usually corresponds to a lessened frequency response, sensitivities greater than those required by the application should be avoided in the interest of maintaining adequate response. 5.12 DYNAMIC CHARACTERISTICS OF SIMPLIFIED MECHANICAL SYSTEMS By making certain simplifying modeling assumptions, we may place the dynamic charac- teristics of most measuring systems in one of several categories. The basic assumptions are + that any restoring element (such as a spring) is linear, that damping is viscous, and that the system may be approximated as a single-degree-of-freedom system. 5.13 SINGLE-DEGREE-OF-FREEDOM SPRING-MASS-DAMPER SYSTEMS. Figure 5.12 shows asimple single-degree-of-freedom mechanical system. Itis single degree because only one coordinate of motion is necessary to completely define the motion of the system. We will also assume a general form of excitation, F (1), which may or may not be periodic. Forces acting on the mass will result from the spring, damping, and the external force, F(¢). Using Newton’s second law, we can write ds m {ds Fits =" (FS) 68) Note that the spring force will always oppose the displacement and that the damping is pro- portional to velocity and opposite the velocity direction. This relationship can be rearranged Spring FIGURE 5.12: Mechanical model of a force-excited second-order system.174 Chapter5 The Response of Measuring Systems where Fo = the amplitude of the forcing function, 9 = the circular frequency of the forcing function in radians per second ‘The solution of Eq. (5.15) yields Folk ey ud vi+ Ga? cos(@t — 4) (5.16) where A\ = a constant whose value depends on the initial conditions, 1 = the time constant = ic 1 2S _) 2a 1 > = tan“! 17) BT (5.17) Qn T = J = the period of excitation cycle in seconds @ = the phase lag = tan ‘We see that the first term on the right side of Eq, (5.16), the complementary function, is transient and after a period of several time constants becomes very small. The second term. is the steady-state relationship and, except for the short initial period, we can write cos(Qt — $) (5.18) or _ 20s(2t = $) TF OP and Sd 1 5 TE ERE 1 7" = (6.19) Vi+Qar/T? where Sq = the maximum amplitude of the periodic dynamic displacement and HET ‘The quantity s, is the static deflection that would occur should'the force amplitude Fo be applied as a static force. The ratio s¢/s, is often called the amplification ratio. For analogous situations, Eq. (5.19) may be written Pa 1 Pe i+ @axt/Ty where P represents the magnitude of the applicable process. (5.19a)56. 57. 58. 59. 5.10. Sul. 5.13, 5.4. 5.15. 5.16. 5.47. 5.18. Problems 197 If the load cell of Problem 5.5 is assumed to be a second-order system with negligible damping, determine the practical frequency range over which it can measure dynamic loads with an inherent error of less than 5%. Ifthe load cell of Problem 5.5 actually has a damping ratio of 0.707, determine the practical frequency range over which it can measure dynamic loads with an inherent error of less than 5%, ‘An RC circuit as shown in Fig. 5.22is required to have atime constant of 1 ms. Determine three different combinations of R and C to accomplish this. A lank containing an initial volume of water Vo, dischasges water from an opening at the bottom of the tank. If the discharge rate is directly proportional to the volume of water in the tank, determine the time constant for this system. For the tank of Problem 5.9, determine the time constant if the initial discharge rate is Q (liters/min). ‘A temperature sensor is expected to measure an input having frequency components as high as 50 Hz with an error no greater than 5%, What is the maximum time constant for the temperature sensor that will permit this measurement? - A thermocouple with a time constant of 0.05 s is considered to behave as a first-order system. Over what frequency range can the thermocouple measure dynamic temperature fluctuations (assumed to be harmonic) with an error less than 5%? ‘A 100-1F capacitor is charged to a voltage of 100 V. At time r = 0, it is discharged through a 1.0-MQ resistor. Determine the time for the voltage across the capacitor to reach 10 V. A pressure wansducer behaves as a second-order system. If the undamped natural fre- quency is 4000 Hz and the damping is 75% of critical, determine the frequency range(s) over which the measurement error is not greater than 5%. ‘What will be the frequency range(s) for Problem 5.14 if the damping ratio is changed to 0.5? Consider the pressure transducer of Problem 5.14 to be damaged such that its viscous damping ratio is unknown. When the transducer is subjected to a harmonic input of 2400 Hz, the phase angle between the output and input is measured as 45°. With this in mind, determine the error when the transducer is used to measure a harmonic pressure signal of 1800 Hz. What is the phase angle between the input and output a this frequency? A fore transducer behaves as a second-order system. If the undamped natural frequency of the transducer is 1800 Hz and its damping is 30% of critical, determine the error in the measured force fora harmonic input of 950 Hz. What is the magnitude of the phase angle? Consider a second-order system with a damping ratio of 0.70 and a undamped natural frequency of 50 Hz. If the value of k is 100 Ibf/in., determine the steady-state output if the forcing is as shown in Fig. 5.26.200 Chapter 6 Sensors This cannot be more than unity, because the pickup cannot generate information but can only receive and process it. Obviously, as high a transfer efficiency as possible is desirable. Sensitivity may be expressed as lou alin ” (1c) Very often sensitivity approximates a constant; that is, the output is the lincer function of the input. 6.2 LOADING OF THE SIGNAL SOURCE Energy will always be taken from the signal source by the measuring system, which means that the information source will always be changed by the act of measurement. This is an axiom of measurement. This effect is referred to as loading. The smaller the load placed on the signal source by the measuring system, the better. ‘Of course, the problem of loading occurs not only in the first stage, but throughout the entire chain of elements, While the first-stage detector-transducer loads the input source, the second stage loads the first stage, and finally the third stage loads the second stage. In fact, the loading problem may be carried right down to the basic elements themselves. In measuring systems made up primarily of electrical elements, the loading of the signal source is almost exclusively a function of the detector. Intermediate modifying devices and output indicators or recorders receive most of the energy necessary for their functioning from sources other than the signal source. A measure of the quality of the first stage, therefore, is its ability to provide a usable output without draining an undue amount of energy from the signal. 6.3 THE SECONDARY TRANSDUCER Asan example of a system of mechanical elements only, consider the Bourdon-tube pressure gage, shown in Fig. 6.1. The primary detecting-transducing element consists of a circular tube of approximately elliptical cross section. When pressure is introduced, the section of the flattened tube tends toward a more circular form. This in turn causes the free end A to move outward and the resulting motion is transmitted by link B to sector gear C and hence to pinion D, thereby causing the indicator hand to move over the scale. In this example, the tube serves as the primary detector-transducer, changing pressure into near linear displacement. The linkage-gear arrangement acts as a secondary transducer (linear to rotary motion) and as an amplifier, yielding a magnified output. A modification of this basic arrangement is to replace the linkage-gear arrangement with either a differential transformer (Section 6.11) or a voltage-dividing potentiometer (Section 6.6). In either case the electrical device serves as a secondary transducer, trans- forming displacement to voltage. ‘As another example, let us analyze a simplified compression-type force-measuring load ceil consisting of a short column or strut, with electrical resistance-type strain gages (sec Section 6.7) attached (Fig, 6.2). When an applied force deflects or strains the block, the force effect is transduced to deflection (we are interested in the unit deflection in this case). The load is transduced to strain. In turn, the strain is transformed into an electrical resistance change, with the strain gages serving as secondary transducers.204 Chapter Sensors TABLE 6.1: (continued) Element Operation HL, Electrical A. Resistive (Sections 6.5-6.8) 1. Contacting Displacement to resistance change 2. Variable-length conductor 3. Variable-area conductor 4, Variable dimensions of conductor 5, Variable resistivity of conductor B. Inductive (Sections 6,10-6.12) 1, Variable coil dimensions 2. Variable air gap 3. Changing core material 4. Changing core positions 5. Changing coil positions 6. Moving coil 7. Moving permanent magnet 8. Moving core C. Capacitive (Section 6.13) 1. Changing air gap 2. Changing plate areas 3. Changing diclectric constant D. Piezoelectric (Section 6.14) E, Semiconductor junction (Section 6.15) 1. Junction threshold voltage 2. Photodiode current F. Photoelectric (Section 6.16) 1, Photovoltaic 2. Photoconductive 3. Photoemissive G. Hall Effect (Section 6.17) Displacement to resistance change Displacement to resistance change Strain to resistance change ‘Temperature to resistance change Displacement to change in inductance Displacement to change in inductance Displacement to change in inductance Displacement to change in inductance Displacement to change in inductance Velocity to change in induced voltage Velocity to change in induced voltage Velocity to change in induced voltage Displacement to change in capacitance Displacement to change in capacitance Displacement to change in capacitance Displacement to voltage and/or voltage to displacement ‘Temperature to voltage change Light intensity to current Light intensity to voltage” Light intensity to resistance change* Light intensity to current* Displacement to voltage * Also sensitive to wavelength of light.208 Chapter6 Sensors. These devices have two basic applications: (1) as temperature-detecting elements used for the purpose of measurement or control, and (2) as electric-power-sensing devices wherein the thermistor temperature—and hence resistance—are a function of the power being dissipated by the device. The second application is particularly useful for measuring radio frequency power. Further discussion of thermistors is given in Section 16.4.3. 6.9 THE THERMOCOUPLE While two dissimilar metals are in contact, an electromotive force exists whose magnitude isa function of several factors, including temperature. Junctions of this sort, when used to measure temperature, are called thermocouples. Often the junction is formed by twisting and welding together two wires. Because of its small size, its reliability, and its relatively large range of usefulness, the thermocouple is a very important primary sensing element. Further discussion of its application is reserved for Chapter 16 (see especially Section 16.5). 6.10 VARIABLE-INDUCTANCE TRANSDUCERS Inductive transducers are based on the voltage output of an inductor (or coil) whose induc- tance changes in response to changes in the measurand. The coil is often driven by an ac excitation, although in dynamic measurements the motion of the coil relative to apermanent magnet may create sufficient voltage. A classification of inductive transducers, based on the fundamental principle used, is as follows. 1. Variable self-inductance (a) Single coil (simple variable reluctance) (b) Two coil (or single coil with center tap) connected for inductance ratio 2. Variable mutual inductance (a) Simple two coil {b) Three coil (using series opposition) 3, Variable reluctance with permanent magnet (@) Moving iron (b) Moving coil (©) Moving magnet The inductance of acoil, Lis influenced by a number of factors, including the number of turns in the coil, the coil size, and especially the permeability of the magnetic flux path that passes through the center of the coil. The magnetic flux path forms a closed loop that extends outside the coil. Often, a magnetic material, such as iron, will be used in the fiux path, commonly in conjunction with one or more air gaps. Because the air gaps have a much lower magnetic permeability than the iron, they control the inductance of the coil. Thus, the variation in the thickness of an air gap is often the primary measurand sensed by a variable-inductance device.212 Chapter6 Sensors Excitation Air woo 22 Armature movement FIGURE 6.8: A mutual-inductance transducer. Coil A is the energizing coil and B is the pickup coil. As the armature is moved, thereby altering the air gap, the output from coil B is changed, and this change may be used as a measure of armature movement, changed by the relative proximity? of a permeable material. A variation of this has been used in a ansducer for measuring small inside diameters [10]. In that case the coupling is varied by relative movement between the two coils. Excitation Measured distance To stage-two circutry FIGURE 6.9: Two-coil inductive pickup for “an electronic micrometer.” Sensors that detect the presence or position of a nearby ebject are often called proximity sensors216 Chapter6 Sensors Position of teeth for minimum value of capacitance Position of teeth for maximum value of capacitance FIGURE 6.13: Section showing relative arrangement of teeth in capacitance-type torque meter. Changing Area Capacitance change depending on changing effective area has been used for the secondary transducing element of a torque meter [12]. The device uses a sleeve with teeth or serrations cut axially, and a matching internal member or shaft with similar axially cut teeth, Figure 6.13 illustrates the arrangement. A clearance is provided between the tips of the teeth, as shown. Torque carried by an elastic member causes a shift in the relative positions of the teeth, thereby changing the effective area. The resulting capacitance change is calibrated in terms of torque. Changing Distance Varying the distance between the plates of a capacitor is undoubtedly the most com- mon method for using capacitance in a pickup, Figure 6.14 illustrates a capacitive-type pressure transducer, wherein the capacitance between the diaphragm to which the pressure is applied and the electrode foot is used as a measure-of the diaphragm’ relative position [13-15]. Flexing of the diaphragm under pressure alters the distance between it and the electrode.220 Chapter6 — Sensors 6.15.2 is formed and will move in the direction opposite the electron. Both negatively charged electrons and positively charged holes contribute to the flow of current in a semiconductor. The number of charge carriers (electrons or holes) in a semiconductor, nc, is a strong function of temperature, T.. Typically, (6.8) constant ne = number per unit volume x T?/7 exp (=) T Since the resistivity of a material is proportional to 1/nc, a semiconductor’s resistance decreases rapidly with increasing temperature. For silicon near room. temperature, the resistivity decreases by about 8% /°C [18] Greater control over a semiconductor's electrical behavior is obtained by doping it with impurity atoms. These atoms may be either electron donors or electron acceptors. Electron donor atoms (such as phosphorus or arsenic) raise the number of free electrons in the material. Electron acceptor atoms (such as gallium and aluminum) hold electrons, thus raising the number of holes. Since the number of doping atomsis usually large relative to the umber of free electrons in the undoped material, the dopant sets the majority current carrier of the material. Specifically, doping with donor atoms creates a predominance of negative charge carriers (electrons), giving an n-type semiconductor, Doping with acceptor atoms creates a predominance of positive charge carriers (holes), giving a p-type semiconductor. Semiconductors, either doped or undoped, are useful as temperature sensors. For undoped semiconductors, the number of carriers increases rapidly with temperature (Eq. (6.8)], so that the resistance is.a strongly decreasing function of temperature. Thermis- tors (Sections 6.8. 16.4.3) are based on this effect. Because such sensors have a negative temperature coefficient of resistance, they are sometimes called NTC sensors. When semi- conductors are heavily doped, the mobility of the carriers decreases with increasing tem- perature, so that the resistance increases with temperature; these positive-remperature- coefficient devices are called PTC sensors. Semiconductors also respond to strain, For example, a p-type region diffused into an n-type base functions as a resistor whose resistance increases strongly when it is strained (this behavior is called piezoresistivity). Such resistors are the basis of semiconductor strain gages, semiconductor diaphragm pressure sensors (Fig. 6.16, Section 14.5.3), and semiconductor accelerometers [19,20]. prJunctions Most semiconductor devices involve a junction, at which n-type and p-type doping meet (Fig. 6.17). Current flows easily from the p-type to the n-type material, since holes (-+charge) easily enter the 2-type material and electrons.(—charge) easily enter the p-type ‘material, Current flow in the opposite direction meets much greater resistance. Thus, this junction behaves like a diode. When a voltage is applied to the junction, the current through it varies as shown in Fig. 6.18. When V is positive, we say that the junction is forward biased. ‘The current becomes very large once the voltage reaches a threshold level. If the voltage is instead negative, the junction is reverse biased, and only a very small current flows. As the reverse- bias voltage is raised, the current quickly reaches a value —o, which is nearly independent of V. The current Ip is called the reverse saturation current. Typical values of Io are on the order of nanoamperes for silicon and microamperes for germanium. ‘The reverse saturation224 Chapter6 Sensors temperature, sensitivity can be improved by cooling the diode to very low temperatures. For example, in high-performatice infrared sensing, photodiodes may be operated at liquid- nitrogen temperatures or even liquid-helium temperatures (— 198°C or —269°C, respec- tively) (23). Some bulk semiconductors, without a pn-junction, also respond to light. Photons create additional electron-hole pairs in,the material, thereby reducing its resistance. Com- mon examples of such photoconductive materials include cadmium sulphide (CdS) and cadmium selenide (CdSe). For high-performance infrared sensing, low-temperature doped germanium may be used. Phiotodetectors are discussed further in the next section. 6.16 LIGHT-DETECTING TRANSDUCERS 6.16.1 Light-sensitive transducers, or photosensors, are used to detect light of all types: thermal radiation from warm objects, laser light, light emitted by diodes, or even sunshine. These transducers may be categorized as either thermal detectors or photon detectors. The thermal detectors use a temperature-sensitive clement which is heated by incident light. The photon detectors respond directly to absorbed photons, either by emitting an electron from a surface (the photoelectric effect) or by creating additional electron-hole pairs in a semiconductor (as discussed in Section 6.15.3). Among the issues to be considered in selecting a photodetector are the wavelength to be sensed, the speed of response needed, and the sensitivity required. In general, ther- mal detectors are much slower than photon detectors but respond to a broader range of wavelengths. For any detector, the speed of response will also depend upon the supporting circuitry. For visible and near-infrared light, semiconductor detectors are commonly used. Photoemissive detectors can be sensitive well into the ultraviolet range. To detect long-wave infrared light, which is the heat emitted by objects near room temperature, either thermal detectors or cryogenically cooled semiconductors may be used. Thermal Detectors ‘Thermal detectors create a temperature change in the detecting element when light heats the detector. Some other property of the detecting element changes in response to the temperature change, and that property is measured to determine the light intensity [24]. Some examples follow. Thermopile detecting elements use several thermocouples in series to produce a volt- age output proportional to the detector temperature (see Section 16.5.6). Pneumatic detect- ing elements use a chamber containing a gas. When the gas is heated, its pressure rises, and the pressure is measured to determine the light intensity (see Section 16.3). Bolometers use temperature-dependent electrical resistance as a serisor (sce Section 16.4). Pyroeleciric detectors respond to temperature changes by generating an electric charge (see Section 6. 14); they are widely. used as infrared motion sensors for automotic light switches and intruder alarms [24,25]. The response times of typical thermal detectors range from a few milliseconds to several seconds, depending mainly upon the size and configuration of the detecting element. Specialized pyroelectric sensors can achieve response times below Ins. Thermal detectors are discussed further in Section 16.8.228 Chapter 6 Sensors ‘The optoisolator is used to match low-impedance current circuits to high-impedance voltage circuits, or vice vérsa. It also provides a high-impedance isolation between circuits, which is an important feature in some forms of health-related electronics [29]. 6.17 HALL-EFFECT SENSORS ‘The Halt effect is the appearance of a transverse voltage difference on a conductor carrying acurrent perpendicular to a magnetic field [30]. This voltage is directly proportional to the magnetic field strength. If the magnetic field is made to vary with the position of a nearby object, the Hall effect can be the basis of a proximity sensor. In Fig. 6.22(a), a conductor carries current in the x-direction, so that electrons flow in the x-direction with a velocity vg. The magnetic field runs in the y-direction. Because the electrons carry a charge —g, they experience a magnetic force Fg in the z-direction: Fp = -gw xB (a) Fa f} — 1 j ——*.8 ©) FIGURE 6.22: The Hall effect: (a) A conductor carries current in a perpendicular magnetic field; (b)electrons are driven upward by magnetic force, creating an opposing electric field.232 Chapter6 Sensors 6.18.2 At this point, itis appropriate to make an additional observation. Weight is basically a ‘force; hence we should express the calibration in newtons rather than in kilograms. Should ‘we wish a scale calibrated in kilograms, then, to be completely correct we should include the assumed value of gravitational acceleration on the faceplate. The standard acceleration due to gravity is 9.80665 m/s?, and the kilogram range corresponding to our 50 N range becomes 0 to 50/9.80665 ~ 5.1 kg. (Use of the non-S] symbol kgf is discouraged.) It should be clear that the procedures used in the preceding example are applicable to ‘most elastic transducer configurations, as well as to many other tolerance problems. Some Temperature-Related Problems ‘An ideal measuring system will react to the design signal only and ignore allelse. Of course, this ideal is never completely fulfilled. One of the more insidious adverse stimuli affecting instrument operation is temperature. It is insidious in that itis almost impossible to maintain a constant-temperature environment for a general-purpose measuring system. The usual solution is to accept the tempersture variation and to devise methods to compensate for it. ‘Temperature variations cause dimensional changes and changes inphysicel properties, both elastic and electrical, resulting in deviations (bias error) referred to as zero shift and scale error (see Fig. 3.2) [35]. Zero shift, as the name implies, results in a change in the no- input reading. Factors other than temperature may cause zero shift; however, temperature is probably the most common cause. In most applications the zero indication on the output scale would be made to correspond to the no-input condition. For example, the indicator or the spring scales referred to earlier should be set at zero when there is no weight in the pan. If the temperature changes after the scale has been set to zero, there may be a differential dimensional change between spring and scale, altering the no-load reading. This change would be referred to as zero shift. Zero shitt is primarily a function of linear dimensional change caused by expansion or contraction with changing temperature. Dimensional changes are expressed in terms of the coefficient of expansion by the following familiar relations: file. wi 6.12) and Ly = Lo(l +a AT) (6.13) where @ = the coefficient of linear expansion (ppm/deg temp. x 10~®), L/Lo = the unit change in length, AT = the change in temperature, T) — To, Lg = the length dimension at the reference temperature To, Ly = the length dimension at any other temperature 7) In addition to causing zero shift, temperature changes usually affect scale calibration when resilient load-carrying members are involved. The coil and wire diameters of our spring would be altered with temperature change, and so too would the modulus of elasticity246 Chapter? Signal Conditioning Demodulated signal (a) continued: FIGURE 7.1: (a) Amplitude modulation, whereby the envelope of the carrier contains the signal information; (b) frequency modulation, whereby the signal information is contained in the frequency variation of the carrier. the capacitive pickup for liquid level (Fig. 6.12) is an example of the latter. In addition, some resistive-type sensors use ac excitation Extracting the signal information from the modulated carrier is required. When AM js used, this operation may take several forms. The simplest is merely to display the entire signal using an oscilloscope or oscillograph, and then to “read” the result from the envelope ofthe carrier. More commonly, the mixed signal and carrier are demodulated by rectification and filtering, as shown in Fig. 7.1(a). FM demodulation is a more complex operation and may be accomplished through the use of frequency discrimination, ratio detection, or IC phase-locked loops. Further discussion is beyond the scope of this text. 7.4 INPUT CIRCUITRY Electrical detector-transducers are of two general types: (I) passive, those requiring an auxiliary source of energy in order to produce a signal; and (2) active, those that are self-powering. The simple bonded strain gage is an example of the former, whereas the piezoelectric accelerometer is an example of the latter.Section 7.12 Electronic Amplification or Gain 269 One should remember that the decibel is fundamentally a power ratio and that “for- getting” the R's in the preceding equations is strictly legitimate only if the two loads, with and without amplification, are equal. Nevertheless, ootput/input ratios are often described using the decibel even when no load is directly involved, and one frequently sees voltage ratios expressed in decibels as dB = 2010839 (2) (7.28b) Another common use of the decibel is in constructing a Bade plot of frequency response. In such a graph, the gain in decibels is plotted against a logarithmic frequency axis, rather than showing ¢,/é; versus on linear coordinates (compare Figs. 7.26 and 7.27). ‘Amplification calculations based on the decibel offer two important advantages: (1) reasonably small numbers are involved, and (2) combining the effects of various stages of a system may be accomplished by simple addition. ‘Voltmeters often carry a decibel scale. When using such a scale one must always be cognizant of three important factors: (1) in reality the measurement is not in decibels, but in voltage; (2) because the decibel is a ratio, the scale must be based on some reference voltage; and (3) reference to Eq, (7.28) shows that the scale must assume a reference load. Most voltmeter scales are based on a reference of | mW across 600 2 or pao hence, e = (PR)'/? = (0.001 x 600)? = 0.7746 V which means that zero on the decibel scale has been arbitrarily set to correspond to 0.7746 V. In some instances the references are indicated directly on the meter face. Often the abbre- viation dBm is used to indicate the aforementioned conventions. Why the 600-2 load rather than something else? The answer's that this is a long-established industrial standard, predating the field of electronics and originated by telegraph and telephone practices. Suppose we use a voltmeter to indicate decibels. Suppose also that the signal source impedance is R, rather than R,, where the latter is the reference. What correction should be applied? The following provides the proper result: Brome) = dBentnt + 10108 ( (7.286) EXAMPLE 7.1 Suppose a reading of 50 dBm is obtained across a 16-@ load, using a voltmeter with scale referenced to 600 ©. What is the true dB value? Solution 6 AB eonected) = 50+ 10s (F) = 65.7 4B214 Chapter? Signal Conditioning ‘Open-loop gain 2 x 10° (depends on frequency) Input impedance 10? 9 Input offset voltage 0.8 mv Input offset voltage drift TuvPC Input offset current 25pA Input bias current 50 pA CMRR 100 dB Maximum output current 25mA Slew rate 1S V/us Maximum power supply voltages £18 V Power supply current 18mA Maximum input voltage range +15V Maximum differential input £30 V Short-circuit output time Indefinite 7.142. Applications of the Op Amp Operational amplifiers may be used as the basic components of linear voltage amplifiers, differential amplifiers, integrators and differentiators, voltage comparators, function gen- erators, filters, impedance transformers, and many other devices. They are not power amplifiers, nor do they have exceptionally wide bandwidth capabilities. Undistorted fre- quency response is typically limited to about | MHz when the circuit gain is low, and it decreases as the gain is raised. In general, an op amp’s maximum voltage output is limited by the supply voltage. Since the number of applications of the op amp to mechanical measurementsis almost limitless, we can describe here only a few. Yet this will give the reader some idea of the tremendous versatility of the device and will suggest additional uses (see also the Suggested Readings at the end of the chapter). One feature common to most op-amp circuits is a negative feedback loop. Because op-amp gain is so high, even a slight input-voltage difference will drive the amplifier to saturation. ‘To prevent this, a connection is made between the output terminal and the inverting (—) terminal. With this connection in place, an increase in ¢, will be fed back to ¢_, reducing the input voltage-difference. The net effect is to produce a circuit that holds e~ © es, preventing saturation. Example 7.2 describes a circuit with no feedback, and several subsequent examples treat circuits having feedback loops. EXAMPLE 7.2 The open-loop configuration® has the following characteristi 1, No feedback loop. Ry. is the load resistance powered by ¢9. The circuit may be free floating or grounded. 2. Amplifier is run wide open: Any input other than zero will drive the amplifier to saturation (i.c., a very small input will drive the output to the limit permitted by the power supply). iis conventional in op-amp circuit diagrams to show only those terminals that are used in the particulir configuration. Power supply inputs are always required, whether shown of not. Null adjustment is offen not shown, although it may be required for optimal performance (see Example 7.7),1. 78. 79, 740. 71. 742, 7.13. 7a4. Problems 297 FIGURE 7.38: Circuit for Problem 7.7. ‘The voltage-dividing potentiometer shown in Fig. 7.5 is modified as shown in Fig, 7.38. Determine the relationship for e, /e; as afunction of &. Compare the results with Eg. (7.9). ‘What advantages or disadvantages does this circuit have over the general voltage-dividing potentiometer? Write a spreadsheet template to solve Eq, (7.15a), permitting each term to be varied by a delta amount, [Suggestion: Rewrite the equation, multiplying each term by (1 + &), where & is the delta plus/minus term—for example, Rj(1 + &j).] ‘A simple Wheatstone bridge as shown in Fig. 7.10 is used to determine accurately the value of an unknown resistance Rj located in leg 1. If upon initial null balance Ry is 127.5 @ and if, when Xz and Rg ase interchanged, null balance is achieved when R3 is 157.9 Q, what is the value of the unknown resistance Rj? Consider the voltage-sensitive bridge shown in Fig. 7.10, If thermistor whose resistance is governed by Eq. (16.3) is placed in leg 1 of the bridge while Rp = R3 = Ry = Ré, determine the bridge output when T= 400°C if Ro = 1000 2 at Ty = 27°C and B = 3500. Plot the bridge output from T = 27°C to T = 500°C and determine the ‘maximum deviation from linearity in this temperature range. Referring to Fig. 7.10, show that if initially R} = Ro = Ri = Ra = R and if AR) = —ARz, the bridge output will belinear. (Nore: This bridge configuration is very commonly used when strain gages are applied to a beam in bending situations; see Table 12.4.) Referring to Fig. 7.10, initially let R) = Ry = Ra = Ry = R. In addition, assume that ARa/R = ~AR,/R. Demonstrate the nonlinearity of the bridge output by plotting ey /¢; over the range 0 = AR,/R < 0.1. (Suggestion: Use a spreadsheet program.) A resistive element of a force cell forms one Jeg of a Wheatstone bridge. If the no-load resistance is 500 © and the sensitivity of the cell is 0.5 Q/N, what will be the bridge outputs for applied loads of 100, 200, and 350 N if the bridge excitation is 10 V and cach atm of the bridge is initially 500 827 Figure 7.39 shows adifferential shunt bridge configuration, One or more of the resistances, Rj, may be resistance-type transducers (thermistor, resistance thermometer, strain gage, etc.), with the remaining resistances fixed. Resistance Rg is a conventional voltage dividing potentiometer, usually of the multium variety. It may be used either for306 Chapter8 _Digital Techniques in Mechanical Measurements e nt) Tt —t f) ) 1 byte Next byte-etc. | fe seer =e ' Previous byte © FIGURE 8.1: Examples of voltage-time relationships for analog signals (a) and (b), and a digital signal (c). Digital informationis transmitted and processed in the form of bits [Fig. 8.1(c)}, each bit being defined by (a) one or the other of two predefined “logic levels” and (b) the time interval assigned to it, called a bit interval. The most common basis for the two logic states is predetermined voltage levels, say 0 and 5 V de. Current or shifts in carrier frequency are also used. The time rate of the bits is closely controlled, commonly by a crystal-controlled oscillator (or clock). ‘The information is then carried by specific bit groupings, coded in predetermined sequences; forexample, alphanumeric data may be handled by sequences of three, four, or more bits sent in the various possible combinations, with each combination or group forming a word of information (see Section 8.6). The term byte is applied to an 8-bit word, whereas the term word may be applied to any unit of digital information. A 16-bit word is two bytes in length. A 4-bit word is sometimes referred to as a nibble. Figure 8.1(c) shows one possible combination of bits grouped to form one byte. In this case, the sequence of bit values is 1010 0110. From this we see that a bit need not be a pulse in the sense that it must be a compleied oft/on/off sequence. Indeed a byte of information could well be 0000 0000 or 1111 1111, in which case no bit-to-bit changes ‘occur throughout the byte. One bit corresponds to either of two different logic states held constant during the one-bit interval. Because most measurement inputs originate in analog form, some type of analog-to- digital (A/D) converter (or ADC) is usually required (Section 8.11). In certain instances it may also be desirable or necessary to use a digital-to-analog (D/A) converter (or DAC) somewhere in the measurement chain. A sophisticated example of digital information handling is pulse-code modulation (PCM) of the human voice, which is used in essentially316 Chapter8 Digital Techniques in Mechanical Measurements +8Vdc a 4 +7 Output {| — et 6 a4 FIGURE 8.8: A 555 astable/monostable oscillator or multivibrator. is controlled by the clock signal. For example, flip-flop can be designed so that when an input is changed, the change in output is delayed until the clock signal changes; the D- type and J-K flip-flops in Table 8.1 (7474 and 74109, respectively) both have this feature. Synchronous operation usually simplifies circuit design. Circuits that are not synchronous are called asynchronous. The output of a 555-based clock circuit may be stable to about 1% or so but it is susceptible to drift. When higher accuracy is required, a quartz-crystal oscillator is used instead. The heartof sucha circuit is asmall piece of quartz cut to precise dimensions, which resonates at a highly stable frequency. Owing to the piezoelectric properties of quartz (see Section 6.14), the resonating crystal is easily coupled to an electrical circuit. Quartz-crystal oscillators are available as off-the-shelf ICs at low cost with frequencies from roughly 10kHz to 100 MHz. Frequency stability 10 a few parts per million is easily obtained. These crystals are used in all microprocessors and microcomputer systems. Crystals resonating at 32.768 kHz are at the heart of digital wristwatches: When additional digital circuits are used to divide this frequency by 2'°, the result is a 1-Hz signal for counting seconds [5]. Note that an error of 10 parts per million in the crystal frequency would correspond to about one second per day of error in a watch, 8.4.5 Digital Displays In some situations we may require only simple, single lights for an indication or readout, perhaps to show that a device is powered or ready for use. In other cases we may use combinations of discrete lights or indicators to produce an alphanumeric display. Alphanumeric readout elements are usually of either the liquid-crystal diode (LCD) type or the light-emitting diode (LED) type. The liquid crystal has a decided advantage in some cases, as, for example, in a digital watch requiring very low power consumption.Section 8.6 BinaryCodes 321 TABLE €.4: Four-Bit Binary Numbers and Gray Code Decimal Digit Binary Gray Code 0 0000 0000 1 0001 0001 2 0010 O01 3 oo1l 0010 4 0100 0110 5 0101 on 6 O10 0101 7 oul 0100 8 1000 1100 9 1001 1101 10 1010 Wit i 1011 1110 12 1100 1010 13 101 1011 4 1110 1001 15 wit 1000 001112 to 01000; (i.e, from 710 to 810). IF just one bit were to change too soon or too late, the position could be sensed as 000002 or 011112, corresponding to either the first or the sixteenth level! Gray code isa binary code that eliminates this type of error, and itis thus preferred for use in position encoders. It is not a numerical code, in the sense that there is no positional weightto the digits. Instead, the digits of the code are sequenced so that only one bit changes in going from one level in the sequence to the next. If one track’s sensor changes too early or too late, the error is no more than one level—the drastic errors possible with direct binary encoding are avoided. A 4-bit Gray code code is shown in Table 8.4. Figure 8.11 illustrates a circular card for a rolatary shaft position encoder based upon four-bit Gray code. 8.6.3 Alphanumeric Codes ‘The preceding codes are for the transmission and processing of numeric data, Alphanumeric information must also include provisions for the letters of the alphabet and perhaps certain other symbols, such as punctuation marks. One of the simplest binary codes for transmission of general information (as opposed to numeric data only) is the International Morse Code, which uses pulse-duration modu- ation, or PDM (also sometimes called pulse-length or pulse-width modulation). The two different pulse widths, the “dot” and the “dash,” are used in various combinations to trans mit the alphabet, the decimal digits, and certain other special-purpose telegraphic symbols. International Morse Code is an wneven-length code in that various time intervals are required for the transmission of the various characters. A related code is the Baudor, or conimon teletypewriter, code, which is a five-unit, even-length code. All characters are formed by aSection 8.7 Some Simple Digital Circuitry 325 word is even. If 1 bit in the received word'is incorrect, the iumber of ones would be odd, thus enabling the receiver to determine that a transmission error has occurred. In an odd parity scheme, the total number of ones would be odd. The parity method clearly cannot detect errors in which 2 bits of one word are incorrect; however, in situations where even a single bit error is improbable, a 2-bit error will be extremely improbable and is thus of little concern. 8.6.4 Bar Codes Everyone is familiar with at least some of the applications of bar codes. The familiar black/white lines are arranged to codify data of various sorts—addresses, inventory, and virtually any other information that can be put into a binary representation. A low-intensity laser may be reflected off of the bar pattern, with the signal received by an appropriate photodetector. The output is electronically processed and sent on to the cash register or inventory tracking system. A number of standards have been used to create bar codes, the most common being the Universal Product Code (UPC) and the European Article Numbering (EAN) system. In atypical implementation, a black line corresponds to 1 and a white line corresponds to 0. Seven lines are used for each character; however, the first and last lines are always ‘opposite each other (black and white or white and black, respectively) in order that adjacent characters can be distinguished. The five remaining bits offer 32 possible combinations, but not all are used. Instead, only combinations yielding just two black bars and two white bars are employed, where a bar's width may be one, two, three, or four lines. There are only 20 such combinations. These combinations are assigned to the ten decimal digits from 010 9 intwo different sequences, one “left handed” and the other “right handed,” Depending upon the particular type of bar code, as many as 14 digits may be encoded. In addition to the bars for the digits, the codes usually include guard bars at lefi, center, and right. The numerical code itself is assigned to particular companies and products by a standardization agency. In addition to the basic one-dimensional bar code, various two-dimensional matrix style coding systems are also in use. These systems can encode greater amounts of alphanu- meric data. They are sometimes used on shipping labels from commercial delivery services. 8.7 SOME SIMPLE DIGITAL CIRCUITRY 8.7.1 Events Counter Figure 8.13 shows the outward simplicity of one form of digital events’ counter. Two consecutive stages, or decades, are shown. Each decade consists of a seven-segment LED display, two IC chips, and a few current-limiting resistors. The 7490 chip is called a decade counter (see Table 8.1). It accepts input pulses in serial form through pin 14 and sends as output parallel BCD pulses through pins d, c, b, and a, where d corresponds to the most significant bit and a to the least significant. In addition, this chip provides an output pulse at the end of the ninth input pulse. This pulse is available from pin 11 and can be used as the input to the next higher counting decade. We can see that additional decades may be cascaded easily. As we will note later, this arrangement obviously also provides a “divide-by-10” capabi330 Chapter8 Digital Techniques in Mechanical Measurements a8 me ita Ground g z = z i Es E_ weg . wf po] & i ge ‘Inverter 23 sad 38 5G = 8s c a dcob t et Selector we FIGURE 8.16: A multiplexer—demultiplexer circuit. Why concern ourselves with multiplexers and demultiplexers? In many cases con- tinuous monitoring or recording of a given data source is not necessary: Periodic sampling will suffice. We can see that by sequencing the binary control through 0 to 15, we can consecutively connect 16 different inputs to a given readout/recording/computing system. This approach is economical in many cases Multiplexing may also be used, in certain instances, within the circuitry of a single instrument. Recall the events counter we discussed in Section 8.7.1. Each decade required 2 seven-segment decoder driver plus seven current-limiting resistors. More sophisticated ‘counter circuitry uses a multiplexer-demultiplexer combination arranged so that each read- out element is sequentially connected to a single driver-resistor combination. By time sharing in this manner we can reduce the number of circuit elements and realize quite a saving in cost. The sequencing rate is sufficiently high that the readout appears to be illuminated continuously. Data processing may take place at quite some distance from the data source, as in a large industrial complex such as a refinery or power plant. By using multiplexer demultiplexer combinations, we may also use single, rather than separate, circuits to connet the two positions (Fig. 8.16). Wehastento add that this may notbe precisely the case because it would also be necessary to synchronize our binary control circuits at the two locations. A simple solution is to run four additional wires connecting ports d, c, b, and a. Thus we would have 5 circuits instead of 16 (for the particular combination cited). There is, however, another possibility through use of more sophisticated ICs. The universal asynchronous receiver/transmitter (UART) contains whatamounts toa multiplexer338 Chapter8 —_ Digital Techniques in Mechanical Measurements 8.11.2 to the least significant bit, bo; switch 7 corresponds to the most significant bit, by; and so on. By closing selected switches, the output voltage can be made proportional to any particular 8-bit number. For example, if switches 0, 5, and 6 are closed, the output voltage is proportional to0110 00012 (= 616 = 9710). Since Re sets the constant of proportionality, the output gain can be scaled to provide an appropriate range of analog values. Additional output circuitry could be used to invert the signal or to add offset voltages. An Analog-to-Digital Converter One typical A/D converter is the parallel encoder (Fig. 8.20). This circuit uses a set of voltage comparators (Example 7.8, Section 7.14.2) and a series of resistors to compare Ent Voltage Binary encoder we Sit binary output Eanaiog FIGURE 8.20: A 3-bit paralle! A/D converter.10.3.2 Section 10.3 Stroboscopy and High-Speed Imaging 385 ,o P ° ee ° FIGURE 104: Photo obtained by “open-shutter” camera technique and using a Strobotac set at a flash rate of 20 Hz (Courtesy: GenRad, Inc., Concord, Massachusetts). High-Speed Imaging Stroboscopy is closely related to high-speed photography or movie making. By using short exposures taken in quick succession, itis possible to capture the dynamics of motion that might otherwise pass in the blink of an eye. For example, Fig. 10.5 shows a sequence of images of a water-filled balloon as it drops onto a surgical scapel and bursts. These images were recorded by a high-speed digital video camera at rate of 10,000 frames per second, ‘The entire process shown took just 0.070 s. ‘The basic requirements of high-speed imaging are that the exposure time be short enough to capture an unblurred image of the moving object and that images be taken frequently enough to track the changing features of the process. The faster the object moves, the shorter must be the exposure and the higher the frame rate. A rule of thumb is that if an object requires a time T to move its own length, the exposure time should be 0.001 7 in order to obtain adequate focus [7]. Framing rates of up to 1000 frames per second (fps) with exposure times down to microseconds are adequate for a wide range of relatively low-speed applications, including studies of moving machinery and crash testing. In studies of ballistics and explosives, the velocities are much higher, and exposures of just 10 to 100 ns with frame rates of up to 10° fps may be required. While film cameras were long used for high-speed movie making, most high speed imaging today is done using digital video cameras. ‘The advantages of digital systems over film systems are considerable. These include immediate availability of the image, the ability to use computers to process the images and extract data from them, ease of operation, longer recording times, and the elimination of chemical film processing. Film retains some advantages, including lower equipment cost, potentially larger image format or resolution, and, in some cases, significantly higher framing rates. High-speed digital video cameras with frame rates from 1000 fps to more than 100,000 fps are available. Not suprisingly, the cost tends to rise as the speed rises. These high-speed cameras are one to two orders of magnitude more expensive than the common digital video camera used for taking home movies, which runs at just 30 fps. By using780 Index [Intemational Temperature Scale of 1990, 2 Intemational Standards Orgarizaton (050), 18, 344 Inerval elas, 84 confidence, $9, 70,71 Torizatcn pressure gage, 525 J Joeephson junction, 42 TPEG, 344 K Kelvin, Lord, Witiam Thomson) 39, 432 Kelvin, iemperature scale, 38 Kiel tae, $73 Kilogram, Inemationa Protoype, 32 Kirchhoft’s lws of curent and vole, 188 Kischhof's lew of canton, 82 Kistler, W. P, 281 L Laer, 410, 411 ‘coherence length of, 11 displacement sensors. 413 holography, 417 inded Aoresence, $86 imererometet, $11 ‘elocity epson, $81, 689, wavelengths of $10 [Laser Doppler aemometer (LDA), 581 frequency shifting of, $84 sealing fx, 384 Latch, 312 Law of intermediate mets 613. Law of intermediate temperatures, 613 Lead wire compensation, 263,442, s-409 ‘Least squares (linear regression), 108, 770 ‘confidence intervals fr, 72 correlation vefcien of, 108,770 ollie and, 772 statistical ests of, 770 Length sundard, 38 Lig ‘and thermal radiation, 631 ‘monochrematic $10 Speed of, 36, $6-57, 122 wavelengas of 10,630 Light detecting tratsducers, 224 avalanche photodiode, 225 cdarkcurent of, 223, Photocondactor, 224,225, 640 photodiodes, 22, 25, 637 [Photomuhipier tbe (PMT), 223, 582 photon detectors, 225, 630, 640 hototransistors, 225 pyroclecuic, 219, 224, 636,639 thermal detects, 224, 630, 635 Linearity of diferentialwansformer, 419 of potentiometer, 207 Linear phase shift device, 680, 652 Line fiting, 101,198, 770 Liquid crystal temperature sersct, 641 Loading error 39, 200, 22, ogarithmic decrement, 683-684 Logic ekmens, 310 Lagi tables. 310 Loadeess, 715 Loadoee level, 716 NDT, se Differential tamsformer M Manormetr, 504, 07, 58 Mass 14,473 standard, 21,475 MeLzod gage, 523 Mean aera, defition of for populin, 63 for sample 67 Measurad, 13 ang 123,303 igi, 305 Measurement eiion of, 18 eneaizedcysiem of, 2 etbds of. 20 Menor, ‘anom acess RAM), 334 readonly (RON), 334 Mendebal ede, 33 Mee, defen of 36 Maer 26,287 muneter (current), 356-358, analog, 355 sigial, 361 EPUT, 351,382 frequency, 327,328 obsstion, 546 ohmmeter, 360 sound level, 719 Aime interval, 351, 382 voltmeter, 356,361 voltohm-millismmeer, 360, 32 Metric system of units, see Stsystem of nis Michelson, AA., 411 Michelson interferonteter. 411-412 Microcomputer, 331,33 Micromanometer $13 Microphone, 717 cealtration of, $31,732 placement of, 30-732 Selection factor, 719 summary of characteristics, 720 Microprocessor, 332 Microscope. 413 digial, 417 Microineh, definition, 404 Microstain,dfiniéon of, 429 Miner, 572,395, Modulation, 245 amplitude (AM), 245 frequency (FM). 245 pulse code, 308 Medias ‘of elasticity (definition), 429-420, ‘of elasticity (wales), 234, 754 resisors, 489| Mabe sree, 761 Moieé technique, 432 Moment (of forces) 473, ‘Morse ede, 121 “Motion measirement, 671-688 MPEG, 344 Maltpiexer, 329 Malevibator, 316, N ‘Naional Bureau of Standards (NBS), 18, 2 ‘National Insite cf Standards and ‘Technology (NIST), 18,32 Newton's laws of gavitaion, 473, of motion. 44574 Nibble, 308 Noise, 21,168: sce alr Sound Normal probebiity plot, 83-85 Nettle flow, $46,549, $56 Number systems, 317, 746 binary, 318, 746 ‘comsersion between, 748 hexadecimal, 319, 747 eta 319,737 oO (Ohm's aw, 188 (Operational amplifiers, 71 feedhack with, 228 ‘ypicaleiteuits, 274 ‘ypicalspectteations, 273, use milters, 288 ‘use m integrators and differemitors, 288 Optointeruptor, 227 harmonic, 126 of electrical systems, 188-193, of mecharical systema, 170-186 Orifice towmetr, $46,548, $56 GPS signi, 388 mechanical, 388 radio signal, 389 voltage-controled (¥CO), 308, 361, 372, 396 Oxillosope (CRO), 382 digi, 368, ‘input coupling of, 364 recerding, 169 storage, 68 twiggering of, 366 Outi, 64, 98, 110,772Yi il La SLO | | } in india, Bangladesh, Bhu‘an, Pakistan, Nepal, Sti Lanka and the Maldives,